1 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
2 Alan Hayward <alan.hayward@arm.com>
3 David Sherwood <david.sherwood@arm.com>
5 * target.def (return_pops_args): Treat both the input and output
6 sizes as poly_int64s rather than HOST_WIDE_INTS.
7 * targhooks.h (default_return_pops_args): Update accordingly.
8 * targhooks.c (default_return_pops_args): Likewise.
9 * doc/tm.texi: Regenerate.
10 * emit-rtl.h (incoming_args): Change pops_args, size and
11 outgoing_args_size from int to poly_int64_pod.
12 * function.h (expr_status): Change x_pending_stack_adjust and
13 x_stack_pointer_delta from int to poly_int64.
14 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
15 (ARGS_SIZE_RTX): Update accordingly.
16 * calls.c (highest_outgoing_arg_in_use): Change from int to
18 (stack_usage_watermark, stored_args_watermark): New variables.
19 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
20 (emit_call_1): Change the stack_size and rounded_stack_size
21 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
23 (save_fixed_argument_area): Check stack_usage_watermark.
24 (initialize_argument_information): Change old_pending_adj from
25 a HOST_WIDE_INT * to a poly_int64_pod *.
26 (compute_argument_block_size): Return the size as a poly_int64
28 (finalize_must_preallocate): Track polynomial argument sizes.
29 (compute_argument_addresses): Likewise.
30 (internal_arg_pointer_based_exp): Track polynomial offsets.
31 (mem_overlaps_already_clobbered_arg_p): Rename to...
32 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
33 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
34 Check stored_args_used_watermark.
35 (load_register_parameters): Update accordingly.
36 (check_sibcall_argument_overlap_1): Likewise.
37 (combine_pending_stack_adjustment_and_call): Take the unadjusted
38 args size as a poly_int64 rather than an int. Return a bool
39 indicating whether the optimization was possible and return
40 the new adjustment by reference.
41 (check_sibcall_argument_overlap): Track polynomail argument sizes.
42 Update stored_args_watermark.
43 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
44 (expand_call): Likewise. Maintain stack_usage_watermark and
45 stored_args_watermark. Update calls to
46 combine_pending_stack_adjustment_and_call.
47 (emit_library_call_value_1): Handle polynomial argument sizes.
48 Call stack_region_maybe_used_p and mark_stack_region_used.
49 Maintain stack_usage_watermark.
50 (store_one_arg): Likewise. Update call to
51 mem_overlaps_already_clobbered_arg_p.
52 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
54 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
55 * config/microblaze/microblaze.c (microblaze_function_prologue):
57 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
58 TARGET_RETURN_POPS_ARGS interface.
59 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
61 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
62 * config/i386/i386.c (ix86_return_pops_args): Update for new
63 TARGET_RETURN_POPS_ARGS interface.
64 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
65 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
66 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
67 TARGET_RETURN_POPS_ARGS interface.
68 * config/vax/vax.c (vax_return_pops_args): Likewise.
69 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
70 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
71 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
72 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
73 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
74 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
75 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
76 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
77 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
78 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
79 and x_stack_pointer_delta from int to poly_int64.
80 * dojump.c (do_pending_stack_adjust): Update accordingly.
81 * explow.c (allocate_dynamic_stack_space): Handle polynomial
83 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
84 (pad_to_arg_alignment): Track polynomial offsets.
85 (assign_parm_find_stack_rtl): Likewise.
86 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
87 * toplev.c (output_stack_usage): Update reference to
88 current_function_pushed_stack_size.
90 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
91 Alan Hayward <alan.hayward@arm.com>
92 David Sherwood <david.sherwood@arm.com>
94 * function.c (in_arg_offset, var_offset, dynamic_offset)
95 (out_arg_offset, cfa_offset): Change from int to poly_int64.
96 (instantiate_new_reg): Return the new offset as a poly_int64_pod
97 rather than a HOST_WIDE_INT.
98 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
99 (instantiate_virtual_regs_in_insn): Likewise.
101 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
102 Alan Hayward <alan.hayward@arm.com>
103 David Sherwood <david.sherwood@arm.com>
105 * rtl.h (get_args_size, add_args_size_note): New functions.
106 (find_args_size_adjust): Return a poly_int64 rather than a
108 (fixup_args_size_notes): Likewise. Make the same change to the
109 end_args_size parameter.
110 * rtlanal.c (get_args_size, add_args_size_note): New functions.
111 * builtins.c (expand_builtin_trap): Use add_args_size_note.
112 * calls.c (emit_call_1): Likewise.
113 * explow.c (adjust_stack_1): Likewise.
114 * cfgcleanup.c (old_insns_match_p): Update use of
115 find_args_size_adjust.
116 * combine.c (distribute_notes): Track polynomial arg sizes.
117 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
118 end_true_args_size, beg_delay_args_size and end_delay_args_size
119 from HOST_WIDE_INT to poly_int64.
120 (add_cfi_args_size): Take the args_size as a poly_int64 rather
121 than a HOST_WIDE_INT.
122 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
123 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
124 polynomial arg sizes.
125 * emit-rtl.c (try_split): Use get_args_size.
126 * recog.c (peep2_attempt): Likewise.
127 * reload1.c (reload_as_needed): Likewise.
128 * expr.c (find_args_size_adjust): Return the adjustment as a
129 poly_int64 rather than a HOST_WIDE_INT.
130 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
131 to a poly_int64 and change the return type in the same way.
132 (emit_single_push_insn): Track polynomial arg sizes.
134 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
135 Alan Hayward <alan.hayward@arm.com>
136 David Sherwood <david.sherwood@arm.com>
138 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
139 from HOST_WIDE_INT to poly_int64.
140 * expr.c (push_block, emit_push_insn): Likewise.
142 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
143 Alan Hayward <alan.hayward@arm.com>
144 David Sherwood <david.sherwood@arm.com>
146 * function.h (frame_space): Change start and length from HOST_WIDE_INT
148 (get_frame_size): Return the size as a poly_int64 rather than a
150 (frame_offset_overflow): Take the offset as a poly_int64 rather
151 than a HOST_WIDE_INT.
152 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
153 (assign_stack_temp): Likewise for the size.
154 * function.c (get_frame_size): Return a poly_int64 rather than
156 (frame_offset_overflow): Take the offset as a poly_int64 rather
157 than a HOST_WIDE_INT.
158 (try_fit_stack_local): Take the start, length and size as poly_int64s
159 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
160 rather than a HOST_WIDE_INT.
161 (add_frame_space): Take the start and end as poly_int64s rather than
163 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
164 (assign_stack_temp): Likewise for the size.
165 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
167 (find_temp_slot_from_address): Handle polynomial offsets.
168 (combine_temp_slots): Likewise.
169 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
171 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
172 poly_int64 rather than a HOST_WIDE_INT.
173 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
174 than a HOST_WIDE_INT.
175 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
176 polynomial frame offsets.
177 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
178 as a poly_int64 rather than an int.
179 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
180 * config/v850/v850-protos.h (compute_frame_size): Likewise.
181 * config/v850/v850.c (compute_frame_size): Likewise.
182 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
183 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
184 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
185 * config/pa/pa.c (pa_compute_frame_size): Likewise.
186 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
187 rather than a HOST_WIDE_INT.
188 * explow.c (get_dynamic_stack_base): Likewise.
189 * final.c (final_start_function): Use the constant lower bound
190 of the frame size for -Wframe-larger-than.
191 * ira.c (do_reload): Adjust for new get_frame_size return type.
192 * lra.c (lra): Likewise.
193 * reload1.c (reload): Likewise.
194 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
195 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
196 * rtlanal.c (get_initial_register_offset): Return the offset as
197 a poly_int64 rather than a HOST_WIDE_INT.
199 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
200 Alan Hayward <alan.hayward@arm.com>
201 David Sherwood <david.sherwood@arm.com>
203 * reload1.c (elim_table): Change initial_offset, offset and
204 previous_offset from HOST_WIDE_INT to poly_int64_pod.
205 (offsets_at): Change the target array's element type from
206 HOST_WIDE_INT to poly_int64_pod.
207 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
208 (elimination_costs_in_insn, update_eliminable_offsets)
209 (verify_initial_elim_offsets, set_offsets_for_label)
210 (init_eliminable_invariants): Update after above changes.
212 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
213 Alan Hayward <alan.hayward@arm.com>
214 David Sherwood <david.sherwood@arm.com>
216 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
217 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
218 (decomposition): Change start and end from HOST_WIDE_INT
220 (decompose, immune_p): Update accordingly.
221 (find_inc_amount): Return a poly_int64 rather than an int.
222 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
225 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
226 Alan Hayward <alan.hayward@arm.com>
227 David Sherwood <david.sherwood@arm.com>
229 * tree.h (get_inner_reference): Return the bitsize and bitpos
230 as poly_int64_pods rather than HOST_WIDE_INT.
231 * fold-const.h (ptr_difference_const): Return the pointer difference
232 as a poly_int64_pod rather than a HOST_WIDE_INT.
233 * expr.c (get_inner_reference): Return the bitsize and bitpos
234 as poly_int64_pods rather than HOST_WIDE_INT.
235 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
237 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
238 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
239 (optimize_bit_field_compare): Update call to get_inner_reference.
240 (decode_field_reference): Likewise.
241 (fold_unary_loc): Track polynomial offsets and sizes.
242 (split_address_to_core_and_offset): Return the bitpos as a
243 poly_int64_pod rather than a HOST_WIDE_INT.
244 (ptr_difference_const): Likewise for the pointer difference.
245 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
246 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
247 * dbxout.c (dbxout_expand_expr): Likewise.
248 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
249 (loc_list_from_tree_1, fortran_common): Likewise.
250 * gimple-laddress.c (pass_laddress::execute): Likewise.
251 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
252 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
253 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
254 * tree-affine.c (tree_to_aff_combination): Likewise.
255 (get_inner_reference_aff): Likewise.
256 * tree-data-ref.c (split_constant_offset_1): Likewise.
257 (dr_analyze_innermost): Likewise.
258 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
259 * tree-sra.c (ipa_sra_check_caller): Likewise.
260 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
261 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
262 (instrument_bool_enum_load, instrument_object_size): Likewise.
263 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
264 to get_inner_reference.
265 * hsa-gen.c (gen_hsa_addr): Likewise.
266 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
267 * tsan.c (instrument_expr): Likewise.
268 * match.pd: Update call to ptr_difference_const.
270 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
271 Alan Hayward <alan.hayward@arm.com>
272 David Sherwood <david.sherwood@arm.com>
274 * fold-const.c (fold_comparison): Track sizes and offsets as
275 poly_int64s rather than HOST_WIDE_INTs when folding address
278 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
279 Alan Hayward <alan.hayward@arm.com>
280 David Sherwood <david.sherwood@arm.com>
282 * expr.h (get_bit_range): Return the bitstart and bitend as
283 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
284 as a poly_int64 rather than a HOST_WIDE_INT.
285 * expr.c (get_bit_range): Likewise.
286 (expand_assignment): Update call accordingly.
287 * fold-const.c (optimize_bit_field_compare): Likewise.
289 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
290 Alan Hayward <alan.hayward@arm.com>
291 David Sherwood <david.sherwood@arm.com>
293 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
294 (aarch64_legitimate_address_p): Use it instead of an rtx code,
295 as an optional final parameter.
296 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
297 (aarch64_legitimate_address_p): Likewise.
298 (aarch64_print_address_internal): Take an aarch64_addr_query_type
299 instead of an rtx code.
300 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
301 (aarch64_legitimate_address_hook_p): Likewise.
302 (aarch64_print_ldpstp_address): Likewise.
303 (aarch64_print_operand_address): Likewise.
304 (aarch64_address_cost): Likewise.
305 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
306 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
307 (aarch64_mem_pair_lanes_operand): Likewise.
309 2017-12-20 Richard Biener <rguenther@suse.de>
311 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
312 update_stmt_if_modified.
314 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
316 PR tree-optimization/83491
317 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
318 before walking uses. Improve coding style and comments.
320 2017-12-20 Tom de Vries <tom@codesourcery.com>
322 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
324 2017-12-20 Jakub Jelinek <jakub@redhat.com>
327 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
329 (pass_ipa_free_fn_summary): Add small_p private data member,
330 initialize to false in the ctor.
331 (pass_ipa_free_fn_summary::clone,
332 pass_ipa_free_fn_summary::set_pass_param,
333 pass_ipa_free_fn_summary::gate): New methods.
334 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
335 | TODO_dump_symtab if small_p.
336 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
337 entry and add another instance of the pass with false parm after
339 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
340 ipa_free_fn_summary here.
342 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
344 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
346 2017-12-20 Martin Sebor <msebor@redhat.com>
349 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
350 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
351 (expand_builtin_strncmp): Same.
353 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
356 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
357 allowed debug stmts before labels.
358 (expand_gimple_basic_block): Likewise.
359 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
360 * gimple-iterator.h (gsi_after_labels): Likewise.
361 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
362 rename reused variable, and simplify using gsi_move_before.
363 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
364 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
365 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
366 (gimple_verify_flow_info, gimple_block_label): Likewise.
367 (make_blocks): Move debug markers after adjacent labels.
368 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
369 allowed debug insns outside blocks.
370 * df-scan.c (df_insn_delete): Likewise.
371 * lra-constraints.c (update_ebb_live_info): Likewise.
372 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
373 (vt_initialize, delete_vta_debug_insns): Likewise.
374 (reemit_marker_as_note): Drop BB parm. Adjust callers.
376 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
377 Alan Hayward <alan.hayward@arm.com>
378 David Sherwood <david.sherwood@arm.com>
380 * poly-int-types.h (round_down_to_byte_boundary): New macro.
381 (round_up_to_byte_boundary): Likewise.
382 * expr.h (get_bit_range): Add temporary shim.
383 * gimple-ssa-store-merging.c (store_operand_info): Change the
384 bitsize, bitpos, bitregion_start and bitregion_end fields from
385 unsigned HOST_WIDE_INT to poly_uint64.
386 (merged_store_group): Likewise load_align_base.
387 (compatible_load_p, compatible_load_p): Update accordingly.
388 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
389 (split_group, imm_store_chain_info::output_merged_store): Likewise.
390 (mem_valid_for_store_merging): Return the bitsize, bitpos,
391 bitregion_start and bitregion_end as poly_uint64s rather than
392 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
393 (handled_load): Take the bitsize, bitpos,
394 bitregion_start and bitregion_end as poly_uint64s rather than
395 unsigned HOST_WIDE_INTs.
396 (pass_store_merging::process_store): Update call to
397 mem_valid_for_store_merging.
399 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
400 Alan Hayward <alan.hayward@arm.com>
401 David Sherwood <david.sherwood@arm.com>
403 * builtins.c (get_object_alignment_2): Track polynomial offsets
404 and sizes. Update the alignment handling.
406 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
407 Alan Hayward <alan.hayward@arm.com>
408 David Sherwood <david.sherwood@arm.com>
410 * tree.h (get_inner_reference): Add a version that returns the
411 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
412 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
413 the case in which bitpos is not associated with the first byte.
415 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
416 Alan Hayward <alan.hayward@arm.com>
417 David Sherwood <david.sherwood@arm.com>
419 * tree-affine.h (get_inner_reference_aff): Return the size as a
421 * tree-affine.c (get_inner_reference_aff): Likewise.
422 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
423 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
425 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
426 Alan Hayward <alan.hayward@arm.com>
427 David Sherwood <david.sherwood@arm.com>
429 * fold-const.c (pointer_may_wrap_p): Take the offset as a
430 HOST_WIDE_INT rather than a poly_int64.
432 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
433 Alan Hayward <alan.hayward@arm.com>
434 David Sherwood <david.sherwood@arm.com>
436 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
437 HOST_WIDE_INT to poly_int64_pod.
438 (perform_symbolic_merge): Update accordingly.
439 (bswap_replace): Likewise.
441 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
442 Alan Hayward <alan.hayward@arm.com>
443 David Sherwood <david.sherwood@arm.com>
445 * tree-affine.h (aff_tree::offset): Change from widest_int
447 (wide_int_ext_for_comb): Delete.
448 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
449 constants as poly_widest_int rather than widest_int.
450 (aff_combination_constant_multiple_p): Return the multiplier
451 as a poly_widest_int.
452 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
454 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
455 version static and add an overload for poly_widest_int.
456 (aff_combination_const, aff_combination_add_cst)
457 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
458 the constants as poly_widest_int rather than widest_int.
459 (tree_to_aff_combination): Generalize INTEGER_CST case to
461 (aff_combination_to_tree): Track offsets as poly_widest_ints.
462 (aff_combination_add_product, aff_combination_mult): Handle
464 (aff_combination_constant_multiple_p): Return the multiplier
465 as a poly_widest_int.
466 * tree-predcom.c (determine_offset): Return the offset as a
468 (split_data_refs_to_components, suitable_component_p): Update
470 (valid_initializer_p): Update call to
471 aff_combination_constant_multiple_p.
472 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
473 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
474 as a poly_int64 rather than a HOST_WIDE_INT.
475 (get_address_cost): Handle polynomial offsets.
476 (iv_elimination_compare_lt): Likewise.
477 (rewrite_use_nonlinear_expr): Likewise.
479 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
480 Alan Hayward <alan.hayward@arm.com>
481 David Sherwood <david.sherwood@arm.com>
483 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
484 as a poly_int64_pod rather than a HOST_WIDE_INT.
485 (get_addr_base_and_unit_offset): Likewise.
486 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
487 (get_addr_base_and_unit_offset): Likewise.
488 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
489 to poly_int64 in example.
490 * fold-const.c (fold_binary_loc): Update call to
491 get_addr_base_and_unit_offset.
492 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
493 (maybe_canonicalize_mem_ref_addr): Likewise.
494 (gimple_fold_stmt_to_constant_1): Likewise.
495 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
497 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
498 * match.pd: Likewise.
499 * omp-low.c (lower_omp_target): Likewise.
500 * tree-sra.c (build_ref_for_offset): Likewise.
501 (build_debug_ref_for_model): Likewise.
502 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
503 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
504 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
505 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
506 (constant_pointer_difference): Likewise.
507 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
508 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
509 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
510 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
511 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
512 (set_ssa_val_to): Likewise.
513 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
514 (maybe_diag_stxncpy_trunc): Likewise.
515 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
516 * tree.c (build_simple_mem_ref_loc): Likewise.
517 (array_at_struct_end_p): Likewise.
519 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
520 Alan Hayward <alan.hayward@arm.com>
521 David Sherwood <david.sherwood@arm.com>
523 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
524 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
525 (get_ref_base_and_extent_hwi): Declare.
526 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
527 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
528 (get_ref_base_and_extent_hwi): New function.
529 * cfgexpand.c (expand_debug_expr): Update call to
530 get_ref_base_and_extent.
531 * dwarf2out.c (add_var_loc_to_decl): Likewise.
532 * gimple-fold.c (get_base_constructor): Return the offset as a
533 poly_int64_pod rather than a HOST_WIDE_INT.
534 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
535 * ipa-polymorphic-call.c
536 (ipa_polymorphic_call_context::set_by_invariant)
537 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
538 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
539 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
540 rather than get_ref_base_and_extent.
541 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
542 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
543 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
545 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
546 call to get_ref_base_and_extent.
547 * tree-sra.c (create_access, get_access_for_expr): Likewise.
548 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
549 (stmt_kills_ref_p): Likewise.
550 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
551 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
553 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
554 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
555 when calling native_encode_expr.
556 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
557 call to get_ref_base_and_extent.
558 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
559 get_ref_base_and_extent.
560 * var-tracking.c (track_expr_p): Likewise.
562 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
563 Alan Hayward <alan.hayward@arm.com>
564 David Sherwood <david.sherwood@arm.com>
566 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
567 HOST_WIDE_INT to poly_int64_pod.
568 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
569 polynomail parameter offsets.
571 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
572 Alan Hayward <alan.hayward@arm.com>
573 David Sherwood <david.sherwood@arm.com>
575 * gengtype.c (main): Handle poly_int64_pod.
576 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
577 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
578 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
579 from HOST_WIDE_INT to poly_int64_pod.
580 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
581 (copy_cfa): New function.
582 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
583 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
584 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
585 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
586 (initial_return_save): Treat offsets as poly_ints.
587 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
588 offset is nonconstant.
589 (reg_save): Take the offset as a poly_int64. Fall back to
590 DW_CFA_expression for nonconstant offsets.
591 (queue_reg_save): Take the offset as a poly_int64.
592 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
594 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
595 Alan Hayward <alan.hayward@arm.com>
596 David Sherwood <david.sherwood@arm.com>
598 * rtl.h (operand_subword, operand_subword_force): Take the offset
599 as a poly_uint64 an unsigned int.
600 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
602 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
603 Alan Hayward <alan.hayward@arm.com>
604 David Sherwood <david.sherwood@arm.com>
606 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
607 'p' format code. Use INT_LIST rather than SUBREG as the example of
608 a code with an XINT and an XEXP. Remove the implication that
609 accessing an rtx field using XINT is expected to work.
610 * rtl.def (SUBREG): Change format from "ei" to "ep".
611 * rtl.h (rtunion::rt_subreg): New field.
612 (XCSUBREG): New macro.
613 (SUBREG_BYTE): Use it.
614 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
615 Update constructor accordingly.
616 (subreg_shape::operator ==): Update accordingly.
617 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
618 than an unsigned int.
619 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
620 a poly_uint64 rather than an unsigned int.
621 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
622 than an unsigned int.
623 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
624 (subreg_size_highpart_offset): Return a poly_uint64 rather than
625 an unsigned int. Take the sizes as poly_uint64s.
626 (subreg_offset_from_lsb): Return a poly_uint64 rather than
627 an unsigned int. Take the shift as a poly_uint64 rather than
629 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
630 as a poly_uint64 rather than an unsigned int.
631 (simplify_subreg_regno): Likewise.
632 (byte_lowpart_offset): Return the memory offset as a poly_int64
634 (subreg_memory_offset): Likewise. Take the subreg offset as a
635 poly_uint64 rather than an unsigned int.
636 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
637 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
638 poly_uint64 rather than an unsigned int.
639 * rtl.c (rtx_format): Describe 'p' in comment.
640 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
641 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
642 offset as a poly_uint64 rather than an unsigned int.
643 (byte_lowpart_offset): Return the memory offset as a poly_int64
645 (subreg_memory_offset): Likewise. Take the subreg offset as a
646 poly_uint64 rather than an unsigned int.
647 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
648 mode sizes as poly_uint64s rather than unsigned ints. Return a
649 poly_uint64 rather than an unsigned int.
650 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
651 (copy_insn_1): Handle 'p'.
652 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
653 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
654 an unsigned int. Return the shift in the same way.
655 (subreg_lsb): Return the shift as a poly_uint64 rather than an
657 (subreg_size_offset_from_lsb): Take the sizes and shift as
658 poly_uint64s rather than unsigned ints. Return the offset as
660 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
661 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
663 * rtlhash.c (add_rtx): Handle 'p'.
664 * genemit.c (gen_exp): Likewise.
665 * gengenrtl.c (type_from_format, gendef): Likewise.
666 * gensupport.c (subst_pattern_match, get_alternatives_number)
667 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
668 (subst_dup): Likewise.
669 * gengtype.c (adjust_field_rtx_def): Likewise.
670 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
671 (match_pattern_2): Likewise.
672 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
673 (rtx_test::subreg_field): New function.
674 (operator ==, safe_to_hoist_p, transition_parameter_type)
675 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
676 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
677 * genpeep.c (match_rtx): Likewise.
678 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
679 (rtx_writer::print_rtx_operand): Handle 'p'.
680 (print_value): Handle SUBREG.
681 * read-rtl.c (apply_int_iterator): Likewise.
682 (rtx_reader::read_rtx_operand): Handle 'p'.
683 * alias.c (rtx_equal_for_memref_p): Likewise.
684 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
685 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
687 * calls.c (expand_call): Likewise.
688 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
689 (make_extraction, gen_lowpart_for_combine): Likewise.
690 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
692 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
693 rather than an unsigned int. Treat subreg offsets as poly_ints.
694 (exp_equiv_p): Handle 'p'.
695 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
696 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
697 * dse.c (find_shift_sequence): Likewise.
698 * dwarf2out.c (rtl_for_decl_location): Likewise.
699 * expmed.c (extract_low_bits): Likewise.
700 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
701 (expand_expr_real_2): Likewise.
702 * final.c (alter_subreg): Likewise.
703 (leaf_renumber_regs_insn): Handle 'p'.
704 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
705 Treat subreg offsets as poly_ints.
706 * fwprop.c (forward_propagate_and_simplify): Likewise.
707 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
708 * ira.c (get_subreg_tracking_sizes): Likewise.
709 * ira-conflicts.c (go_through_subreg): Likewise.
710 * ira-lives.c (process_single_reg_class_operands): Likewise.
711 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
712 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
713 as a poly_uint64 rather than an unsigned int.
714 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
715 subreg offsets as poly_ints.
716 * lra-constraints.c (operands_match_p): Handle 'p'.
717 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
718 * lra-spills.c (assign_mem_slot): Likewise.
719 * postreload.c (move2add_valid_value_p): Likewise.
720 * recog.c (general_operand, indirect_operand): Likewise.
721 * regcprop.c (copy_value, maybe_mode_change): Likewise.
722 (copyprop_hardreg_forward_1): Likewise.
723 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
724 (record_subregs_of_mode): Likewise.
725 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
726 * reload.c (operands_match_p): Handle 'p'.
727 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
728 * reload1.c (alter_reg, choose_reload_regs): Likewise.
729 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
730 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
731 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
732 (simplify_const_poly_int_tests<N>::run): Likewise.
733 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
734 a poly_uint64 rather than an unsigned int.
735 * valtrack.c (debug_lowpart_subreg): Likewise.
736 * var-tracking.c (var_lowpart): Likewise.
737 (loc_cmp): Handle 'p'.
739 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
740 Alan Hayward <alan.hayward@arm.com>
741 David Sherwood <david.sherwood@arm.com>
743 * ira.c (get_subreg_tracking_sizes): New function.
744 (init_live_subregs): Take an integer size rather than a register.
745 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
746 to init_live_subregs.
748 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
749 Alan Hayward <alan.hayward@arm.com>
750 David Sherwood <david.sherwood@arm.com>
752 * expr.c (store_constructor_field): Change bitsize from a
753 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
754 HOST_WIDE_INT to a poly_int64.
755 (store_constructor): Change size from a HOST_WIDE_INT to
757 (store_field): Likewise bitsize and bitpos.
759 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
760 Alan Hayward <alan.hayward@arm.com>
761 David Sherwood <david.sherwood@arm.com>
763 * expmed.h (store_bit_field): Change bitregion_start and
764 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
765 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
766 (store_bit_field_1, store_integral_bit_field, store_bit_field)
767 (store_fixed_bit_field, store_split_bit_field): Likewise.
768 * expr.c (store_constructor_field, store_field): Likewise.
769 (optimize_bitfield_assignment_op): Likewise. Make the same change
770 to bitsize and bitpos.
771 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
772 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
773 same change in the constructor arguments.
774 (get_best_mode): Change bitregion_start and bitregion_end from
775 unsigned HOST_WIDE_INT to poly_uint64.
776 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
777 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
779 (bit_field_mode_iterator::next_mode): Update for new types
780 of m_bitregion_start and m_bitregion_end.
781 (get_best_mode): Change bitregion_start and bitregion_end from
782 unsigned HOST_WIDE_INT to poly_uint64.
784 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
785 Alan Hayward <alan.hayward@arm.com>
786 David Sherwood <david.sherwood@arm.com>
788 * rtl.h (simplify_gen_subreg): Add a temporary overload that
789 accepts poly_uint64 offsets.
790 * expmed.h (extract_bit_field): Take bitsize and bitnum as
791 poly_uint64s rather than unsigned HOST_WIDE_INTs.
792 * expmed.c (lowpart_bit_field_p): Likewise.
793 (extract_bit_field_as_subreg): New function, split out from...
794 (extract_bit_field_1): ...here. Take bitsize and bitnum as
795 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
796 extractions, check that BITSIZE matches the size of the extracted
797 value and that BITNUM is an exact multiple of that size.
798 If all else fails, try forcing the value into memory if
799 BITNUM is variable, and adjusting the address so that the
800 offset is constant. Split the part that can only handle constant
801 bitsize and bitnum out into...
802 (extract_integral_bit_field): ...this new function.
803 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
804 rather than unsigned HOST_WIDE_INTs.
806 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
807 Alan Hayward <alan.hayward@arm.com>
808 David Sherwood <david.sherwood@arm.com>
810 * expmed.h (store_bit_field): Take bitsize and bitnum as
811 poly_uint64s rather than unsigned HOST_WIDE_INTs.
812 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
813 that returns the byte size.
814 (store_bit_field_1): Take bitsize and bitnum as
815 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
816 to simple_mem_bitfield_p. Split the part that can only handle
817 constant bitsize and bitnum out into...
818 (store_integral_bit_field): ...this new function.
819 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
820 than unsigned HOST_WIDE_INTs.
821 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
823 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
824 Alan Hayward <alan.hayward@arm.com>
825 David Sherwood <david.sherwood@arm.com>
827 * lra-int.h (lra_reg): Change offset from int to poly_int64.
828 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
830 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
831 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
832 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
833 offset as a poly_int64 rather than an int.
834 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
835 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
836 * lra-constraints.c (equiv_address_substitution): Track offsets
838 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
839 (curr_insn_transform): Handle the new form of sp_offset.
840 * lra-eliminations.c (lra_elim_table): Change previous_offset
841 and offset from HOST_WIDE_INT to poly_int64.
842 (print_elim_table, update_reg_eliminate): Update accordingly.
843 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
844 (get_elimination): Update accordingly.
845 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
846 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
847 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
848 poly_int64 offsets generally.
849 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
850 (mark_not_eliminable, init_elimination): Update accordingly.
851 (remove_reg_equal_offset_note): Return a bool and pass the new
852 offset back by pointer as a poly_int64.
853 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
854 rather than a HOST_WIDE_INT.
855 (do_remat): Track offsets poly_int64s.
856 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
858 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
859 Alan Hayward <alan.hayward@arm.com>
860 David Sherwood <david.sherwood@arm.com>
862 * rtl.h (mem_attrs): Add a default constructor. Change size and
863 offset from HOST_WIDE_INT to poly_int64.
864 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
865 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
866 (widen_memory_access): Take the sizes and offsets as poly_int64s
867 rather than HOST_WIDE_INTs.
868 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
869 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
871 (adjust_offset_for_component_ref): Change the offset from a
872 HOST_WIDE_INT to a poly_int64.
873 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
874 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
875 * dce.c (find_call_stack_args): Likewise.
876 * dse.c (record_store): Likewise.
877 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
878 * print-rtl.c (rtx_writer::print_rtx): Likewise.
879 * read-rtl-function.c (test_loading_mem): Likewise.
880 * rtlanal.c (may_trap_p_1): Likewise.
881 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
882 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
883 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
884 (mem_attrs::mem_attrs): New function.
885 (set_mem_attributes_minus_bitpos): Change bitpos from a
886 HOST_WIDE_INT to poly_int64.
887 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
888 (clear_mem_offset, clear_mem_size, change_address)
889 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
890 initialize mem_attrs.
891 (set_mem_offset, set_mem_size, adjust_address_1)
892 (adjust_automodify_address_1, offset_address, widen_memory_access):
893 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
895 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
896 Alan Hayward <alan.hayward@arm.com>
897 David Sherwood <david.sherwood@arm.com>
899 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
900 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
901 rather than 0 to represent an unknown size. Assert that the size
902 is known when the mode isn't BLKmode.
903 (may_trap_p_1): Use -1 for unknown sizes.
904 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
906 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
907 Alan Hayward <alan.hayward@arm.com>
908 David Sherwood <david.sherwood@arm.com>
910 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
911 to poly_int64. Update commentary for positions_needed.large.
912 (read_info_type): Change offset and width from HOST_WIDE_INT
914 (set_usage_bits): Likewise.
915 (canon_address): Return the offset as a poly_int64 rather than
916 a HOST_WIDE_INT. Use strip_offset_and_add.
917 (set_all_positions_unneeded, any_positions_needed_p): Use
918 positions_needed.large to track stores with non-constant widths.
919 (all_positions_needed_p): Likewise. Take the offset and width
920 as poly_int64s rather than ints. Assert that rhs is nonnull.
921 (record_store): Cope with non-constant offsets and widths.
922 Nullify the rhs of an earlier store if we can't tell which bytes
924 (find_shift_sequence): Take the access_size and shift as poly_int64s
926 (get_stored_val): Take the read_offset and read_width as poly_int64s
927 rather than HOST_WIDE_INTs.
928 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
929 non-constant offsets and widths.
931 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
932 Alan Hayward <alan.hayward@arm.com>
933 David Sherwood <david.sherwood@arm.com>
935 * inchash.h (inchash::hash::add_poly_int): New function.
936 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
937 Use poly_int64 rather than HOST_WIDE_INT.
938 (ao_ref::max_size_known_p): New function.
939 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
940 rather than HOST_WIDE_INT.
941 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
942 to temporaries until its interface is adjusted to match.
943 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
944 (aliasing_component_refs_p, decl_refs_may_alias_p)
945 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
946 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
947 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
949 * alias.c (ao_ref_from_mem): Likewise.
950 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
951 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
952 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
953 (maybe_trim_complex_store, maybe_trim_constructor_store)
954 (live_bytes_read, dse_classify_store): Likewise.
955 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
956 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
957 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
958 (vn_reference_lookup_3): Likewise.
959 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
961 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
962 Alan Hayward <alan.hayward@arm.com>
963 David Sherwood <david.sherwood@arm.com>
965 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
966 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
967 instead of ranges_overlap_p.
969 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
970 Alan Hayward <alan.hayward@arm.com>
971 David Sherwood <david.sherwood@arm.com>
973 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
974 sizes as poly_int64s rather than HOST_WIDE_INTs.
976 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
977 Alan Hayward <alan.hayward@arm.com>
978 David Sherwood <david.sherwood@arm.com>
980 * gimple-fold.h (fold_ctor_reference): Take the offset and size
981 as poly_uint64 rather than unsigned HOST_WIDE_INT.
982 * gimple-fold.c (fold_ctor_reference): Likewise.
984 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
985 Alan Hayward <alan.hayward@arm.com>
986 David Sherwood <david.sherwood@arm.com>
988 * target.def (dwarf_poly_indeterminate_value): New hook.
989 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
990 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
991 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
992 * doc/tm.texi: Regenerate.
993 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
994 offset as a poly_int64.
995 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
996 offset as a poly_int64.
997 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
998 Take the offset as a poly_int64.
999 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
1000 (frame_pointer_fb_offset): Change to a poly_int64.
1001 (int_loc_descriptor): Take the offset as a poly_int64. Use
1002 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
1003 (based_loc_descr): Take the offset as a poly_int64.
1004 Use strip_offset_and_add to handle (plus X (const)).
1005 Use new_reg_loc_descr instead of an open-coded version of the
1006 previous implementation.
1007 (mem_loc_descriptor): Handle CONST_POLY_INT.
1008 (compute_frame_pointer_to_fb_displacement): Take the offset as a
1009 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
1011 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1012 Alan Hayward <alan.hayward@arm.com>
1013 David Sherwood <david.sherwood@arm.com>
1015 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
1016 (gen_rtx_REG_offset): Take the offset as a poly_int64.
1017 * inchash.h (inchash::hash::add_poly_hwi): New function.
1018 * gengtype.c (main): Register poly_int64.
1019 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
1020 offset as a poly_int.
1021 (reg_attr_hasher::equal): Use must_eq to compare offsets.
1022 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
1023 offset as a poly_int64.
1024 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
1025 * print-rtl.c (print_poly_int): New function.
1026 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
1028 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
1030 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
1031 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
1032 (vt_get_decl_and_offset): Return the offset as a poly_int64.
1033 Enforce track_offset_p for parts of a PARALLEL.
1034 (vt_add_function_parameter): Use const_offset for the final
1035 offset to track. Use get_tracked_reg_offset for the parts
1038 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1039 Alan Hayward <alan.hayward@arm.com>
1040 David Sherwood <david.sherwood@arm.com>
1042 * target.def (truly_noop_truncation): Take poly_uint64s instead of
1043 unsigned ints. Change default to hook_bool_puint64_puint64_true.
1044 * doc/tm.texi: Regenerate.
1045 * hooks.h (hook_bool_uint_uint_true): Delete.
1046 (hook_bool_puint64_puint64_true): Declare.
1047 * hooks.c (hook_bool_uint_uint_true): Delete.
1048 (hook_bool_puint64_puint64_true): New function.
1049 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
1050 instead of unsigned ints.
1051 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
1052 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
1054 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1055 Alan Hayward <alan.hayward@arm.com>
1056 David Sherwood <david.sherwood@arm.com>
1058 * optabs.h (expand_operand): Add an int_value field.
1059 (create_expand_operand): Add an int_value parameter and use it
1060 to initialize the new expand_operand field.
1061 (create_integer_operand): Replace with a declaration of a function
1062 that accepts poly_int64s. Move the implementation to...
1063 * optabs.c (create_integer_operand): ...here.
1064 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
1065 the mode preserves the value of int_value, instead of calling
1066 const_int_operand on the rtx. Use gen_int_mode to generate
1069 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1070 Alan Hayward <alan.hayward@arm.com>
1071 David Sherwood <david.sherwood@arm.com>
1073 * dumpfile.h (dump_dec): Declare.
1074 * dumpfile.c (dump_dec): New function.
1075 * pretty-print.h (pp_wide_integer): Turn into a function and
1076 declare a poly_int version.
1077 * pretty-print.c (pp_wide_integer): New function for poly_ints.
1079 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1080 Alan Hayward <alan.hayward@arm.com>
1081 David Sherwood <david.sherwood@arm.com>
1083 * doc/generic.texi (POLY_INT_CST): Document.
1084 * tree.def (POLY_INT_CST): New tree code.
1085 * treestruct.def (TS_POLY_INT_CST): New tree layout.
1086 * tree-core.h (tree_poly_int_cst): New struct.
1087 (tree_node): Add a poly_int_cst field.
1088 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
1089 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
1090 instead of a wide_int_ref.
1091 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
1093 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1094 instead of an unsigned HOST_WIDE_INT.
1095 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
1096 (ptrdiff_tree_p): Declare.
1097 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
1098 extern inline implementations if the target doesn't use POLY_INT_CST.
1099 (poly_int_tree_p): New function.
1100 (wi::unextended_tree): New class.
1101 (wi::int_traits <unextended_tree>): New override.
1102 (wi::extended_tree): Add a default constructor.
1103 (wi::extended_tree::get_tree): New function.
1104 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
1105 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
1106 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
1107 (wi::tree_to_poly_wide_ref): New typedefs.
1108 (wi::ints_for): Provide overloads for extended_tree and
1110 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
1111 (wi::to_wide): New functions.
1112 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
1113 * tree.c (poly_int_cst_hasher): New struct.
1114 (poly_int_cst_hash_table): New variable.
1115 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
1116 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
1118 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
1119 (init_ttree): Initialize poly_int_cst_hash_table.
1120 (build_int_cst, build_int_cst_type, build_invariant_address): Take
1121 a poly_int64 instead of a HOST_WIDE_INT.
1122 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1123 instead of an unsigned HOST_WIDE_INT.
1124 (wide_int_to_tree): Rename to...
1125 (wide_int_to_tree_1): ...this.
1126 (build_new_poly_int_cst, build_poly_int_cst): New functions.
1127 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
1128 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
1129 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
1130 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
1131 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
1133 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
1134 (streamer_read_tree_body): Likewise.
1135 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
1136 (streamer_write_tree_body): Likewise.
1137 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
1138 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
1139 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
1140 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
1141 * gimple-expr.h (is_gimple_constant): Likewise.
1142 * gimplify.c (maybe_with_size_expr): Likewise.
1143 * print-tree.c (print_node): Likewise.
1144 * tree-data-ref.c (data_ref_compare_tree): Likewise.
1145 * tree-pretty-print.c (dump_generic_node): Likewise.
1146 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1147 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
1148 * tree-vrp.c (compare_values_warnv): Likewise.
1149 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
1150 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
1151 (force_expr_to_var_cost): Likewise.
1152 * tree-ssa-loop.c (for_each_index): Likewise.
1153 * fold-const.h (build_invariant_address, size_int_kind): Take a
1154 poly_int64 instead of a HOST_WIDE_INT.
1155 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
1156 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
1158 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
1159 (int_const_binop_2): New function, split out from...
1160 (int_const_binop_1): ...here. Handle POLY_INT_CST.
1161 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
1162 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
1163 * gimple-ssa-strength-reduction.c (slsr_process_add)
1164 (slsr_process_mul): Check for INTEGER_CSTs before using them
1166 * stor-layout.c (bits_from_bytes): New function.
1167 (bit_from_pos): Use it.
1168 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
1169 by BITS_PER_UNIT to get the TYPE_SIZE.
1170 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
1171 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
1173 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1174 Alan Hayward <alan.hayward@arm.com>
1175 David Sherwood <david.sherwood@arm.com>
1177 * doc/rtl.texi (const_poly_int): Document. Also document the
1178 rtl sharing behavior.
1179 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
1180 * rtl.h (const_poly_int_def): New struct.
1181 (rtx_def::u): Add a cpi field.
1182 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
1183 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
1184 (wi::rtx_to_poly_wide_ref): New typedef
1185 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
1186 (poly_int_rtx_p): New functions.
1187 (trunc_int_for_mode): Declare a poly_int64 version.
1188 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1189 (immed_wide_int_const): Take a poly_wide_int_ref rather than
1191 (strip_offset): Declare.
1192 (strip_offset_and_add): New function.
1193 * rtl.def (CONST_POLY_INT): New rtx code.
1194 * rtl.c (rtx_size): Handle CONST_POLY_INT.
1195 (shared_const_p): Use poly_int_rtx_p.
1196 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
1198 (gen_int_shift_amount): Likewise.
1199 * emit-rtl.c (const_poly_int_hasher): New class.
1200 (const_poly_int_htab): New variable.
1201 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
1202 (const_poly_int_hasher::hash): New function.
1203 (const_poly_int_hasher::equal): Likewise.
1204 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
1205 (immed_wide_int_const): Rename to...
1206 (immed_wide_int_const_1): ...this and make static.
1207 (immed_wide_int_const): New function, taking a poly_wide_int_ref
1208 instead of a wide_int_ref.
1209 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
1210 (gen_lowpart_common): Handle CONST_POLY_INT.
1211 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
1212 * cselib.c (cselib_hash_rtx): Likewise.
1213 * dwarf2out.c (const_ok_for_output_1): Likewise.
1214 * expr.c (convert_modes): Likewise.
1215 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
1216 * rtlhash.c (add_rtx): Likewise.
1217 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
1218 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1219 Handle existing CONST_POLY_INT rtxes.
1220 * expmed.h (expand_shift): Take a poly_int64 instead of a
1222 * expmed.c (expand_shift): Likewise.
1223 * rtlanal.c (strip_offset): New function.
1224 (commutative_operand_precedence): Give CONST_POLY_INT the same
1225 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
1227 * rtl-tests.c (const_poly_int_tests): New struct.
1228 (rtl_tests_c_tests): Use it.
1229 * simplify-rtx.c (simplify_const_unary_operation): Handle
1231 (simplify_const_binary_operation): Likewise.
1232 (simplify_binary_operation_1): Fold additions of symbolic constants
1233 and CONST_POLY_INTs.
1234 (simplify_subreg): Handle extensions and truncations of
1236 (simplify_const_poly_int_tests): New struct.
1237 (simplify_rtx_c_tests): Use it.
1238 * wide-int.h (storage_ref): Add default constructor.
1239 (wide_int_ref_storage): Likewise.
1240 (trailing_wide_ints): Use GTY((user)).
1241 (trailing_wide_ints::operator[]): Add a const version.
1242 (trailing_wide_ints::get_precision): New function.
1243 (trailing_wide_ints::extra_size): Likewise.
1245 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1246 Alan Hayward <alan.hayward@arm.com>
1247 David Sherwood <david.sherwood@arm.com>
1249 * emit-rtl.h (gen_int_shift_amount): Declare.
1250 * emit-rtl.c (gen_int_shift_amount): New function.
1251 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
1253 * calls.c (shift_return_value): Likewise.
1254 * cse.c (fold_rtx): Likewise.
1255 * dse.c (find_shift_sequence): Likewise.
1256 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
1257 (expand_shift, expand_smod_pow2): Likewise.
1258 * lower-subreg.c (shift_cost): Likewise.
1259 * optabs.c (expand_superword_shift, expand_doubleword_mult)
1260 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
1261 (expand_vec_perm_var): Likewise.
1262 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1263 (simplify_binary_operation_1): Likewise.
1264 * combine.c (try_combine, find_split_point, force_int_to_mode)
1265 (simplify_shift_const_1, simplify_shift_const): Likewise.
1266 (change_zero_ext): Likewise. Use simplify_gen_binary.
1268 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1270 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
1272 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1274 * doc/invoke.texi (ARM Options): Document accepted extension options
1275 for -march=armv8.3-a.
1277 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
1280 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
1281 or --with-float=softfp, set the default CPU to arm10e.
1283 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
1285 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
1286 * config/visium/predicates.md (const_shift_operand): Likewise.
1287 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
1288 (visium_legitimize_reload_address): Likewise.
1290 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1293 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
1295 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1298 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
1300 2017-12-20 Julia Koval <julia.koval@intel.com>
1302 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
1303 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
1304 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
1305 * config.gcc: Include vpclmulqdqintrin.h.
1306 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
1307 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
1309 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
1310 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
1311 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
1312 * config/i386/i386.c (isa2_opts): Add -mcx16.
1313 (isa_opts): Add -mpclmulqdq, remove -mcx16.
1314 (ix86_option_override_internal): Move mcx16 to flags2.
1315 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
1316 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
1317 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
1318 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
1319 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
1320 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
1321 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
1322 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
1323 * doc/invoke.texi: Add -mvpclmulqdq.
1325 2017-12-20 Tom de Vries <tom@codesourcery.com>
1328 * config/i386/i386.c (ix86_static_chain): Move DECL_STATIC_CHAIN test ...
1329 * calls.c (rtx_for_static_chain): ... here. New function.
1330 * calls.h (rtx_for_static_chain): Declare.
1331 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
1332 instead of targetm.calls.static_chain.
1333 * df-scan.c (df_get_entry_block_def_set): Same.
1335 2017-12-19 Tom de Vries <tom@codesourcery.com>
1337 PR tree-optimization/83493
1338 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
1341 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1343 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
1344 inform with hardcoded english plural handling.
1346 2017-12-18 Jeff Law <law@redhat.com>
1348 PR tree-optimization/83477
1349 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
1350 a non-virtual PHI, always push a new range.
1352 2017-12-19 Martin Sebor <msebor@redhat.com>
1355 * builtins.c (compute_objsize): Handle non-constant offsets.
1357 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1359 PR tree-optimization/83444
1360 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
1361 character load case, if get_stridx on MEM_REF's operand doesn't
1362 look usable, retry with get_addr_stridx.
1364 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
1367 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
1368 (variable_tracking_main_1): Keep markers even when VTA fails.
1371 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
1372 even if there are markers before it.
1373 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
1375 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1377 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
1378 typical order conditions.
1379 * sel-sched.c (extract_new_fences_from): Likewise.
1380 * config/visium/constraints.md (J, K, L): Likewise.
1381 * config/visium/predicates.md (const_shift_operand): Likewise.
1382 * config/visium/visium.c (visium_legitimize_address,
1383 visium_legitimize_reload_address): Likewise.
1384 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
1385 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
1386 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
1387 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
1388 SET_DUMP_DETAIL): Likewise.
1389 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
1390 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
1391 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
1392 avr_set_core_architecture, avr_set_current_function,
1393 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
1394 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
1395 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
1396 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
1397 avr_map_decompose, avr_fold_builtin): Likewise.
1398 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
1399 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
1400 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
1401 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
1402 * config/m32c/m32c.c (m32c_conditional_register_usage,
1403 m32c_address_cost): Likewise.
1404 * config/m32c/predicates.md (shiftcount_operand,
1405 longshiftcount_operand): Likewise.
1406 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
1407 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
1408 can_use_cdx_ldstw): Likewise.
1409 * config/nios2/nios2.h (CDX_REG_P): Likewise.
1410 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
1412 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
1413 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
1414 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
1415 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
1416 * config/vax/vax.c (adjacent_operands_p): Likewise.
1417 * config/ft32/constraints.md (L, b, KA): Likewise.
1418 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
1420 * cfgexpand.c (expand_stack_alignment): Likewise.
1421 * gcse.c (insert_expr_in_table): Likewise.
1422 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
1423 * cgraphunit.c (cgraph_node::expand): Likewise.
1424 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
1425 * emit-rtl.c (add_insn): Likewise.
1426 * input.c (dump_location_info): Likewise.
1427 * passes.c (NEXT_PASS): Likewise.
1428 * read-rtl-function.c (parse_note_insn_name,
1429 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
1431 * sched-rgn.c (sched_rgn_init): Likewise.
1432 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
1433 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
1434 if_then_else_cond, simplify_shift_const_1, simplify_comparison): Likewise.
1435 * explow.c (eliminate_constant_term): Likewise.
1436 * final.c (leaf_renumber_regs_insn): Likewise.
1437 * cfgrtl.c (print_rtl_with_bb): Likewise.
1438 * genhooks.c (emit_init_macros): Likewise.
1439 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
1440 * tree-data-ref.c (conflict_fn): Likewise.
1441 * selftest.c (assert_streq): Likewise.
1442 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
1443 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
1444 fold_binary_loc, multiple_of_p): Likewise.
1445 * reload.c (push_reload, find_equiv_reg): Likewise.
1446 * et-forest.c (et_nca, et_below): Likewise.
1447 * dbxout.c (dbxout_symbol_location): Likewise.
1448 * reorg.c (relax_delay_slots): Likewise.
1449 * dojump.c (do_compare_rtx_and_jump): Likewise.
1450 * gengtype-parse.c (type): Likewise.
1451 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
1452 simplify_const_relational_operation): Likewise.
1453 * reload1.c (do_output_reload): Likewise.
1454 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
1455 * gengtype.c (type_for_name): Likewise.
1456 * gimple-ssa-sprintf.c (format_directive): Likewise.
1458 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1461 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
1462 accessing it. Adjust comment.
1464 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1467 * predict.c (zero_one_minusone): New function.
1468 (apply_return_prediction): Avoid return prediction for functions
1469 returning only -1, 0 and 1 values, unless they only return -1 and 0
1472 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1474 * config/arc/arc.c (legitimate_scaled_address_p): Clean
1475 fall-through warning.
1476 (arc_compute_frame_size): Remove unused variables.
1477 (arc_print_operand): Fix fprintif format.
1478 (arc_can_follow_jump): Clean fall-through warning.
1480 2017-12-19 Marek Polacek <polacek@redhat.com>
1483 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
1486 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1488 * config/arc/arc.c (overriderregs): New variable.
1489 (arc_override_options): Track fixed/call saved/call options.
1490 (arc_conditional_register_usage): Check against overrideregs
1491 variable whenever we change register properties.
1493 2017-12-19 Nathan Sidwell <nathan@acm.org>
1495 * opts.c (finish_options): Don't prefix dump_base_name if it
1496 already contains directories.
1498 2017-12-19 Martin Liska <mliska@suse.cz>
1500 PR rtl-optimization/82675
1501 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
1502 more element in sbitmap.
1504 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
1506 * gimplify.c (gimplify_expr): Use error_operand_p.
1508 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
1511 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
1514 2017-12-19 Daniel Cederman <cederman@gaisler.com>
1516 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
1517 the jump is to a label.
1519 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1521 PR tree-optimization/83444
1522 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
1527 * ipa-inline.c (flatten_remove_node_hook): New function.
1528 (ipa_inline): Keep only nodes with flatten attribute at the end of
1529 the array in the order from ipa_reverse_postorder, only walk that
1530 portion of array for flattening, if there is more than one such
1531 node, temporarily register a removal hook and ignore removed nodes.
1533 PR tree-optimization/80631
1534 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
1535 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
1536 IFN_REDUC_MAX or IFN_REDUC_MIN.
1538 2017-12-18 Jeff Law <law@redhat.com>
1540 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
1541 record symbolic equivalences from backedges in the CFG.
1544 2017-11-19 Jeff Law <law@redhat.com>
1546 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1547 of degenerates resulting from ignoring an edge.
1549 2017-12-18 Martin Sebor <msebor@redhat.com>
1552 PR tree-optimization/78450
1553 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
1554 (handle_builtin_strlen): Call it.
1556 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
1558 PR rtl-optimization/83424
1559 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
1561 2017-12-18 Marek Polacek <polacek@redhat.com>
1564 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1565 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
1568 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1570 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
1571 (umaddsidi4, umaddsidi_split): Likewise.
1573 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1575 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
1578 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1581 * config/s390/s390.c (s390_option_override): Avoid strncpy.
1583 2017-12-18 Richard Biener <rguenther@suse.de>
1585 PR tree-optimization/81877
1586 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
1587 (outermost_indep_loop): Adjust.
1588 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
1589 (can_sm_ref_p): Adjust.
1591 2017-12-18 Richard Biener <rguenther@suse.de>
1594 * tree.c (array_at_struct_end_p): Return true if the underlying
1595 object has space for at least one element in excess of what
1596 the array domain specifies.
1598 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1600 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
1601 example to avoid overfull hbox.
1602 * doc/invoke.texi (Option Summary): Add missing @gol.
1603 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
1606 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1607 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1609 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
1610 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
1611 (AArch64 Options): Document -mverbose-cost-dump.
1612 (ARM Options): Likewise, plus -mflip-thumb.
1614 2017-12-17 Martin Sebor <msebor@redhat.com>
1617 * gimple-ssa-warn-restrict.c
1618 (builtin_memref::offset_out_of_bounds): Correct the handling of
1621 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1623 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
1625 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
1627 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
1629 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
1631 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
1634 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
1636 * doc/invoke.texi: Fix some typos.
1638 2017-12-16 Martin Sebor <msebor@redhat.com>
1640 PR tree-optimization/78918
1641 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
1642 * builtins.c (check_sizes): Rename...
1643 (check_access): ...to this. Rename function arguments for clarity.
1644 (check_memop_sizes): Adjust names.
1645 (expand_builtin_memchr, expand_builtin_memcpy): Same.
1646 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
1647 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
1648 (check_strncat_sizes, expand_builtin_strncat): Same.
1649 (expand_builtin_strncpy, expand_builtin_memset): Same.
1650 (expand_builtin_bzero, expand_builtin_memcmp): Same.
1651 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
1652 (maybe_emit_sprintf_chk_warning): Same.
1653 (expand_builtin_strcpy): Adjust.
1654 (expand_builtin_stpcpy): Same.
1655 (expand_builtin_with_bounds): Detect out-of-bounds accesses
1656 in pointer-checking forms of memcpy, memmove, and mempcpy.
1657 (gcall_to_tree_minimal, max_object_size): Define new functions.
1658 * builtins.h (max_object_size): Declare.
1659 * calls.c (alloc_max_size): Call max_object_size instead of
1660 hardcoding ssizetype limit.
1661 (get_size_range): Handle new argument.
1662 * calls.h (get_size_range): Add a new argument.
1663 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
1664 * doc/invoke.texi (-Wrestrict): Adjust, add example.
1665 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
1667 (gimple_fold_builtin_memory_chk): Same.
1668 (gimple_fold_builtin_stxcpy_chk): New function.
1669 * gimple-ssa-warn-restrict.c: New source.
1670 * gimple-ssa-warn-restrict.h: New header.
1671 * gimple.c (gimple_build_call_from_tree): Propagate location.
1672 * passes.def (pass_warn_restrict): Add new pass.
1673 * tree-pass.h (make_pass_warn_restrict): Declare.
1674 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
1676 (handle_builtin_strcat): Same.
1677 (strlen_optimize_stmt): Rename...
1678 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
1679 stpncpy, strncpy, and their checking forms.
1681 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
1683 PR rtl-optimization/82849
1684 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
1685 and get_max_loop_iterations_int.
1687 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1688 Alan Hayward <alan.hayward@arm.com>
1689 David Sherwood <david.sherwood@arm.com>
1691 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
1692 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
1693 size as a poly_uint64.
1694 (mode_for_vector, mode_for_int_vector): Take the number of vector
1695 elements as a poly_uint64.
1696 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
1697 size as a poly_uint64.
1698 (mode_for_vector, mode_for_int_vector): Take the number of vector
1699 elements as a poly_uint64.
1701 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1702 Alan Hayward <alan.hayward@arm.com>
1703 David Sherwood <david.sherwood@arm.com>
1705 * machmode.h (MACRO_MODE): New macro.
1706 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
1707 * targhooks.c (default_libcall_value, default_secondary_reload)
1708 (default_memory_move_cost, default_register_move_cost)
1709 (default_class_max_nregs): Likewise.
1711 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1712 Alan Hayward <alan.hayward@arm.com>
1713 David Sherwood <david.sherwood@arm.com>
1715 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
1716 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
1718 * genautomata.c (main): Likewise.
1719 * genconditions.c (write_header): Likewise.
1720 * genemit.c (main): Likewise.
1721 * genextract.c (print_header): Likewise.
1722 * genopinit.c (main): Likewise.
1723 * genoutput.c (output_prologue): Likewise.
1724 * genpeep.c (main): Likewise.
1725 * genpreds.c (write_insn_preds_c): Likewise.
1726 * genrecog.c (writer_header): Likewise.
1727 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
1728 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
1729 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
1730 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
1731 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
1732 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
1733 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
1734 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
1735 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
1736 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
1737 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
1738 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
1739 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
1740 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
1741 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
1742 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
1743 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
1744 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
1745 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
1746 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
1747 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
1748 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
1749 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
1750 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
1751 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
1752 * config/darwin.c (IN_TARGET_CODE): Likewise.
1753 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
1754 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
1755 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
1756 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
1757 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
1758 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
1759 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
1760 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
1761 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
1762 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
1763 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
1764 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
1765 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
1766 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
1767 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
1768 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
1769 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
1770 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
1771 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
1772 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
1773 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
1774 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
1775 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
1776 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
1777 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
1778 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
1779 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
1780 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
1781 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
1782 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
1783 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
1784 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
1785 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
1786 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
1787 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
1788 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
1789 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
1790 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
1791 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
1792 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
1793 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
1794 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
1795 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
1796 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
1797 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
1798 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
1799 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
1800 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
1801 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
1802 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
1803 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
1804 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
1805 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
1806 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
1807 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
1808 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
1809 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
1810 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
1811 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
1812 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
1813 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1814 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
1815 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
1816 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
1817 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
1818 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
1819 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
1820 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
1821 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
1822 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
1823 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
1824 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1825 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
1826 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
1827 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
1828 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
1829 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
1830 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
1831 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
1832 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
1833 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
1834 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
1835 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
1836 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
1837 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
1838 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
1839 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
1840 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
1841 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
1842 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
1843 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
1844 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
1845 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
1846 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
1847 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
1848 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
1849 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
1850 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
1851 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
1852 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
1853 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
1854 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
1855 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
1856 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
1857 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
1858 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
1860 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1861 Alan Hayward <alan.hayward@arm.com>
1862 David Sherwood <david.sherwood@arm.com>
1864 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
1867 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1868 Alan Hayward <alan.hayward@arm.com>
1869 David Sherwood <david.sherwood@arm.com>
1871 * doc/generic.texi (VEC_SERIES_EXPR): Document.
1872 * doc/md.texi (vec_series@var{m}): Document.
1873 * tree.def (VEC_SERIES_EXPR): New tree code.
1874 * tree.h (build_vec_series): Declare.
1875 * tree.c (build_vec_series): New function.
1876 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
1877 * tree-pretty-print.c (dump_generic_node): Likewise.
1878 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
1879 * tree-inline.c (estimate_operator_cost): Likewise.
1880 * expr.c (expand_expr_real_2): Likewise.
1881 * optabs-tree.c (optab_for_tree_code): Likewise.
1882 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
1883 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
1884 * expmed.c (make_tree): Handle VEC_SERIES.
1885 * optabs.def (vec_series_optab): New optab.
1886 * optabs.h (expand_vec_series_expr): Declare.
1887 * optabs.c (expand_vec_series_expr): New function.
1888 * tree-vect-generic.c (expand_vector_operations_1): Check that
1889 the operands also have vector type.
1891 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1892 Alan Hayward <alan.hawyard@arm.com>
1893 David Sherwood <david.sherwood@arm.com>
1895 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
1896 (VEC_COND_EXPR): Add missing @tindex.
1897 * doc/md.texi (vec_duplicate@var{m}): Document.
1898 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
1899 * tree.c (build_vector_from_val): Add stubbed-out handling of
1900 variable-length vectors, using VEC_DUPLICATE_EXPR.
1901 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
1902 * cfgexpand.c (expand_debug_expr): Likewise.
1903 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
1904 * tree-inline.c (estimate_operator_cost): Likewise.
1905 * tree-pretty-print.c (dump_generic_node): Likewise.
1906 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
1907 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
1908 (test_vec_duplicate_folding): New function.
1909 (fold_const_c_tests): Call it.
1910 * optabs.def (vec_duplicate_optab): New optab.
1911 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
1912 * optabs.h (expand_vector_broadcast): Declare.
1913 * optabs.c (expand_vector_broadcast): Make non-static. Try using
1914 vec_duplicate_optab.
1915 * expr.c (store_constructor): Try using vec_duplicate_optab for
1917 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
1919 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
1922 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
1923 div/mod latencies a bit.
1925 2017-12-15 Jeff Law <law@redhat.com>
1927 PR tree-optimization/36550
1928 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
1929 (mark_threaded_blocks): Rewrite code to avoid block copying when
1930 optimizing for size. Don't pessimize blocks which will be
1931 copied, but all the statements will be dead.
1933 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
1935 PR tree-optimization/81165
1936 * tree-ssa-threadupdate.c (uses_in_bb): New.
1937 (estimate_threading_killed_stmts): New.
1938 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
1939 * tree-ssa-threadedge.c
1940 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
1943 2017-12-15 Jeff Law <law@redhat.com>
1945 PR tree-optimization/83410
1946 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
1947 threads when parallelizing loops.
1949 2017-12-15 Jakub Jelinek <jakub@redhat.com>
1951 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
1953 * config/alpha/alpha.c (vms_attribute_table): Swap
1954 affects_type_identity and handler fields, adjust comments.
1955 * config/mips/mips.c (mips_attribute_table): Likewise.
1956 * config/visium/visium.c (visium_attribute_table): Likewise.
1957 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
1958 * config/microblaze/microblaze.c (microblaze_attribute_table):
1960 * config/spu/spu.c (spu_attribute_table): Likewise.
1961 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
1962 * config/arc/arc.c (arc_attribute_table): Likewise.
1963 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
1964 * config/v850/v850.c (v850_handle_interrupt_attribute,
1965 v850_handle_data_area_attribute): Formatting fixes.
1966 (v850_attribute_table): Swap affects_type_identity and handler
1967 fields, adjust comments.
1968 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
1969 * config/arm/arm.c (arm_attribute_table): Likewise.
1970 * config/avr/avr.c (avr_attribute_table): Likewise.
1971 * config/s390/s390.c (s390_attribute_table): Likewise.
1972 * config/sh/sh.c (sh_attribute_table): Likewise.
1973 * config/i386/i386.c (ix86_handle_cconv_attribute,
1974 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
1975 (ix86_attribute_table): Swap affects_type_identity and handler
1976 fields, adjust comments.
1977 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1978 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
1979 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
1980 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
1981 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
1982 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
1983 * config/rx/rx.c (rx_attribute_table): Likewise.
1984 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
1985 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
1986 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
1987 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
1988 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1989 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
1990 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
1991 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
1992 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
1993 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
1994 * doc/plugins.texi (user_attr): Likewise. Add NULL for
1996 * attribs.c (empty_attribute_table): Swap affects_type_identity and
1998 (register_scoped_attributes, decl_attributes): Formatting fixes.
2000 PR tree-optimization/83269
2001 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
2002 subtraction in arg0's type if type is signed and arg0 is unsigned.
2006 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
2007 convert? on @0 instead of convert. Check type of @1, not @0.
2008 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
2009 convert? on @0 instead of convert. Check type of @1, not @0.
2010 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
2011 only optimize if either both @1 and @2 types are narrower
2012 precision, or both are wider or equal precision, and in the former
2013 case only if both have undefined overflow.
2015 2017-12-15 Richard Biener <rguenther@suse.de>
2018 * internal-fn.def (IFN_NOP): Add.
2019 * internal-fn.c (expand_NOP): Do nothing.
2020 * lto-streamer-in.c (input_function): Instead of removing
2021 sanitizer calls replace them with IFN_NOP calls.
2023 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
2024 Alan Hayward <alan.hayward@arm.com>
2025 David Sherwood <david.sherwood@arm.com>
2027 * dse.c (store_info, read_info_type): Replace begin and end with
2029 (print_range): New function.
2030 (set_all_positions_unneeded, any_positions_needed_p)
2031 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
2033 (record_store): Likewise. Optimize the case in which all positions
2035 (get_stored_val): Replace read_begin and read_end with read_offset
2037 (replace_read): Update call accordingly.
2039 2017-12-15 Bin Cheng <bin.cheng@arm.com>
2041 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
2042 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
2043 (loop_cand::loop_cand): Initialize above members.
2044 (loop_cand::supported_operations): Delete.
2045 (loop_cand::can_interchange_p): Inline above function.
2046 (loop_cand::classify_simple_reduction): Record number of constant
2047 initialized simple reductions.
2048 (should_interchange_loops): New parameters. Check stmt cost of loops
2050 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
2051 Update call to should_interchange_loops.
2052 (should_interchange_loop_nest): Update call to
2053 should_interchange_loops.
2055 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
2058 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
2059 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
2060 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
2062 2017-12-15 Julia Koval <julia.koval@intel.com>
2064 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
2065 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
2066 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
2067 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
2068 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
2070 2017-12-15 Julia Koval <julia.koval@intel.com>
2072 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
2073 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
2074 * config/i386/sse.md (vaesenc_<mode>): New pattern.
2075 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
2076 _mm_aesenc_epi128): New intrinsics.
2078 2017-12-15 Julia Koval <julia.koval@intel.com>
2080 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
2081 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
2082 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
2083 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
2084 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
2086 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2088 * gimple-ssa-strength-reduction.c (analyze_increments):
2089 Distinguish replacement costs for constant strides from those for
2092 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2094 * var-tracking.c (variable_tracking_main_1): Formatting fix.
2096 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2098 * doc/invoke.texi: Document -Wcast-function-type.
2099 * recog.h (stored_funcptr): Change signature.
2100 * tree-dump.c (dump_node): Avoid warning.
2101 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
2103 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
2106 * gimple-fold.c (get_range_strlen): Add the handling of non-member
2109 2017-12-14 David Malcolm <dmalcolm@redhat.com>
2111 PR tree-optimization/83312
2112 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
2113 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
2114 "val" param, and to cope with arbitrary basic blocks.
2115 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
2116 handle NULL_TREE for "val", dropping "bb" param.
2117 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
2118 drop "bb" param. Handle NULL_TREE for "val".
2119 (find_case_label_for_value): Make "switch_stmt" param const.
2120 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
2122 (vrp_prop::check_all_array_refs): Reimplement as...
2123 (check_array_bounds_dom_walker::before_dom_children): ...this new
2124 vfunc. Replace linear search through BB block list, excluding
2125 those with non-executable in-edges via dominator walk.
2127 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2129 * config/arm/arm.opt (mverbose-cost-dump): New option.
2130 * config/arm/arm.c (arm_rtx_costs): Use it.
2132 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
2135 * reload1.c (emit_input_reload_insns): Skip debug markers.
2137 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
2139 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
2143 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
2144 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
2145 * config/sh/sh.c: Skip debug insns besides notes.
2146 * config/sh/sh.md: Likewise.
2147 * config/sh/sh_treg_combine.cc: Likewise.
2148 * config/sh/sync.md: Likewise.
2150 2017-12-14 Tom de Vries <tom@codesourcery.com>
2152 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2153 Add item for weak_undefined.
2155 2017-12-14 Richard Biener <rguenther@suse.de>
2157 PR tree-optimization/67842
2158 * sese.h (bb_in_region): Remove #if 0'ed code.
2160 2017-12-14 Richard Biener <rguenther@suse.de>
2162 PR tree-optimization/83326
2163 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2164 may_be_zero parameter and handle it by not marking the first
2165 peeled copy as not exiting the loop.
2166 (try_peel_loop): Likewise.
2167 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
2168 to handle the case of constant or zero iterations and perform
2169 loop header copying on-the-fly.
2171 2017-12-14 Richard Biener <rguenther@suse.de>
2173 PR tree-optimization/83418
2174 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
2175 Instead of asserting we don't get unfolded comparisons deal with
2178 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2181 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
2182 basic blocks. Assert debug bind insns don't appear outside of bbs,
2183 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
2186 PR tree-optimization/83198
2187 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
2188 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
2189 value if arg is a REAL_CST with incompatible type.
2191 2017-12-14 Sudakshina Das <sudi.das@arm.com>
2192 Bin Cheng <bin.cheng@arm.com>
2195 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
2197 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
2200 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2202 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
2203 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
2205 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
2206 Alan Hayward <alan.hayward@arm.com>
2207 David Sherwood <david.sherwood@arm.com>
2209 * poly-int.h: New file.
2210 * poly-int-types.h: Likewise.
2211 * coretypes.h: Include them.
2212 (POLY_INT_CONVERSION): Define.
2213 * target.def (estimated_poly_value): New hook.
2214 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
2215 * doc/tm.texi: Regenerate.
2216 * doc/poly-int.texi: New file.
2217 * doc/gccint.texi: Include it.
2218 * doc/rtl.texi: Describe restrictions on subreg modes.
2219 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
2220 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
2221 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
2222 * targhooks.h (default_estimated_poly_value): Declare.
2223 * targhooks.c (default_estimated_poly_value): New function.
2224 * target.h (estimated_poly_value): Likewise.
2225 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
2226 (wi::unary_traits): Delete.
2227 (wi::binary_traits::signed_shift_result_type): Define for
2228 offset_int << HOST_WIDE_INT, etc.
2229 (generic_wide_int::operator <<=): Define for all types and use
2230 wi::lshift instead of <<.
2231 (wi::hwi_with_prec): Add a default constructor.
2232 (wi::ints_for): New class.
2233 (operator <<): Define for all wide-int types.
2234 (operator /): New function.
2235 (operator %): Likewise.
2236 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
2237 (ASSERT_MAYBE_NE_AT): New macros.
2239 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
2240 Dominik Vogt <vogt@linux.vnet.ibm.com>
2243 * emit-rtl.c (init_emit): Remove ??? comment.
2244 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2245 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
2246 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
2247 alignment of 3 virtual registers to BITS_PER_WORD.
2249 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
2251 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
2253 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
2254 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
2256 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2259 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
2261 2017-12-13 Jakub Jelinek <jakub@redhat.com>
2263 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
2264 with the exception of debug begin stmt markers appear before
2268 * final.c (rest_of_handle_final): Call variable_tracking_main only
2269 if !flag_var_tracking.
2271 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2272 Jakub Jelinek <jakub@redhat.com>
2276 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
2277 labels debug stmts that can only appear after labels.
2279 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
2281 PR rtl-optimization/82398
2282 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
2283 EXPR_USEFULNESS in priority comparison.
2285 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
2287 PR rtl-optimization/83393
2288 * combine.c (move_deaths): If reg_stat points to a too new insn in
2289 last_death, do not use it: find the proper insn instead.
2291 2017-12-12 Jeff Law <law@redhat.com>
2293 PR tree-optimization/83298
2294 PR tree-optimization/83362
2295 PR tree-optimization/83383
2296 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
2297 push_value_range a public interface. Add new argument to
2298 record_ranges_from_stmt.
2299 * gimple-ssa-evrp-analyze.c
2300 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
2301 Update comments. Handle recording temporary equivalences.
2302 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
2303 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
2304 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
2305 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
2306 gimple-ssa-evrp-analyze.h.
2307 (record_temporary_equivalences_from_phis): Add new argument. When
2308 the PHI arg is an SSA_NAME, set the result's range to the range
2310 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
2311 from statements too.
2312 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
2313 Pass it down to children as needed.
2314 (thread_outgoing_edges): Likewise.
2315 (thread_across_edge): Likewise. Push/pop range state as needed.
2316 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
2318 2017-12-12 Julia Koval <julia.koval@intel.com>
2320 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
2321 (PTA_CANNONLAKE): Remove PTA_CLWB.
2323 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2326 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
2327 (vcondv2div2df): Likewise.
2328 (vconduv2dfv2di): Likewise.
2330 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2332 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
2333 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
2334 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
2335 ATTR_MATHFN_FPROUNDING_ERRNO.
2337 2017-12-12 Richard Biener <rguenther@suse.de>
2339 PR tree-optimization/83385
2340 * graphite-scop-detection.c (get_order, order): Remove.
2341 (bb_to_rpo): New global.
2343 (build_scops): Sort pbbs in RPO order.
2345 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
2347 * combine.c (simplify_set): Do not transform subregs to zero_extends
2348 if the destination mode is a vector mode.
2350 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2352 PR tree-optimization/83359
2353 * tree-cfg.h (fold_loop_internal_call): Declare.
2354 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
2355 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
2356 (find_loop_dist_alias): New function.
2357 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
2358 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
2361 PR tree-optimization/80631
2362 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
2363 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
2364 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
2365 hardcoding zero as the value if COND_EXPR is never true. For
2366 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
2367 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
2368 hardcoding MAX_EXPR as the reduction operation.
2369 (is_nonwrapping_integer_induction): Allow negative step.
2370 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
2371 vect_create_epilog_for_reduction, if no value is suitable, don't
2372 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
2374 2017-12-12 Richard Biener <rguenther@suse.de>
2376 PR tree-optimization/81889
2377 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
2378 range info from the non-wrapping IV instead of just the range
2381 2017-12-12 Julia Koval <julia.koval@intel.com>
2383 * config.gcc: Add vaesintrin.h.
2384 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2386 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2387 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2389 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2390 * config/i386/immintrin.h: Include vaesintrin.h.
2391 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2392 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2393 _mm_aesdec_epi128): New intrinsics.
2395 2017-12-12 Julia Koval <julia.koval@intel.com>
2397 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
2398 OPTION_MASK_ISA_VAES_UNSET): New.
2399 (ix86_handle_option): Handle -mvaes.
2400 * config/i386/cpuid.h: Define bit_VAES.
2401 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
2402 * config/i386/i386-c.c (__VAES__): New.
2403 * config/i386/i386.c (ix86_target_string): Add -mvaes.
2404 (ix86_valid_target_attribute_inner_p): Ditto.
2405 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
2406 * config/i386/i386.opt: Add -mvaes.
2407 * doc/invoke.texi: Ditto.
2409 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
2411 * debug.h (gcc_debug_hooks): Add inline_entry.
2412 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
2413 * debug.c (do_nothing_debug_hooks): Likewise.
2414 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2415 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
2416 (dwarf2_lineno_debug_hooks): Likewise.
2418 * common.opt (gstatement-frontiers): New, setting
2419 debug_nonbind_markers_p.
2420 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
2421 * toplev.c (process_options): Autodetect value for debug statement
2423 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
2424 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
2426 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
2427 markers. Integrate source bind into debug stmt expand loop.
2428 (pass_expand::execute): Check debug marker limit. Avoid deep
2429 TER and expand debug locations for debug bind insns only.
2430 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
2432 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
2433 * final.c (reemit_insn_block_notes): Take current block from
2434 nonbind markers. Declare note where it's first set.
2435 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
2436 begin stmt markers if enabled.
2437 (notice_source_line): Handle nonbind markers. Fail if their
2438 location is unknown or that of builtins.
2439 (rest_of_handle_final): Convert begin stmt markers to notes if
2440 var-tracking didn't run.
2441 (rest_of_clean_state): Skip begin stmt markers.
2442 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
2444 * function.c (allocate_struct_function): Set begin_stmt_markers.
2445 * function.h (struct function): Add debug_marker_count counter
2446 and debug_nonbind_markers flag.
2447 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
2448 * gimple-low.c (lower_function_body): Adjust
2449 debug_nonbind_markers.
2450 (lower_stmt): Drop or skip gimple debug stmts.
2451 (lower_try_catch): Skip debug stmts.
2452 * gimple.c (gimple_build_debug_begin_stmt): New.
2453 (gimple_copy): Increment debug_marker_count if copying one.
2454 * gimple.h (gimple_build_debug_begin_stmt): Declare.
2455 * gimplify.c (rexpr_location): New.
2456 (rexpr_has_location): New.
2457 (warn_switch_unreachable_r): Handle gimple debug stmts.
2458 (shortcut_cond_r): Call expr_location.
2460 (find_goto_label): New.
2461 (shortcut_cond_expr): Call expr_has_location, expr_location, and
2463 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
2465 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
2466 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
2467 (LANG_HOOKS_INITIALIZER): ... this.
2468 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
2469 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
2471 (update_ebb_live_info): Skip debug insn markers.
2472 * lra.c (debug_insn_static_data): Rename to...
2473 (debug_bind_static_data): ... this.
2474 (debug_marker_static_data): New.
2475 (lra_set_insn_recog_data): Select one of the above depending
2477 (lra_update_isn_regno_info): Don't assume debug insns have
2479 (push_insns): Skip debug insns.
2480 * lto-streamer-in.c (input_function): Drop debug stmts
2481 depending on active options. Adjust debug_nonbind_markers.
2482 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
2483 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
2484 begin stmt marker notes.
2485 (print_insn): Likewise.
2486 * recog.c (extract_insn): Recognize rtl for debug markers.
2487 * rtl.def (DEBUG_MARKER): New.
2488 * tree-inline.c: Include params.h.
2489 (remap_gimple_stmt): Handle nonbind markers.
2490 (maybe_move_debug_stmts_to_successors): Likewise.
2491 (copy_debug_stmt): Likewise.
2492 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
2493 markers regardless of no side effects.
2494 (tsi_link_before): Don't update container's side effects when adding
2495 a begin stmt marker.
2496 (tsi_link_after): Likewise.
2497 (expr_first): Skip begin stmt markers.
2498 (expr_last): Likewise.
2499 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
2500 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2501 Disregard nonbind markers.
2502 * tree.c (make_node_stat): Don't set side effects for begin stmt
2504 (build1_stat): Likewise.
2505 * tree.def (DEBUG_BEGIN_STMT): New.
2506 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
2507 * var-tracking.c (delete_debug_insns): Renamed to...
2508 (delete_vta_debug_insns): ... this.
2509 (reemit_marker_as_note): New.
2510 (vt_initialize): Reemit markers.
2511 (delete_vta_debug_insns): Likewise.
2512 (vt_debug_insns_local): Reemit or delete markers.
2513 (variable_tracking_main_1): Likewise.
2514 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
2515 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
2516 (gimple_debug_nonbind_marker_p): New.
2517 (gimple_build_debug_bind): Adjust.
2518 (gimple_build_debug_begin_stmt): New.
2519 * doc/invoke.texi (max-debug-marker-count): New param.
2520 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
2521 (debug_parameter_ref, debug_marker): New.
2522 (NOTE_INSN_BEGIN_STMT): New.
2523 (DEBUG_INSN): Describe begin stmt markers.
2525 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
2526 without debug insns, we wouldn't, but clean up debug insns
2527 after a control flow insn nevertheless.
2529 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
2530 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
2531 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
2532 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
2533 (rtl_tidy_fallthru_edge): Likewise.
2534 (rtl_verify_fallthru): Likewise.
2535 (rtl_verify_bb_layout): Likewise.
2536 (skip_insns_after_block): Likewise.
2537 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
2538 * dwarf2out.c: Include print-rtl.h.
2539 (dwarf2out_next_real_insn): New.
2540 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
2541 Dump debug binds in asm comments.
2542 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
2543 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
2544 callers to use gsi_start_nondebug_bb instead.
2545 (gsi_after_labels): Skip gimple debug stmts.
2546 (gsi_start_nondebug): New.
2547 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
2548 (proper_loop_form_for_interchange): Adjust.
2549 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
2550 * gimple.h (gimple_seq_last_nondebug_stmt): New.
2551 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
2552 (collect_fallthrough_labels): Likewise.
2553 (should_warn_for_implicit_fallthrough): Likewise.
2554 (warn_implicit_fallthrough_r): Likewise.
2555 (expand_FALLTHROUGH_r): Likewise.
2556 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
2557 (graphite_copy_stmts_from_block): Skip nonbind markers.
2558 * haifa-sched.c (sched_extend_bb): Skip debug insns.
2559 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
2560 * jump.c (clean_barriers): Skip debug insns.
2561 * omp-expand.c (expand_parallel_call): Skip debug insns.
2562 (expand_task_call): Likewise.
2563 (remove_exit_barrier): Likewise.
2564 (expand_omp_taskreg): Likewise.
2565 (expand_omp_for_init_counts): Likewise.
2566 (expand_omp_for_generic): Likewise.
2567 (expand_omp_for_static_nochunk): Likewise.
2568 (expand_omp_for_static_chunk): Likewise.
2569 (expand_omp_simd): Likewise.
2570 (expand_omp_taskloop_for_outer): Likewise.
2571 (expand_omp_taskloop_for_inner): Likewise.
2572 (expand_oacc_for): Likewise.
2573 (expand_omp_sections): Likewise.
2574 (expand_omp_single): Likewise.
2575 (expand_omp_synch): Likewise.
2576 (expand_omp_atomic_load): Likewise.
2577 (expand_omp_atomic_store): Likewise.
2578 (expand_omp_atomic_fetch_op): Likewise.
2579 (expand_omp_atomic_pipeline): Likewise.
2580 (expand_omp_atomic_mutex): Likewise.
2581 (expand_omp_target): Likewise.
2582 (grid_expand_omp_for_loop): Likewise.
2583 (grid_expand_target_grid_body): Likewise.
2584 (build_omp_regions_1): Likewise.
2585 * omp-low.c (check_combined_parallel): Skip debug stmts.
2586 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
2587 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
2588 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
2590 * tree-cfg.c (make_blobs_1): Skip debug stmts.
2591 (make_edges): Likewise.
2592 (cleanup_dead_labels): Likewise.
2593 (gimple_can_merge_blocks_p): Likewise.
2594 (stmt_starts_bb_p): Likewise.
2595 (gimple_block_label): Likewise.
2596 (gimple_redirect_edge_and_branch): Likewise.
2597 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
2599 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
2601 * tree-pretty-print (print_declaration): Omit initializer in slim
2603 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
2605 (eliminate_unnecessary_stmts): Stabilize block removal order.
2606 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
2607 * var-tracking.c (get_first_insn): New.
2608 (vt_emit_notes): Call it.
2609 (vt_initialize): Walk any insns before the first BB.
2610 (delete_debug_insns): Likewise.
2612 * gimple.h (enum gimple_debug_subcode): Add
2613 GIMPLE_DEBUG_BEGIN_STMT.
2614 (gimple_debug_begin_stmt_p): New.
2615 (gimple_debug_nonbind_marker_p): New.
2616 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
2617 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
2618 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
2619 * insn-notes.def (BEGIN_STMT): New.
2620 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
2621 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
2622 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
2623 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
2624 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
2625 (INSN_DEBUG_MARKER_KIND): New.
2626 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
2627 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
2628 (INSN_VAR_LOCATION_PTR): New.
2629 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
2631 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
2632 deps only if debug bind insns are enabled.
2633 (pass_expand::execute): Avoid deep TER and expand
2634 debug locations for debug bind insns only.
2635 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
2636 debug stmts special handling down to debug bind stmts.
2637 * combine.c (try_combine): Narrow debug insns special handling
2638 down to debug bind insns.
2639 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
2640 Narrow debug insns preexisting special handling down to debug
2642 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
2643 handling down to debug bind insns.
2644 * function.c (instantiate_virtual_regs): Skip debug markers,
2645 adjust handling of debug binds.
2646 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
2647 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
2648 * haifa-sched.c (schedule_insn): Narrow special handling of debug
2649 insns to debug bind insns.
2650 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
2651 special handling of debug stmts to debug bind stmts.
2652 * ipa-split.c (split_function): Likewise.
2653 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
2654 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
2656 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
2657 * regrename.c (build_def_use): Likewise.
2658 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
2659 (pass_cprop_hardreg): Narrow special casing of debug insns to
2661 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
2662 * reload1.c (reload): Likewise.
2663 * sese.c (sese_insert_phis_for_liveouts): Narrow special
2664 casing of debug stmts to debug bind stmts.
2665 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
2666 * ssa-iterators.h (num_imm_uses): Likewise.
2667 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
2668 debug stmts to debug bind stmts.
2669 * tree-inline.c (tree_function_versioning): Narrow special casing
2670 of debug stmts to debug bind stmts.
2671 * tree-loop-distribution.c (generate_loops_for_partition):
2672 Narrow special casing of debug stmts to debug bind stmts.
2673 * tree-sra.c (analyze_access_subtree): Narrow special casing
2674 of debug stmts to debug bind stmts.
2675 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
2676 stmts to debug bind stmts.
2677 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
2678 casing of debug stmts to debug bind stmts.
2679 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
2680 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
2681 casing of debug stmts to debug bind stmts.
2682 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2684 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
2685 debug stmts to debug bind stmts.
2686 (gimple_replace_ssa_lhs): Likewise.
2687 (insert_debug_temp_for_var_def): Likewise.
2688 (insert_debug_temps_for_defs): Likewise.
2689 (reset_debug_uses): Likewise.
2690 * tree-ssanames.c (release_ssa_name_fn): Likewise.
2691 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
2692 (adjust_debug_stmts): Likewise.
2693 (adjust_phi_and_debug_stmts): Likewise.
2694 (vect_do_peeling): Likewise.
2695 * tree-vect-loop.c (vect_transform_loop): Likewise.
2696 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
2697 * var-tracking.c (adjust_mems): Narrow special casing of debug
2698 insns to debug bind insns.
2699 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
2700 (compute_bb_dataflow, vt_find_locations): Likewise.
2701 (vt_expand_loc, emit_notes_for_changes): Likewise.
2702 (vt_init_cfa_base): Likewise.
2703 (vt_emit_notes): Likewise.
2704 (vt_initialize): Likewise.
2705 (vt_finalize): Likewise.
2707 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
2708 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
2709 (next_nonnote_nondebug_insn_bb): New.
2710 (prev_nonnote_nondebug_insn_bb): New.
2711 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
2712 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
2713 (prev_nonnote_nondebug_insn_bb): Declare.
2714 (next_nonnote_nondebug_insn_bb): Declare.
2715 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
2716 * cfgrtl.c (get_last_bb_insn): Likewise.
2717 * lra.c (push_insns): Likewise.
2719 2017-12-11 David Malcolm <dmalcolm@redhat.com>
2722 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
2724 * selftest.c (selftest::test_runner::test_runner): New ctor.
2725 (selftest::test_runner::~test_runner): New dtor.
2726 * selftest.h (class selftest::test_runner): New class.
2728 2017-12-11 Carl Love <cel@us.ibm.com>
2730 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
2731 vec_extract_fp32_from_shortl]): Add #defines.
2732 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
2733 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
2734 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
2735 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
2736 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
2737 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
2738 * doc/extend.texi: Add documentation for the added builtins.
2740 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
2742 PR rtl-optimization/80693
2743 PR rtl-optimization/81019
2744 PR rtl-optimization/81020
2745 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
2746 are not mentioned in i3. Place the REG_UNUSED note on i2,
2747 possibly modified to REG_DEAD, if it did not originate in i3.
2749 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2751 * recog.c (store_data_bypass_p_1): New function.
2752 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
2753 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
2755 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
2757 PR rtl-optimization/83361
2758 * ifcvt.c (if_convert): Call fixup_partitions.
2760 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
2762 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2763 early folding of splat_u{8,16,32}.
2765 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2767 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
2768 output_operand_lossage first argument with capital letter.
2769 (aarch64_override_options): Don't start error and sorry first argument
2770 with capital letter.
2772 2017-12-11 Andi Kleen <ak@linux.intel.com>
2774 PR gcov-profile/83355
2775 * auto-profile.c (string_table::get_index_by_decl): Don't
2776 recurse when abstract origin points to itself.
2778 2017-12-11 Bin Cheng <bin.cheng@arm.com>
2780 PR tree-optimization/83320
2781 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
2782 (prune_datarefs_not_in_loop): Ditto.
2784 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
2786 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
2788 2017-12-10 Jakub Jelinek <jakub@redhat.com>
2790 PR tree-optimization/83337
2791 * gimple-loop-interchange.cc (compute_access_stride): Handle
2792 bitfield DRs properly.
2794 2017-12-09 Jakub Jelinek <jakub@redhat.com>
2796 PR tree-optimization/83338
2797 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
2798 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
2799 vector of unsigned integers to vector of signed integers.
2801 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
2803 PR rtl-optimization/83317
2804 * lra-constraints.c (process_address_1): Add insn code check.
2806 2017-12-08 Michael Matz <matz@suse.de>
2808 Fix PR tree-optimization/83323
2809 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
2810 head-controlled loops and loop BBs.
2811 * common.opt (funroll-and-jam): Remove, instead ...
2812 (floop-unroll-and-jam): ... reuse this option.
2813 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
2814 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
2815 (-floop-unroll-and-jam): ... this option.
2817 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2819 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
2820 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
2821 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
2823 2017-12-08 Richard Biener <rguenther@suse.de>
2826 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
2827 handle accesses outside of zero-sized vars.
2829 2017-12-08 Martin Jambor <mjambor@suse.cz>
2831 PR tree-optimization/83141
2832 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
2833 test for MEM_REFs implicitely changing types with padding. Remove
2835 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
2837 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2839 * config/arc/arc.c (arc_attribute_table): Add exclusions to
2841 * config/avr/avr.c (avr_attribute_table): Likewise.
2842 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2843 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2844 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2845 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
2846 of struct attribute_spec.
2847 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2849 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
2852 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
2854 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
2857 * profile-count.c (profile_count::from_gcov_type): Move from
2858 profile-count.h; handle overflow.
2859 * profile-count.h (profile_count::from_gcov_type): Move offline.
2861 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
2863 PR rtl-optimization/83304
2864 * combine.c (move_deaths): If we do not know where a register died,
2867 2017-12-08 Richard Biener <rguenther@suse.de>
2869 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
2870 Provide -fopt-info-loop feedback when we interchange in a nest.
2872 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2874 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
2875 for armv6 ARM CPU IDs.
2877 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2879 * common/config/arm/arm-common.c: Include <algorithm>.
2880 (INCLUDE_VECTOR): Define.
2881 (compare_opt_names): New function.
2882 (arm_rewrite_selected_arch): Only strip out extensions that can be
2883 expressed through -mfpu. Sort the remaining extensions
2886 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2888 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
2889 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
2890 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
2891 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
2893 2017-06-08 Tristan Gingold <gindold@adacore.com>
2896 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
2897 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
2899 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2901 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
2902 of struct attribute_spec.
2904 2017-12-08 Julia Koval <julia.koval@intel.com>
2906 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
2907 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
2908 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
2909 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
2910 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
2911 _mm_maskz_dpwssds_epi32): Ditto.
2913 2017-12-08 Richard Biener <rguenther@suse.de>
2915 PR tree-optimization/81303
2916 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
2917 conditions try to create a comparison vector type matching
2918 the data vector type.
2919 (vectorizable_condition): Adjust.
2920 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
2921 Leave invariant conditions alone in case we can vectorize those.
2923 2017-12-08 Julia Koval <julia.koval@intel.com>
2925 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
2926 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
2927 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
2928 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
2929 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
2931 2017-12-08 Julia Koval <julia.koval@intel.com>
2933 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
2934 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
2935 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
2936 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
2937 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
2938 _mm_maskz_dpbusds_epi32): New intrinsics.
2940 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
2942 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
2944 (ld<bh>io_signed): Likewise.
2945 (st<bhw>io): Likewise.
2946 * config/nios2/predicates.md (ldstio_memory_operand): Allow
2947 SMALL_INT12 constant integer operand.
2949 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
2951 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
2953 (nios2_large_constant_allowed): ...this. Adjust uses.
2954 (nios2_plus_symbolic_constant_p): Rename to...
2955 (nios2_plus_large_constant_p): ...this. Adjust uses.
2956 (nios2_legitimate_address_p): Correct CONST_INT handling.
2957 (nios2_symbolic_memory_operand_p): Rename to...
2958 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
2959 (nios2_large_constant_p): Check for large constant integers too.
2960 (nios2_split_large_constant): Handle constant integers.
2961 (nios2_split_symbolic_memory_operand): Rename to...
2962 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
2963 (nios2_legitimize_constant_address): Handle constant integers.
2964 (r0rel_constant_p): Handle small constant integers.
2965 (nios2_print_operand_address): Handle r0-relative integer addresses.
2966 * config/nios2/nios2-protos.h: Adjust for renamed functions.
2967 * config/nios2/nios2.md: Adjust for renamed functions.
2969 2017-12-07 Andrew Waterman <andrew@sifive.com>
2971 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
2972 (TARGET_HAVE_SRODATA_SECTION): New define.
2973 (riscv_select_section): New function.
2975 2017-12-08 Joseph Myers <joseph@codesourcery.com>
2976 Alexander Monakov <amonakov@ispras.ru>
2977 Jakub Jelinek <jakub@redhat.com>
2980 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
2982 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2984 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
2985 a bool success value. Don't call output_operand_lossage here.
2986 (aarch64_print_ldpstp_address): Return a bool success value.
2987 (aarch64_print_operand_address): Call output_addr_const if
2988 aarch64_print_address_internal fails.
2989 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
2990 'y'; call output_operand_lossage instead. Call output_operand_lossage
2991 if aarch64_print_ldpstp_address fails.
2993 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2995 * tree-vector-builder.h
2996 (tree_vector_builder::binary_encoded_nelts): Declare.
2997 * tree-vector-builder.c
2998 (tree_vector_builder::binary_encoded_nelts): New function.
2999 * fold-const.c (negate_expr_p): Likewise.
3000 (operand_equal_p, fold_checksum_tree): Likewise.
3001 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
3002 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
3003 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
3004 (uniform_vector_p): Likewise.
3005 * varasm.c (const_hash_1, compare_constant): Likewise.
3006 * tree-ssa-ccp.c: Include tree-vector-builder.h.
3007 (valid_lattice_transition): Operate directly on the VECTOR_CST
3009 * ipa-icf.c: Include tree-vector-builder.h.
3010 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
3011 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
3013 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3015 * tree.c (build_vector): Delete.
3016 * tree.h (build_vector): Make static and move into the self-testing
3019 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3021 * vector-builder.h (vector_builder::derived): New const overload.
3022 (vector_builder::elt): New function.
3023 * tree-vector-builder.h (tree_vector_builder::type): New function.
3024 (tree_vector_builder::apply_step): Declare.
3025 * tree-vector-builder.c (tree_vector_builder::apply_step): New
3027 * gimple-fold.h (tree_vector_builder): Declare.
3028 (gimple_build_vector): Take a tree_vector_builder instead of a
3029 type and vector of elements.
3030 * gimple-fold.c (gimple_build_vector): Likewise.
3031 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
3033 (get_initial_defs_for_reduction): Likewise.
3034 (vectorizable_induction): Likewise.
3036 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3038 * tree-vector-builder.h
3039 (tree_vector_builder::new_binary_operation): Declare.
3040 * tree-vector-builder.c
3041 (tree_vector_builder::new_binary_operation): New function.
3042 * fold-const.c (fold_relational_const): Use it.
3043 (const_binop): Likewise. Check that both input vectors have
3044 the same number of elements, thus excluding things like WIDEN_SUM.
3045 Check whether it is possible to operate directly on the encodings
3048 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3050 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
3051 new_unary_operation, operating only on the encoded elements.
3052 (const_unop): Likewise.
3053 (exact_inverse): Likewise.
3054 (distributes_over_addition_p): New function.
3055 (const_binop): Use tree_vector_builder and new_unary_operation
3056 for combinations of VECTOR_CST and INTEGER_CST. Operate only
3057 on the encoded elements unless the encoding is strided and the
3058 operation does not distribute over addition.
3059 (fold_convert_const): Use tree_vector_builder and
3060 new_unary_operation. Operate only on the encoded elements
3061 for truncating integer conversions, or for non-stepped encodings.
3063 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3065 * config/sparc/sparc.c: Include tree-vector-builder.h.
3066 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
3067 * expmed.c: Include tree-vector-builder.h.
3068 (make_tree): Use tree_vector_builder instead of build_vector.
3069 * fold-const.c: Include tree-vector-builder.h.
3070 (const_binop): Use tree_vector_builder instead of build_vector.
3071 (const_unop): Likewise.
3072 (native_interpret_vector): Likewise.
3073 (fold_vec_perm): Likewise.
3074 (fold_ternary_loc): Likewise.
3075 * gimple-fold.c: Include tree-vector-builder.h.
3076 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
3078 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
3079 (simplify_vector_constructor): Use tree_vector_builder instead
3081 * tree-vect-generic.c: Include tree-vector-builder.h.
3082 (add_rshift): Use tree_vector_builder instead of build_vector.
3083 (expand_vector_divmod): Likewise.
3084 (optimize_vector_constructor): Likewise.
3085 * tree-vect-loop.c: Include tree-vector-builder.h.
3086 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
3087 of build_vector. Explicitly use a stepped encoding for
3089 * tree-vect-slp.c: Include tree-vector-builder.h.
3090 (vect_get_constant_vectors): Use tree_vector_builder instead
3092 (vect_transform_slp_perm_load): Likewise.
3093 (vect_schedule_slp_instance): Likewise.
3094 * tree-vect-stmts.c: Include tree-vector-builder.h.
3095 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
3096 (vect_gen_perm_mask_any): Likewise.
3097 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
3098 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
3100 (build_vector_from_val): Likewise. Explicitly use a duplicate
3103 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
3105 * doc/generic.texi (VECTOR_CST): Describe new representation of
3107 * vector-builder.h: New file.
3108 * tree-vector-builder.h: Likewise.
3109 * tree-vector-builder.c: Likewise.
3110 * Makefile.in (OBJS): Add tree-vector-builder.o.
3111 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
3112 * tree-core.h (tree_base): Add a vector_cst field to the u union.
3113 (tree_vector): Change the number of elements to
3114 vector_cst_encoded_nelts.
3115 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
3116 (VECTOR_CST_ELTS): Delete.
3117 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
3118 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
3119 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
3120 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
3121 (VECTOR_CST_ENCODED_ELT): Likewise.
3122 (vector_cst_encoded_nelts): New function.
3123 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3124 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3125 (vector_cst_int_elt, vector_cst_elt): Declare.
3126 * tree.c: Include tree-vector-builder.h.
3127 (tree_code_size): Abort if passed VECTOR_CST.
3128 (tree_size): Update for new VECTOR_CST layout.
3129 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3130 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3131 (build_vector): Use tree_vector_builder.
3132 (vector_cst_int_elt, vector_cst_elt): New functions.
3133 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
3134 encoded elements and then create the vector in the canonical form.
3135 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
3136 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
3137 (tree_c_tests): Call test_vector_cst_patterns.
3138 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
3140 (hash_tree): Likewise.
3141 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
3142 (streamer_write_tree_header): Likewise.
3143 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
3144 (streamer_alloc_tree): Likewise. Update call to make_vector.
3145 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
3147 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3149 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
3150 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
3151 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
3153 2017-12-07 Bin Cheng <bin.cheng@arm.com>
3154 Richard Biener <rguenther@suse.de>
3156 PR tree-optimization/81303
3157 * Makefile.in (gimple-loop-interchange.o): New object file.
3158 * common.opt (floop-interchange): Reuse the option from graphite.
3159 * doc/invoke.texi (-floop-interchange): Ditto. New document for
3160 -floop-interchange and mention it for -O3.
3161 * opts.c (default_options_table): Enable -floop-interchange at -O3.
3162 * gimple-loop-interchange.cc: New file.
3163 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
3164 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
3165 * passes.def (pass_linterchange): New pass.
3166 * timevar.def (TV_LINTERCHANGE): New time var.
3167 * tree-pass.h (make_pass_linterchange): New declaration.
3168 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
3169 interchange. Record IV before/after increment in new parameters.
3170 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
3171 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
3173 (check_reduction_path): ...New function here.
3174 * tree-vectorizer.h (check_reduction_path): New declaration.
3176 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
3179 PR rtl-optimization/80818
3180 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
3181 always early clobbered.
3182 * lra-lives.c (process_bb_lives): Check input hard regs for early
3183 clobbered non-operand hard reg.
3185 2017-12-07 Jakub Jelinek <jakub@redhat.com>
3188 * tree-cfg.c (verify_gimple_assign_binary): Don't require
3189 types_compatible_p, just that TYPE_MODE is the same.
3191 2017-12-07 Martin Sebor <msebor@redhat.com>
3194 * attribs.c (empty_attribute_table): Initialize new member of
3195 struct attribute_spec.
3196 (decl_attributes): Add argument. Handle mutually exclusive
3197 combinations of attributes.
3198 (selftests::test_attribute_exclusions): New function.
3199 (selftests::attribute_c_tests): Ditto.
3200 * attribs.h (decl_attributes): Add default argument.
3201 * selftest.h (attribute_c_tests): Declare.
3202 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
3203 * tree-core.h (attribute_spec::exclusions, exclude): New type and
3205 * doc/extend.texi (Common Function Attributes): Update const and pure.
3206 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
3207 of struct attribute_spec.
3208 * config/arc/arc.c (arc_attribute_table): Same.
3209 * config/arm/arm.c (arm_attribute_table): Same.
3210 * config/avr/avr.c ( avr_attribute_table): Same.
3211 * config/bfin/bfin.c (bfin_attribute_table): Same.
3212 * config/cr16/cr16.c (cr16_attribute_table): Same.
3213 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
3214 * config/h8300/h8300.c (h8300_attribute_table): Same.
3215 * config/i386/i386.c (ix86_attribute_table): Same.
3216 * config/ia64/ia64.c (ia64_attribute_table): Same.
3217 * config/m32c/m32c.c (m32c_attribute_table): Same.
3218 * config/m32r/m32r.c (m32r_attribute_table): Same.
3219 * config/m68k/m68k.c (m68k_attribute_table): Same.
3220 * config/mcore/mcore.c (mcore_attribute_table): Same.
3221 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
3222 * config/mips/mips.c (mips_attribute_table): Same.
3223 * config/msp430/msp430.c (msp430_attribute_table): Same.
3224 * config/nds32/nds32.c (nds32_attribute_table): Same.
3225 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
3226 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
3227 * config/rl78/rl78.c (rl78__attribute_table): Same.
3228 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
3229 * onfig/rx/rx.c (rx_attribute_table): Same.
3230 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
3231 * config/sh/sh.c (sh_attribute_table): Same.
3232 * config/sparc/sparc.c (sparc_attribute_table): Same.
3233 * config/spu/spu.c (spu_attribute_table): Same.
3234 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
3235 * config/v850/v850.c (v850_attribute_table): Same.
3236 * config/visium/visium.c (visium_attribute_table): Same.
3238 2017-12-07 Tamar Christina <tamar.christina@arm.com>
3241 * config/arm/arm.c (INCLUDE_STRING): Define.
3242 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
3243 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
3246 2017-12-07 Michael Matz <matz@suse.de>
3248 Add unroll and jam pass
3250 * gimple-loop-jam.c: New file.
3251 * Makefile.in (OBJS): Add gimple-loop-jam.o.
3252 * common.opt (funroll-and-jam): New option.
3253 * opts.c (default_options_table): Add unroll-and-jam at -O3.
3254 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
3255 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
3256 * passes.def: Add pass_loop_jam.
3257 * timevar.def (TV_LOOP_JAM): Add.
3258 * tree-pass.h (make_pass_loop_jam): Declare.
3259 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
3260 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
3261 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
3262 to flow_loop_tree_node_add.
3263 (duplicate_subloops, copy_loops_to): Append to sibling list.
3264 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
3265 * doc/invoke.texi (-funroll-and-jam): Document new option.
3266 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
3268 2017-12-07 Richard Biener <rguenther@suse.de>
3270 PR tree-optimization/83296
3271 PR tree-optimization/67769
3272 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
3273 flow sensitive info in an unrelated BB.
3274 (value_replacement): Use reset_flow_sensitive_info.
3275 (minmax_replacement): Reset flow sensitive info on the def
3276 we move. Do not reset flow sensitive info in the whole BB
3277 we move the stmt to.
3278 (abs_replacement): Likewise.
3280 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3283 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3284 rs6000_cpu to the given -mcpu=, or to the default processor.
3286 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3288 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
3289 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
3290 instead of rs6000_cpu_attr.
3291 (rs6000_adjust_cost): Ditto.
3292 (is_microcoded_insn): Ditto.
3293 (rs6000_adjust_priority): Ditto.
3294 (rs6000_issue_rate): Ditto.
3295 (rs6000_use_sched_lookahead): Ditto.
3296 (rs6000_use_sched_lookahead_guard): Ditto.
3297 (rs6000_sched_reorder): Ditto.
3298 (force_new_group): Ditto.
3299 * config/rs6000/rs6000.md (cpu attribute): Ditto.
3300 (group_ending_nop): Ditto.
3302 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3304 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
3305 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
3306 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
3307 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
3309 (rs6000_reassociation_width): Ditto.
3310 (rs6000_emit_epilogue): Ditto.
3311 (rs6000_adjust_cost): Ditto.
3312 (is_microcoded_insn): Ditto.
3313 (is_cracked_insn): Ditto.
3314 (rs6000_adjust_priority): Ditto.
3315 (rs6000_sched_reorder): Ditto.
3316 (rs6000_sched_reorder2): Ditto.
3317 (insn_must_be_first_in_group): Ditto.
3318 (insn_must_be_last_in_group): Ditto.
3319 (rs6000_register_move_cost): Ditto.
3320 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
3323 2017-12-07 Julia Koval <julia.koval@intel.com>
3325 * config.gcc: Add vaesintrin.h.
3326 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3328 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3329 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3331 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3332 * config/i386/immintrin.h: Include vaesintrin.h.
3333 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3334 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3335 _mm_aesdec_epi128): New intrinsics.
3337 2017-12-06 David Malcolm <dmalcolm@redhat.com>
3339 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
3340 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
3341 c-family/c-spellcheck.cc.
3342 (best_macro_match::best_macro_match): Likewise.
3344 (struct edit_distance_traits<cpp_hashnode *>): Move to
3345 c-family/c-spellcheck.h.
3346 (class best_macro_match): Likewise.
3348 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3350 PR tree-optimization/83293
3351 * gimple-ssa-strength-reduction.c (insert_initializers): Use
3352 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
3353 might insert into empty bb.
3356 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
3357 simplify for plus with :c added, and pointer_plus without that.
3358 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
3359 with undefined overflow and the conversion is not widening,
3360 perform negation in utype and only convert to type afterwards.
3361 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
3362 simplify for plus with :c added, and pointer_plus without that.
3363 If type is integral with undefined overflow and the conversion is
3364 not widening, perform minus in utype and only convert to type
3365 afterwards. Move the last pointer_diff_expr simplify into the
3368 2017-12-06 Martin Sebor <msebor@redhat.com>
3370 PR tree-optimization/82646
3371 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
3372 strncpy, not maxlen.
3374 2017-12-06 Martin Sebor <msebor@redhat.com>
3376 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
3379 PR tree-optimization/83075
3380 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
3381 strncat/strncpy don't change length of source string.
3383 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
3386 2017-11-29 Martin Aberg <maberg@gaisler.com>
3388 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3389 to prevent b2bst errata sequence.
3390 (sqrtdf2_fix): Likewise.
3392 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3394 PR tree-optimization/81945
3395 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
3396 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
3397 to dest_cfun has orig_loop_num set, either remap it to the new
3398 loop number if the loop got moved too, or clear it.
3400 2017-12-05 Steve Ellcey <sellcey@cavium.com>
3402 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
3404 (thunderx2t99_nothing): New insn reservation.
3405 (thunderx2t99_mrs): New insn reservation.
3406 (thunderx2t99_multiple): New insn reservation.
3407 (thunderx2t99_alu_basi): Add bfx to reservation.
3408 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
3410 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3413 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
3415 2017-12-05 Bin Cheng <bin.cheng@arm.com>
3417 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
3418 tree-ssa-pre.c::remove_dead_inserted_code.
3419 * tree-ssa-dce.h: New file.
3420 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
3421 (remove_dead_inserted_code): Move and rename to function
3422 tree-ssa-dce.c::simple_dce_from_worklist.
3423 (pass_pre::execute): Update use.
3425 2017-12-05 Richard Biener <rguenther@suse.de>
3427 PR tree-optimization/83277
3428 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
3429 to code-gen liveout vars.
3431 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
3433 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
3434 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
3435 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
3438 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
3440 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
3442 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
3443 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
3446 2017-12-05 Richard Biener <rguenther@suse.de>
3448 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
3449 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
3450 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
3451 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
3453 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3455 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
3457 2017-12-05 Martin Liska <mliska@suse.cz>
3458 Jakub Jelinek <jakub@redhat.com>
3460 * doc/invoke.texi: Document the options.
3461 * flag-types.h (enum sanitize_code): Add
3462 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3463 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
3464 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3465 * opts.c: Define new sanitizer options.
3466 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
3467 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
3469 2017-12-05 Julia Koval <julia.koval@intel.com>
3471 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
3472 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
3473 (ix86_handle_option): Handle -mavx512vnni.
3474 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
3475 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3476 * config/i386/i386-c (__AVX512VNNI__): New.
3477 * config/i386/i386.c (ix86_target_string): Handle new option.
3478 (ix86_valid_target_attribute_inner_p): Handle new option.
3479 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
3480 * config/i386/i386.opt (mavx512vnni): New option.
3482 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3485 * athlon.md: Disable for generic.
3486 * haswell.md: Enable for generic.
3487 * i386.c (ix86_sched_init_global): Add core hooks for generic.
3488 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
3490 (ix86_adjust_cost): Move generic to haswell path.
3492 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
3494 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
3495 instead of MEM_P in a couple more places. Fix formatting issues.
3497 2017-12-04 Jim Wilson <jimw@sifive.com>
3499 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
3500 instead of GP_REG_LAST-1.
3501 (riscv_adjust_libcall_cfi_prologue): Likewise.
3502 (riscv_adjust_libcall_cri_epilogue): Likewise.
3503 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
3506 2017-12-04 Luis Machado <luis.machado@linaro.org>
3508 * ipa-pure-const.c (check_decl): Add missing newline.
3509 (state_from_flags): Likewise.
3511 2017-12-04 Jeff Law <law@redhat.com>
3513 PR tree-optimizatin/78496
3514 * gimple-ssa-evrp-analyze.h
3515 (evrp_range_analyzer::get_vr_values): Simplify.
3516 * gimple-ssa-evrp-analyze.c: Corresponding changes.
3517 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
3518 and gimple-ssa-evrp-analyze.h.
3519 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
3520 (simplify_stmt_for_jump_threading): Copy a blob of code from
3521 tree-vrp.c to use ranges to simplify statements.
3522 (dom_opt_dom_walker::before_dom_children): Call
3523 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
3524 (dom_opt_dom_walker::after_dom_children): Similarly for
3525 evrp_range_analyzer::leave.
3526 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
3529 * gimple-ssa-evrp-analyze.c
3530 (evrp_range_analyzer::extract_range_from_stmt): Always use
3531 vr_values::update_value_range so preexisting range info is
3532 medged with new range info, even if the new range is VR_VARYING.
3534 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3536 * combine.c: Adjust comment.
3537 (use_crosses_set_p): Delete.
3538 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
3539 (try_combine): Ditto.
3541 2017-12-04 Richard Biener <rguenther@suse.de>
3543 PR tree-optimization/83255
3544 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
3545 Re-add zero-iteration check.
3547 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3549 PR rtl-optimization/83245
3550 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
3551 hard registers as earlyclobber, also if not in an asm.
3553 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3557 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3560 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3561 rs6000_cpu based on cpu_index, not tune_index.
3563 2017-12-04 Richard Biener <rguenther@suse.de>
3565 PR tree-optimization/83238
3566 * graphite-scop-detection.c (scop_detection::merge_sese): Make
3567 code match comment, rejecting invalid SESE regions.
3569 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
3571 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
3572 require base operand is a REG_POINTER prior to reload on targets
3573 with non-equivalent space registers.
3575 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3577 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
3578 (update_specialized_profile): Fix updating of counts.
3579 (perhaps_add_new_callers): Likewise.
3581 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3584 * x86-tune.def: Remove obsolette FIXMEs.
3585 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
3586 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
3587 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
3589 (X86_TUNE_PAD_RETURNS): Disable for generic.
3591 2017-12-02 Jakub Jelinek <jakub@redhat.com>
3593 PR tree-optimization/83170
3594 PR tree-optimization/83241
3595 * gimple-ssa-store-merging.c
3596 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
3597 gimple_vuse (ins_stmt) in case it has changed.
3598 (imm_store_chain_info::output_merged_store): Likewise.
3600 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
3604 * tree-cfg.c (pass_warn_function_return::execute): Handle
3605 __builtin_ubsan_handle_missing_return like __builtin_unreachable
3606 with BUILTINS_LOCATION.
3610 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
3611 is BLKmode for vector field with vector raw mode, use TYPE_MODE
3612 instead of DECL_MODE.
3614 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
3615 last argument to rtx pointer.
3616 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
3617 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
3618 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
3619 depending on the chosen ISAs.
3620 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
3621 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
3622 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
3624 * config/i386/sse.md (mov<mode>_internal): Likewise.
3625 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
3627 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3629 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
3631 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3633 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
3634 parameters from prototype.
3635 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
3636 parameters. Don't print a comment.
3637 (emit_fusion_gpr_load): Adjust.
3638 (emit_fusion_load_store): Adjust.
3639 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
3640 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
3641 comment on the second line.
3643 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3646 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3647 rs6000_cpu based on cpu_index, not tune_index.
3649 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3651 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
3652 which_alternative instead of which_alternative + 1.
3653 (output_asm_insn): Print an extra tab if the template is short.
3655 2017-12-01 Jim Wilson <jimw@sifive.com>
3657 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
3659 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
3660 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
3662 * doc/tm.texi: Regenerate.
3664 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
3667 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
3668 whether we can allocate pseudos before trying to fix an address.
3669 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
3670 memory address is indexed or indirect.
3671 (floatuns_<mode>si2_hw2): Likewise.
3673 2017-12-01 Jason Merrill <jason@redhat.com>
3675 * Makefile.in (TAGS): Add c-family/*.cc.
3677 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
3679 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
3680 (call_value_insn): Likewise.
3681 (sibcall_insn): Likewise.
3682 (sibcall_value_insn): Likewise.
3683 (movsi_aarch64): Likewise.
3684 (movdi_aarch64): Likewise.
3685 (add_losym_): Likewise.
3686 (ldr_got_small_): Likewise.
3687 (ldr_got_small_sidi): Likewise.
3688 (ldr_got_small_28k_): Likewise.
3689 (ldr_got_small_28k_sidi): Likewise.
3690 * config/aarch64/aarch64.c (aarch64_print_address_internal):
3691 Move output_addr_const to symbolic case. Add error check.
3693 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3695 * config/s390/predicates.md (plus16_Q_operand): New predicate.
3696 * config/s390/s390.md: Disable MVC merging peephole if it would
3697 disable operand forwarding.
3698 (new peephole2): Split MVCs if it would turn them into up to 2
3701 2017-12-01 Richard Biener <rguenther@suse.de>
3703 PR tree-optimization/83232
3704 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
3705 detection of same access. Instead of breaking the group here
3706 do not consider the duplicate. Add comment explaining real fix.
3708 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
3710 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
3712 2017-12-01 Sudakshina Das <sudi.das@arm.com>
3714 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
3716 2017-12-01 Jakub Jelinek <jakub@redhat.com>
3718 * function.h (struct function): Remove cilk_frame_decl,
3719 is_cilk_function and calls_cilk_spawn fields.
3720 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
3722 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
3723 cilk_elemental field.
3724 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
3725 * target.def: Adjust comment.
3726 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
3727 Don't test cilk_elemental.
3729 PR tree-optimization/83233
3730 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
3731 bswap_stat name for the struct.
3734 * tree.h (SWITCH_BREAK_LABEL_P): Define.
3735 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
3736 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
3737 SWITCH_BREAK_LABEL_P set on the label.
3738 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
3739 added for default case if it was missing and not all cases covered.
3740 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
3741 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
3743 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
3744 to avoid -Wimplicit-fallthrough warning.
3745 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
3746 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
3748 PR tree-optimization/83221
3749 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
3751 (init_reassoc): Formatting fix.
3754 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
3755 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
3759 * tree-cfg.c: Include asan.h.
3760 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
3762 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3764 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
3766 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3768 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
3769 vpcmpeqd instruction.
3771 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3773 * config/i386/i386.c (standard_sse_constant_opcode): Fix
3774 registers type for 128bit mode.
3776 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3778 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
3780 * gengtype-state.c (read_state_pair): Likewise.
3781 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
3782 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
3783 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
3784 * ifcvt.c (noce_try_store_flag_constants): Likewise.
3785 * tree-ssa-ccp.c (ccp_finalize): Likewise.
3786 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
3787 * builtins.c (fold_builtin_3): Likewise.
3788 * graphite-scop-detection.c
3789 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
3790 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
3793 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
3796 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
3797 parameter. Return true if ignore_decl_rtl_set_p is true and other
3798 conditions are satisfied.
3799 * asan.h (asan_protect_global): Add new parameter.
3800 * varasm.c (categorize_decl_for_section): Pass true as second parameter
3801 to asan_protect_global calls.
3803 2017-11-30 Jim Wilson <jimw@sifive.com>
3805 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
3806 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
3807 -mstrict-align, add info on default value. Delete redundant lines for
3808 -mabi. Add missing -mexplicit-relocs docs.
3810 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3812 * config/arc/arc.md (trap): New pattern.
3814 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3816 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
3817 instruction to end into a delay slot.
3818 * config/arc/arc.md (cond_delay_insn): Check if the instruction
3819 can be placed into a delay slot against reg_note.
3821 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3823 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
3824 labels number of usages.
3826 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3828 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
3830 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
3832 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3835 * internal-fn.c (expand_mul_overflow): Optimize unsigned
3836 multiplication by power of 2 constant into two shifts + comparison.
3838 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
3841 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
3843 2017-11-30 Richard Biener <rguenther@suse.de>
3845 PR tree-optimization/83202
3846 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3847 allow_peel argument and guard peeling.
3848 (canonicalize_loop_induction_variables): Likewise.
3849 (canonicalize_induction_variables): Pass false.
3850 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
3851 peeling from cunrolli.
3853 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3855 * combine.c (try_combine): Print a message to dump file whenever
3856 I0, I1, or I2 cannot be combined into I3.
3858 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3860 PR rtl-optimization/83156
3861 PR rtl-optimization/82621
3862 * combine.c (try_combine): Don't split an I2 if one of the dests is
3863 set again before I3. Allow unused dests.
3865 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3867 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
3869 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
3871 PR rtl-optimization/80818
3872 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
3873 recursively. Use insn code for clobber.
3874 (lra_set_insn_recog_data): Pass the new arg to
3875 collect_non_operand_hard_regs.
3876 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
3878 (lra_update_insn_regno_info): Pass insn to
3879 add_regs_to_insn_regno_info.
3881 2017-11-29 Jim Wilson <jimw@sifive.com>
3882 Andrew Waterman <andrew@sifive.com>
3884 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
3885 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
3886 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
3887 use SINGLE_SHIFT_COST.
3888 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
3889 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
3891 2017-11-29 Julia Koval <julia.koval@intel.com>
3893 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
3894 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
3895 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
3896 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
3897 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
3898 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
3899 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
3900 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
3901 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
3902 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
3903 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
3904 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
3905 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
3906 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
3907 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
3908 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
3909 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
3910 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
3911 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
3912 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
3913 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
3914 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
3915 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
3916 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
3917 __builtin_ia32_vpshldv_v2di_mask,
3918 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
3919 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
3920 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
3922 2017-11-29 Julia Koval <julia.koval@intel.com>
3924 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
3925 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
3926 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
3927 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
3928 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
3929 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
3930 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
3931 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
3932 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
3933 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
3934 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
3935 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
3936 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
3937 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
3938 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
3939 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
3940 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
3941 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
3942 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3943 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
3944 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
3946 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3948 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
3949 movsi_pic_gotdata_op instruction as a load for the UT699 errata
3952 2017-11-29 Martin Aberg <maberg@gaisler.com>
3954 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3955 to prevent b2bst errata sequence.
3956 (sqrtdf2_fix): Likewise.
3958 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3960 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
3961 (div_sqrt_insn_p): New function.
3962 (sparc_do_work_around_errata): Insert NOP instructions to
3963 prevent sequences that could trigger the TN-0013 errata for
3964 certain LEON3 processors.
3965 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
3966 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
3967 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
3968 (in_branch_delay): Prevent div and sqrt in delay slot if
3970 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
3972 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3974 * config/sparc/sparc.c (atomic_insn_p): New function.
3975 (sparc_do_work_around_errata): Insert NOP instructions to
3976 prevent sequences that could trigger the TN-0010 errata for
3978 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
3979 instruction referable in atomic_insns_p.
3981 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3983 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
3984 (atomic_compare_and_swap_leon3_1): Likewise.
3987 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3989 * config/sparc/sparc.c (fpop_insn_p): New function.
3990 (sparc_do_work_around_errata): Insert NOP instructions to
3991 prevent sequences that could trigger the TN-0012 errata for
3993 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
3994 * config/sparc/sparc.md (fix_gr712rc): New attribute.
3995 (in_branch_annul_delay): Prevent floating-point instructions
3996 in delay slot of annulled integer branch.
3998 2017-11-29 Richard Biener <rguenther@suse.de>
4000 PR tree-optimization/83202
4001 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
4003 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
4004 nodes vectorized to the same stmts multiple times.
4005 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
4006 (vect_analyze_slp_instance): Adjust.
4007 (scalar_stmts_to_slp_tree_map_t): New typedef.
4008 (vect_schedule_slp_instance): Add a map recording the SLP node
4009 representing the vectorized stmts for a set of scalar stmts.
4010 Avoid code-generating redundancies.
4011 (vect_schedule_slp): Allocate map and pass it down.
4013 2017-11-29 Nathan Sidwell <nathan@acm.org>
4016 * tree.c (build_complex_type): Fix canonicalization. Only fill in
4019 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
4021 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
4022 cases for printing LDP/STP memory addresses.
4023 (aarch64_print_address_internal): Renamed from
4024 aarch64_print_operand_address, added parameter, add Pmode check.
4025 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
4026 (aarch64_print_operand_address): Indirect to
4027 aarch64_print_address_internal.
4028 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
4029 'y' operand output specifier.
4031 2017-11-29 Jakub Jelinek <jakub@redhat.com>
4034 * tree.c (build_simple_mem_ref_loc): Handle
4035 get_addr_base_and_unit_offset returning a MEM_REF.
4038 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
4041 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
4044 2017-11-28 David Malcolm <dmalcolm@redhat.com>
4046 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
4048 (layout::move_to_column): Likewise.
4050 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4053 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
4054 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
4056 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4057 Martin Jambor <mjambor@suse.cz>
4060 * tree.h (expr_type_first_operand_type_p): Declare
4061 * tree.c (expr_type_first_operand_type_p): New function.
4062 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
4063 (ipa_value_from_jfunc): Adjust declaration.
4064 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
4065 Use it as result type for arithmetics, unless it is NULL in which case
4066 be more conservative.
4067 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
4068 ipa_get_jf_pass_through_result.
4069 (propagate_vals_across_pass_through): Likewise.
4070 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
4071 is to propagate_vals_across_pass_through.
4072 (propagate_constants_across_call): Pass PARM_TYPE to
4073 propagate_scalar_across_jump_function.
4074 (find_more_scalar_values_for_callers_subset): Pass parameter type to
4075 ipa_value_from_jfunc.
4076 (cgraph_edge_brings_all_scalars_for_node): Likewise.
4077 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
4078 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
4079 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
4080 target_type, pass it to ipa_value_from_jfunc.
4081 (update_indirect_edges_after_inlining): Pass parameter type to
4082 try_make_edge_direct_simple_call.
4084 2017-11-28 Jeff Law <law@redhat.com>
4086 * gimple-ssa-evrp-analyze.c
4087 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
4088 refine ranges if scev_initialized_p returns true.
4089 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
4091 2017-11-28 Julia Koval <julia.koval@intel.com>
4093 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
4094 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
4095 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
4096 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
4097 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
4098 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
4099 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
4100 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
4101 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
4102 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
4103 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
4104 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
4105 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
4106 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
4107 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
4108 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
4109 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
4110 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
4111 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
4112 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
4113 __builtin_ia32_vpshrd_v2di_mask): New builtins.
4114 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
4116 2017-11-28 Julia Koval <julia.koval@intel.com>
4118 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
4119 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
4120 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
4121 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
4122 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
4123 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
4124 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
4125 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
4126 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
4127 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
4128 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
4129 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
4130 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
4131 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
4132 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
4133 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
4134 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
4135 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
4136 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
4137 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
4138 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
4139 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
4140 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
4141 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
4142 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
4143 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
4144 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
4145 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
4146 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
4147 __builtin_ia32_vpshld_v2di_mask): New builtins.
4148 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4149 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
4151 2017-11-28 Richard Biener <rguenther@suse.de>
4153 PR tree-optimization/80776
4154 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
4156 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
4158 (evrp_range_analyzer::record_ranges_from_incoming_edges):
4159 If the incoming edge is an effective fallthru because the other
4160 edge only reaches a __builtin_unreachable () then record ranges
4161 derived from the controlling condition in SSA info.
4162 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
4163 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4165 2017-11-28 Olivier Hainque <hainque@adacore.com>
4167 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
4168 on mingw build hosts.
4170 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4173 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
4175 2017-11-28 Richard Biener <rguenther@suse.de>
4178 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
4179 copies generated from memcpy use a character array as reference
4182 2017-11-28 Julia Koval <julia.koval@intel.com>
4183 Sebastian Peryt <sebastian.peryt@intel.com>
4185 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
4186 c-family/c-cilkplus.o, c-family/array-notation-common.o,
4187 cilk-common.o, cilk.h, cilk-common.c): Remove.
4189 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
4190 * builtins.c (is_builtin_name): Remove cilkplus condition.
4191 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
4192 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
4193 cilk-builtins.def, cilkplus.def): Remove.
4194 * cif-code.def (CILK_SPAWN): Remove.
4195 * cilk-builtins.def: Delete.
4196 * cilk-common.c: Ditto.
4198 * cilkplus.def: Ditto.
4199 * config/darwin.h (fcilkplus): Delete.
4200 * cppbuiltin.c: Ditto.
4201 * doc/extend.texi: Remove cilkplus doc.
4202 * doc/generic.texi: Ditto.
4203 * doc/invoke.texi: Ditto.
4204 * doc/passes.texi: Ditto.
4205 * gcc.c (fcilkplus): Remove.
4206 * gengtype.c (cilk.h): Remove.
4207 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
4209 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
4211 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
4213 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
4214 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
4215 cilkplus conditions.
4216 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
4217 inline_read_section): Ditto.
4218 * ipa-inline-analysis.c (cilk.h): Remove.
4219 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
4220 * lto-wrapper.c (merge_and_complain, append_compiler_options,
4221 append_linker_options): Remove condition for fcilkplus.
4222 * lto/lto-lang.c (cilk.h): Remove.
4223 (lto_init): Remove condition for fcilkplus.
4224 * omp-expand.c (expand_cilk_for_call): Delete.
4225 (expand_omp_taskreg, expand_omp_for_static_chunk,
4226 expand_omp_for): Remove cilkplus
4228 (expand_cilk_for): Delete.
4229 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
4230 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
4231 execute_lower_omp, diagnose_sb_0): Ditto.
4232 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
4233 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4234 * tree-nested.c: Ditto.
4235 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
4236 (dump_generic_node): Ditto.
4237 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4238 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
4240 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
4242 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4244 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
4245 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
4246 for vector mode and !TARGET_SIMD.
4248 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4250 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
4252 * tree.h (SWITCH_LABELS): Remove.
4253 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
4254 assert SWITCH_BODY is non-NULL.
4255 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
4257 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
4259 PR tree-optimization/80788
4260 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
4261 has TREE_OVERFLOW set, call drop_tree_overflow.
4263 2017-11-28 Richard Biener <rguenther@suse.de>
4265 PR tree-optimization/83158
4266 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
4268 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
4271 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
4272 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
4274 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
4276 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
4277 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
4279 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4282 * varasm.c (compare_constant): Take the mode of the constants into
4283 account when comparing floating point constants.
4285 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
4287 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
4288 from explicit instantiation of debug_helper.
4289 * vec.h (DEFINE_DEBUG_VEC): Ditto.
4291 2017-11-27 Richard Biener <rguenther@suse.de>
4293 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
4296 2017-11-27 Richard Biener <rguenther@suse.de>
4298 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
4299 (type_hash_canon): Also clear int_cst_hash_table entry for
4301 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
4303 2017-11-27 Tamar Christina <tamar.christina@arm.com>
4305 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4306 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4307 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4309 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4311 * hash-map.h (gt_cleare_cache): Avoid UB.
4313 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
4315 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
4316 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
4317 (decide_unroll_runtime_iterations): Likewise.
4318 (decide_unroll_stupid): Likewise.
4320 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4323 * config/i386/i386.md: Add a loop with incssp.
4325 2017-11-27 Martin Jambor <mjambor@suse.cz>
4327 PR tree-optimization/81248
4328 * tree-sra.c (splice_param_accesses): Remove size check.
4329 (decide_one_param_reduction): Fix size check.
4330 * gimple-pretty-print.c (dump_profile): Silence warning.
4331 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
4333 2017-11-27 Jakub Jelinek <jakub@redhat.com>
4336 * dbxout.c (lastlineno): New variable.
4337 (dbx_debug_hooks): Use dbxout_switch_text_section as
4338 switch_text_section debug hook.
4339 (dbxout_function_end): Switch to current_function_section
4340 rather than function_section. If crtl->has_bb_partition,
4341 output just one N_FUN, depending on in_cold_section_p.
4342 (dbxout_source_line): Remember last lineno in lastlineno.
4343 (dbxout_switch_text_section): New function.
4344 (dbxout_function_decl): Adjust dbxout_block caller.
4345 (dbx_block_with_cold_children): New function.
4346 (dbxout_block): Return true if any LBRAC/RBRAC have been
4347 emitted. Use dbx_block_with_cold_children at depth == 0
4348 in second partition. Add PARENT_BLOCKNUM argument, pass
4349 it optionally adjusted to children. Output LBRAC/RBRAC
4350 around recursive call only if the block is in the current
4351 partition, if not and anything was output, emit empty
4353 * final.c (final_scan_insn): Compute cold_function_name
4354 before calling switch_text_section debug hook. Call
4355 that hook even if dwarf2out_do_frame if not emitting
4359 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
4360 TREE_READONLY decls.
4362 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4364 PR rtl-optimization/82488
4365 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
4367 2017-11-26 Julia Koval <julia.koval@intel.com>
4369 * config/i386/i386.c (processor_target_table): Add skylake_cost for
4371 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
4374 2017-11-26 Julia Koval <julia.koval@intel.com>
4376 * config/i386/driver-i386.c (host_detect_local_cpu):
4377 Detect skylake-avx512.
4379 2017-11-26 Julia Koval <julia.koval@intel.com>
4381 * config.gcc: Add -march=cannonlake.
4382 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
4383 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
4384 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
4385 (PTA_CANNONLAKE): New.
4386 (processor_target_table): Add cannonlake.
4387 (ix86_option_override_internal): Ditto.
4388 (fold_builtin_cpu): Ditto.
4389 (get_builtin_code_for_version): Handle cannonlake.
4390 (M_INTEL_COREI7_CANNONLAKE): New.
4391 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
4392 * doc/invoke.texi: Add -march=cannonlake.
4394 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
4396 * plugin.c (add_new_plugin): Use platform-specific library extensions.
4397 (try_init_one_plugin): Alternative implementation for MinGW.
4398 * Makefile.in (plugin_implib): New.
4399 (gengtype-lex.c): Fix broken AIX workaround.
4400 * configure: Regenerate.
4401 * doc/plugins.texi: Document support for MinGW.
4403 2017-11-25 Jakub Jelinek <jakub@redhat.com>
4405 PR rtl-optimization/81553
4406 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
4407 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
4408 is a shift where C1 has different mode than the whole shift, use C1's
4409 mode for MULT rather than the shift's mode.
4412 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
4413 builtins not enabled in the currently selected ISA.
4415 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
4417 PR tree-optimization/71026
4418 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
4419 (insert_reciprocals): Change to insert reciprocals before a division
4420 by a square and to insert the square of a reciprocal.
4421 (execute_cse_reciprocals_1): Change to consider division by a square.
4422 (register_division_in): Add importance parameter.
4424 2017-11-24 Richard Biener <rguenther@suse.de>
4426 PR tree-optimization/82402
4427 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
4428 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
4430 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
4432 * match.pd (0-ptr): New transformation.
4434 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
4437 * ipa-inline.c (inline_small_functions): Set current badnes correctly
4438 when skipping checking.
4440 2017-11-24 Richard Biener <rguenther@suse.de>
4442 PR tree-optimization/83128
4443 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
4444 (vn_reference_lookup_3): Likewise.
4446 2017-11-24 Jakub Jelinek <jakub@redhat.com>
4449 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
4450 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
4453 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
4454 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
4455 check for the latter with gimple_call_builtin_p. Do not handle
4456 BUILT_IN_STPNCPY_CHK which is not a pass through call.
4458 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
4460 * config/arm/arm_neon.h: Fix pragma GCC push_options before
4463 2017-11-23 Julia Koval <julia.koval@intel.com>
4465 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
4466 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
4467 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
4468 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
4469 _mm512_maskz_expandloadu_epi16): New intrinsics.
4470 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
4471 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
4472 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
4473 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
4474 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
4475 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
4476 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
4477 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
4478 _mm256_maskz_expandloadu_epi8): New intrinsics.
4479 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
4480 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
4481 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
4482 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
4483 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
4484 * config/i386/sse.md (VI248_VLBW): New iterator.
4485 (expand<mode>_mask, expand<mode>_maskz): New patterns.
4487 2017-11-23 Julia Koval <julia.koval@intel.com>
4489 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
4490 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
4491 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
4492 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
4493 _mm512_mask_compressstoreu_epi16): New.
4494 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
4495 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
4496 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
4497 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
4498 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
4499 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
4500 _mm256_mask_compressstoreu_epi8): New.
4501 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
4502 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
4503 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
4504 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
4505 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
4506 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
4507 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
4508 __builtin_ia32_compresshi128_mask,
4509 __builtin_ia32_compressstoreuqi512_mask,
4510 __builtin_ia32_compressstoreuhi512_mask,
4511 __builtin_ia32_compressstoreuqi256_mask,
4512 __builtin_ia32_compressstoreuqi128_mask,
4513 __builtin_ia32_compressstoreuhi256_mask,
4514 __builtin_ia32_compressstoreuhi128_mask): New builtins.
4515 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
4517 (ix86_expand_special_args_builtin): Handle new types.
4518 (s4fma_expand): Handle new builtin array.
4519 * config/i386/immintrin.h: Include new headers.
4520 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
4521 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
4523 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4525 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
4526 fixes. Declare temp and g variables at the top in order to avoid
4527 {} in most of the cases.
4529 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
4531 * match.pd (ptr-0): New transformation.
4533 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4535 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
4536 (struct addr_mode_cost_table): New.
4537 (struct tune_params): Add field addr_mode_costs.
4538 * config/arm/arm.c (generic_addr_mode_costs): New.
4539 (arm_slowmul_tune): Initialise addr_mode_costs field.
4540 (arm_fastmul_tune): Likewise.
4541 (arm_strongarm_tune): Likewise.
4542 (arm_xscale_tune): Likewise.
4543 (arm_9e_tune): Likewise.
4544 (arm_marvell_pj4_tune): Likewise.
4545 (arm_v6t2_tune): Likewise.
4546 (arm_cortex_tune): Likewise.
4547 (arm_cortex_a8_tune): Likewise.
4548 (arm_cortex_a7_tune): Likewise.
4549 (arm_cortex_a15_tune): Likewise.
4550 (arm_cortex_a35_tune): Likewise.
4551 (arm_cortex_a53_tune): Likewise.
4552 (arm_cortex_a57_tune): Likewise.
4553 (arm_exynosm1_tune): Likewise.
4554 (arm_xgene1_tune): Likewise.
4555 (arm_cortex_a5_tune): Likewise.
4556 (arm_cortex_a9_tune): Likewise.
4557 (arm_cortex_a12_tune): Likewise.
4558 (arm_cortex_a73_tune): Likewise.
4559 (arm_v7m_tune): Likewise.
4560 (arm_cortex_m7_tune): Likewise.
4561 (arm_v6m_tune): Likewise.
4562 (arm_fa726te_tune): Likewise.
4563 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
4565 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4567 * config/arm/arm.c (arm_mem_costs): New function.
4568 (arm_rtx_costs_internal): Use arm_mem_costs.
4570 2017-11-23 Mark Wielaard <mark@klomp.org>
4572 * dwarf2out.c (init_sections_and_labels): Use generation to create
4573 unique ranges_section_label and ranges_base_label. Return generation.
4574 (output_rnglists): Add generation argument. Use generation to create
4575 unique ranges labels.
4576 (dwarf2out_finish): Get generation from init_sections_and_labels
4577 and pass generation to output_rnglists.
4579 2017-11-23 Mike Stump <mikestump@comcast.net>
4580 Eric Botcazou <ebotcazou@adacore.com>
4582 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
4583 * cfgloop.h (struct loop): Add unroll field.
4584 * function.h (struct function): Add has_unroll bitfield.
4585 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
4586 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
4587 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
4588 (pass_rtl_unroll_loops::gate): Likewise.
4589 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
4590 for which loop->unroll==1.
4591 (decide_unroll_constant_iterations): Use note for consistency and
4592 take loop->unroll into account. Return early if loop->unroll is set.
4593 Fix thinko in existing test.
4594 (decide_unroll_runtime_iterations): Use note for consistency and
4595 take loop->unroll into account.
4596 (decide_unroll_stupid): Likewise.
4597 * lto-streamer-in.c (input_cfg): Read loop->unroll.
4598 * lto-streamer-out.c (output_cfg): Write loop->unroll.
4599 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
4601 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
4602 (print_loop): Print loop->unroll if set.
4603 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
4604 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
4605 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
4607 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
4608 loop->unroll is set and smaller than the trip count. Otherwise bypass
4609 entirely the heuristics if loop->unroll is set. Remove dead note.
4610 Fix off-by-one bug in other note.
4611 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
4612 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
4614 (tree_unroll_loops_completely): Make static.
4615 (pass_complete_unroll::execute): Use correct type for variable.
4616 (pass_complete_unrolli::execute): Fix formatting.
4617 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
4619 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
4621 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
4622 enable when TARGET_PREFER_AVX128 is set.
4624 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4626 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
4628 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
4629 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
4631 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4633 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
4634 frequencies. Fix estimation of aggregate parameters.
4636 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4638 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
4641 2017-11-23 Tom de Vries <tom@codesourcery.com>
4643 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
4645 2017-11-23 Tom de Vries <tom@codesourcery.com>
4647 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
4650 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
4653 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
4654 sibcall_value_pcrel_fdpic): Use local variable instead of
4656 (calli_tbr_rel): Add missing operand 2.
4657 (call_valuei_tbr_rel): Add missing operand 3.
4659 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4662 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
4663 bitpos/bitsize covering the whole destination, use store_expr only if
4664 the complex mode is the same. Otherwise, use expand_normal and if
4665 it returns CONCAT, subreg each part separately instead of trying to
4666 subreg the whole result.
4668 2017-11-23 Richard Biener <rguenther@suse.de>
4670 PR tree-optimization/23094
4671 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
4672 come here from walking over backedges in the first iteration.
4673 (vn_reference_lookup_3): Skip clobbers that store the same value.
4675 2017-11-23 Richard Biener <rguenther@suse.de>
4677 PR tree-optimization/81403
4678 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
4679 a block we need a leader relative to.
4680 (phi_translate_1): For nary processing require a leader from
4681 get_representative_for given we run expression simplification
4682 using match-and-simplify. Remove previous fix.
4684 2017-11-22 Jeff Law <law@redhat.com>
4686 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
4687 Use new method allocate_value_range rather than accessing the
4688 vrp_value_range_pool data member directly.
4689 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
4690 to use extract_range_from_stmt method to avoid need for
4691 extract_range_from_assignment method.
4692 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
4693 method rather than setting values_propgated data member directly.
4694 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
4695 and values propagated data members and extract_range_from_assignment
4696 method. Reorder private data members to conform to standards.
4697 Add new methods set_lattice_propagation_complete and
4698 allocate_value_range.
4700 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
4702 PR rtl-optimization/83030
4703 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
4704 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
4705 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
4706 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
4707 CROSSING_JUMP_P flag is already set before setting it.
4708 * cfgrtl.c (fixup_partition_crossing): Likewise.
4709 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
4712 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4714 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
4715 Handle the case where both arguments are using gen_const_vec_series.
4717 2017-11-22 David Malcolm <dmalcolm@redhat.com>
4720 * pretty-print.c (pp_format): Move quoting implementation to
4721 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
4722 to pass address of "quote" local.
4723 (pp_begin_quote): New function.
4724 (pp_end_quote): New function.
4725 * pretty-print.h (printer_fn): Convert penultimate param from bool
4727 (pp_begin_quote): New decl.
4728 (pp_end_quote): New decl.
4729 * tree-diagnostic.c (default_tree_printer): Convert penultimate
4730 param from bool to bool *.
4731 * tree-diagnostic.h (default_tree_printer): Likewise.
4733 2017-11-22 Jeff Law <law@redhat.com>
4735 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
4736 Initialize vr_values.
4737 (evrp_range_analyzer::try_find_new_range): Call methods attached to
4738 vr_values via vr_values class instance rather than delegators.
4739 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
4740 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4741 (evrp_range_analyzer::push_value_range): Likewise.
4742 (evrp_range_analyzer::pop_value_range): Likewise.
4743 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
4744 most delegators. Those remaining are exposed as public interfaces.
4745 Make vr_values a pointer and private.
4746 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
4748 (evrp_range_analyzer::get_vr_value): New method.
4749 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
4750 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
4751 (class evrp_dom_walker): Attach evrp_folder class, initialize
4752 it in the ctor. Remove temporary delegators.
4753 (evrp_dom_walker::before_dom_children): Call methods in attached
4754 evrp_range_analyzer class via class instance pointer. Use
4755 free value_range_constant_singleton to remove need for
4756 op_with_constant_singleton_value delegator method. Do not
4757 create a vrp_prop class instance for every call! Narrow
4758 scope of a couple variables.
4759 (evrp_dom_walker::cleanup): Call methods in attached
4760 evrp_range_analyzer class via class instance pointer.
4761 * vr-values.h (class vr_values): Privatize many methods and
4764 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4766 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
4767 * cfgexpand.c (expand_debug_expr): Remove handling for them.
4768 * expr.c (expand_expr_real_2): Likewise.
4769 * fold-const.c (const_unop): Likewise.
4770 * optabs-tree.c (optab_for_tree_code): Likewise.
4771 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
4772 * tree-inline.c (estimate_operator_cost): Likewise.
4773 * tree-pretty-print.c (dump_generic_node): Likewise.
4774 (op_code_prio): Likewise.
4775 (op_symbol_code): Likewise.
4776 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
4777 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
4778 * internal-fn.c (direct_internal_fn_optab): New function.
4779 (direct_internal_fn_array, direct_internal_fn_supported_p
4780 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
4781 * fold-const-call.c (fold_const_reduction): New function.
4782 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
4784 * tree-vect-loop.c: Include internal-fn.h.
4785 (reduction_code_for_scalar_code): Rename to...
4786 (reduction_fn_for_scalar_code): ...this and return an internal
4788 (vect_model_reduction_cost): Take an internal_fn rather than
4790 (vect_create_epilog_for_reduction): Likewise. Build calls rather
4792 (vectorizable_reduction): Use internal functions rather than tree
4793 codes for the reduction operation. Update calls to the functions
4795 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
4796 Use calls to internal functions rather than REDUC tree codes.
4797 * config/aarch64/aarch64-simd.md: Update comment accordingly.
4799 2017-11-22 Olivier Hainque <hainque@adacore.com>
4801 * config/vxworks.c (vxworks_override_options): Pick default
4802 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
4803 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
4804 DWARF_GNAT_ENCODINGS_DEFAULT.
4805 * config/vxworksae.h: Likewise.
4807 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
4809 PR tree-optimization/83104
4810 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
4813 2017-11-22 Marek Polacek <polacek@redhat.com>
4814 H.J. Lu <hongjiu.lu@intel.com>
4815 Jason Merrill <jason@redhat.com>
4820 * calls.c (initialize_argument_information): Call
4821 warn_parameter_passing_abi target hook.
4822 (store_one_arg): Use 0 for empty record size. Don't push 0 size
4823 argument onto stack.
4824 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
4825 * common.opt: Update -fabi-version description.
4826 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
4827 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
4829 (ix86_is_empty_record): New function.
4830 (ix86_warn_parameter_passing_abi): New function.
4831 (TARGET_EMPTY_RECORD_P): Redefine.
4832 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
4833 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
4834 * doc/tm.texi: Regenerated.
4835 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
4836 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
4837 * dwarf2out.c (get_ultimate_context): Move to tree.c.
4838 * explow.c (hard_function_value): Call arg_int_size_in_bytes
4839 instead of int_size_in_bytes.
4840 * expr.c (copy_blkmode_to_reg): Likewise.
4841 * function.c (aggregate_value_p): Return 0 for empty types.
4842 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
4844 (locate_and_pad_parm): Call arg size_in_bytes instead
4846 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
4847 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
4848 * target.def (empty_record_p, warn_parameter_passing_abi): New target
4850 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
4851 (std_gimplify_va_arg_expr): Skip empty records. Call
4852 arg_size_in_bytes instead size_in_bytes.
4853 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
4854 * tree-core.h (tree_type_common): Add empty_flag.
4855 (tree_decl_common): Update comments.
4856 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
4858 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4859 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
4861 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4862 * tree.c (default_is_empty_type): New function.
4863 (default_is_empty_record): New function.
4864 (arg_int_size_in_bytes): New function.
4865 (arg_size_in_bytes): New function.
4866 (get_ultimate_context): New function.
4867 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
4868 TRANSLATION_UNIT_WARN_EMPTY_P.
4869 (default_is_empty_record, arg_int_size_in_bytes,
4870 arg_size_in_bytes, get_ultimate_context): Declare.
4872 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4874 * config/arm/arm.c (cmse_clear_registers): New function.
4875 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
4876 code by call to cmse_clear_registers.
4877 (cmse_nonsecure_entry_clear_before_return): Likewise.
4879 2017-11-22 Tamar Christina <tamar.christina@arm.com>
4881 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
4882 (vdot_s32, vdotq_s32): New.
4883 (vdot_lane_u32, vdotq_lane_u32): New.
4884 (vdot_lane_s32, vdotq_lane_s32): New.
4887 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4890 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
4891 for unsigned values with fewer HWIs than the precision.
4892 (test_overflow): New function.
4893 (wide_int_cc_tests): Call it.
4895 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4897 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
4898 has the same mode class as Pmode.
4900 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4902 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
4903 padding_bits_to_clear_ptr.
4904 (cmse_nonsecure_entry_clear_before_return): Likewise.
4906 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4908 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
4909 auto_sbitap instead of integer bitfield to control register needing
4912 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4914 PR tree-optimization/83044
4915 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
4916 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
4917 that need the upper bound. Subtract offset from
4918 get_addr_base_and_unit_offset only if positive and subtract it
4919 before division by eltsize rather than after it.
4922 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
4923 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
4924 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
4928 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
4929 expand_binop with *mul_widen_optab, make sure at least one of the
4930 operands doesn't have VOIDmode.
4933 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
4935 PR rtl-optimization/82044
4936 PR tree-optimization/82042
4937 * dse.c (record_store): Check for overflow.
4938 (check_mem_read_rtx): Properly check for overflow if width == -1, call
4939 add_wild_read instead of clear_rhs_from_active_local_stores on
4940 overflow and log it into dump_file.
4942 2017-11-22 Richard Biener <rguenther@suse.de>
4944 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
4945 fake edges to exit when looking for a place to insert.
4946 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
4948 (insert_into_preds_of_block): Commit edge insertion immediately,
4949 assert that doesn't require new BBs.
4950 (fini_pre): Release expressions.
4951 (pass_pre::execute): Shuffle things around a bit, if the fn
4952 is too large do not compute AVAIL either as this is really the
4955 2017-11-22 Richard Biener <rguenther@suse.de>
4957 PR tree-optimization/83089
4958 * tree-if-conv.c (pass_if_conversion::execute): If anything
4959 changed reset SCEV and free the number of iteration estimates.
4961 2017-11-21 Martin Sebor <msebor@redhat.com>
4963 PR tree-optimization/82945
4964 * calls.h (warn_nonstring_bound): Remove unused function.
4966 2017-11-21 Martin Sebor <msebor@redhat.com>
4968 PR tree-optimization/82945
4969 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
4970 * calls.h (maybe_warn_nonstring_arg): Declare new function.
4971 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
4973 (initialize_argument_information): Call maybe_warn_nonstring_arg.
4974 * calls.h (get_attr_nonstring_decl): Declare new function.
4975 * doc/extend.texi (attribute nonstring): Update.
4976 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
4977 get_attr_nonstring_decl and handle it.
4978 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
4979 detection of nul-termination.
4980 (strlen_to_stridx): Change to a pointer.
4981 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
4982 (pass_strlen::execute): Same.
4984 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
4986 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
4987 for the new option -mprefer-vector-width=[none|128|256|512].
4988 * config/i386/i386.c (ix86_target_string): remove old style options
4989 -mprefer-avx256 and make -mprefer-avx128 as alias.
4990 (ix86_option_override_internal): Apply defaults for the
4991 -mprefer-vector-width=[128|256] option.
4992 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
4993 Implement macros to work with -mprefer-vector-width=.
4994 * config/i386/i386.opt: Implemented option
4995 -mprefer-vector-width=[none|128|256|512].
4996 * doc/invoke.texi: Documentation for
4997 -mprefer-vector-width=[none|128|256|512].
4999 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
5001 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
5003 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
5005 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
5007 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
5009 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
5011 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
5012 * expr.c (expand_expr_real_2): Likewise.
5013 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
5014 fold_binary_loc): Likewise.
5015 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
5016 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
5017 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
5018 MINUS_EXPR transformations.
5019 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
5020 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
5021 * tree-inline.c (estimate_operator_cost): Likewise.
5022 * tree-pretty-print.c (dump_generic_node, op_code_prio,
5023 op_symbol_code): Likewise.
5024 * tree-vect-stmts.c (vectorizable_operation): Likewise.
5025 * vr-values.c (extract_range_from_binary_expr): Likewise.
5026 * varasm.c (initializer_constant_valid_p_1): Likewise.
5027 * tree.def: New tree code POINTER_DIFF_EXPR.
5029 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
5031 * config/i386/i386.md (*bswap<mode>2_movbe): Add
5032 integer suffix to movbe mnemonic.
5033 (*bswaphi2_movbe): Ditto.
5034 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
5036 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5039 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
5040 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
5041 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
5043 2017-11-21 Martin Liska <mliska@suse.cz>
5045 * tree-inline.c (expand_call_inline): Remove not needed
5048 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
5049 Jakub Jelinek <jakub@redhat.com>
5052 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
5053 Remove static keyword from f variable.
5055 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5057 PR tree-optimization/83086
5058 * gimple-ssa-store-merging.c
5059 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
5060 rather than n.base_addr.
5062 2017-11-21 Martin Liska <mliska@suse.cz>
5064 PR rtl-optimization/82044
5065 PR tree-optimization/82042
5066 * dse.c (check_mem_read_rtx): Check for overflow.
5068 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5070 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
5073 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5075 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
5076 and its corresponding call arg location note.
5078 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5079 Andrew Burgess <andrew.burgess@embecosm.com>
5081 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
5083 (arc_return_slot_offset): Likewise.
5084 (arc_eh_return_address_location): New declaration.
5085 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
5086 (MUST_SAVE_REGISTER): Add exception handler case.
5087 (MUST_SAVE_RETURN_ADDR): Likewise.
5088 (arc_frame_pointer_required): Likewise.
5089 (arc_frame_pointer_needed): New function.
5090 (arc_compute_frame_size): Changed.
5091 (arc_expand_prologue): Likewise.
5092 (arc_expand_epilogue): Likewise.
5093 (arc_initial_elimination_offset): Likewise.
5094 (arc_return_slot_offset): Delete.
5095 (arc_eh_return_address_location): New function.
5096 (arc_builtin_setjmp_frame_value): Likewise.
5097 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
5098 (EH_RETURN_STACKADJ_RTX): Define.
5099 (EH_RETURN_HANDLER_RTX): Likewise.
5100 * config/arc/arc.md (eh_return): Delete.
5102 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5104 * print-tree.h (debug_vec_tree): Remove prototype.
5105 * gdbinit.in (pvt): Remove macro.
5107 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5109 PR tree-optimization/83047
5110 * gimple-ssa-store-merging.c
5111 (imm_store_chain_info::output_merged_store): If the loads with the
5112 same vuse are in different basic blocks, for load_gsi pick a load
5113 location that is dominated by the other loads.
5116 * config/i386/i386.c (ix86_memmodel_check): Start
5117 -Winvalid-memory-model diagnostics with lowercase letter.
5120 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
5121 set in_cold_section_p to the partition containing loc_list->first.
5122 When seeing loc_list->last_before_switch node, update secname and
5123 perform range_across_switch second partition handling only after that.
5126 * run-rtl-passes.c: Include debug.h.
5127 (run_rtl_passes): Call debug_hooks->assembly_start.
5128 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
5132 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
5133 OPTAB_DIRECT in calls to expand_simple_binop.
5135 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5138 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
5140 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5143 * spellcheck.h (best_match::blithely_get_best_candidate): New
5146 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5148 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
5149 with lower case letter.
5151 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
5153 * config/i386/i386.md (bswaphi2): New expander.
5154 (*bswaphi2_movbe): New insn pattern.
5155 (bswaphi -> rorhi pepehole2): New peephole pattern.
5157 2017-11-20 Jeff Law <law@redhat.com>
5159 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
5160 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5161 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
5162 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
5163 gimple-ssa-evrp-analyze.h.
5165 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
5166 set BB_VISITED here.
5167 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
5169 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5171 * tree-predcom.c: Add general comment on Store-Store chains.
5172 (split_data_refs_to_components): Postpone clearing eliminate_store_p
5174 (get_chain_last_ref_at): Rename into...
5175 (get_chain_last_write_at): ...this.
5176 (get_chain_last_write_before_load): New function.
5177 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
5178 CT_STORE_STORE when write reference is added.
5179 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
5180 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
5181 (initialize_root_vars_store_elim_1): Ditto.
5182 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
5183 definition is created.
5184 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
5185 chain by replacing it with dominant stored value.
5187 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5189 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
5191 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
5193 * vr-values.c (extract_range_from_binary_expr): Use a full range
5196 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
5198 * config/arm/arm.md (R4_REGNUM): Define constant.
5199 (nonsecure_call_internal): Remove r4 clobber.
5200 (nonsecure_call_value_internal): Likewise.
5201 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
5202 clobber and resequence match_operands.
5203 (nonsecure_call_value_reg_thumb1_v5): Likewise.
5204 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
5205 (nonsecure_call_value_reg_thumb2): Likewise.
5207 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5209 PR tree-optimization/78821
5210 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
5211 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
5212 to the address of the base rather than the base itself.
5213 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
5214 (find_bswap_or_nop_finalize): New function.
5215 (find_bswap_or_nop): Use it.
5216 (bswap_replace): Return a tree rather than bool, change first
5217 argument from gimple * to gimple_stmt_iterator, allow inserting
5218 into an empty sequence, allow ins_stmt to be NULL - then emit
5219 all stmts into gsi. Fix up MEM_REF address gimplification.
5220 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
5221 (struct store_immediate_info): Add N and INS_STMT non-static
5223 (store_immediate_info::store_immediate_info): Initialize them
5224 from newly added ctor args.
5225 (merged_store_group::apply_stores): Formatting fixes. Sort by
5227 (stmts_may_clobber_ref_p): For stores call also
5228 refs_anti_dependent_p.
5229 (gather_bswap_load_refs): New function.
5230 (imm_store_chain_info::try_coalesce_bswap): New method.
5231 (imm_store_chain_info::coalesce_immediate_stores): Use it.
5232 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
5233 (imm_store_chain_info::output_merged_store): Fail if number of
5234 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
5235 and NOP_EXPR rhs_code.
5236 (pass_store_merging::process_store): Compute n and ins_stmt, if
5237 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
5238 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
5240 (pass_store_merging::execute): Calculate dominators.
5242 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
5243 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
5244 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
5245 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5246 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
5247 class pass_optimize_bswap, bswap_replace,
5248 pass_optimize_bswap::execute): Moved to ...
5249 * gimple-ssa-store-merging.c: ... this file.
5250 Include optabs-tree.h.
5251 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
5252 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5253 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
5254 anonymous namespace, remove static keywords.
5255 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
5256 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
5258 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
5261 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
5263 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
5265 * vec.h (debug_helper): New function.
5266 (DEFINE_DEBUG_VEC): New macro.
5267 * hash-set.h (debug_helper): New function.
5268 (DEFINE_DEBUG_HASH_SET): New macro.
5269 * cfg.c (debug_slim (edge)): New function.
5270 Call DEFINE_DEBUG_VEC for edges.
5271 Call DEFINE_DEBUG_HASH_SET for edges.
5272 * cfghooks.c (debug_slim (basic_block)): New function.
5273 Call DEFINE_DEBUG_VEC for basic blocks.
5274 Call DEFINE_DEBUG_HASH_SET for basic blocks.
5275 * print-tree.c (debug_slim): New function to handle trees.
5276 Call DEFINE_DEBUG_VEC for trees.
5277 Call DEFINE_DEBUG_HASH_SET for trees.
5278 (debug (vec<tree, va_gc>) &): Remove.
5279 (debug (<vec<tree, va_gc>) *): Remove.
5280 * print-rtl.c (debug_slim): New function to handle const_rtx.
5281 Call DEFINE_DEBUG_VEC for rtx_def.
5282 Call DEFINE_DEBUG_VEC for rtx_insn.
5283 Call DEFINE_DEBUG_HASH_SET for rtx_def.
5284 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
5285 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
5286 (debug (vec<rtx_insn *> *ptr): Remove.
5287 (debug_insn_vector): Remove.
5288 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
5290 2017-11-20 Tom de Vries <tom@codesourcery.com>
5292 PR rtl-optimization/82020
5293 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
5294 IF_THEN_ELSE condition.
5296 2017-11-19 Jeff Law <law@redhat.com>
5298 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
5299 of degenerates resulting from ignoring an edge.
5301 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5304 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
5306 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5309 * profile-count.c (profile_count::to_sreal_scale): Fix return value
5310 for uninitialied counts.
5312 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5315 * tree-inline.c (estimate_num_insns): Set to 1 at least.
5317 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5320 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
5322 2017-11-19 Tom de Vries <tom@codesourcery.com>
5324 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
5327 2017-11-19 Tom de Vries <tom@codesourcery.com>
5329 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
5332 2017-11-19 Tom de Vries <tom@codesourcery.com>
5334 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5337 2017-11-19 Tom de Vries <tom@codesourcery.com>
5339 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
5341 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
5343 2017-11-19 Tom de Vries <tom@codesourcery.com>
5345 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
5347 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
5348 semicolon after MCORE_EXPORT_NAME call.
5350 2017-11-19 Tom de Vries <tom@codesourcery.com>
5353 * vmsdbgout.c (vmsdbgout_early_finish): New function.
5354 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
5356 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
5358 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
5359 (cgraph_edge::clone): Cleanup updating of profile.
5360 * ipa-cp.c (update_profiling_info): Likewise.
5361 * ipa-inline-transform.c (inline_transform): Likewise.
5362 * ipa-inline.c (inline_small_functions): Add missing space to dump.
5363 * ipa-split.c (execute_split_functions): Do not split when function
5365 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
5366 * profile-count.c (profile_count::dump): Add global0.
5367 (profile_count::to_cgraph_frequency): Do not ICE when entry is
5369 (profile_count::to_sreal_scale): Likewise.
5370 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
5371 (profile_count::combine_with_ipa_count): New function.
5372 * profile-count.h (profile_guessed_global0adjusted): New.
5373 (profile_count::adjusted_zero): New.
5374 (profile_count::global0adjusted): New.
5375 (profile_count::combine_with_ipa_count): New.
5376 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
5377 correct profile of return block of split functions.
5378 (copy_cfg_body): Remove unused profile_count.
5379 (copy_body): Likewise.
5380 (expand_call_inline): Update.
5381 (tree_function_versioning): Update.
5383 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
5385 * hash-set.h (hash_set::empty): New.
5386 * tree-ssa-threadbackward.h: Delete.
5387 * tree-ssa-threadbackward.c (class thread_jumps): New.
5388 Move max_threaded_paths into class.
5389 (fsm_find_thread_path): Remove arguments that are now in class.
5390 (profitable_jump_thread_path): Rename to...
5391 (thread_jumps::profitable_jump_thread_path): ...this.
5392 (convert_and_register_jump_thread_path): Rename to...
5393 (thread_jumps::convert_and_register_current_path): ...this.
5394 (check_subpath_and_update_thread_path): Rename to...
5395 (thread_jumps::check_subpath_and_update_thread_path): ...this.
5396 (register_jump_thread_path_if_profitable): Rename to...
5397 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
5398 (handle_phi): Rename to...
5399 (thread_jumps::handle_phi): ...this.
5400 (handle_assignment): Rename to...
5401 (thread_jumps::handle_assignment): ...this.
5402 (fsm_find_control_statement_thread_paths): Rename to...
5403 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
5404 (find_jump_threads_backwards): Rename to...
5405 (thread_jumps::find_jump_threads_backwards): ...this.
5406 Initialize path local data.
5407 (pass_thread_jumps::execute): Call find_jump_threads_backwards
5408 from within thread_jumps class.
5409 (pass_early_thread_jumps::execute): Same.
5411 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5413 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
5415 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5417 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
5418 to canonical location.
5420 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5423 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
5425 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
5427 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
5429 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
5430 rcrt1.o%s/grcrt1.o%s for -static-pie.
5432 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5434 * i386.c (ix86_multiplication_cost, ix86_division_cost,
5435 ix86_shift_rotate_cost): Break out from ...
5436 (ix86_rtx_costs): ... here.
5437 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
5440 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5442 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
5443 when entry block was promoted unlikely.
5444 (estimate_bb_frequencies): Increase frequency scale.
5445 * profile-count.h (profile_count): Export precision info.
5447 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5449 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
5450 disturbin profile of entry block.
5452 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5454 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
5457 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5459 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
5460 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
5462 2017-11-17 Jeff Law <law@redhat.com>
5464 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
5465 from evrp_dom_walker class. Various methods moved into new class.
5466 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
5467 (evrp_range_analyzer::enter): New method.
5468 (evrp_range_analyzer::leave): New method.
5469 (evrp_dom_walker): Remove delegators no longer needed by this class.
5470 Replace vr_values data member with evrp_range_analyzer
5472 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
5473 method extracted from evrp_dom_walker::before_dom_children.
5474 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
5475 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
5477 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
5478 Add private copy constructor and move assignment operators.
5479 Privatize methods and class data where trivially possible.
5480 (evrp_dom_walker::cleanup): New function, extracted from
5481 execute_early_vrp. Simplify access to class data.
5483 * vr-values.h (get_output_for_vrp): Prototype.
5484 * vr-values.c (get_output_for_vrp): New function extracted from
5485 vrp_visit_assignment_or_call and extract_range_from_stmt.
5486 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
5488 2017-11-17 Luis Machado <luis.machado@linaro.org>
5490 * config/aarch64/aarch64.c
5491 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
5492 (qdf24xx_tunings) <autoprefetcher_model>: Set to
5493 tune_params::AUTOPREFETCHER_WEAK.
5495 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5498 * config/arm/arm.c (arm_valid_target_attribute_rec):
5499 Parse "arch=" and "+<ext>".
5500 (arm_valid_target_attribute_tree): Re-init global options.
5501 (arm_option_override): Make non-static.
5502 (arm_options_perform_arch_sanity_checks): Make errors fatal.
5503 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
5504 (__ARM_FEATURE_CRC32): Support undef.
5505 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
5506 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
5508 2017-11-17 David Malcolm <dmalcolm@redhat.com>
5510 * gdbinit.in (break-on-diagnostic): New command.
5512 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5514 * config/i386/i386.c (ix86_expand_epilogue): Change simple
5515 return to indirect jump for EH return if control-flow
5516 protection is enabled. Change explicit 'false' argument in
5517 pro_epilogue_adjust_stack with a value of
5519 * config/i386/i386.md (simple_return_indirect_internal):
5520 Remove SImode restriction to support 64-bit.
5522 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5524 * combine.c (added_notes_insn): New.
5525 (try_combine): Handle added_notes_insn like added_links_insn.
5526 Rewrite return value code.
5527 (distribute_notes): Set added_notes_insn to the earliest insn we added
5530 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5532 PR rtl-optimization/82621
5533 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
5534 dest of one of those SETs is unused.
5536 2017-11-17 Richard Biener <rguenther@suse.de>
5539 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
5540 * tree-pretty-print.c (dump_generic_node): Handle
5541 annot_expr_parallel_kind.
5542 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
5543 * gimplify.c (gimple_boolify): Likewise.
5545 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5547 * config.gcc (extra_headers): Add cet.h for x86 targets.
5548 * config/i386/cet.h: New file.
5549 * doc/install.texi: Add --enable-cet/--disable-cet.
5551 2017-11-17 Richard Biener <rguenther@suse.de>
5553 PR tree-optimization/83017
5554 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
5555 (gen_parallel_loop): Properly count iterations.
5556 (parallelize_loops): Handle loop->can_be_parallel independent
5557 of flag_loop_parallelize_all. Make static profitability test match
5559 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
5560 * invoke.texi (parloops-min-per-thread): Document.
5562 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
5564 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
5565 upstreaming review comments.
5567 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5569 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
5570 with fast unaligned access.
5571 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
5573 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
5575 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
5576 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
5578 2017-11-17 Richard Biener <rguenther@suse.de>
5580 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
5581 folding of references.
5583 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
5586 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
5587 of replacing call to strncmp with corresponding call to strcmp when
5590 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
5592 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
5593 option prefer-avx256 for skylake-avx512 configuration.
5594 * config/i386/i386.c (ix86_option_override_internal): Ditto.
5595 (get_builtin_code_for_version): Ditto.
5597 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5598 Monk Chiang <sh.chiang04@gmail.com>
5600 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
5601 (FIXED_REGISTERS): Reserve more register numbers.
5602 (CALL_USED_REGISTERS): Likewise.
5603 (REG_ALLOC_ORDER): Likewise.
5604 (REG_CLASS_CONTENTS): Likewise.
5605 (REGISTER_NAMES): Likewise.
5607 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5608 Kito Cheng <kito.cheng@gmail.com>
5610 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
5612 * config/nds32/iterators.md: Add vector mode iterators and attributes.
5614 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
5616 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
5617 parameter list for vec_splats.
5619 2017-11-16 Joseph Myers <joseph@codesourcery.com>
5621 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
5623 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
5625 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
5628 * doc/install.texi: Document incompatibility of Perl >=5.6.26
5629 with the required version of automake 1.11.6.
5631 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
5633 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
5635 (DU_C2_3_power9): Correct reservation combinations.
5636 (FP_DIV_power9, VEC_DIV_power9): New.
5637 (power9-alu): Split out rotate/shift...
5638 (power9-rot): ...to here, correct dispatch resource.
5639 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
5641 (power9-fp): Correct latency.
5642 (power9-sdiv): Add div/sqrt resource.
5643 (power9-ddiv): Correct latency, add div/sqrt resource.
5644 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
5645 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
5647 (power9-qpdiv, power9-qpmul): Adjust resource usage.
5649 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
5651 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
5652 switch statement mapping KF built-ins to TF built-ins if we don't
5653 have the proper ISA 3.0 assembler support.
5655 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5657 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
5658 (lower_emutls_function_body): Do not compute it.
5660 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5662 * ipa-split.c (split_bb_info): Turn time to sreal.
5663 (split_point): Likewise.
5664 (dump_split_point): Likewise.
5665 (fine_split_points): Likewise.
5666 (execute_split_functions): Only zero split_bbs; turn time to sreals.
5668 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5670 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
5673 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5675 * predict.c (combine_predictions_for_bb): Preserve zero predicted
5677 (expensive_function_p): Remove useless assert.
5678 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
5680 2017-11-16 Martin Sebor <msebor@redhat.com>
5682 PR tree-optimization/82588
5683 PR tree-optimization/82583
5684 * tree-vrp.c (check_array_ref): Handle flexible array members,
5685 string literals, and inner indices.
5686 (search_for_addr_array): Add detail to diagnostics.
5688 2017-11-16 Nathan Sidwell <nathan@acm.org>
5692 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
5693 (SET_DECL_ASSEMBLER_NAME): Forward to
5694 overwrite_decl_assembler_name.
5695 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
5696 (overwrite_decl_assembler_name): Declare.
5697 * tree.c (overwrite_decl_assembler_name): New.
5698 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
5699 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
5700 (LANG_HOOKS_INITIALIZER): Add it.
5701 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
5702 * langhooks.c (lhd_set_decl_assembler_name): Use
5703 SET_DECL_ASSEMBLER_NAME.
5704 (lhd_overwrite_decl_assembler_name): Default implementation.
5706 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5707 Jackson Woodruff <jackson.woodruff@arm.com>
5709 PR tree-optimization/71026
5710 * match.pd: Canonicalize constant multiplies in division.
5712 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5714 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
5715 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
5716 Remove OPT_fomit_frame_pointer entry.
5717 * common/config/arc/arc-common.c: Likewise.
5718 * common/config/arm/arm-common.c: Likewise.
5719 * common/config/avr/avr-common.c: Likewise.
5720 * common/config/c6x/c6x-common.c: Likewise.
5721 * common/config/cr16/cr16-common.c: Likewise.
5722 * common/config/cris/cris-common.c: Likewise.
5723 * common/config/epiphany/epiphany-common.c: Likewise.
5724 * common/config/fr30/fr30-common.c: Likewise.
5725 * common/config/frv/frv-common.c: Likewise.
5726 * common/config/ia64/ia64-common.c: Likewise.
5727 * common/config/iq2000/iq2000-common.c: Likewise.
5728 * common/config/lm32/lm32-common.c: Likewise.
5729 * common/config/m32r/m32r-common.c: Likewise.
5730 * common/config/mcore/mcore-common.c: Likewise.
5731 * common/config/microblaze/microblaze-common.c: Likewise.
5732 * common/config/mips/mips-common.c: Likewise.
5733 * common/config/mmix/mmix-common.c: Likewise.
5734 * common/config/mn10300/mn10300-common.c: Likewise.
5735 * common/config/nios2/nios2-common.c: Likewise.
5736 * common/config/pa/pa-common.c: Likewise.
5737 * common/config/pdp11/pdp11-common.c: Likewise.
5738 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
5739 * common/config/riscv/riscv-common.c: Likewise.
5740 * common/config/rs6000/rs6000-common.c: Likewise.
5741 * common/config/rx/rx-common.c: Likewise.
5742 * common/config/s390/s390-common.c: Likewise.
5743 * common/config/sh/sh-common.c: Likewise.
5744 * common/config/sparc/sparc-common.c: Likewise.
5745 * common/config/tilegx/tilegx-common.c: Likewise.
5746 * common/config/tilepro/tilepro-common.c: Likewise.
5747 * common/config/v850/v850-common.c: Likewise.
5748 * common/config/visium/visium-common.c: Likewise.
5749 * common/config/xstormy16/xstormy16-common.c: Likewise.
5750 * common/config/xtensa/xtensa-common.c: Likewise.
5751 * invoke.texi (-fomit-frame-pointer): Update documentation.
5753 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5755 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
5757 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5759 * cfg.c (scale_bbs_frequencies_int,
5760 cale_bbs_frequencies_gcov_type): Remove.
5761 * cfg.h (scale_bbs_frequencies_int,
5762 cale_bbs_frequencies_gcov_type): Remove.
5764 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5766 * tree-ssa-loop-manip.c
5767 (scale_dominated_blocks_in_loop): Update to profile counts.
5768 (tree_transform_and_unroll_loop): Likewise.
5770 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5772 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
5773 scale_bbs_frequencies_int.
5775 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5777 * final.c (compute_alignments): Use counts rather than frequencies.
5779 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5781 * cfgloopanal.c: Include sreal.h
5782 (average_num_loop_insns): Use counts and sreal for accounting.
5784 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5786 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
5789 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5790 Kito Cheng <kito.cheng@gmail.com>
5792 * config/nds32/constraints.md: Provide more constraints.
5793 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
5794 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
5795 support constraints usage.
5797 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5799 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
5801 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5802 Kito Cheng <kito.cheng@gmail.com>
5804 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
5805 * config/nds32/nds32.opt: Refine the layout.
5806 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
5807 TARGET_EXT_STRING): Support new options.
5808 * config/nds32/nds32.h: Likewise.
5809 * config/nds32/nds32.md: Likewise.
5810 * config/nds32/nds32-predicates.c: Likewise.
5811 * config/nds32/constraints.md: Likewise.
5812 * common/config/nds32/nds32-common.c: Likewise.
5814 2017-11-16 Julia Koval <julia.koval@intel.com>
5817 * config/i386/gfniintrin.h: Add sse check.
5818 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
5820 2017-11-16 Julia Koval <julia.koval@intel.com>
5822 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
5823 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
5824 (ix86_handle_option): Handle -mavx512vbmi2.
5825 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
5826 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
5827 * config/i386/i386-c.c (__AVX512VBMI2__): New.
5828 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
5829 (ix86_valid_target_attribute_inner_p): Ditto.
5830 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
5831 * config/i386/i386.opt (mavx512vbmi2): New option.
5832 * doc/invoke.texi: Add new option.
5834 2017-11-16 Julia Koval <julia.koval@intel.com>
5836 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
5837 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
5838 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
5839 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
5840 _mm512_gf2p8mul_epi8): New intrinsics.
5841 * config/i386/i386-builtin-types.def
5842 (V64QI_FTYPE_V64QI_V64QI): New type.
5843 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
5844 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
5845 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
5846 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
5847 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
5848 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
5850 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
5852 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
5853 explicitly as a stream of bytes.
5855 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5857 * config/rs6000/altivec.h (vec_xst_be): New #define.
5858 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
5859 and externalize from *altivec_vperm_<mode>_internal.
5860 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
5862 (XL_BE_V8HI): Likewise.
5863 (XL_BE_V4SI): Likewise.
5864 (XL_BE_V4SI): Likewise.
5865 (XL_BE_V2DI): Likewise.
5866 (XL_BE_V4SF): Likewise.
5867 (XL_BE_V2DF): Likewise.
5868 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
5869 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
5870 all array entries with these keys: VSX_BUILTIN_VEC_XL,
5871 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
5872 VSX_BUILTIN_VEC_XST_BE.
5873 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
5874 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
5876 (altivec_init_builtins): Replace conditional calls to def_builtin
5877 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
5878 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
5879 with unconditional calls. Remove calls to def_builtin for
5880 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
5881 __builtin_vec_xst_be.
5882 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
5883 to define_expand, and add alternate RTL generation for P8.
5884 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
5885 vsx_ld_elemrev_v8hi.
5886 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
5887 add alternate RTL generation for P8.
5888 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
5889 vsx_ld_elemrev_v16qi.
5890 (vsx_st_elemrev_v8hi): Convert define_insn
5891 to define_expand, and add alternate RTL generation for P8.
5892 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
5893 vsx_st_elemrev_v8hi.
5894 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
5895 add alternate RTL generation for P8.
5896 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
5897 vsx_st_elemrev_v16qi.
5899 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
5902 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
5903 TARGET_AVX512ER check.
5904 (ix86_option_override_internal): Set MASK_VZEROUPPER if
5905 neither -mzeroupper nor -mno-zeroupper is used and
5906 TARGET_EMIT_VZEROUPPER is set.
5907 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
5908 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
5910 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
5912 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5913 folding of vector compares.
5914 (fold_build_vec_cmp): New helper function.
5915 (fold_compare_helper): New helper function.
5916 (builtin_function_type): Add compare builtins to the list of functions
5917 having unsigned arguments. Cosmetic updates to comment indentation.
5918 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
5919 the not+eq combination.
5921 2017-11-15 Bin Cheng <bin.cheng@arm.com>
5923 PR tree-optimization/82726
5924 PR tree-optimization/70754
5925 * tree-predcom.c (order_drefs_by_pos): New function.
5926 (combine_chains): Move code setting has_max_use_after to...
5927 (try_combine_chains): ...here. New parameter. Sort combined chains
5928 according to position information.
5929 (tree_predictive_commoning_loop): Update call to above function.
5930 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
5932 2017-11-15 Bin Cheng <bin.cheng@arm.com>
5934 PR tree-optimization/82726
5936 2017-01-23 Bin Cheng <bin.cheng@arm.com>
5938 PR tree-optimization/70754
5939 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
5940 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
5941 combined stmt before it if not NULL.
5942 (combine_chains): Process refs reversely and compute dominance point
5946 2017-02-23 Bin Cheng <bin.cheng@arm.com>
5948 PR tree-optimization/79663
5949 * tree-predcom.c (combine_chains): Process refs in reverse order
5950 only for ZERO length chains, and add explaining comment.
5952 2017-11-15 Tamar Christina <tamar.christina@arm.com>
5954 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
5955 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
5956 * doc/invoke.texi (ARM Options): Add armv8.3-a.
5958 2017-11-15 Tamar Christina <tamar.christina@arm.com>
5960 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
5962 2017-11-15 Martin Liska <mliska@suse.cz>
5964 * tree-cfg.c (pass_warn_function_return::execute):
5965 Compare warn_return_type for greater than zero.
5967 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
5971 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
5972 to return true on Xeon and not on Xeon Phi.
5973 (ix86_check_avx256_register): Changed to ...
5974 (ix86_check_avx_upper_register): ... this. Add extra check for
5975 VALID_AVX512F_REG_OR_XI_MODE.
5976 (ix86_avx_u128_mode_needed): Changed
5977 ix86_check_avx256_register to ix86_check_avx_upper_register.
5978 (ix86_check_avx256_stores): Changed to ...
5979 (ix86_check_avx_upper_stores): ... this. Changed
5980 ix86_check_avx256_register to ix86_check_avx_upper_register.
5981 (ix86_avx_u128_mode_after): Changed
5982 avx_reg256_found to avx_upper_reg_found. Changed
5983 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
5984 (ix86_avx_u128_mode_entry): Changed
5985 ix86_check_avx256_register to ix86_check_avx_upper_register.
5986 (ix86_avx_u128_mode_exit): Ditto.
5987 * config/i386/i386.h: (host_detect_local_cpu): New define.
5989 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5991 * config/arm/xgene1.md (xgene1): Split into automatons
5992 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
5993 (xgene1_f_load): Adjust reservations and/or types.
5994 (xgene1_f_store): Likewise.
5995 (xgene1_load_pair): Likewise.
5996 (xgene1_store_pair): Likewise.
5997 (xgene1_fp_load1): Likewise.
5998 (xgene1_load1): Likewise.
5999 (xgene1_store1): Likewise.
6000 (xgene1_move): Likewise.
6001 (xgene1_alu): Likewise.
6002 (xgene1_simd): Likewise.
6003 (xgene1_bfm): Likewise.
6004 (xgene1_neon_load1): Likewise.
6005 (xgene1_neon_store1): Likewise.
6006 (xgene1_neon_logic): Likewise.
6007 (xgene1_neon_st1): Likewise.
6008 (xgene1_neon_ld1r): Likewise.
6009 (xgene1_alu_cond): Added.
6010 (xgene1_shift_reg): Likwise.
6011 (xgene1_bfx): Likewise.
6012 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
6014 2017-11-15 Jakub Jelinek <jakub@redhat.com>
6017 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
6019 (can_widen_mult_without_libcall): New function.
6020 (expand_mul_overflow): If only checking unsigned mul overflow,
6021 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
6022 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
6023 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
6024 (expand_DIVMOD): Formatting fix.
6025 * expmed.h (expand_mult): Add NO_LIBCALL argument.
6026 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
6027 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
6029 PR tree-optimization/82977
6030 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
6031 constructed temporary to strlen_to_stridx.put.
6033 2017-11-15 Martin Liska <mliska@suse.cz>
6035 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
6036 * configure: Regenerate.
6038 2017-11-15 Martin Liska <mliska@suse.cz>
6041 * config/sh/sh-mem.cc: Use proper probability for
6044 2017-11-14 Jeff Law <law@redhat.com>
6046 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
6047 the red zone for stack_clash_protection_final_dynamic_probe targets
6048 when the total dynamic stack size is zero bytes.
6050 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
6051 blocks is post order.
6053 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
6055 * dumpfile.h (TDF_COMPARE_DEBUG): New.
6056 * final.c (rest_of_clean_state): Set it for the
6057 -fcompare-debug dump.
6058 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
6059 class when TDF_COMPARE_DEBUG is set.
6061 * dwarf2out.c (gen_producer_string): Discard
6064 2017-11-15 Joseph Myers <joseph@codesourcery.com>
6067 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
6068 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
6069 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
6070 (__TGMATH_REAL_2_3): Remove macros.
6071 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
6072 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
6074 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
6076 (remquo): Define using __TGMATH_REAL_3.
6078 2017-11-14 Jeff Law <law@redhat.com>
6080 * vr-values.c: New file with contents extracted from tree-vrp.c.
6081 * Makefile.in (OBJS): Add vr-values.o
6082 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
6083 (set_value_range, set_and_canonicalize_value_range): Likewise.
6084 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
6085 (value_range_constant_singleton, symbolic_range_p): Likewise.
6086 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
6087 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
6088 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
6089 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
6090 (find_case_label_range, find_case_label_index): Likewise.
6091 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
6092 (range_int_cst_singleton_p, value_inside_range): Likewise.
6093 (get_single_symbol): Likewise.
6094 (switch_update): Move structure definition here.
6095 (to_remove_edges, to_update_switch_stmts): Provide externs.
6096 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
6097 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
6098 (vrp_val_is_min, set_value_range): Likewise.
6099 (set_and_canonicalize_value_range, copy_value_range): Likewise.
6100 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
6101 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
6102 (range_is_nonnull, range_int_cst_p): Likewwise.
6103 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
6104 (get_single_symbol, operand_less_p): Likewise
6105 (compare_values_warnv, compare_values): Likewise.
6106 (value_inside_range, value_range_constant_singleton): Likewise.
6107 (zero_nonzero_bitgs_from_vr): Likewise.
6108 (extract_range_from_binary_expr_1): Likewise.
6109 (overflow_comparison_p): Likewise.
6110 (to_remove_edges, to_update_switch_stmts): Likewise.
6111 (find_case_label-index, find_case_label_range): Likewise.
6112 (switch_update, set_value_range_to_nonnegative): Remove.
6113 (set_value_range_to_truthvalue): Likewise.
6114 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
6115 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
6116 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
6117 (find_case_label_ranges, test_for_singularity): Likewise.
6118 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
6119 (x_vr_values): Move to its remaining use site.
6121 2017-11-10 Jeff Law <law@redhat.com>
6123 * vr-values.h (VR_INITIALIZER): Move #define here.
6124 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
6125 * Makefile.in (OBJS): Add tree-evrp.o
6126 * tree-vrp.h (assert_info): Move structure definition here.
6127 (set_value_range_to_varying): Prototype.
6128 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6129 (infer_value_range, register_edge_assert_for): Likewise.
6130 (stmt_interesting_for_vrp): Likewise.
6131 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
6132 (set_value_range_to_varying): No longer static.
6133 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6134 (infer_value_range, register_edge_assert_for): Likewise.
6136 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6138 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
6139 generate the XXBRD instruction.
6141 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
6142 (rs6000_builtin_type_compatible): Treat _Float128 and long double
6143 as being compatible if -mabi=ieeelongdouble.
6144 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
6145 to setup float128 built-ins with hardware support.
6146 (BU_FLOAT128_HW_2): Likewise.
6147 (BU_FLOAT128_HW_3): Likewise.
6148 (BU_FLOAT128_HW_VSX_1): Likewise.
6149 (BU_FLOAT128_HW_VSX_2): Likewise.
6150 (scalar_extract_expq): Change float128 built-in functions to
6151 accommodate having both KFmode and TFmode functions. Use the
6152 KFmode variant as the default.
6153 (scalar_extract_sigq): Likewise.
6154 (scalar_test_neg_qp): Likewise.
6155 (scalar_insert_exp_q): Likewise.
6156 (scalar_insert_exp_qp): Likewise.
6157 (scalar_test_data_class_qp): Likewise.
6158 (sqrtf128_round_to_odd): Delete processing the round to odd
6159 built-in functions as special built-in functions, and define them
6160 as float128 built-ins. Use the KFmode variant as the default.
6161 (truncf128_round_to_odd): Likewise.
6162 (addf128_round_to_odd): Likewise.
6163 (subf128_round_to_odd): Likewise.
6164 (mulf128_round_to_odd): Likewise.
6165 (divf128_round_to_odd): Likewise.
6166 (fmaf128_round_to_odd): Likewise.
6167 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6168 support for KFmode and TFmode xststdcqp calls.
6169 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
6170 point, switch the built-in handlers for the get/set float128
6171 exponent, get float128 mantissa, float128 test built-ins, and the
6172 float128 round to odd built-in functions. Eliminate creating the
6173 float128 round to odd built-in functions as special built-ins.
6174 (rs6000_init_builtins): Eliminate special creation of the float128
6175 round to odd built-in functions.
6176 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
6177 function insns to support both TFmode and KFmode variants.
6178 (xsxsigqp_<mode>): Likewise.
6179 (xsiexpqpf_<mode>): Likewise.
6180 (xsiexpqp_<mode>): Likewise.
6181 (xststdcqp_<mode>): Likewise.
6182 (xststdcnegqp_<mode>): Likewise.
6183 (xststdcqp_<mode>): Likewise.
6185 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6187 * tree-ssa-threadupdate.c (compute_path_counts): Remove
6188 unused path_in_freq_ptr parameter.
6189 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
6191 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6193 * ipa-inline.c (edge_badness): Dump sreal frequency.
6194 (compute_inlined_call_time): Match natural implementaiton ...
6195 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
6196 forgotten division by CGRAPH_FREQ_BASE.
6198 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6200 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
6201 Solaris 11. Update comment.
6202 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
6204 * config/sol2.h (STARTFILE_SPEC): Likewise.
6205 * configure: Regenerate.
6207 2017-11-14 Carl Love <cel@us.ibm.com>
6209 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
6210 le_ and be_ prefixes to swap* variables. Remove
6211 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
6214 2017-11-14 Jason Merrill <jason@redhat.com>
6216 Support GTY((cache)) on hash_map.
6217 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
6218 (ggc_cache_remove): Override it instead of ggc_mx.
6219 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
6220 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
6221 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
6222 (simple_cache_map_traits): Override maybe_mx.
6223 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
6224 (hash_map): Friend gt_cleare_cache.
6225 (gt_cleare_cache): New.
6226 * tree.h (tree_cache_traits): New hash_map traits class.
6227 (tree_cache_map): New typedef.
6229 2017-11-14 Richard Biener <rguenther@suse.de>
6231 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
6232 paramter and handling.
6233 (cleanup_control_flow_bb): Likewise.
6234 (cleanup_control_flow_pre): New helper performing a DFS walk
6235 to call cleanup_control_flow_bb in PRE order.
6236 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
6237 via cleanup_control_flow_pre.
6239 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
6241 * config/aarch64/aarch64-simd.md
6242 (aarch64_simd_bsl<mode>_internal): Remove DImode.
6243 (*aarch64_simd_bsl<mode>_alt): Likewise.
6244 (aarch64_simd_bsldi_internal): New.
6245 (aarch64_simd_bsldi_alt): Likewise.
6247 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6249 * tracer.c (better_p): Do not compare frequencies.
6250 * reg-stack.c (better_edge): Likewise.
6251 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
6254 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6256 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
6257 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
6258 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
6259 * lto-streamer-in.c (input_function): Use update_max_bb_count.
6260 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
6261 * predict.c (maybe_hot_frequency_p): Inline to ...
6262 (maybe_hot_count_p): ... here; rewrite to counts.
6263 (counts_to_freqs): Rename to ...
6264 (update_max_bb_count): ... this one.
6265 (expensive_function_p): Use counts.
6266 (estimate_bb_frequencies): Update.
6267 (rebuild_frequencies): Update.
6268 * predict.h (counts_to_freqs): Rename to ...
6269 (update_max_bb_count): ... this one.
6270 * profile.c (compute_branch_probabilities): Add debug info
6271 * tree-inline.c (expand_call_inline): Update debug info.
6272 (optimize_inline_calls): Use update_max_bb_count..
6273 (tree_function_versioning): Use update_max_bb_count..
6274 * value-prof.c (gimple_value_profile_transformations):
6275 Do not use update_max_bb_count.
6277 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6279 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
6280 always use frequencies.
6282 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6284 * bb-reorder.c: Remove frequencies from comments.
6285 (better_edge_p): Use profile counts.
6286 (find_traces): Dump profile counts.
6287 (rotate_loop): Use profile counts.
6288 (find_traces_1_round): Likewise.
6289 (connect_better_edge_p): Use counts instead of probabilities for
6291 (copy_bb_p): Drop early check for non-0 frequency.
6292 (sanitize_hot_paths): Update comments.
6294 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6296 * ipa-split.c (struct split_point): Add count.
6297 (consider_split): Do not compute incoming frequency; compute incoming
6298 count and store it to split_point.
6299 (split_function): Set count of the call to split part correctly.
6301 2017-11-13 Carl Love <cel@us.ibm.com>
6303 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
6305 2017-11-13 Tom Tromey <tom@tromey.com>
6307 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
6309 2017-11-13 Carl Love <cel@us.ibm.com>
6311 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
6312 Add support for builtins:
6313 unsigned int vec_first_{,miss}_match_{,or_eos}index,
6314 vector {un,}signed {char,int,short},
6315 vector {un,}signed {char,int,short}) arguments.
6316 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
6317 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
6318 Add BU_P9V_AV_2 expansions for the builtins.
6319 * config/rs6000/altivec.h (vec_first_match_index,
6320 vec_first_mismatch_index, vec_first_match_or_eos_index,
6321 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
6322 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
6323 new extern declaration.
6324 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
6325 * config/rs6000/vsx.md (first_match_index_<mode>,
6326 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
6327 first_mismatch_or_eos_index_<mode>): Add define expand.
6328 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
6329 * doc/extend.texi: Update the built-in documenation file for the new
6332 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
6334 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
6335 into the min/max operations for _Float<N> and _Float<N>X types.
6337 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6340 * dwarf2out.c (do_eh_frame): New static variable.
6341 (dwarf2out_begin_prologue): Set it.
6342 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
6344 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6346 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
6348 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
6350 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6353 * debug.h (dwarf2out_do_eh_frame): Declare.
6354 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
6355 (dwarf2out_do_frame): Use it.
6356 (dwarf2out_do_cfi_asm): Likewise.
6357 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
6358 (dwarf2out_assembly_start): Likewise.
6359 (dwarf2out_begin_prologue): Fix comment.
6360 * toplev.c (compile_file): Always call dwarf2out_frame_finish
6361 if the target needs either debug or unwind DWARF2 info.
6362 * lto-opts.c (lto_write_options): Do not save -fexceptions,
6363 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
6364 -ftrapping-math, -ftrapv and -fwrapv.
6366 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6368 * cgraph.c (cgraph_edge::sreal_frequency): New function.
6369 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
6370 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
6371 (estimate_edge_size_and_time): Likewise.
6372 (ipa_merge_fn_summary_after_inlining): Likewise.
6373 * ipa-inline.c (cgraph_freq_base_rec): Remove.
6374 (compute_uninlined_call_time): Use sreal_frequency.
6375 (compute_inlined_call_time): Likewise.
6376 (ipa_inline): Do not initialize cgraph_freq_base_rec.
6377 * profile-count.c: Include sreal.h.
6378 (profile_count::to_sreal_scale): New.
6379 * profile-count.h: Forward declare sreal.
6380 (profile_count::to_sreal_scale): Declare.
6382 2017-11-13 Nathan Sidwell <nathan@acm.org>
6384 * diagnostic.c (maybe_line_and_column): New.
6385 (diagnostic_get_location_text): Use it.
6386 (diagnostic_report_current_module): Likewise.
6387 (test_diagnostic_get_location_text): Add tests.
6389 2017-11-13 Luis Machado <luis.machado@linaro.org>
6391 * doc/md.texi (Specifying processor pipeline description): Fix
6392 incorrect latency for the div instruction example.
6394 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6396 PR tree-optimization/78821
6397 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
6398 that bit_not_p is the same.
6399 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
6400 (split_group): Count precisely bit_not_p bits in each statement.
6401 (invert_op): New function.
6402 (imm_store_chain_info::output_merged_store): Use invert_op to
6403 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
6404 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
6406 2017-11-13 Martin Liska <mliska@suse.cz>
6408 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
6409 (struct source_info): Likewise.
6410 (add_branch_counts): Likewise.
6411 (add_line_counts): Likewise.
6412 (function_summary): Likewise.
6413 (output_intermediate_line): Likewise.
6414 (generate_results): Likewise.
6416 2017-11-13 Martin Liska <mliska@suse.cz>
6418 * gcov.c (struct block_info): Remove typedef for block_t.
6419 (struct line_info): Likewise.
6420 (line_info::has_block): Likewise.
6421 (EXIT_BLOCK): Likewise.
6422 (unblock): Likewise.
6423 (circuit): Likewise.
6424 (get_cycles_count): Likewise.
6425 (process_file): Likewise.
6426 (read_graph_file): Likewise.
6427 (solve_flow_graph): Likewise.
6428 (find_exception_blocks): Likewise.
6429 (add_line_counts): Likewise.
6430 (accumulate_line_info): Likewise.
6431 (output_line_details): Likewise.
6433 2017-11-13 Martin Liska <mliska@suse.cz>
6435 * gcov.c (struct arc_info): Remove typedef for arc_t.
6436 (struct line_info): Likewise.
6437 (add_branch_counts): Likewise.
6438 (output_branch_count): Likewise.
6439 (function_info::~function_info): Likewise.
6440 (circuit): Likewise.
6441 (output_intermediate_line): Likewise.
6442 (read_graph_file): Likewise.
6443 (solve_flow_graph): Likewise.
6444 (find_exception_blocks): Likewise.
6445 (add_line_counts): Likewise.
6446 (accumulate_line_info): Likewise.
6447 (output_line_details): Likewise.
6448 (output_function_details): Likewise.
6450 2017-11-13 Martin Liska <mliska@suse.cz>
6452 * gcov.c (struct function_info): Remove typedef for function_t.
6453 (struct source_info): Likewise.
6454 (source_info::get_functions_at_location): Likewise.
6455 (solve_flow_graph): Likewise.
6456 (find_exception_blocks): Likewise.
6457 (add_line_counts): Likewise.
6458 (output_intermediate_file): Likewise.
6459 (process_file): Likewise.
6460 (generate_results): Likewise.
6461 (release_structures): Likewise.
6462 (read_graph_file): Likewise.
6463 (read_count_file): Likewise.
6464 (accumulate_line_counts): Likewise.
6465 (output_lines): Likewise.
6467 2017-11-13 Martin Liska <mliska@suse.cz>
6469 * gcov.c (function_info::function_info): Remove num_counts
6470 and add vector<gcov_type>.
6471 (function_info::~function_info): Use the vector.
6472 (process_file): Likewise.
6473 (read_graph_file): Likewise.
6474 (read_count_file): Likewise.
6475 (solve_flow_graph): Likewise.
6477 2017-11-13 Martin Liska <mliska@suse.cz>
6479 * gcov.c (function_info::is_artificial): New function.
6480 (process_file): Erase all artificial early.
6481 (generate_results): Skip as all artificial are already
6484 2017-11-13 Martin Liska <mliska@suse.cz>
6486 * gcov.c (read_graph_file): Store to global vector of functions.
6487 (read_count_file): Iterate the vector.
6488 (process_file): Likewise.
6489 (generate_results): Likewise.
6490 (release_structures): Likewise.
6492 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6494 PR tree-optimization/82954
6495 * gimple-ssa-store-merging.c
6496 (imm_store_chain_info::coalesce_immediate_stores): If
6497 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
6499 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
6501 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6502 Upddate call to ENDIAN_LANE_N.
6503 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
6504 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
6505 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
6506 and use aarch64_endian_lane_rtx.
6507 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
6509 2017-11-12 Tom de Vries <tom@codesourcery.com>
6511 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
6513 2017-11-12 Tom de Vries <tom@codesourcery.com>
6515 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
6517 2017-11-12 Tom de Vries <tom@codesourcery.com>
6519 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
6521 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6522 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6523 * defaults.h (ASM_OUTPUT_LABELREF): Same.
6525 2017-11-11 Martin Sebor <msebor@redhat.com>
6528 * doc/extend.texi (attribute nonstring): Remove spurious argument.
6531 * prefic.c (translate_name): Replace strncpy with memcpy to
6532 avoid -Wstringop-truncation.
6534 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6536 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
6538 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6540 * predict.c (maybe_hot_frequency_p): Do not use cfun.
6542 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6544 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
6547 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
6549 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
6550 enable generating XXBRH if the value is in a vector register.
6551 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
6552 value is in a vector register.
6553 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
6554 register to register bswap64's instead of doing the GPR sequence
6555 used on previous machines.
6556 (bswapdi2_xxbrd): New insn.
6557 (bswapdi2_reg): Disallow on ISA 3.0.
6558 (register to register bswap64 splitter): Do not split the insn on
6559 ISA 3.0 systems that use XXBRD.
6561 2017-11-10 Martin Sebor <msebor@redhat.com>
6564 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
6566 (find_subframework_file): Same.
6568 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6570 * auto-profile.c (afdo_indirect_call): Drop frequency.
6571 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
6572 (cgraph_node::create_edge): Drop frequency argument.
6573 (cgraph_node::create_indirect_edge): Drop frequency argument.
6574 (cgraph_edge::make_speculative): Drop frequency arguments.
6575 (cgraph_edge::resolve_speculation): Do not update frequencies
6576 (cgraph_edge::dump_edge_flags): Do not dump frequency.
6577 (cgraph_node::dump): Check consistency in IPA mode.
6578 (cgraph_edge::maybe_hot_p): Use IPA counter.
6579 (cgraph_edge::verify_count_and_frequency): Rename to ...
6580 (cgraph_edge::verify_count): ... this one; drop frequency checking.
6581 (cgraph_node::verify_node): Update.
6582 * cgraph.h (struct cgraph_edge): Drop frequency.
6583 (cgraph_edge::frequency): New function.
6584 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
6586 (cgraph_edge::rebuild_edges): Likewise.
6587 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
6588 (duplicate_thunk_for_node): Do not pass frequency.
6589 (cgraph_node::create_clone): Scale only counts.
6590 (cgraph_node::create_virtual_clone): Do not pass frequency.
6591 (cgraph_node::create_edge_including_clones): Do not pass frequency.
6592 (cgraph_node::create_version_clone): Do not pass frequency.
6593 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
6594 (cgraph_node::expand_thunk): Do not pass frequency.
6595 (cgraph_node::create_wrapper): Do not pass frequency.
6596 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
6598 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
6599 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
6600 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
6601 (gather_caller_stats): Use frequency function.
6602 (ipcp_cloning_candidate_p): Use frequency function.
6603 (ipcp_propagate_stage): Use frequency function.
6604 (get_info_about_necessary_edges): Use frequency function.
6605 (update_profiling_info): Update only IPA profile.
6606 (update_specialized_profile): Use frequency functoin.
6607 (perhaps_add_new_callers): Update only IPA profile.
6608 * ipa-devirt.c (ipa_devirt): Use IPA profile.
6609 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
6610 (dump_ipa_call_summary): Use frequency function.
6611 (estimate_edge_size_and_time): Use frequency function.
6612 (ipa_merge_fn_summary_after_inlining): Use frequency function.
6613 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
6614 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
6615 (update_noncloned_counts): ... ths one; scale counts only.
6616 (clone_inlined_nodes): Do not scale frequency.
6617 (inline_call): Do not pass frequency.
6618 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
6619 (compute_inlined_call_time): Use IPA profile.
6620 (want_inline_small_function_p): Use IPA profile.
6621 (want_inline_self_recursive_call_p): Use IPA profile.
6622 (edge_badness): Use IPA profile.
6623 (lookup_recursive_calls): Use IPA profile.
6624 (recursive_inlining): Do not pass frequency.
6625 (resolve_noninline_speculation): Do not update frequency.
6626 (inline_small_functions): Collect max of IPA profile.
6627 (dump_overall_stats): Dump IPA porfile.
6628 (dump_inline_stats): Dump IPA porfile.
6629 (ipa_inline): Collect IPA stats.
6630 * ipa-inline.h (clone_inlined_nodes): Update prototype.
6631 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
6632 (ipa_propagate_frequency): Use frequency function.
6633 (ipa_profile): Cleanup.
6634 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
6635 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
6636 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
6637 (input_node): Do not stream frequency.
6638 (input_edge): Do not stream frequency.
6639 (merge_profile_summaries): Scale only IPA profiles.
6640 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
6641 * predict.c (drop_profile): Do not recompute frequency.
6642 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
6643 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
6644 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
6645 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
6646 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
6647 * tree-inline.c (copy_bb): Do not scale frequency.
6648 (expand_call_inline): Do not scale frequency.
6649 (tree_function_versioning): Do not scale frequency.
6650 * ubsan.c (ubsan_create_edge): Do not pass frequency.
6652 2017-11-10 Julia Koval <julia.koval@intel.com>
6654 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
6655 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
6656 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
6657 (_mm256_maskz_gf2p8affine_epi64_epi8)
6658 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
6659 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
6660 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
6661 (__builtin_ia32_vgf2p8affineqb_v32qi)
6662 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
6663 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
6665 2017-11-10 Tamar Christina <tamar.christina@arm.com>
6669 (arm_option_override): Refactor.
6670 (arm_option_reconfigure_globals): New.
6671 (arm_options_perform_arch_sanity_checks): New.
6672 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
6674 (arm_options_perform_arch_sanity_checks): Likewise
6676 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
6678 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
6679 (power9-qpmul): New.
6680 * rs6000/rs6000.md ("type" attr): Add qmul.
6681 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
6682 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
6683 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
6685 2017-11-10 Martin Sebor <msebor@redhat.com>
6688 * builtins.c (compute_objsize): Handle arrays that
6689 compute_builtin_object_size likes to fail for. Make extern.
6690 * builtins.h (compute_objsize): Declare.
6691 (check_strncpy_sizes): New function.
6692 (expand_builtin_strncpy): Call check_strncpy_sizes.
6693 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
6694 -Wstringop-truncation.
6695 (gimple_fold_builtin_strncat): Same.
6696 * gimple.c (gimple_build_call_from_tree): Set call location.
6697 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
6698 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
6699 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
6700 (handle_builtin_strlen): Use strlen_to_stridx.
6701 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
6703 Use strlen_to_stridx.
6704 (pass_strlen::execute): Release strlen_to_stridx.
6705 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
6706 (-Wstringop-truncation): Document new option.
6708 2017-11-10 Martin Liska <mliska@suse.cz>
6710 PR gcov-profile/82702
6711 * gcov.c (main): Handle intermediate files in a different
6713 (get_gcov_intermediate_filename): New function.
6714 (output_gcov_file): Remove support of intermediate files.
6715 (generate_results): Allocate intermediate file.
6716 (release_structures): Clean-up properly fn_end.
6717 (output_intermediate_file): Start iterating with line 1.
6719 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6721 PR tree-optimization/82929
6722 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6723 ops_swapped_p non-static data member.
6724 (store_immediate_info::store_immediate_info): Clear it.
6725 (imm_store_chain_info::coalesce_immediate_stores): If swapping
6726 ops set ops_swapped_p.
6727 (count_multiple_uses): Handle ops_swapped_p.
6729 2017-11-10 Martin Liska <mliska@suse.cz>
6731 * coverage.c (coverage_init): Stream information about
6732 support of has_unexecuted_blocks.
6733 * doc/gcov.texi: Document that.
6734 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
6735 * gcov.c (read_graph_file): Likewise.
6736 (output_line_beginning): Fix a small issue with
6739 2017-11-10 Bin Cheng <bin.cheng@arm.com>
6741 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
6742 reference of trivial component.
6744 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6747 * gimple-ssa-store-merging.c
6748 (pass_store_merging::terminate_all_aliasing_chains): For
6749 gimple_store_p stmts also call refs_output_dependent_p.
6751 PR rtl-optimization/82913
6752 * compare-elim.c (try_merge_compare): Punt if def_insn is not
6755 2017-11-09 Jeff Law <law@redhat.com>
6757 * vr-values.h: New file with vr_values class.
6758 * tree-vrp.c: Include vr-values.h
6759 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
6760 data objects into the vr_values class.
6761 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
6762 (get_value_range): Make it a member function within vr_values class.
6763 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
6764 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
6765 (op_with_constant_singleton_value_range): Likewise.
6766 (extract_range_for_var_from_comparison_expr): Likewise.
6767 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
6768 (extract_range_from_binary_expr): Likewise.
6769 (extract_range_from_unary_expr): Likewise.
6770 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
6771 (check_for_binary_op_overflow, extract_range_basic): Likewise.
6772 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
6773 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
6774 (compare_name_with_value, compare_names): Likewise.
6775 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
6776 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
6777 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
6778 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
6779 (extract_range_from_phi_node): Likewise.
6780 (simplify_truth_ops_using_ranges): Likewise.
6781 (simplify_div_or_mod_using_ranges): Likewise.
6782 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
6783 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
6784 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
6785 (simplify_float_conversion_using_ranges): Likewise.
6786 (simplify_internal_call_using_ranges): Likewise.
6787 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
6788 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
6789 poitner via x_vr_values for calls into gimple folder.
6790 (vrp_initialize_lattice): Make this the vr_values ctor.
6791 (vrp_free_lattice): Make this the vr_values dtor.
6792 (set_vr_value): New function.
6793 (class vrp_prop): Add vr_values data member. Add various member
6794 functions as well as member functions that delegate to vr_values.
6795 (check_array_ref): Make a member function within vrp_prop class.
6796 (search_for_addr_array, vrp_initialize): Likewise.
6797 (vrp_finalize): Likewise. Revamp to avoid direct access to
6798 vr_value, values_propagated, etc.
6799 (check_array_bounds): Extract vrp_prop class instance pointer from
6800 walk info structure. Use it to call member functions.
6801 (check_all_array_refs): Make a member function within vrp_prop class.
6802 Smuggle class instance pointer via walk info structure.
6803 (x_vr_values): New local static.
6804 (vrp_valueize): Use x_vr_values to get class instance.
6805 (vr_valueize_1): Likewise.
6806 (class vrp_folder): Add vr_values data member. Add various member
6807 functions as well as member functions that delegate to vr_values.
6808 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
6809 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
6810 class instance from vr_values. Use it to call member functions.
6811 (vrp_dom_walker): Add vr_values data member.
6812 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
6813 instance via x_vr_values.
6814 (identify_jump_threads): Accept vr_values as argument. Store
6815 it into the walker structure.
6816 (evrp_dom_walker): Add vr_values class data member. Add various
6818 (evrp_dom_walker::try_find_new_range): Use vr_values data
6819 member to access the memory allocator.
6820 (evrp_dom_walker::before_dom_children): Store vr_values class
6821 instance into the vrp_folder class.
6822 (evrp_dom_walker::push_value_range): Rework to avoid direct
6823 access to num_vr_values and vr_value.
6824 (evrp_dom_walker::pop_value_range): Likewise.
6825 (execute_early_vrp): Remove call to vrp_initialize_lattice.
6826 Use vr_values to get to dump_all_value_ranges member function.
6827 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
6828 and simplify_cond_using_ranges_2 via vrp_prop class instance.
6829 Pass vr_values class instance down to identify_jump_threads.
6830 Remove call to vrp_free_lattice.
6831 (debug_all_value_ranges): Remove.
6833 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
6834 (vrp_folder): Likewise.
6836 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
6837 Get it from the existing bitmap instead.
6838 (vrp_intersect_ranges_1): Likewise.
6840 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6842 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6844 (store_immediate_info::store_immediate_info): Add bitnotp argument,
6845 set bit_not_p to it.
6846 (imm_store_chain_info::coalesce_immediate_stores): Break group
6847 if bit_not_p is different.
6848 (count_multiple_uses, split_group,
6849 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
6850 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
6851 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
6852 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
6853 to store_immediate_info ctor.
6855 2017-11-09 Jim Wilson <jimw@sifive.com>
6857 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
6858 (scan_prog_file): Likewise.
6860 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
6862 * bb-reorder.c (max_entry_frequency): Remove.
6863 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
6864 connect_traces, push_to_next_round_p): Remove prototypes.
6865 (find_traces_1_round): Use counts only.
6866 (push_to_next_round_p): Likewise.
6867 (find_traces): Likewise.
6868 (rotate_loop): Likewise.
6869 (find_traces_1_round): Likewise.
6870 (connect_traces): Likewise.
6871 (edge_order): Likewise.
6873 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
6875 * config/arm/arm.c (output_return_instruction): Add comments to
6876 indicate requirement for cmse_nonsecure_entry return to account
6877 for the size of clearing instruction output here.
6878 (thumb_exit): Likewise.
6879 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
6880 return in hardfloat mode.
6882 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
6884 * config/rs6000/rs6000.c (machine_function): Add a bool,
6885 "toc_is_wrapped_separately".
6886 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
6887 if it wasn't explicitly set or unset, we are optimizing for speed, and
6888 doing separate shrink-wrapping.
6889 (rs6000_get_separate_components): Enable the TOC component if
6890 saving the TOC register in the prologue.
6891 (rs6000_components_for_bb): Handle the TOC component.
6892 (rs6000_emit_prologue_components): Store the TOC register where needed.
6893 (rs6000_set_handled_components): Mark TOC as handled, if handled.
6894 (rs6000_emit_prologue): Don't save the TOC if that is already done.
6896 2017-11-09 Martin Jambor <mjambor@suse.cz>
6898 * ipa-param-manipulation.c: New file.
6899 * ipa-param-manipulation.h: Likewise.
6900 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
6901 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
6902 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
6903 (ipa_parm_adjustment): Likewise.
6904 (ipa_parm_adjustment_vec): Likewise.
6905 (ipa_get_vector_of_formal_parms): Moved declaration to
6906 ipa-param-manipulation.h.
6907 (ipa_get_vector_of_formal_parm_types): Likewise.
6908 (ipa_modify_formal_parameters): Likewise.
6909 (ipa_modify_call_arguments): Likewise.
6910 (ipa_combine_adjustments): Likewise.
6911 (ipa_dump_param_adjustments): Likewise.
6912 (ipa_modify_expr): Likewise.
6913 (ipa_get_adjustment_candidate): Likewise.
6914 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
6915 ipa-param-manipulation.c.
6916 (ipa_get_vector_of_formal_parm_types): Likewise.
6917 (ipa_modify_formal_parameters): Likewise.
6918 (ipa_modify_call_arguments): Likewise.
6919 (ipa_modify_expr): Likewise.
6920 (get_ssa_base_param): Likewise.
6921 (ipa_get_adjustment_candidate): Likewise.
6922 (index_in_adjustments_multiple_times_p): Likewise.
6923 (ipa_combine_adjustments): Likewise.
6924 (ipa_dump_param_adjustments): Likewise.
6925 * tree-sra.c: Also include ipa-param-manipulation.h
6926 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
6929 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6930 Alan Hayward <alan.hayward@arm.com>
6931 David Sherwood <david.sherwood@arm.com>
6933 * doc/sourcebuild.texi (vect_masked_store): Document.
6935 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6936 Alan Hayward <alan.hayward@arm.com>
6937 David Sherwood <david.sherwood@arm.com>
6939 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
6941 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6942 Alan Hayward <alan.hayward@arm.com>
6943 David Sherwood <david.sherwood@arm.com>
6945 * doc/sourcebuild.texi (vect_variable_length): Document.
6947 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6948 Alan Hayward <alan.hayward@arm.com>
6949 David Sherwood <david.sherwood@arm.com>
6951 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
6953 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6954 Alan Hayward <alan.hayward@arm.com>
6955 David Sherwood <david.sherwood@arm.com>
6957 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
6959 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6960 Alan Hayward <alan.hayward@arm.com>
6961 David Sherwood <david.sherwood@arm.com>
6963 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
6964 previously undocumented selectors.
6965 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
6967 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6969 * doc/rtl.texi (const_vector): Say that elements can be
6970 const_wide_ints too.
6971 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
6972 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
6973 (gen_vec_duplicate): Use it instead of CONSTANT_P.
6974 * optabs.c (expand_vector_broadcast): Likewise.
6976 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6977 Alan Hayward <alan.hayward@arm.com>
6978 David Sherwood <david.sherwood@arm.com>
6980 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
6981 scaled index even if the unscaled address was invalid.
6982 Don't increase the complexity of using a scale in that case.
6984 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6985 Alan Hayward <alan.hayward@arm.com>
6986 David Sherwood <david.sherwood@arm.com>
6988 * doc/rtl.texi: Rewrite the subreg rules so that they partition
6989 the inner register into REGMODE_NATURAL_SIZE bytes rather than
6990 UNITS_PER_WORD bytes.
6991 * emit-rtl.c (validate_subreg): Divide subregs into blocks
6992 based on REGMODE_NATURAL_SIZE of the inner mode.
6993 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
6994 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
6995 * expmed.c (lowpart_bit_field_p): Divide the value up into
6996 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
6997 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
6998 whether something is likely to occupy more than one register.
7000 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7003 * ipa-inline-transform.c (update_noncloned_frequencies): Use
7004 profile_count::adjust_for_ipa_scaling.
7005 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
7006 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
7008 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
7010 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7012 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
7013 (split_group): Add total_orig and total_new arguments, estimate the
7014 number of statements related to the store group without store merging
7015 and with store merging.
7016 (imm_store_chain_info::output_merged_store): Adjust split_group
7017 callers, punt if estimated number of statements with store merging
7018 is not smaller than estimated number of statements without it.
7020 (handled_load): Remove has_single_use checks.
7021 (pass_store_merging::process_store): Likewise.
7023 2017-11-09 Richard Biener <rguenther@suse.de>
7025 PR tree-optimization/82902
7026 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
7028 2017-11-09 Martin Liska <mliska@suse.cz>
7031 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
7034 2017-11-09 Martin Liska <mliska@suse.cz>
7036 PR tree-optimization/82669
7037 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
7039 2017-11-09 Martin Liska <mliska@suse.cz>
7041 PR gcov-profile/48463
7042 * coverage.c (coverage_begin_function): Output also end locus
7043 of a function and information whether the function is
7045 * gcov-dump.c (tag_function): Parse and print the information.
7046 * gcov.c (INCLUDE_MAP): Add include.
7047 (INCLUDE_SET): Likewise.
7048 (struct line_info): Move earlier in the source file because
7049 of vector<line_info> in function_info structure.
7050 (line_info::line_info): Likewise.
7051 (line_info::has_block): Likewise.
7052 (struct source_info): Add new member index.
7053 (source_info::get_functions_at_location): New function.
7054 (function_info::group_line_p): New function.
7055 (output_intermediate_line): New function.
7056 (output_intermediate_file): Use the mentioned function.
7057 (struct function_start): New.
7058 (struct function_start_pair_hash): Likewise.
7059 (process_file): Add code that identifies group functions.
7060 Assign lines either to global or function scope.
7061 (generate_results): Skip artificial functions.
7062 (find_source): Assign index for each source file.
7063 (read_graph_file): Read new flag artificial and end_line.
7064 (add_line_counts): Assign it either to global of function scope.
7065 (accumulate_line_counts): Isolate core of the function to
7066 accumulate_line_info and call it for both function and global
7068 (accumulate_line_info): New function.
7069 (output_line_beginning): Fix GNU coding style.
7070 (print_source_line): New function.
7071 (output_line_details): Likewise.
7072 (output_function_details): Likewise.
7073 (output_lines): Iterate both source (global) scope and function
7075 (struct function_line_start_cmp): New class.
7076 * doc/gcov.texi: Reflect changes in documentation.
7078 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7081 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
7082 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
7083 and similarly for not instead of neg.
7085 2017-11-08 Andi Kleen <ak@linux.intel.com>
7087 * config/i386/i386.opt: Add -mforce-indirect-call.
7088 * config/i386/predicates.md: Check for flag_force_indirect_call.
7089 * doc/invoke.texi: Document -mforce-indirect-call
7091 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
7093 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
7095 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
7096 config/riscv/riscv.c (predict.h): New include.
7097 (riscv_slow_unaligned_access_p): No longer static.
7098 (riscv_block_move_straight): Add require.
7099 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7101 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7104 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
7105 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
7106 nonimmediate_operand predicate for operand 1 instead of
7109 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7111 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
7113 * config/aarch64/constraints.md (Uml): New constraint.
7114 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
7117 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7119 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
7120 of two vec_duplicates into a vec_concat.
7122 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7124 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7125 Simplify vec_merge of vec_duplicate and vec_concat.
7126 * config/aarch64/constraints.md (Utq): New constraint.
7127 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
7130 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7132 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7133 Simplify vec_merge of vec_duplicate and const_vector.
7134 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
7136 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
7137 mode iterator. Update predicate on operand 1 to
7138 handle non-const_vec constants. Delete constraints.
7139 (*aarch64_combinez_be<mode>): Likewise for operand 2.
7141 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7143 PR tree-optimization/78821
7144 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
7146 (store_operand_info::store_operand_info): Initialize it to false.
7147 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
7148 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
7149 store in the group, and if chain_info is non-NULL, to ignore altogether
7151 (compatible_load_p): Fail if bit_not_p does not match.
7152 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
7153 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
7154 (pass_store_merging::process_store): Adjust
7155 terminate_all_aliasing_chains calls to pass NULL in all current spots,
7156 call terminate_all_aliasing_chains newly when adding a store into
7157 a chain with non-NULL chain_info.
7159 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7161 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
7163 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7165 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7167 (aarch64_layout_frame): Initialise emit_frame_chain.
7168 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
7169 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
7171 2017-11-08 Martin Liska <mliska@suse.cz>
7173 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
7174 of gimple_call_internal_p.
7176 2017-11-07 Tom de Vries <tom@codesourcery.com>
7178 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
7180 2017-11-07 Tom de Vries <tom@codesourcery.com>
7182 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
7185 2017-11-08 Martin Liska <mliska@suse.cz>
7188 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
7190 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7192 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
7194 2017-11-07 Jakub Jelinek <jakub@redhat.com>
7197 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
7198 (*cmp<mode>_ccz_1): New insn with $k alternative.
7201 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
7202 RTX_COMM_COMPARE as commutative as well.
7203 (ix86_binary_operator_ok): Formatting fix.
7204 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
7205 *<code><mode>3<mask_name><round_saeonly_name>,
7206 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
7207 *<s>mul<mode>3_highpart<mask_name>,
7208 *vec_widen_umult_even_v16si<mask_name>,
7209 *vec_widen_umult_even_v8si<mask_name>,
7210 *vec_widen_umult_even_v4si<mask_name>,
7211 *vec_widen_smult_even_v16si<mask_name>,
7212 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
7213 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
7214 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
7215 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
7216 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
7217 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
7218 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
7219 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
7220 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
7221 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
7222 ix86_binary_operator_ok. Formatting fixes.
7223 (*<plusminus_insn><mode>3<mask_name><round_name>,
7224 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
7227 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
7229 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
7230 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
7231 eq and ne if TARGET_ISEL.
7232 (cmp): New code_iterator.
7233 (UNS, UNSU_, UNSIK): New code_attrs.
7234 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
7235 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
7237 ("eq<mode>3"): ... this.
7238 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
7240 ("ne<mode>3"): ... this.
7242 2017-11-07 Julia Koval <julia.koval@intel.com>
7245 * common/config/i386/i386-common.c
7246 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
7247 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
7248 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
7249 * config/i386/i386.opt: Ditto.
7250 * config/i386/i386.c (ix86_target_string): Ditto.
7251 (ix86_option_override_internal): Ditto.
7252 (ix86_init_mpx_builtins): Move MPX to args2.
7253 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
7254 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7255 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7256 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7257 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7258 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7259 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
7261 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
7264 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
7265 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
7266 (zero-extendsidi peephole2): Remove peephole.
7268 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7271 * stor-layout.c: Include attribs.h.
7272 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
7273 explicit lookup of "aligned" attribute.
7275 2017-11-07 Andrew Waterman <andrew@sifive.com>
7277 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
7278 (riscv_expand_block_move): Likewise.
7279 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
7281 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
7282 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
7283 * config/riscv/riscv.c (riscv_block_move_straight): New function.
7284 (riscv_adjust_block_mem): Likewise.
7285 (riscv_block_move_loop): Likewise.
7286 (riscv_expand_block_move): Likewise.
7287 * config/riscv/riscv.md (movmemsi): New pattern.
7289 2017-11-07 Michael Clark <michaeljclark@mac.com>
7291 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
7292 (MUSL_DYNAMIC_LINKER): Likewise.
7294 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
7296 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
7297 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
7300 2017-11-07 Richard Biener <rguenther@suse.de>
7302 * match.pd: Fix build.
7304 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
7305 Jackson Woodruff <jackson.woodruff@arm.com>
7307 PR tree-optimization/71026
7308 * match.pd: Canonicalize negate in division.
7310 2017-11-07 Sudakshina Das <sudi.das@arm.com>
7313 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
7315 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7317 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
7318 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
7320 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7322 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
7323 non-scalar integral types.
7324 * match.pd (negate_expr_p): Handle MINUS_EXPR.
7325 (-(A-B), -(~A)): New transformations.
7327 2017-11-07 Tom de Vries <tom@codesourcery.com>
7329 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
7330 semicolon after "do {} while (0)".
7331 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7332 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7333 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7334 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7335 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7337 2017-11-07 Tom de Vries <tom@codesourcery.com>
7339 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
7340 after "do {} while (0)".
7341 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7342 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7343 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7344 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7345 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7347 2017-11-07 Tom de Vries <tom@codesourcery.com>
7350 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
7352 (arm_rtx_costs_internal): Add missing semicolon after
7353 HANDLE_NARROW_SHIFT_ARITH call.
7355 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7357 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7358 disable isel if it was not set explicitly.
7360 2017-11-06 James Bowman <james.bowman@ftdichip.com>
7362 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
7363 (add_type_attribute) likewise.
7365 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
7367 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
7368 of struct ix86_frame.
7369 (ix86_initial_elimination_offset): Likewise.
7370 (ix86_expand_split_stack_prologue): Likewise.
7372 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
7374 * tree-vrp.h (enum value_range_type): Update stale comment.
7376 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7377 Alan Hayward <alan.hayward@arm.com>
7378 David Sherwood <david.sherwood@arm.com>
7380 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
7381 (aarch64_expand_vec_perm_const): Take the number of units too.
7382 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
7383 (aarch64_expand_vec_perm_const): Likewise.
7384 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
7385 (vec_perm<mode>): Update accordingly.
7387 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7388 Alan Hayward <alan.hayward@arm.com>
7389 David Sherwood <david.sherwood@arm.com>
7391 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
7392 Take the number of units too.
7393 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
7394 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
7395 but check for a vector mode before rather than after the call.
7396 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7397 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
7398 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
7399 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
7400 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
7401 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
7402 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
7403 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
7404 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
7405 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
7406 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
7407 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
7408 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
7409 (aarch64_sqdmull2_n<mode>): Update accordingly.
7411 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7412 Alan Hayward <alan.hayward@arm.com>
7413 David Sherwood <david.sherwood@arm.com>
7415 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
7416 the number of units too.
7417 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
7418 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
7419 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
7420 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
7421 (vec_store_lanesxi<mode>): Update accordingly.
7423 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7424 Alan Hayward <alan.hayward@arm.com>
7425 David Sherwood <david.sherwood@arm.com>
7427 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
7428 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
7429 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
7430 of units rather than the mode.
7431 * config/aarch64/iterators.md (nunits): New mode attribute.
7432 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
7433 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
7434 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
7435 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
7436 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
7437 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
7438 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
7439 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
7440 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
7441 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
7442 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
7443 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
7444 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
7445 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
7446 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
7447 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
7448 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
7449 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
7450 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
7451 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
7452 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
7453 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
7454 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
7455 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
7456 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
7457 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
7458 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
7459 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
7460 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
7461 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
7462 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
7463 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
7464 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
7465 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
7466 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
7467 (aarch64_simd_vec_setv2di): Likewise.
7469 2017-11-06 Carl Love <cel@us.ibm.com>
7471 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
7473 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
7474 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
7476 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
7478 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
7479 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
7480 Add power 8 macro expansions.
7481 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
7482 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
7483 power 8 instructions. (VSX_XXBR): Add iterator.
7485 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
7487 * config/arm/arm.md (predicable_short_it): Change default to "no",
7488 improve documentation, remove uses that are identical to the default.
7489 (enabled_for_depr_it): Rename to enabled_for_short_it.
7490 * gcc/config/arm/arm-fixed.md (predicable_short_it):
7491 Remove default uses.
7492 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
7493 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
7494 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
7495 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
7497 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
7500 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
7501 float128 helper macros, which are no longer used after deleting
7502 the old 'q' built-in functions, and moving the round to odd
7503 built-in functions to being special built-in functions.
7504 (BU_FLOAT128_2): Likewise.
7505 (BU_FLOAT128_1_HW): Likewise.
7506 (BU_FLOAT128_2_HW): Likewise.
7507 (BU_FLOAT128_3_HW): Likewise.
7508 (FABSQ): Delete old 'q' built-in functions.
7509 (COPYSIGNQ): Likewise.
7510 (SQRTF128_ODD): Move round to odd built-in functions to be
7511 special built-in functions, so that we can handle
7512 -mabi=ieeelongdouble.
7513 (TRUNCF128_ODD): Likewise.
7514 (ADDF128_ODD): Likewise.
7515 (SUBF128_ODD): Likewise.
7516 (MULF128_ODD): Likewise.
7517 (DIVF128_ODD): Likewise.
7518 (FMAF128_ODD): Likewise.
7519 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
7520 built-in names to 'f128'.
7521 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
7522 old 'q' built-in functions, as the machine independent code for
7523 'f128' built-in functions handles this.
7524 (rs6000_expand_builtin): Add expansion for float128 round to odd
7525 functions, keying off on -mabi=ieeelongdouble of whether to use
7526 the KFmode or TFmode variant.
7527 (rs6000_init_builtins): Initialize the _Float128 round to odd
7529 * doc/extend.texi (PowerPC Built-in Functions): Document the old
7530 _Float128 'q' built-in functions are now mapped into the new
7531 'f128' built-in functions.
7533 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
7535 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
7536 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
7538 2017-11-06 David Malcolm <dmalcolm@redhat.com>
7541 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
7542 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
7543 * toplev.c: Include "ipa-fnsummary.h".
7544 (toplev::finalize): Call ipa_fnsummary_c_finalize.
7546 2017-11-06 Jakub Jelinek <jakub@redhat.com>
7548 PR tree-optimization/82838
7549 * gimple-ssa-store-merging.c
7550 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
7551 on a separate gimple_seq which is then appended to seq.
7553 2017-11-06 Jeff Law <law@redhat.com>
7556 * config/i386/i386.c (PROBE_INTERVAL): Remove.
7557 (get_probe_interval): New functions.
7558 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
7559 (ix86_adjust_stack_and_probe): Likewise.
7560 (output_adjust_stack_and_probe): Likewise.
7561 (ix86_emit_probe_stack_range): Likewise.
7562 (ix86_expand_prologue): Likewise.
7564 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7566 PR tree-optimization/82816
7567 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
7568 if the modes of the two types are the same.
7569 (convert_plusminus_to_widen): Likewise.
7571 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7573 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
7575 (usadv16qi): New define_expand.
7576 (usadv8hi): New define_expand.
7578 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
7581 * ipa-inline-transform.c (update_noncloned_frequencies): Always
7583 (inline_transform): Likewise.
7584 * predict.c (counts_to_freqs): Remove useless conditional.
7585 * profile-count.h (profile_count::apply_scale): Move sanity check.
7586 * tree-inline.c (copy_bb): Always scale.
7587 (copy_cfg_body): Likewise.
7589 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
7592 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
7594 (*cmp_ite0): Add enabled_for_depr_it attribute.
7595 (*cmp_ite1): Likewise.
7597 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7599 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
7602 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7604 * tree-vrp.c (vrp_int_const_binop): Return true on success and
7605 return the value by pointer.
7606 (extract_range_from_multiplicative_op_1): Update accordingly.
7607 Return as soon as an operation fails.
7609 2017-11-05 Tom de Vries <tom@codesourcery.com>
7612 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
7613 (DEF_SANITIZER_BUILTIN): ... here.
7614 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
7615 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
7617 2017-11-05 Tom de Vries <tom@codesourcery.com>
7620 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
7622 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
7623 ASM_OUTPUT_BEFORE_CASE_LABEL call.
7624 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
7626 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7627 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7628 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7630 2017-11-05 Tom de Vries <tom@codesourcery.com>
7633 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
7636 2017-11-04 Michael Clark <michaeljclark@mac.com>
7638 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
7639 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
7641 (*addsi3_extended): Likewise.
7642 (*addsi3_extended2): Likewise.
7643 (<optab>si3): Likewise.
7644 (<optab>di3): Likewise.
7645 (<optab><mode>3): Likewise.
7646 (<*optabe>si3_internal): Likewise.
7647 (zero_extendqi<SUPERQI:mode>2): Likewise.
7648 (*add<mode>hi3): Likewise.
7649 (*xor<mode>hi3): Likewise.
7650 (<optab>di3): Likewise.
7651 (*<optab>si3_extend): Likewise.
7652 (*sge<u>_<X:mode><GPR:mode>): Likewise.
7653 (*slt<u>_<X:mode><GPR:mode>): Likewise.
7654 (*sle<u>_<X:mode><GPR:mode>): Likewise.
7656 2017-11-04 Andrew Waterman <andrew@sifive.com>
7658 * config/riscv/riscv.c (riscv_option_override): Conditionally set
7659 TARGET_STRICT_ALIGN based upon -mtune argument.
7661 2017-11-04 Andrew Waterman <andrew@sifive.com>
7663 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
7665 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
7667 * config/i386/i386.c (choose_basereg): Use optional scratch
7668 register and add assertion.
7669 (x86_emit_outlined_ms2sysv_save): Use scratch register when
7670 needed, and don't allocate stack.
7671 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
7672 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
7673 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
7675 2017-11-03 Jeff Law <law@redhat.com>
7677 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
7678 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
7679 to probe at the start of a noreturn function.
7681 2017-11-03 Jakub Jelinek <jakub@redhat.com>
7683 PR tree-optimization/78821
7684 * gimple-ssa-store-merging.c: Update the file comment.
7685 (MAX_STORE_ALIAS_CHECKS): Define.
7686 (struct store_operand_info): New type.
7687 (store_operand_info::store_operand_info): New constructor.
7688 (struct store_immediate_info): Add rhs_code and ops data members.
7689 (store_immediate_info::store_immediate_info): Add rhscode, op0r
7690 and op1r arguments to the ctor, initialize corresponding data members.
7691 (struct merged_store_group): Add load_align_base and load_align
7693 (merged_store_group::merged_store_group): Initialize them.
7694 (merged_store_group::do_merge): Update them.
7695 (merged_store_group::apply_stores): Pick the constant for
7696 encode_tree_to_bitpos from one of the two operands, or skip
7697 encode_tree_to_bitpos if neither operand is a constant.
7698 (class pass_store_merging): Add process_store method decl. Remove
7699 bool argument from terminate_all_aliasing_chains method decl.
7700 (pass_store_merging::terminate_all_aliasing_chains): Remove
7701 var_offset_p argument and corresponding handling.
7702 (stmts_may_clobber_ref_p): New function.
7703 (compatible_load_p): New function.
7704 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
7705 if there is overlap and rhs_code is not INTEGER_CST. For
7706 non-overlapping stores terminate group if rhs is not mergeable.
7707 (get_alias_type_for_stmts): Change first argument from
7708 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
7709 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
7710 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
7712 (get_location_for_stmts): Change first argument from
7713 auto_vec<gimple *> & to vec<gimple *> &.
7714 (struct split_store): Remove orig_stmts data member, add orig_stores.
7715 (split_store::split_store): Create orig_stores rather than orig_stmts.
7716 (find_constituent_stmts): Renamed to ...
7717 (find_constituent_stores): ... this. Change second argument from
7718 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
7719 to info structures rather than the statements.
7720 (split_group): Rename ALLOW_UNALIGNED argument to
7721 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
7722 it. Adjust find_constituent_stores caller.
7723 (imm_store_chain_info::output_merged_store): Handle rhs_code other
7724 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
7725 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
7726 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
7727 (mem_valid_for_store_merging): New function.
7728 (handled_load): New function.
7729 (pass_store_merging::process_store): New method.
7730 (pass_store_merging::execute): Use process_store method. Adjust
7731 terminate_all_aliasing_chains caller.
7733 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7735 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
7736 Return true for more constants, symbols and label references.
7737 (aarch64_valid_floating_const): Remove unused function.
7739 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7742 * config/aarch64/aarch64.c (aarch64_layout_frame):
7743 Undo forcing of LR at bottom of frame.
7745 2017-11-03 Jeff Law <law@redhat.com>
7748 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
7749 for int_registers_saved.
7751 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
7752 extracted from tree-ssa-dom.c.
7753 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
7754 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
7755 (record_equivalences_from_incoming_edge): Add additional argument
7756 to single_pred_edge_ignoring_loop_edges call.
7757 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
7758 (uncprop_dom_walker::before_dom_children): Add additional argument
7759 to single_pred_edge_ignoring_loop_edges call.
7760 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
7761 single_pred_edge_ignoring_loop_edges rather than open coding.
7762 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
7764 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
7766 * match.pd (-(-A)): Rewrite.
7768 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7770 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
7771 (rs6000_emit_int_cmove): New declaration.
7772 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
7773 (rs6000_emit_sISEL): Delete.
7774 (rs6000_emit_int_cmove): Make non-static.
7775 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
7776 instead of rs6000_emit_sISEL.
7778 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
7780 * asan.c (create_cond_insert_point): Maintain profile.
7781 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
7783 * basic-block.h (struct basic_block_def): Remove frequency.
7784 (EDGE_FREQUENCY): Use to_frequency
7785 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
7787 (find_traces): Update to use to_frequency.
7788 (find_traces_1_round): Likewise; use only IPA counts.
7789 (bb_to_key): Likewise.
7790 (connect_traces): Use IPA counts only.
7791 (copy_bb_p): Update to use to_frequency.
7792 (fix_up_crossing_landing_pad): Likewise.
7793 (sanitize_hot_paths): Likewise.
7794 * bt-load.c (basic_block_freq): Likewise.
7795 * cfg.c (init_flow): Set count_max to uninitialized.
7796 (check_bb_profile): Remove frequencies; check counts.
7797 (dump_bb_info): Do not dump frequencies.
7798 (update_bb_profile_for_threading): Update counts only.
7799 (scale_bbs_frequencies_int): Likewise.
7800 (MAX_SAFE_MULTIPLIER): Remove.
7801 (scale_bbs_frequencies_gcov_type): Update counts only.
7802 (scale_bbs_frequencies_profile_count): Update counts only.
7803 (scale_bbs_frequencies): Update counts only.
7804 * cfg.h (struct control_flow_graph): Add count-max.
7805 (update_bb_profile_for_threading): Update prototype.
7806 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
7807 (find_many_sub_basic_blocks): Likewise.
7808 * cfgcleanup.c (try_forward_edges): Likewise.
7809 (try_crossjump_to_edge): Likewise.
7810 * cfgexpand.c (expand_gimple_cond): Likewise.
7811 (expand_gimple_tailcall): Likewise.
7812 (construct_init_block): Likewise.
7813 (construct_exit_block): Likewise.
7814 * cfghooks.c (verify_flow_info): Check consistency of counts.
7815 (dump_bb_for_graph): Do not dump frequencies.
7816 (split_block_1): Do not update frequencies.
7817 (split_edge): Do not update frequencies.
7818 (make_forwarder_block): Do not update frequencies.
7819 (duplicate_block): Do not update frequencies.
7820 (account_profile_record): Do not update frequencies.
7821 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
7822 for global heuristics.
7823 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
7824 (expected_loop_iterations_unbounded): Use counts only.
7825 * cfgloopmanip.c (scale_loop_profile): Simplify.
7826 (create_empty_loop_on_edge): Simplify
7828 (duplicate_loop_to_header_edge): Simplify
7829 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
7830 (update_br_prob_note): Take care of removing note when profile
7832 (relink_block_chain): Do not dump frequency.
7833 (rtl_account_profile_record): Use to_frequency.
7834 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
7835 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
7836 (cgraph_update_edges_for_call_stmt_node): Likewise.
7837 (cgraph_edge::verify_count_and_frequency): Update.
7838 (cgraph_node::verify_node): Temporarily disable frequency verification.
7839 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
7840 to_cgraph_frequency.
7841 (cgraph_edge::rebuild_edges): Convert to ipa counts.
7842 * cgraphunit.c (init_lowered_empty_function): Do not initialize
7844 (cgraph_node::expand_thunk): Update profile.
7845 * except.c (dw2_build_landing_pads): Do not update frequency.
7846 * final.c (compute_alignments): Use to_frequency.
7847 (dump_basic_block_info): Do not dump frequency.
7848 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
7849 (dump_gimple_bb_header): Do not dump frequency.
7850 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
7852 * gimple-streamer-in.c (input_bb): Do not stream frequency.
7853 * gimple-streamer-out.c (output_bb): Do not stream frequency.
7854 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
7855 (init_before_recovery): Do not update frequency.
7856 (sched_create_recovery_edges): Do not update frequency.
7857 * hsa-gen.c (convert_switch_statements): Do not update frequency.
7858 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
7859 (ipa_cp_c_finalize): Set max_count to uninitialized.
7860 * ipa-fnsummary.c (get_minimal_bb): Use counts.
7861 (param_change_prob): Use counts.
7862 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
7864 * ipa-split.c (consider_split): Use to_frequency.
7865 (split_function): Use to_frequency.
7866 * ira-build.c (loop_compare_func): Likewise.
7867 (mark_loops_for_removal): Likewise.
7868 (mark_all_loops_for_removal): Likewise.
7869 * loop-doloop.c (doloop_modify): Do not update frequency.
7870 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
7872 * lto-streamer-in.c (input_function): Update count_max.
7873 * omp-expand.c (expand_omp_taskreg): Update count_max.
7874 * omp-simd-clone.c (simd_clone_adjust): Update profile.
7875 * predict.c (maybe_hot_frequency_p): Use to_frequency.
7876 (maybe_hot_count_p): Use ipa counts only.
7877 (maybe_hot_bb_p): Simplify.
7878 (maybe_hot_edge_p): Simplify.
7879 (probably_never_executed): Do not take frequency argument.
7880 (probably_never_executed_bb_p): Do not pass frequency.
7881 (probably_never_executed_edge_p): Likewise.
7882 (combine_predictions_for_bb): Check that profile is nonzero.
7883 (propagate_freq): Do not set frequency.
7884 (drop_profile): Simplify.
7885 (counts_to_freqs): Simplify.
7886 (expensive_function_p): Use to_frequency.
7887 (propagate_unlikely_bbs_forward): Simplify.
7888 (determine_unlikely_bbs): Simplify.
7889 (estimate_bb_frequencies): Add hack to silence graphite issues.
7890 (compute_function_frequency): Use ipa counts.
7891 (pass_profile::execute): Update.
7892 (rebuild_frequencies): Use counts only.
7893 (force_edge_cold): Use counts only.
7894 * profile-count.c (profile_count::dump): Dump new count types.
7895 (profile_count::differs_from_p): Check compatiblity.
7896 (profile_count::to_frequency): New function.
7897 (profile_count::to_cgraph_frequency): New function.
7898 * profile-count.h (struct function): Declare.
7899 (enum profile_quality): Add profile_guessed_local and
7900 profile_guessed_global0.
7901 (class profile_proability): Decrease number of bits to 29;
7902 update from_reg_br_prob_note and to_reg_br_prob_note.
7903 (class profile_count: Update comment; decrease number of bits
7904 to 61. Check compatibility.
7905 (profile_count::compatible_p): New private member function.
7906 (profile_count::ipa_p): New member function.
7907 (profile_count::operator<): Handle global zero correctly.
7908 (profile_count::operator>): Handle global zero correctly.
7909 (profile_count::operator<=): Handle global zero correctly.
7910 (profile_count::operator>=): Handle global zero correctly.
7911 (profile_count::nonzero_p): New member function.
7912 (profile_count::force_nonzero): New member function.
7913 (profile_count::max): New member function.
7914 (profile_count::apply_scale): Handle IPA scalling.
7915 (profile_count::guessed_local): New member function.
7916 (profile_count::global0): New member function.
7917 (profile_count::ipa): New member function.
7918 (profile_count::to_frequency): Declare.
7919 (profile_count::to_cgraph_frequency): Declare.
7920 * profile.c (OVERLAP_BASE): Delete.
7921 (compute_frequency_overlap): Delete.
7922 (compute_branch_probabilities): Do not use compute_frequency_overlap.
7923 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
7924 * sched-ebb.c (rank): Use counts only.
7925 * shrink-wrap.c (handle_simple_exit): Use counts only.
7926 (try_shrink_wrapping): Use counts only.
7927 (place_prologue_for_one_component): Use counts only.
7928 * tracer.c (find_best_predecessor): Use to_frequency.
7929 (find_trace): Use to_frequency.
7930 (tail_duplicate): Use to_frequency.
7931 * trans-mem.c (expand_transaction): Do not update frequency.
7932 * tree-call-cdce.c: Do not update frequency.
7933 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
7934 (gimple_merge_blocks): Likewise.
7935 (gimple_split_edge): Likewise.
7936 (gimple_duplicate_sese_region): Likewise.
7937 (gimple_duplicate_sese_tail): Likewise.
7938 (move_sese_region_to_fn): Likewise.
7939 (gimple_account_profile_record): Likewise.
7940 (insert_cond_bb): Likewise.
7941 * tree-complex.c (expand_complex_div_wide): Likewise.
7942 * tree-eh.c (lower_resx): Update profile.
7943 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
7945 (initialize_cfun): Do not initialize frequencies
7946 (freqs_to_counts): Delete.
7947 (copy_cfg_body): Ignore count parameter.
7948 (copy_body): Update.
7949 (expand_call_inline): Update count_max.
7950 (optimize_inline_calls): Update count_max.
7951 (tree_function_versioning): Update count_max.
7952 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
7953 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
7955 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
7956 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
7957 (try_peel_loop): Likewise.
7958 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
7960 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
7961 (tree_transform_and_unroll_loop): Do not use frequencies
7962 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
7963 Use reliable prediction only.
7964 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
7965 * tree-ssa-sink.c (select_best_block): Use to_frequency.
7966 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
7967 probability scaling.
7968 * tree-ssa-threadupdate.c (create_block_for_threading): Do
7969 not update frequency
7970 (any_remaining_duplicated_blocks): Likewise.
7971 (update_profile): Likewise.
7972 (estimated_freqs_path): Delete.
7973 (freqs_to_counts_path): Delete.
7974 (clear_counts_path): Delete.
7975 (ssa_fix_duplicate_block_edges): Likewise.
7976 (duplicate_thread_path): Likewise.
7977 * tree-switch-conversion.c (gen_inbound_check): Use counts.
7978 * tree-tailcall.c (decrease_profile): Do not update frequency.
7979 (eliminate_tail_call): Likewise.
7980 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
7981 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
7982 (optimize_mask_stores): Likewise.
7983 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
7984 * ubsan.c (ubsan_expand_null_ifn): Update profile.
7985 (ubsan_expand_ptr_ifn): Update profile.
7986 * value-prof.c (gimple_ic): Simplify.
7987 * value-prof.h (gimple_ic): Update prototype.
7988 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
7989 * ipa-inline.c (compute_uninlined_call_time): Be sure that
7991 (want_inline_self_recursive_call_p): Likewise.
7992 (resolve_noninline_speculation): Only cummulate defined counts.
7993 (inline_small_functions): Use nonzero_p.
7994 (ipa_inline): Do not access freed node.
7996 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7998 * config/aarch64/aarch64.c (aarch64_override_options_internal):
7999 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
8001 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
8003 * config/riscv/riscv.c (riscv_legitimize_move): Handle
8004 non-legitimate address.
8006 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8008 * config/rs6000/rs6000.md (*lt0_disi): Delete.
8009 (*lt0_<mode>di, *lt0_<mode>si): New.
8011 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8013 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
8014 TARGET_PAIRED_FLOAT.
8016 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
8017 Jim Wilson <jim.wilson@linaro.org>
8019 * config/aarch64/aarch64-cores.def (saphira): New CPU.
8020 * config/aarch64/aarch64-tune.md: Regenerated.
8021 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
8022 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
8024 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
8026 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
8027 (arc_expand_prologue): Restore blink for millicode.
8028 * config/arc/linux.h (LINK_EH_SPEC): Defined.
8030 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
8033 * config/i386/i386.c (ix86_vector_duplicate_value): Use
8034 gen_vec_duplicate after forcing the scalar into a register.
8036 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
8038 * combine (try_combine): Print the insns input to try_combine to the
8041 2017-11-02 Steve Ellcey <sellcey@cavium.com>
8044 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
8045 Remove second argument from aarch64_process_target_attr call.
8046 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
8048 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
8050 (aarch64_handle_attr_arch): Remove second argument.
8051 (aarch64_handle_attr_cpu): Ditto.
8052 (aarch64_handle_attr_tune): Ditto.
8053 (aarch64_handle_attr_isa_flags): Ditto.
8054 (aarch64_process_one_target_attr): Ditto.
8055 (aarch64_process_target_attr): Ditto.
8056 (aarch64_option_valid_attribute_p): Remove second argument.
8057 on aarch64_process_target_attr call.
8059 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8061 * diagnostic.c: Include "selftest-diagnostic.h".
8062 (selftest::assert_location_text): New function.
8063 (selftest::test_diagnostic_get_location_text): New function.
8064 (selftest::diagnostic_c_tests): Call it.
8066 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8068 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
8069 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
8070 (class selftest::test_diagnostic_context): Move to...
8071 * selftest-diagnostic.c: New file.
8072 * selftest-diagnostic.h: New file.
8074 2017-11-02 James Bowman <james.bowman@ftdichip.com>
8076 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
8077 offset range for FT32B.
8078 * config/ft32/ft32.h: option "mcompress" enables relaxation.
8079 * config/ft32/ft32.md: Add TARGET_NOPM.
8080 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
8081 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
8083 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
8085 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
8087 2017-11-02 Jeff Law <law@redhat.com>
8089 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
8090 virtual keyword on FINAL OVERRIDE members.
8092 * tree-ssa-propagate.h (ssa_propagation_engine): Group
8093 virtuals together. Add virtual destructor.
8094 (substitute_and_fold_engine): Similarly.
8096 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
8098 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
8100 2017-11-02 Richard Biener <rguenther@suse.de>
8102 PR tree-optimization/82795
8103 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
8105 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8107 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
8109 (gcc_GAS_CHECK_FEATURE): Remove.
8110 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
8111 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
8112 * configure: Regenerate.
8114 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
8116 * config/arc/arc.c (hwloop_optimize): Account for empty
8119 2017-11-02 Richard Biener <rguenther@suse.de>
8122 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
8123 Truncate ARRAY_REF index and element size.
8125 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8127 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
8129 2017-11-01 Jeff Law <law@redhat.com>
8131 * tree-ssa-ccp.c (ccp_folder): New class derived from
8132 substitute_and_fold_engine.
8133 (ccp_folder::get_value): New member function.
8134 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
8135 (ccp_fold_stmt): Remove prototype.
8136 (ccp_finalize): Call substitute_and_fold from the ccp_class.
8137 * tree-ssa-copy.c (copy_folder): New class derived from
8138 substitute_and_fold_engine.
8139 (copy_folder::get_value): Renamed from get_value.
8140 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
8141 * tree-vrp.c (vrp_folder): New class derived from
8142 substitute_and_fold_engine.
8143 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
8144 (vrp_folder::get_value): New member function.
8145 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
8146 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
8147 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
8148 provide a class interface to folder/substitute routines.
8149 (ssa_prop_fold_stmt_fn): Remove typedef.
8150 (ssa_prop_get_value_fn): Likewise.
8151 (subsitute_and_fold): Remove prototype.
8152 (replace_uses_in): Likewise.
8153 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
8154 Renamed from replace_uses_in. Call the virtual member function
8155 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
8156 (substitute_and_fold_dom_walker): Remove initialization of
8157 data member entries for calbacks. Add substitute_and_fold_engine
8158 member and initialize it.
8159 (substitute_and_fold_dom_walker::before_dom_children0: Use the
8160 member functions for get_value, replace_phi_args_in c
8161 replace_uses_in, and fold_stmt calls.
8162 (substitute_and_fold_engine::substitute_and_fold): Renamed from
8163 substitute_and_fold. Remove assert. Update ctor call.
8165 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
8166 (ssa_prop_visit_phi_fn): Likewise.
8167 (class ssa_propagation_engine): New class to provide an interface
8169 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
8171 (ssa_prop_visit_phi): Likewise.
8172 (ssa_propagation_engine::simulate_stmt): Moved into class.
8173 Call visit_phi/visit_stmt from the class rather than via
8174 file scoped static variables.
8175 (ssa_propagation_engine::simulate_block): Moved into class.
8176 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
8177 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
8178 set file scoped statics for the visit_stmt/visit_phi callbacks.
8179 * tree-complex.c (complex_propagate): New class derived from
8180 ssa_propagation_engine.
8181 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
8182 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
8183 (tree_lower_complex): Call ssa_propagate via the complex_propagate
8185 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
8186 ssa_propagation_engine.
8187 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
8188 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
8189 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
8190 * tree-ssa-copy.c (copy_prop): New class derived from
8191 ssa_propagation_engine.
8192 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
8193 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
8194 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
8195 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
8196 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
8197 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
8198 (execute_vrp): Call ssa_propagate from the vrp_prop class.
8200 2017-11-01 Jakub Jelinek <jakub@redhat.com>
8202 PR rtl-optimization/82778
8203 PR rtl-optimization/82597
8204 * compare-elim.c (struct comparison): Add in_a_setter field.
8205 (find_comparison_dom_walker::before_dom_children): Remove killed
8206 bitmap and df_simulate_find_defs call, instead walk the defs.
8207 Compute last_setter and initialize in_a_setter. Merge definitions
8208 with first initialization for a few variables.
8209 (try_validate_parallel): Use insn_invalid_p instead of
8210 recog_memoized. Return insn rather than just the pattern.
8211 (try_merge_compare): Fix up comment. Don't uselessly test if
8212 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
8214 (execute_compare_elim_after_reload): Remove df_chain_add_problem
8217 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8218 Alan Hayward <alan.hayward@arm.com>
8219 David Sherwood <david.sherwood@arm.com>
8221 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
8222 aarch64_hard_regno_nregs to get the number of registers
8225 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8226 Alan Hayward <alan.hayward@arm.com>
8227 David Sherwood <david.sherwood@arm.com>
8229 * config/aarch64/constraints.md (Upl): Rename to...
8231 * config/aarch64/aarch64.md
8232 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
8235 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8236 Alan Hayward <alan.hayward@arm.com>
8237 David Sherwood <david.sherwood@arm.com>
8239 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
8240 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
8243 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8244 Alan Hayward <alan.hayward@arm.com>
8245 David Sherwood <david.sherwood@arm.com>
8247 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
8248 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
8249 (aarch64_evpc_dup): Generate rtl direcly, rather than using
8251 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
8252 of a single element.
8253 * config/aarch64/iterators.md: Add a comment above the permute
8254 unspecs to say that they are generated directly by
8255 aarch64_expand_vec_perm_const.
8256 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
8258 2017-11-01 Nathan Sidwell <nathan@acm.org>
8260 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
8262 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8264 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
8265 and medany code models, and describe what they do.
8267 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8269 Revert accidental duplicate:
8271 * combine.c (can_change_dest_mode): Reject changes in
8272 REGMODE_NATURAL_SIZE.
8274 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
8276 PR rtl-optimization/64682
8277 PR rtl-optimization/69567
8278 PR rtl-optimization/69737
8279 PR rtl-optimization/82683
8280 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
8281 register mentioned in the note, drop the note, unless it came from I3,
8282 in which case it should go to I3 again.
8284 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8286 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
8287 and return false if not.
8288 (clear_bytes_written_by, live_bytes_read): Update accordingly.
8290 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8292 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
8293 range is known to be empty.
8295 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8296 Alan Hayward <alan.hayward@arm.com>
8297 David Sherwood <david.sherwood@arm.com>
8299 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
8300 and CONST_VECTOR_NUNITS instead of computing the number of units from
8301 the byte sizes of the vector and element.
8302 (simplify_binary_operation_1): Likewise.
8303 (simplify_const_binary_operation): Likewise.
8304 (simplify_ternary_operation): Likewise.
8306 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8307 Alan Hayward <alan.hayward@arm.com>
8308 David Sherwood <david.sherwood@arm.com>
8310 * var-tracking.c (INT_MEM_OFFSET): Replace with...
8311 (int_mem_offset): ...this new function.
8312 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
8313 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
8314 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
8317 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8318 Alan Hayward <alan.hayward@arm.com>
8319 David Sherwood <david.sherwood@arm.com>
8321 * lower-subreg.c (interesting_mode_p): New function.
8322 (compute_costs, find_decomposable_subregs, decompose_register)
8323 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
8324 (resolve_clobber, dump_choices): Use it.
8326 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8327 Alan Hayward <alan.hayward@arm.com>
8328 David Sherwood <david.sherwood@arm.com>
8330 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
8332 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8333 Alan Hayward <alan.hayward@arm.com>
8334 David Sherwood <david.sherwood@arm.com>
8336 * alias.c (find_base_value, find_base_term): Only process integer
8337 truncations. Check the precision rather than the size.
8339 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8340 Alan Hayward <alan.hayward@arm.com>
8341 David Sherwood <david.sherwood@arm.com>
8343 * machmode.h (is_narrower_int_mode): New function
8344 * optabs.c (expand_float, expand_fix): Use it.
8345 * dwarf2out.c (rotate_loc_descriptor): Likewise.
8347 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8348 Alan Hayward <alan.hayward@arm.com>
8349 David Sherwood <david.sherwood@arm.com>
8351 * rtl.h (narrower_subreg_mode): New function.
8352 * ira-color.c (update_costs_from_allocno): Use it.
8354 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8355 Alan Hayward <alan.hayward@arm.com>
8356 David Sherwood <david.sherwood@arm.com>
8358 * optabs-query.h (convert_optab_p): New function, split out from...
8359 (convert_optab_handler): ...here.
8360 (widening_optab_handler): Delete.
8361 (find_widening_optab_handler): Remove permit_non_widening parameter.
8362 (find_widening_optab_handler_and_mode): Likewise. Provide an
8363 override that operates on mode class wrappers.
8364 * optabs-query.c (widening_optab_handler): Delete.
8365 (find_widening_optab_handler_and_mode): Remove permit_non_widening
8366 parameter. Assert that the two modes are the same class and that
8367 the "from" mode is narrower than the "to" mode. Use
8368 convert_optab_handler instead of widening_optab_handler.
8369 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
8370 instead of widening_optab_handler.
8371 * expr.c (expand_expr_real_2): Update calls to
8372 find_widening_optab_handler.
8373 * optabs.c (expand_widen_pattern_expr): Likewise.
8374 (expand_binop_directly): Take the insn_code as a parameter.
8375 (expand_binop): Only call find_widening_optab_handler for
8376 conversion optabs; use optab_handler otherwise. Update calls
8377 to find_widening_optab_handler and expand_binop_directly.
8378 Use convert_optab_handler instead of widening_optab_handler.
8379 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
8380 find_widening_optab_handler and use scalar_mode rather than
8382 (convert_plusminus_to_widen): Likewise.
8384 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8385 Alan Hayward <alan.hayward@arm.com>
8386 David Sherwood <david.sherwood@arm.com>
8388 * machmode.h (fixed_size_mode): New class.
8389 * rtl.h (get_pool_mode): Return fixed_size_mode.
8390 * gengtype.c (main): Add fixed_size_mode.
8391 * target.def (get_raw_result_mode): Return a fixed_size_mode.
8392 (get_raw_arg_mode): Likewise.
8393 * doc/tm.texi: Regenerate.
8394 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
8395 * targhooks.c (default_get_reg_raw_mode): Likewise.
8396 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
8397 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
8398 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
8399 (msp430_get_raw_result_mode): Likewise.
8400 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
8401 * dbxout.c (dbxout_parms): Require fixed-size modes.
8402 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
8403 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8404 * omp-low.c (lower_oacc_reductions): Likewise.
8405 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
8406 (simplify_subreg): Update accordingly.
8407 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
8408 (force_const_mem): Update accordingly. Return NULL_RTX for modes
8409 that aren't fixed-size.
8410 (get_pool_mode): Return a fixed_size_mode.
8411 (output_constant_pool_2): Take a fixed_size_mode.
8413 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8414 Alan Hayward <alan.hayward@arm.com>
8415 David Sherwood <david.sherwood@arm.com>
8417 * doc/rtl.texi (vec_series): Document.
8418 (const): Say that the operand can be a vec_series.
8419 * rtl.def (VEC_SERIES): New rtx code.
8420 * rtl.h (const_vec_series_p_1): Declare.
8421 (const_vec_series_p): New function.
8422 * emit-rtl.h (gen_const_vec_series): Declare.
8423 (gen_vec_series): Likewise.
8424 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
8425 (gen_vec_series): Likewise.
8426 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
8427 * simplify-rtx.c (simplify_unary_operation): Handle negations
8429 (simplify_binary_operation_series): New function.
8430 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
8431 (test_vector_ops_series): New function.
8432 (test_vector_ops): Call it.
8433 * config/powerpcspe/altivec.md (altivec_lvsl): Use
8434 gen_const_vec_series.
8435 (altivec_lvsr): Likewise.
8436 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
8438 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8439 Alan Hayward <alan.hayward@arm.com>
8440 David Sherwood <david.sherwood@arm.com>
8442 * doc/rtl.texi (const): Update description of address constants.
8443 Say that vector constants are allowed too.
8444 * common.md (E, F): Use CONSTANT_P instead of checking for
8446 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
8447 checking for CONST_VECTOR.
8448 * expmed.c (make_tree): Use build_vector_from_val for a CONST
8450 * expr.c (expand_expr_real_2): Check for vector modes instead
8451 of checking for CONST_VECTOR.
8452 * rtl.h (const_vec_p): New function.
8453 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
8454 (unwrap_const_vec_duplicate): Handle them here too.
8456 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8457 David Malcolm <dmalcolm@redhat.com>
8458 Alan Hayward <alan.hayward@arm.com>
8459 David Sherwood <david.sherwood@arm.com>
8461 * rtl.h (vec_duplicate_p): New function.
8462 * selftest-rtl.c (assert_rtx_eq_at): New function.
8463 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
8464 (assert_rtx_eq_at): Declare.
8465 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
8466 * selftest-run-tests.c (selftest::run_tests): Call it.
8467 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
8468 (simplify_unary_operation_1): Recursively handle vector duplicates.
8469 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
8471 (simplify_subreg): Handle subregs of vector duplicates.
8472 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
8473 (selftest::simplify_rtx_c_tests): New functions.
8475 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8476 Alan Hayward <alan.hayward@arm.com>
8477 David Sherwood <david.sherwood@arm.com>
8479 * emit-rtl.h (gen_const_vec_duplicate): Declare.
8480 (gen_vec_duplicate): Likewise.
8481 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
8483 (gen_const_vector): ...here.
8484 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
8485 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
8486 whose elements are all equal.
8487 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
8488 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
8489 (simplify_relational_operation): Likewise.
8490 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
8492 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
8493 (aarch64_expand_vector_init): Likewise.
8494 * config/arm/arm.c (neon_vdup_constant): Likewise.
8495 (neon_expand_vector_init): Likewise.
8496 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
8497 (arm_block_set_unaligned_vect): Likewise.
8498 (arm_block_set_aligned_vect): Likewise.
8499 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
8500 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
8501 (expand_vec_perm_even_odd_pack): Likewise.
8502 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
8503 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
8504 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
8505 gen_const_vec_duplicate.
8506 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
8507 * config/mips/mips.c (mips_gen_const_int_vector): Use
8508 gen_const_vec_duplicate.
8509 (mips_expand_vector_init): Use CONST0_RTX.
8510 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
8511 (define_split): Use gen_const_vec_duplicate.
8512 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
8513 (define_split): Use gen_const_vec_duplicate.
8514 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
8515 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
8516 * config/spu/spu.c (spu_const): Likewise.
8518 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8519 Alan Hayward <alan.hayward@arm.com>
8520 David Sherwood <david.sherwood@arm.com>
8522 * combine.c (can_change_dest_mode): Reject changes in
8523 REGMODE_NATURAL_SIZE.
8525 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
8527 * configure.ac (--enable-libssp): New.
8528 (gcc_cv_libc_provides_ssp): Check for explicit setting before
8529 trying to determine target-specific default. Adjust indentation.
8530 * configure: Regenerated.
8531 * doc/install.texi (Configuration): Expand --disable-libssp
8534 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
8536 config/i386/i386.c (ix86_expand_epilogue): Correct stack
8539 2017-10-31 Martin Jambor <mjambor@suse.cz>
8542 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
8544 2017-10-31 David Malcolm <dmalcolm@redhat.com>
8546 * auto-profile.c (autofdo_source_profile::read): Use
8547 UNKNOWN_LOCATION rather than 0.
8548 * diagnostic-core.h (warning_at_rich_loc): Rename to...
8549 (warning_at): ...this overload.
8550 (warning_at_rich_loc_n): Rename to...
8551 (warning_n): ...this overload.
8552 (error_at_rich_loc): Rename to...
8553 (error_at): ...this overload.
8554 (pedwarn_at_rich_loc): Rename to...
8555 (pedwarn): ...this overload.
8556 (permerror_at_rich_loc): Rename to...
8557 (permerror): ...this overload.
8558 (inform_at_rich_loc): Rename to...
8559 (inform): ...this overload.
8560 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
8561 (diagnostic_n_impl_richloc): Rename to...
8562 (diagnostic_n_impl): ...this rich_location *-based decl.
8563 (inform_at_rich_loc): Rename to...
8564 (inform): ...this, and add an assertion.
8565 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
8566 (warning_at_rich_loc): Rename to...
8567 (warning_at): ...this, and add an assertion.
8568 (warning_at_rich_loc_n): Rename to...
8569 (warning_n): ...this, and add an assertion.
8570 (warning_n): Update location_t-based implementation for removal of
8571 location_t-based diagnostic_n_impl.
8572 (pedwarn_at_rich_loc): Rename to...
8573 (pedwarn): ...this, and add an assertion.
8574 (permerror_at_rich_loc): Rename to...
8575 (permerror): ...this, and add an assertion.
8576 (error_n): Update for removal of location_t-based diagnostic_n_impl.
8577 (error_at_rich_loc): Rename to...
8578 (error_at): ...this, and add an assertion.
8579 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
8580 (driver::do_spec_on_infiles): Likewise.
8581 * substring-locations.c (format_warning_va): Update for renaming
8582 of inform_at_rich_loc.
8584 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
8586 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
8587 _Float<N>X built-in functions so that the variant without the
8588 "__builtin_" prefix is only enabled for the GNU C and Objective C
8589 languages when they are in non-strict ANSI/ISO mode.
8590 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
8591 * target.def (floatn_builtin_p): Add a target hook to control
8592 whether _Float<N> and _Float<N>X built-in functions without the
8593 "__builtin_" prefix are enabled, and return true for C and
8594 Objective C in the default hook. Include langhooks.h in
8596 * targhooks.h (default_floatn_builtin_p): Likewise.
8597 * targhooks.c (default_floatn_builtin_p): Likewise.
8598 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
8599 floatn_builtin_p target hook.
8600 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
8602 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
8603 Eric Botcazou <ebotcazou@adacore.com>
8605 PR rtl-optimization/81803
8606 * lra-constraints.c (curr_insn_transform): Also reload the whole
8607 register for a strict subreg no wider than a word if this is for
8608 a WORD_REGISTER_OPERATIONS target.
8610 2017-10-31 Jason Merrill <jason@redhat.com>
8612 * gdbinit.in: Skip over inlines from timevar.h.
8614 2017-10-31 Martin Liska <mliska@suse.cz>
8616 * doc/gcov.texi: Document new option.
8617 * gcov.c (print_usage): Likewise print it.
8618 (process_args): Support the argument.
8619 (format_count): New function.
8620 (format_gcov): Use the function.
8622 2017-10-31 Martin Liska <mliska@suse.cz>
8624 * gcov.c (struct name_map): do not use typedef.
8625 Define operator== and operator<.
8626 (name_search): Remove.
8627 (name_sort): Remove.
8628 (main): Do not allocate names.
8629 (process_file): Add vertical space.
8630 (generate_results): Use std::find.
8631 (release_structures): Do not release memory.
8632 (find_source): Use std::find.
8634 2017-10-31 Martin Liska <mliska@suse.cz>
8636 * gcov.c (struct line_info): Remove it's typedef.
8637 (line_info::line_info): Add proper ctor.
8638 (line_info::has_block): Do not use a typedef.
8639 (struct source_info): Do not use typedef.
8640 (circuit): Likewise.
8641 (get_cycles_count): Likewise.
8642 (output_intermediate_file): Iterate via vector iterator.
8643 (add_line_counts): Use std::vector methods.
8644 (accumulate_line_counts): Likewise.
8645 (output_lines): Likewise.
8647 2017-10-31 Martin Liska <mliska@suse.cz>
8649 * gcov.c (struct source_info): Remove typedef.
8650 (source_info::source_info): Add proper ctor.
8651 (accumulate_line_counts): Use struct, not it's typedef.
8652 (output_gcov_file): Likewise.
8653 (output_lines): Likewise.
8654 (main): Do not allocate an array.
8655 (output_intermediate_file): Use size of vector container.
8656 (process_file): Resize the vector.
8657 (generate_results): Do not preallocate, use newly added vector
8659 (release_structures): Do not release sources.
8660 (find_source): Use vector methods.
8661 (add_line_counts): Do not use typedef.
8663 2017-10-31 Martin Liska <mliska@suse.cz>
8665 * doc/gcov.texi: Document that.
8666 * gcov.c (add_line_counts): Mark lines with a non-executed
8668 (output_line_beginning): Handle such lines.
8669 (output_lines): Pass new argument.
8670 (output_intermediate_file): Print it in intermediate format.
8672 2017-10-31 Martin Liska <mliska@suse.cz>
8674 * color-macros.h: New file.
8675 * diagnostic-color.c: Factor out color related to macros to
8677 * doc/gcov.texi: Document -k option.
8678 * gcov.c (INCLUDE_STRING): Include string.h.
8679 (print_usage): Add -k option.
8680 (process_args): Parse it.
8681 (pad_count_string): New function.
8682 (output_line_beginning): Likewise.
8683 (DEFAULT_LINE_START): New macro.
8684 (output_lines): Support color output.
8686 2017-10-31 Martin Liska <mliska@suse.cz>
8688 PR gcov-profile/82633
8689 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
8690 their interaction with GCOV infrastructure.
8691 * configure.ac: Add -fkeep-{inline,static}-functions to
8693 * configure: Regenerate.
8695 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
8698 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
8700 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
8703 * config/rs6000/rs6000.md (allocate_stack): Force update interval
8704 into a register if it does not fit into an immediate offset field.
8706 2017-10-31 Olivier Hainque <hainque@adacore.com>
8708 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
8710 2017-10-31 Julia Koval <julia.koval@intel.com>
8712 * config.gcc: Add gfniintrin.h.
8713 * config/i386/gfniintrin.h: New.
8714 * config/i386/i386-builtin-types.def
8715 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8716 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8717 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8718 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8719 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8720 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
8721 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
8722 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
8723 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
8724 V64QI_FTYPE_V64QI_V64QI_INT): New types.
8725 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
8726 * config/i386/immintrin.h: Include gfniintrin.h.
8727 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
8729 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
8731 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
8733 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8735 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
8736 (arm_ashldi3_1bit): Remove pattern.
8737 (ashrdi3): Remove shift by 1 expansion.
8738 (arm_ashrdi3_1bit): Remove pattern.
8739 (lshrdi3): Remove shift by 1 expansion.
8740 (arm_lshrdi3_1bit): Remove pattern.
8741 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
8742 cost of ashldi3 by 1.
8743 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
8744 (<shift>di3_neon): Likewise.
8746 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
8748 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
8749 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
8750 and (*aarch64_simd_mov<VQ:mode>).
8751 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
8752 pattern alternative.
8753 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
8754 attributes to match pattern alternative.
8756 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
8758 * config.gcc (powerpc*-*-*): Add emmintrin.h.
8759 * config/rs6000/emmintrin.h: New file.
8760 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
8762 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8764 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
8765 * (movdi_vfp_cortexa8): Remove pattern.
8767 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8769 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
8771 (Specific, alpha*-dec-osf5.1): Remove.
8772 (Specific, mips-sgi-irix5): Remove.
8773 (Specific, mips-sgi-irix6): Remove.
8775 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8778 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
8779 arguments to clear_bit_region_be.
8781 2017-10-30 Jim Wilson <wilson@tuliptree.org>
8783 * gimplify.c: Include memmodel.h.
8785 2017-10-30 Martin Jambor <mjambor@suse.cz>
8787 * omp-grid.c (grid_attempt_target_gridification): Also insert a
8788 condition whether loop should be executed at all.
8790 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
8792 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
8793 gimple folding of vec_madd() intrinsics.
8794 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
8795 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
8796 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
8798 2017-10-30 Richard Biener <rguenther@suse.de>
8800 PR tree-optimization/82762
8802 2017-10-23 Richard Biener <rguenther@suse.de>
8804 PR tree-optimization/82129
8806 2017-08-01 Richard Biener <rguenther@suse.de>
8808 PR tree-optimization/81181
8809 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8810 (compute_antic): ... end of iteration here.
8812 2017-10-30 Joseph Myers <joseph@codesourcery.com>
8814 * doc/invoke.texi (C Dialect Options): Document -std=c17,
8815 -std=iso9899:2017 and -std=gnu17.
8816 * doc/standards.texi (C Language): Document C17 support.
8817 * doc/cpp.texi (Overview): Mention -std=c17.
8818 (Standard Predefined Macros): Document C11 and C17 values of
8819 __STDC_VERSION__. Do not refer to C99 support as incomplete.
8820 * doc/extend.texi (Inline): Do not list individual options for
8821 standards newer than C99.
8822 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
8824 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
8827 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
8829 * asan.c (asan_finish_file): Align asan globals array by shadow
8832 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8835 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
8836 (struct store_immediate_info): Add bitregion_start and bitregion_end
8838 (store_immediate_info::store_immediate_info): Add brs and bre
8839 arguments and initialize bitregion_{start,end} from those.
8840 (struct merged_store_group): Add bitregion_start, bitregion_end,
8841 align_base and mask fields. Drop unnecessary struct keyword from
8842 struct store_immediate_info. Add do_merge method.
8843 (clear_bit_region_be): Use memset instead of loop storing zeros.
8844 (merged_store_group::do_merge): New method.
8845 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
8846 stores as long as the surrounding bitregions have no gaps.
8847 (merged_store_group::merge_overlapping): Use do_merge.
8848 (merged_store_group::apply_stores): Test that bitregion_{start,end}
8849 is byte aligned, rather than requiring that start and width are
8850 byte aligned. Drop unnecessary struct keyword from
8851 struct store_immediate_info. Allocate and populate also mask array.
8852 Make start of the arrays relative to bitregion_start rather than
8853 start and size them according to bitregion_{end,start} difference.
8854 (struct imm_store_chain_info): Drop unnecessary struct keyword from
8855 struct store_immediate_info.
8856 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
8857 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
8858 struct keyword from struct store_immediate_info.
8859 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
8860 between stores as long as the surrounding bitregions have no gaps.
8862 (struct split_store): Add orig non-static data member.
8863 (split_store::split_store): Initialize orig to false.
8864 (find_constituent_stmts): Return store_immediate_info *, non-NULL
8865 if there is exactly a single original stmt. Change stmts argument
8866 to pointer from reference, if NULL, don't push anything to it. Add
8867 first argument, use it to optimize skipping over orig stmts that
8868 are known to be before bitpos already. Simplify.
8869 (split_group): Return unsigned int count how many stores are or
8870 would be needed rather than a bool. Add allow_unaligned argument.
8871 Change split_stores argument from reference to pointer, if NULL,
8872 only do a dry run computing how many stores would be produced.
8873 Rewritten algorithm to use both alignment and misalign if
8874 !allow_unaligned and handle bitfield stores with gaps.
8875 (imm_store_chain_info::output_merged_store): Set start_byte_pos
8876 from bitregion_start instead of start. Compute allow_unaligned
8877 here, if true, do 2 split_group dry runs to compute which one
8878 produces fewer stores and prefer aligned if equal. Punt if
8879 new count is bigger or equal than original before emitting any
8880 statements, rather than during that. Remove no longer needed
8881 new_ssa_names tracking. Replace num_stmts with
8882 split_stores.length (). Use 32-bit stack allocated entries
8883 in split_stores auto_vec. Try to reuse original store lhs/rhs1
8884 if possible. Handle bitfields with gaps.
8885 (pass_store_merging::execute): Ignore bitsize == 0 stores.
8886 Compute bitregion_{start,end} for the stores and construct
8887 store_immediate_info with that. Formatting fixes.
8889 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
8892 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
8893 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
8895 2017-10-29 Jim Wilson <wilson@tuliptree.org>
8897 * gimplify.c: Include tm_p.h.
8899 * common.opt (gcoff): Re-add as ignored option.
8900 (gcoff1, gcoff2, gcoff3): Likewise.
8902 * Makefile.in (OBJS): Delete sdbout.o.
8903 (GTFILES): Delete $(srcdir)/sdbout.c.
8904 * debug.h: Delete sdb_debug_hooks.
8905 * final.c: Delete sdbout.h include.
8906 (final_scan_insn): Delete SDB_DEBUG check.
8907 (rest_of_clean_state): Likewise.
8908 * output.h: Delete sdb_begin_function_line.
8911 * toplev.c: Delete sdbout.h include.
8912 (process_options): Delete SDB_DEBUG check.
8913 * tree-core.h (tree_type_common): Delete pointer field of
8915 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
8916 TYPE_SYMTAB_POINTER.
8917 * tree.h (TYPE_SYMTAB_POINTER): Delete.
8918 (TYPE_SYMTAB_IS_POINTER): Delete.
8919 (TYPE_SYMTAB_IS_DIE): Renumber.
8920 * xcoffout.c: Refer to former sdbout.c file.
8921 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
8923 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
8924 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
8925 (Debugging Options): Delete -gcoff.
8926 (-gstabs): Delete SDB reference.
8928 (-gcoff@var{level}): Delete.
8929 * doc/passes.texi (Debugging information output): Delete SDB and
8930 sdbout.c references.
8931 * doc/tm.texi: Regenerate.
8932 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
8933 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
8935 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
8936 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
8937 (SDB_DEBUGGING_INFO): Delete.
8938 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
8939 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
8940 * target.def (output_source_filename): Delete COFF reference.
8942 * common.opt (gcoff): Delete.
8943 (gxcoff+): Update Negative chain.
8944 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
8946 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
8947 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
8948 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
8950 (expand_function_start): Change sdb reference to past tense.
8951 (expand_function_end): Change sdb reference to past tense.
8952 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
8953 * opts.c (debug_type_names): Delete coff entry.
8954 (common_handle_option): Delete OPT_gcoff case.
8955 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
8957 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
8958 * config/cris/cris.h: Delete SDB reference in comment.
8959 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
8960 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
8961 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
8962 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
8964 (ix86_expand_prologue): Likewise.
8965 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
8966 * config/ia64/ia64.h: Likewise.
8967 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
8968 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
8970 * config/mmix/mmix.h: Likewise.
8971 * config/nds32/nds32.c: Likewise.
8972 * config/stormy/storym16.h: Likewise.
8973 * config/visium/visium.h: Likewise.
8974 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
8976 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
8978 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
8979 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
8980 FRAME_POINTER_REGNUM point at high end of local var area.
8982 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
8984 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
8985 Move comment around. Do not reset best_edge for a copiable
8986 destination if the copy would cause a partition change.
8987 (better_edge_p): Remove redundant check.
8989 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
8991 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
8993 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
8995 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
8996 for math functions that have _Float<N> and _Float<N>X variants.
8997 (mathfn_built_in_2): Add support for math functions that have
8998 _Float<N> and _Float<N>X variants.
8999 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
9000 (expand_builtin_mathfn_ternary): Add support for fma with
9001 _Float<N> and _Float<N>X variants.
9002 (expand_builtin): Likewise.
9003 (fold_builtin_3): Likewise.
9004 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
9005 create math function _Float<N> and _Float<N>X variants as external
9007 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
9008 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
9009 the __builtin_ prefix and if not strict ansi, without the prefix.
9010 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
9011 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
9012 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
9013 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
9014 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
9015 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
9016 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
9017 function signatures for fma _Float<N> and _Float<N>X variants.
9018 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
9019 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
9020 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
9021 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
9022 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
9023 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
9024 * gencfn-macros.c (print_case_cfn): Add support for math functions
9025 that have _Float<N> and _Float<N>X variants.
9026 (print_define_operator_list): Likewise.
9027 (fltfn_suffixes): Likewise.
9029 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
9030 for math functions that have _Float<N> and _Float<N>X variants.
9031 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
9032 and _Float<N>X variants.
9033 (COPYSIGN): Likewise.
9036 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
9037 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
9039 (integer_valued_read_call_p): Likewise.
9040 * fold-const-call.c (fold_const_call_ss): Likewise.
9041 (fold_const_call_sss): Add support for copysign, fmin, and fmax
9042 _Float<N> and _Float<N>X variants.
9043 (fold_const_call_ssss): Add support for fma _Float<N> and
9044 _Float<N>X variants.
9045 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
9046 support for copysign and fma _Float<N> and _Float<N>X variants.
9047 (backprop::process_builtin_call_use): Likewise.
9048 * tree-call-cdce.c (can_test_argument_range); Add support for
9049 sqrt _Float<N> and _Float<N>X variants.
9050 (edom_only_function): Likewise.
9051 (get_no_error_domain): Likewise.
9052 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
9053 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
9054 copysign _Float<N> and _Float<N>X variants.
9055 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
9056 handled by machine independent code.
9057 (FMAF128): Likewise.
9058 * doc/cpp.texi (Common Predefined Macros): Document defining
9059 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
9060 fma _Float<N> and _Float<N>X variants.
9062 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9065 * config/i386/i386-modes.def (CCFPU): Remove definition.
9066 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
9067 (ix86_cc_modes_compatible): Ditto.
9068 (ix86_expand_carry_flag_compare): Ditto.
9069 (ix86_expand_int_movcc): Ditto.
9070 (ix86_expand_int_addcc): Ditto.
9071 (ix86_reverse_condition): Ditto.
9072 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
9073 Return true/false for unordered/ordered fp comparisons.
9074 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
9075 (ix86_prepare_fp_compare_args): Update for rename.
9076 (ix86_expand_fp_compare): Update for rename. Generate unordered
9077 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
9078 (ix86_expand_sse_compare_and_jump): Ditto.
9079 * config/i386/predicates.md (fcmov_comparison_operator):
9080 Remove CCFPU mode handling.
9081 (ix86_comparison_operator): Ditto.
9082 (ix86_carry_flag_operator): Ditto.
9083 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
9084 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
9085 (*cmpu<mode>_cc_i387): Ditto.
9086 (FPCMP): Remove mode iterator.
9087 (unord): Remove mode attribute.
9088 (unord_subst): New define_subst transformation
9089 (unord): New define_subst attribute.
9091 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
9092 (*cmpi<unord>xf_i387): Ditto.
9093 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
9094 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
9095 using unord_subst transformation.
9096 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
9097 (round_saeonly): Also handle CCFP mode.
9098 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
9099 Remove UNSPEC_SAHF unspec handling.
9101 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9103 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
9105 2017-10-27 Jeff Law <law@redhat.com>
9107 * gimple-ssa-sprintf.c: Include domwalk.h.
9108 (class sprintf_dom_walker): New class, derived from dom_walker.
9109 (sprintf_dom_walker::before_dom_children): New function.
9110 (struct call_info): Moved into sprintf_dom_walker class
9111 (compute_formath_length, handle_gimple_call): Likewise.
9112 (sprintf_length::execute): Call the dominator walker rather
9113 than walking the statements.
9115 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
9116 gimple statement locations.
9117 (check_array_bounds): Corresponding changes. Get the statement's
9118 location directly from wi->stmt.
9120 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
9123 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
9125 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9127 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
9128 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
9130 2017-10-27 Jakub Jelinek <jakub@redhat.com>
9133 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
9134 * config/i386/i386.c (maybe_get_pool_constant): Removed.
9135 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
9136 maybe_get_pool_constant.
9137 * config/i386/predicates.md (zero_extended_scalar_load_operand):
9140 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9142 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
9143 2.26 caveat. Update gas and gld versions.
9144 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
9147 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9149 * cgraph.h (set_malloc_flag): Declare.
9150 * cgraph.c (set_malloc_flag_1): New function.
9151 (set_malloc_flag): Likewise.
9152 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
9153 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
9155 (read_ipa_call_summary): Add support for reading is_return_callee.
9156 (write_ipa_call_summary): Stream is_return_callee.
9157 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
9158 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
9159 ipa-prop.h, ipa-fnsummary.h.
9160 (pure_const_names): Change to static.
9161 (malloc_state_e): Define.
9162 (malloc_state_names): Define.
9163 (funct_state_d): Add field malloc_state.
9164 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
9165 (check_retval_uses): New function.
9166 (malloc_candidate_p): Likewise.
9167 (analyze_function): Add support for malloc attribute.
9168 (pure_const_write_summary): Stream malloc_state.
9169 (pure_const_read_summary): Add support for reading malloc_state.
9170 (dump_malloc_lattice): New function.
9171 (propagate_malloc): New function.
9172 (warn_function_malloc): New function.
9173 (ipa_pure_const::execute): Call propagate_malloc and
9174 ipa_free_fn_summary.
9175 (pass_local_pure_const::execute): Add support for malloc attribute.
9176 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
9177 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
9179 2017-10-27 Martin Liska <mliska@suse.cz>
9181 PR gcov-profile/82457
9182 * doc/invoke.texi: Document that one needs a non-strict ISO mode
9183 for fork-like functions to be properly instrumented.
9185 2017-10-27 Richard Biener <rguenther@suse.de>
9188 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
9189 info when we redirected EH.
9191 2017-10-26 Michael Collison <michael.collison@arm.com>
9193 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
9195 (<optab>_trunchf<GPI:mode>2: New pattern.
9196 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
9197 * config/aarch64/iterators.md (wv): New mode attribute.
9198 (vf, VF): New mode attributes.
9199 (vgp, VGP): New mode attributes.
9200 (s): Update attribute with SImode and DImode prefixes.
9202 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9204 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
9205 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
9206 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
9207 (nios2_option_overide): Initialize it. Don't allow R0-relative
9208 addressing with PIC.
9209 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
9210 (nios2_symbolic_constant_p): Likewise.
9211 (nios2_legitimate_address_p): Likewise.
9212 (nios2_r0rel_section_name_p): New.
9213 (nios2_symbol_ref_in_r0rel_data_p): New.
9214 (nios2_emit_move_sequence): Handle r0rel_constant_p.
9215 (r0rel_constant_p): New.
9216 (nios2_print_operand_address): Handle r0rel_constant_p.
9217 (nios2_cdx_narrow_form_p): Likewise.
9218 * config/nios2/nios2.opt (mr0rel-sec=): New option.
9219 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
9220 (Nios II Options): Document -mr0rel-sec.
9222 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9224 * config/nios2/nios2.c: Include xregex.h.
9225 (nios2_gprel_sec_regex): New.
9226 (nios2_option_overide): Initialize it. Don't allow GP-relative
9227 addressing with PIC.
9228 (nios2_small_section_name_p): Check for regex match.
9229 * config/nios2/nios2.opt (mgprel-sec=): New option.
9230 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
9231 (Nios II Options): Document -mgprel-sec.
9233 2017-10-26 Jim Wilson <wilson@tuliptree.org>
9235 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
9237 2017-10-26 Tom de Vries <tom@codesourcery.com>
9239 PR tree-optimization/82707
9240 * gimple.c (gimple_copy): Fix unsharing of
9241 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
9243 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
9245 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
9246 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
9247 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
9248 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
9249 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
9250 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
9251 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
9252 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
9253 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
9254 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
9255 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
9256 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
9257 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
9258 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
9259 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
9260 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
9261 _mm512_mask_cmpunord_ps_mask): New intrinsics.
9263 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9265 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
9267 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
9268 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
9269 warning to rs6000.c. Remove the Undocumented flag, since it has
9271 (-mabi=ibmlongdouble): Likewise.
9272 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
9273 already set, set the default format for long double.
9274 (rs6000_debug_reg_global): Print whether long double is IBM or
9276 (rs6000_option_override_internal): Rework setting long double
9277 format. Only warn if the user is changing the long double default
9278 and they did not use -Wno-psabi.
9279 * doc/invoke.texi (PowerPC options): Update the documentation for
9280 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
9282 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9283 Alan Hayward <alan.hayward@arm.com>
9284 David Sherwood <david.sherwood@arm.com>
9286 * rtl.h (wider_subreg_mode): New function.
9287 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
9288 rather than an unsigned int *.
9289 * ira-color.c (regno_max_ref_width): Replace with...
9290 (regno_max_ref_mode): ...this new variable.
9291 (coalesced_pseudo_reg_slot_compare): Update accordingly.
9292 Use wider_subreg_mode.
9293 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
9294 rather than an unsigned int *.
9295 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
9296 (process_alt_operands): Likewise.
9297 (invariant_p): Likewise.
9298 * lra-spills.c (assign_mem_slot): Likewise.
9299 (add_pseudo_to_slot): Likewise.
9300 * lra.c (collect_non_operand_hard_regs): Likewise.
9301 (add_regs_to_insn_regno_info): Likewise.
9302 * reload1.c (regno_max_ref_width): Replace with...
9303 (regno_max_ref_mode): ...this new variable.
9304 (reload): Update accordingly. Update call to
9305 ira_sort_regnos_for_alter_reg.
9306 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
9307 (init_eliminable_invariants): Update to use regno_max_ref_mode.
9308 (scan_paradoxical_subregs): Likewise.
9310 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9312 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
9313 (aarch64_frame): Add emit_frame_chain boolean.
9314 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9315 Move eh_return case to aarch64_layout_frame.
9316 (aarch64_layout_frame): Initialize emit_frame_chain.
9317 (aarch64_expand_prologue): Use emit_frame_chain.
9319 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9321 * config/aarch64/aarch64.c (aarch64_layout_frame):
9322 Ensure LR is always stored at the bottom of the callee-saves.
9323 Remove rarely used frame layout which saves callee-saves at top of
9324 frame, so the store of LR can be used as a valid probe in all cases.
9326 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9328 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
9329 Improve unaligned TImode/TFmode base/offset split.
9331 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9332 Alan Hayward <alan.hayward@arm.com>
9333 David Sherwood <david.sherwood@arm.com>
9335 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
9336 * combine.c (find_single_use_1): Likewise.
9337 (expand_field_assignment): Likewise.
9338 (move_deaths): Likewise.
9339 * lra-constraints.c (simplify_operand_subreg): Likewise.
9340 (curr_insn_transform): Likewise.
9341 * lra.c (collect_non_operand_hard_regs): Likewise.
9342 (add_regs_to_insn_regno_info): Likewise.
9343 * rtlanal.c (reg_referenced_p): Likewise.
9344 (covers_regno_no_parallel_p): Likewise.
9346 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9348 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
9349 Don't print any bits outside the precision of the value.
9350 * wide-int.cc (test_printing): Add some new tests.
9352 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9354 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
9355 supports -xbrace_comment option.
9356 * configure: Regenerate.
9357 * config.in: Regenerate.
9358 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
9359 (ASM_CPU_SPEC): Use it.
9361 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9363 * target.def (static_rtx_alignment): New hook.
9364 * targhooks.h (default_static_rtx_alignment): Declare.
9365 * targhooks.c (default_static_rtx_alignment): New function.
9366 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
9367 * doc/tm.texi: Regenerate.
9368 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
9369 instead of targetm.constant_alignment. Remove call to
9371 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9372 (cris_preferred_mininum_alignment): New function, split out from...
9373 (cris_constant_alignment): ...here.
9374 (cris_static_rtx_alignment): New function.
9375 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
9377 (ix86_constant_alignment): ...here.
9378 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9379 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9380 (mmix_static_rtx_alignment): New function.
9381 * config/spu/spu.c (spu_static_rtx_alignment): New function.
9382 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9384 2017-10-26 Tamar Christina <tamar.christina@arm.com>
9387 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
9388 Add flag_trapping_math and flag_fp_int_builtin_inexact.
9390 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
9392 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
9393 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
9394 mark as a sign-extending load.
9395 (local_pic_load_u): Define.
9397 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
9400 * fold-const.c (operand_equal_for_comparison_p): Also return true
9401 if ARG0 is a simple variant of ARG1 with narrower precision.
9402 (fold_ternary_loc): Always pass unstripped operands to the predicate.
9404 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
9406 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
9408 * i386.h (processor_costs): Add gather_static, gather_per_elt,
9409 scatter_static, scatter_per_elt.
9410 * x86-tune-costs.h: Add new cost entries.
9412 2017-10-25 Richard Biener <rguenther@suse.de>
9414 * tree-ssa-sccvn.h (vn_eliminate): Declare.
9415 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
9416 class pass_fre): Move to ...
9417 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
9418 class pass_fre): ... here and adjust for statistics.
9420 2017-10-25 Jakub Jelinek <jakub@redhat.com>
9423 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
9424 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
9425 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
9426 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
9429 2017-10-25 Richard Biener <rguenther@suse.de>
9431 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
9432 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
9433 eliminate_push_avail, eliminate_insert): Move inside...
9434 (class eliminate_dom_walker): ... this class in preparation
9436 (fini_eliminate): Remove by merging with ...
9437 (eliminate): ... this function. Adjust for class changes.
9438 (pass_pre::execute): Remove fini_eliminate call.
9439 (pass_fre::execute): Likewise.
9441 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9444 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
9445 (VPERMI2, VPERMI2I): New mode iterators.
9446 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
9447 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
9449 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
9450 mode iterator. Remove 3 old define_insn patterns.
9451 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
9452 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
9453 VPERMI2 mode iterator, remove the other two expanders.
9454 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
9455 to use VPERMI2 mode iterator, add another alternative for vpermi2*
9456 instructions, remove the other two patterns.
9457 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
9458 mode iterator, remove the other two patterns.
9459 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
9460 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
9461 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
9462 and adjust argument order accordingly.
9463 (ix86_expand_vec_perm): Adjust caller.
9464 (expand_vec_perm_1): Likewise.
9465 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
9466 (expand_vec_perm_vpermt2_vpshub2): ... this.
9467 (ix86_expand_vec_perm_const_1): Adjust caller.
9468 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
9471 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
9472 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
9473 (vec_shl_<mode>): Remove unused expander.
9474 (avx512bw_<shift_insn><mode>3): New define_insn.
9475 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
9476 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
9478 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
9481 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
9484 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9486 PR rtl-optimization/82396
9487 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
9488 (autopref_multipass_init): Simplify initialization.
9489 (autopref_rank_data): Simplify sort order.
9490 * gcc/sched-int.h (autopref_multipass_data_): Remove
9491 multi_mem_insn_p, min_offset and max_offset.
9493 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9496 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9497 Check special value of flag_omit_frame_pointer.
9498 (aarch64_can_eliminate): Likewise.
9499 (aarch64_override_options_after_change_1): Simplify handling of
9500 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
9502 2017-10-24 Richard Biener <rguenther@suse.de>
9504 PR tree-optimization/82697
9505 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
9506 zero for conditional load and unconditional store.
9508 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9510 * doc/install.texi: Document bootstrap-cet.
9512 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9515 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
9516 ENDBR instruction at function entrance if function is only
9519 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9522 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
9523 patterns to better describe from which operation the CF is computed.
9524 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
9525 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
9526 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
9527 is 0, use _0 suffixed expanders instead of emitting a comparison
9530 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
9532 * config/i386/i386.md(*movsf_internal, *movdf_internal):
9533 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
9535 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
9538 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
9539 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
9540 * loop-iv.c (iv_get_reaching_def): Likewise.
9541 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
9542 variable is promoted and the partition contains undefined values.
9544 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9546 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
9548 (nios2_address_cost): Define.
9549 (nios2_legitimize_address): Recognize (exp + constant) directly.
9550 (TARGET_ADDRESS_COST): Define.
9552 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9554 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
9555 (nios2_symbolic_memory_operand_p): Declare.
9556 (nios2_split_large_constant): Declare.
9557 (nios2_split_symbolic_memory_operand): Declare.
9558 * config/nios2/nios2.c: Adjust includes.
9559 (nios2_symbolic_constant_allowed): New.
9560 (nios2_symbolic_constant_p): New.
9561 (nios2_plus_symbolic_constant_p): New.
9562 (nios2_valid_addr_expr_p): Recognize addresses involving
9564 (nios2_legitimate_address_p): Likewise, also LO_SUM.
9565 (nios2_symbolic_memory_operand_p): New.
9566 (nios2_large_constant_p): New.
9567 (nios2_split_large_constant): New.
9568 (nios2_split_plus_large_constant): New.
9569 (nios2_split_symbolic_memory_operand): New.
9570 (nios2_legitimize_address): Code refactoring. Handle addresses
9571 involving symbolic constants.
9572 (nios2_emit_move_sequence): Likewise.
9573 (nios2_print_operand): Improve error output.
9574 (nios2_print_operand_address): Handle LO_SUM.
9575 (nios2_cdx_narrow_form_p): Likewise.
9576 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
9577 operands involving symbolic constants.
9578 (movhi_internal, movsi_internal): Likewise.
9579 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
9580 (extendhisi2, extendqi<mode>2): Likewise.
9582 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9584 * tree-pass.h (PROP_rtl_split_insns): Define.
9585 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
9587 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9589 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
9591 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9594 * target.def (const_not_ok_for_debug_p): Default to
9595 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
9596 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
9597 * targhooks.c (default_const_not_ok_for_debug_p): New function.
9598 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
9599 which targetm.const_not_ok_for_debug_p returned true.
9600 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
9602 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
9604 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
9605 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
9606 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
9608 (ix86_const_not_ok_for_debug_p): New function.
9609 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
9610 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
9612 2017-10-23 David Malcolm <dmalcolm@redhat.com>
9615 * system.h: Conditionally include "unique-ptr.h" if
9616 INCLUDE_UNIQUE_PTR is defined.
9617 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
9618 of defining INCLUDE_UNIQUE_PTR before including "system.h".
9620 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
9622 * config/rl78/rl78.md: New define_expand "subdi3".
9624 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
9627 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
9628 DF_REF_INSN if DF_REF_INSN_INFO is false.
9630 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
9632 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
9633 xmm_move instead of sse_move.
9634 (sse_store_index): New function.
9635 (ix86_register_move_cost): Be more sensible about mismatch stall;
9636 model AVX moves correctly; make difference between sse->integer and
9638 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
9639 moves; make difference between SSE and AVX.
9640 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
9641 and zmm_move. Increase size of sse load and store tables;
9642 add unaligned load and store tables; add ssemmx_to_integer.
9643 * x86-tune-costs.h: Update all entries according to real
9644 move latencies from Agner Fog's manual and chip documentation.
9646 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9649 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
9650 * config/i386/constraints.md (Wf): New constraint.
9651 * config/i386/i386.md (UNSPEC_SBB): New unspec.
9652 (cmp<dwi>_doubleword): Removed.
9653 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
9654 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
9655 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
9656 expand with cmp<dwi>_doubleword. For LTU and GEU use
9657 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
9659 * common.opt (gcolumn-info): Enable by default.
9660 * doc/invoke.texi (gcolumn-info): Document new default.
9662 2017-10-23 Richard Biener <rguenther@suse.de>
9664 PR tree-optimization/82672
9665 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
9666 Fold the stmt if we propagated into it.
9668 2017-10-23 Richard Biener <rguenther@suse.de>
9670 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
9671 (bitmap_remove_expr_from_set): ... this. All callers call this
9672 for non-constant values.
9673 (bitmap_set_subtract): Rename to...
9674 (bitmap_set_subtract_expressions): ... this. Adjust and
9676 (bitmap_set_contains_value): Remove superfluous check.
9677 (bitmap_set_replace_value): Inline into single caller ...
9678 (bitmap_value_replace_in_set): ... here and simplify.
9679 (dependent_clean): Merge into ...
9680 (clean): ... this using an overload. Adjust.
9681 (prune_clobbered_mems): Adjust.
9682 (compute_antic_aux): Likewise.
9683 (compute_partial_antic_aux): Likewise.
9685 2017-10-23 Richard Biener <rguenther@suse.de>
9687 PR tree-optimization/82129
9689 2017-08-01 Richard Biener <rguenther@suse.de>
9691 PR tree-optimization/81181
9692 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9693 (compute_antic): ... end of iteration here.
9695 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9697 * target.def (starting_frame_offset): New hook.
9698 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
9699 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
9700 * doc/tm.texi.in: Regenerate.
9701 * hooks.h (hook_hwi_void_0): Declare.
9702 * hooks.c (hook_hwi_void_0): New function.
9703 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
9704 STARTING_FRAME_OFFSET.
9705 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
9706 * reload1.c (reload): Likewise.
9707 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
9708 instead of STARTING_FRAME_OFFSET.
9709 * function.c (try_fit_stack_local): Likewise.
9710 (assign_stack_local_1): Likewise
9711 (instantiate_virtual_regs): Likewise.
9712 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
9713 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
9714 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
9715 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
9716 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
9717 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
9718 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
9719 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
9720 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
9721 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
9722 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
9723 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
9724 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
9725 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
9726 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
9727 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
9728 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
9729 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
9730 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
9731 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
9732 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
9733 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
9734 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
9735 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
9736 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
9737 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
9738 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
9739 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
9740 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
9741 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
9742 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
9743 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
9744 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
9745 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
9746 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
9747 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
9748 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
9749 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
9750 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
9751 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
9752 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
9753 * config/avr/avr.c (avr_starting_frame_offset): Make static and
9754 return a HOST_WIDE_INT.
9755 (avr_builtin_setjmp_frame_value): Use it instead of
9756 STARTING_FRAME_OFFSET.
9757 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9758 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
9759 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
9761 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9762 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
9763 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
9764 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9765 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
9766 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
9767 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9768 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
9769 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
9770 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9771 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
9772 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
9774 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9775 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
9776 * config/mips/mips.c (mips_compute_frame_info): Refer to
9777 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9778 (mips_starting_frame_offset): New function.
9779 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9780 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
9781 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
9782 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
9783 and return a HOST_WIDE_INT.
9784 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9785 (mmix_initial_elimination_offset): Refer to
9786 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9787 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
9788 * config/pa/pa.c (pa_starting_frame_offset): New function.
9789 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
9790 (pa_expand_prologue): Likewise.
9791 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9792 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
9793 !FRAME_GROWS_DOWNWARD handling to...
9794 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9795 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
9796 !FRAME_GROWS_DOWNWARD handling to...
9797 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9798 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
9799 !FRAME_GROWS_DOWNWARD handling to...
9800 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9801 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
9803 (rs6000_starting_frame_offset): New function.
9804 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
9805 !FRAME_GROWS_DOWNWARD handling to...
9806 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9807 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
9808 !FRAME_GROWS_DOWNWARD handling to...
9809 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9810 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
9811 !FRAME_GROWS_DOWNWARD handling to...
9812 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9813 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
9814 (rs6000_starting_frame_offset): New function.
9815 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
9816 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
9817 * config/vax/vax.c (vax_starting_frame_offset): New function.
9818 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
9819 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9820 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
9821 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
9822 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9823 * system.h (STARTING_FRAME_OFFSET): Poison.
9825 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9827 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
9828 SCALAR_TYPE_MODE instead of TYPE_MODE.
9830 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9831 Alan Hayward <alan.hayward@arm.com>
9832 David Sherwood <david.sherwood@arm.com>
9834 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
9836 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9837 Alan Hayward <alan.hayward@arm.com>
9838 David Sherwood <david.sherwood@arm.com>
9840 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
9842 2017-10-23 Richard Biener <rguenther@suse.de>
9844 PR tree-optimization/82129
9845 * tree-ssa-pre.c (bitmap_set_and): Remove.
9846 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
9847 canonicalizing expressions in the set to those with lowest
9848 ID rather than taking that from the first edge.
9850 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9852 * combine.c (rtx_equal_for_field_assignment_p): Use
9853 byte_lowpart_offset.
9855 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9856 Alan Hayward <alan.hayward@arm.com>
9857 David Sherwood <david.sherwood@arm.com>
9859 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
9860 to a SUBREG_PROMOTED_VAR.
9862 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9863 Alan Hayward <alan.hayward@arm.com>
9864 David Sherwood <david.sherwood@arm.com>
9866 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
9867 (expand_debug_source_expr): Likewise.
9868 * combine.c (combine_simplify_rtx): Likewise.
9869 * cse.c (fold_rtx): Likewise.
9870 * optabs.c (expand_float): Likewise.
9871 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9872 (simplify_binary_operation_1): Likewise.
9874 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9875 Alan Hayward <alan.hayward@arm.com>
9876 David Sherwood <david.sherwood@arm.com>
9878 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
9879 (record_promoted_value): Likewise.
9880 * expr.c (expand_expr_real_2): Likewise.
9881 * ree.c (update_reg_equal_equiv_notes): Likewise.
9882 (combine_set_extension): Likewise.
9883 * rtlanal.c (low_bitmask_len): Likewise.
9884 * simplify-rtx.c (neg_const_int): Likewise.
9885 (simplify_binary_operation_1): Likewise.
9887 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9888 Alan Hayward <alan.hayward@arm.com>
9889 David Sherwood <david.sherwood@arm.com>
9891 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
9892 * regcprop.c (maybe_mode_change): Likewise.
9893 * reload1.c (alter_reg): Likewise.
9895 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9897 * inchash.h (inchash::hash::add_wide_int): New function.
9898 * lto-streamer-out.c (hash_tree): Use it.
9900 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9902 * inchash.h (inchash::hash::add_wide_int): Rename to...
9903 (inchash::hash::add_hwi): ...this.
9904 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
9905 (polymorphic_call_target_hasher::hash): Likewise.
9906 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
9907 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
9908 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
9909 * lto-streamer-out.c (hash_tree): Likewise.
9910 * optc-save-gen.awk: Likewise.
9911 * tree.c (add_expr): Likewise.
9913 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
9916 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
9917 for ordered inequality comparisons even with TARGET_IEEE_FP.
9919 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
9922 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
9923 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
9924 Expand with cmp<dwi>_doubleword.
9926 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9928 * extend.texi: Add x86 specific to 'nocf_check' attribute.
9929 List CET intrinsics.
9930 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
9931 specific to -fcf-protection option.
9933 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9935 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
9936 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
9937 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
9938 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
9939 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
9940 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
9941 (extra_objs): Add cet.o for Linux/x86 targets.
9942 (tmake_file): Add i386/t-cet for Linux/x86 targets.
9943 * config/i386/cet.c: New file.
9944 * config/i386/cetintrin.h: Likewise.
9945 * config/i386/t-cet: Likewise.
9946 * config/i386/cpuid.h (bit_SHSTK): New.
9947 (bit_IBT): Likewise.
9948 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
9949 pass IBT and SHSTK bits.
9950 * config/i386/i386-builtin-types.def
9951 (VOID_FTYPE_UNSIGNED_PVOID): New.
9952 (VOID_FTYPE_UINT64_PVOID): Likewise.
9953 * config/i386/i386-builtin.def: Add CET intrinsics.
9954 * config/i386/i386-c.c (ix86_target_macros_internal): Add
9955 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
9956 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
9957 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
9959 * config/i386/i386.c (rest_of_insert_endbranch): New.
9960 (pass_data_insert_endbranch): Likewise.
9961 (pass_insert_endbranch): Likewise.
9962 (make_pass_insert_endbranch): Likewise.
9963 (ix86_notrack_prefixed_insn_p): Likewise.
9964 (ix86_target_string): Add -mibt, -mshstk flags.
9965 (ix86_option_override_internal): Add flag_cf_protection
9967 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
9968 (ix86_print_operand): Add 'notrack' prefix output.
9969 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
9970 (ix86_expand_builtin): Expand CET intrinsics.
9971 (x86_output_mi_thunk): Add 'endbranch' instruction.
9972 * config/i386/i386.h (TARGET_IBT): New.
9973 (TARGET_IBT_P): Likewise.
9974 (TARGET_SHSTK): Likewise.
9975 (TARGET_SHSTK_P): Likewise.
9976 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
9977 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
9978 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
9979 (builtin_setjmp_setup): New pattern.
9980 (builtin_longjmp): Likewise.
9981 (rdssp<mode>): Likewise.
9982 (incssp<mode>): Likewise.
9983 (saveprevssp): Likewise.
9984 (rstorssp): Likewise.
9985 (wrss<mode>): Likewise.
9986 (wruss<mode>): Likewise.
9987 (setssbsy): Likewise.
9988 (clrssbsy): Likewise.
9989 (nop_endbr): Likewise.
9990 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
9992 * config/i386/immintrin.h: Include <cetintrin.h>.
9993 * config/i386/linux-common.h
9994 (file_end_indicate_exec_stack_and_cet): New prototype.
9995 (TARGET_ASM_FILE_END): New.
9997 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
9999 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
10000 latencies instead of having separate table; make difference between
10001 integer and float costs.
10002 * i386.h (processor_costs): Remove scalar_stmt_cost,
10003 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
10004 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
10006 * x86-tune-costs.h: Remove entries which has been removed in
10007 procesor_costs from all tables; make cond_taken_branch_cost
10008 and cond_not_taken_branch_cost COST_N_INSNS based.
10010 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10012 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
10014 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10016 * config/i386/i386.md (isa): Remove fma_avx512f.
10017 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
10018 <avx512>_fmadd_<mode>_mask3<round_name>,
10019 <avx512>_fmsub_<mode>_mask<round_name>,
10020 <avx512>_fmsub_<mode>_mask3<round_name>,
10021 <avx512>_fnmadd_<mode>_mask<round_name>,
10022 <avx512>_fnmadd_<mode>_mask3<round_name>,
10023 <avx512>_fnmsub_<mode>_mask<round_name>,
10024 <avx512>_fnmsub_<mode>_mask3<round_name>,
10025 <avx512>_fmaddsub_<mode>_mask<round_name>,
10026 <avx512>_fmaddsub_<mode>_mask3<round_name>,
10027 <avx512>_fmsubadd_<mode>_mask<round_name>,
10028 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
10029 (*vec_widen_umult_even_v16si<mask_name>,
10030 *vec_widen_smult_even_v16si<mask_name>): Likewise.
10031 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
10033 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10035 * extend.texi: Add 'nocf_check' documentation.
10036 * gimple.texi: Add second parameter to
10037 gimple_build_call_from_tree.
10038 * invoke.texi: Add -fcf-protection documentation.
10039 * rtl.texi: Add REG_CALL_NOTRACK documenation.
10041 2017-10-20 Richard Biener <rguenther@suse.de>
10043 PR tree-optimization/82473
10044 * tree-vect-loop.c (vectorizable_reduction): Properly get at
10045 the largest input type.
10047 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10049 * c-attribs.c (handle_nocf_check_attribute): New function.
10050 (c_common_attribute_table): Add 'nocf_check' handling.
10051 * gimple-parser.c: Add second argument NULL to
10052 gimple_build_call_from_tree.
10053 * attrib.c (comp_type_attributes): Check nocf_check attribute.
10054 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
10056 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
10057 * common.opt: Add fcf-protection flag.
10058 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
10059 * flag-types.h: Add enum cf_protection_level.
10060 * gimple.c (gimple_build_call_from_tree): Add second parameter.
10061 Add 'nocf_check' attribute propagation to gimple call.
10062 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
10063 (gimple_build_call_from_tree): Update prototype.
10064 (gimple_call_nocf_check_p): New function.
10065 (gimple_call_set_nocf_check): Likewise.
10066 * gimplify.c: Add second argument to gimple_build_call_from_tree.
10067 * ipa-icf.c: Add nocf_check attribute in statement hash.
10068 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
10069 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
10070 * toplev.c (process_options): Add flag_cf_protection handling.
10072 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10074 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
10076 2017-10-20 Richard Biener <rguenther@suse.de>
10078 PR tree-optimization/82603
10079 * tree-if-conv.c (predicate_mem_writes): Make sure to only
10080 remove false predicated stores.
10082 2017-10-20 Richard Biener <rguenther@suse.de>
10084 * graphite-isl-ast-to-gimple.c
10085 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
10086 Remove return value and simplify, dump copied stmt after lhs
10088 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
10089 Reduce dump verbosity.
10090 (gsi_insert_earliest): Likewise.
10091 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
10092 * graphite.c (print_global_statistics): Adjust dumping.
10093 (print_graphite_scop_statistics): Likewise.
10094 (print_graphite_statistics): Do not dump loops here.
10095 (graphite_transform_loops): But here.
10097 2017-10-20 Nicolas Roche <roche@adacore.com>
10099 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
10100 * configure: Regenerate.
10102 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10105 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
10106 functions when optimizing replace GIMPLE_RETURN stmts with
10107 calls to __builtin_unreachable ().
10110 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
10111 for -fsanitize=thread link of executables.
10112 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
10113 link of executables.
10116 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
10117 New mode iterators.
10118 (<shift_insn><mode>3<mask_name>): Change the last of the 3
10119 define_insns for logical vector shifts to use VI248_AVX512BW
10120 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
10121 condition, useless isa and prefix attributes. Change the first
10122 2 of these define_insns to ...
10123 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
10124 define_insn for avx512vl.
10125 (<shift_insn><mode>3): ... and this, new define_insn without
10126 masking for non-avx512vl.
10129 * config/i386/sse.md (*andnot<mode>3,
10130 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
10131 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
10132 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
10133 not applied use empty suffix even for TARGET_AVX512VL.
10134 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
10135 is applied, supply evex,evex or evex,evex,evex instead of just
10138 2017-10-20 Julia Koval <julia.koval@intel.com>
10140 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
10141 (OPTION_MASK_ISA_GFNI_UNSET): New.
10142 (ix86_handle_option): Handle OPT_mgfni.
10143 * config/i386/cpuid.h (bit_GFNI): New.
10144 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
10145 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
10146 * config/i386/i386.c (ix86_target_string): Add -mgfni.
10147 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
10148 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
10149 * config/i386/i386.opt: Add mgfni.
10151 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
10153 * config/msp430/msp430.c (msp430_option_override): Disable
10154 -fdelete-null-pointer-checks.
10155 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
10157 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10159 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
10160 of x87 and SSE instructions.
10162 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10164 * asan.c (create_cond_insert_point): Do not update edge count.
10165 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
10166 (afdo_propagate_circuit): Likewise.
10167 (afdo_calculate_branch_prob): Likewise.
10168 (afdo_annotate_cfg): Likewise.
10169 * basic-block.h (struct edge_def): Remove count.
10170 (edge_def::count): New accessor.
10171 * bb-reorder.c (rotate_loop): Update.
10172 (find_traces_1_round): Update.
10173 (connect_traces): Update.
10174 (sanitize_hot_paths): Update.
10175 * cfg.c (unchecked_make_edge): Update.
10176 (make_single_succ_edge): Update.
10177 (check_bb_profile): Update.
10178 (dump_edge_info): Update.
10179 (update_bb_profile_for_threading): Update.
10180 (scale_bbs_frequencies_int): Update.
10181 (scale_bbs_frequencies_gcov_type): Update.
10182 (scale_bbs_frequencies_profile_count): Update.
10183 (scale_bbs_frequencies): Update.
10184 * cfganal.c (connect_infinite_loops_to_exit): Update.
10185 * cfgbuild.c (compute_outgoing_frequencies): Update.
10186 (find_many_sub_basic_blocks): Update.
10187 * cfgcleanup.c (try_forward_edges): Update.
10188 (try_crossjump_to_edge): Update
10189 * cfgexpand.c (expand_gimple_cond): Update
10190 (expand_gimple_tailcall): Update
10191 (construct_exit_block): Update
10192 * cfghooks.c (verify_flow_info): Update
10193 (redirect_edge_succ_nodup): Update
10194 (split_edge): Update
10195 (make_forwarder_block): Update
10196 (duplicate_block): Update
10197 (account_profile_record): Update
10198 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10199 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10200 * cfgloopmanip.c (scale_loop_profile): Update.
10202 (lv_adjust_loop_entry_edge): Update.
10203 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10204 (force_nonfallthru_and_redirect): Update.
10205 (purge_dead_edges): Update.
10206 (rtl_flow_call_edges_add): Update.
10207 * cgraphunit.c (init_lowered_empty_function): Update.
10208 (cgraph_node::expand_thunk): Update.
10209 * gimple-pretty-print.c (dump_probability): Update.
10210 (dump_edge_probability): Update.
10211 * gimple-ssa-isolate-paths.c (isolate_path): Update.
10212 * haifa-sched.c (sched_create_recovery_edges): Update.
10213 * hsa-gen.c (convert_switch_statements): Update.
10214 * ifcvt.c (dead_or_predicable): Update.
10215 * ipa-inline-transform.c (inline_transform): Update.
10216 * ipa-split.c (split_function): Update.
10217 * ipa-utils.c (ipa_merge_profiles): Update.
10218 * loop-doloop.c (add_test): Update.
10219 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
10220 * lto-streamer-in.c (input_cfg): Update.
10221 (input_function): Update.
10222 * lto-streamer-out.c (output_cfg): Update.
10223 * modulo-sched.c (sms_schedule): Update.
10224 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10225 * predict.c (maybe_hot_edge_p): Update.
10226 (unlikely_executed_edge_p): Update.
10227 (probably_never_executed_edge_p): Update.
10228 (dump_prediction): Update.
10229 (drop_profile): Update.
10230 (propagate_unlikely_bbs_forward): Update.
10231 (determine_unlikely_bbs): Update.
10232 (force_edge_cold): Update.
10233 * profile.c (compute_branch_probabilities): Update.
10234 * reg-stack.c (better_edge): Update.
10235 * shrink-wrap.c (handle_simple_exit): Update.
10236 * tracer.c (better_p): Update.
10237 * trans-mem.c (expand_transaction): Update.
10238 (split_bb_make_tm_edge): Update.
10239 * tree-call-cdce.c: Update.
10240 * tree-cfg.c (gimple_find_sub_bbs): Update.
10241 (gimple_split_edge): Update.
10242 (gimple_duplicate_sese_region): Update.
10243 (gimple_duplicate_sese_tail): Update.
10244 (gimple_flow_call_edges_add): Update.
10245 (insert_cond_bb): Update.
10246 (execute_fixup_cfg): Update.
10247 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
10248 * tree-complex.c (expand_complex_div_wide): Update.
10249 * tree-eh.c (lower_resx): Update.
10250 (unsplit_eh): Update.
10251 (cleanup_empty_eh_move_lp): Update.
10252 * tree-inline.c (copy_edges_for_bb): Update.
10253 (freqs_to_counts): Update.
10254 (copy_cfg_body): Update.
10255 * tree-ssa-dce.c (remove_dead_stmt): Update.
10256 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10257 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10258 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10259 (unloop_loops): Update.
10260 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10261 * tree-ssa-loop-split.c (connect_loops): Update.
10262 (split_loop): Update.
10263 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10264 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10265 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10266 * tree-ssa-reassoc.c (branch_fixup): Update.
10267 * tree-ssa-tail-merge.c (replace_block_by): Update.
10268 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10269 (compute_path_counts): Update.
10270 (update_profile): Update.
10271 (recompute_probabilities): Update.
10272 (update_joiner_offpath_counts): Update.
10273 (estimated_freqs_path): Update.
10274 (freqs_to_counts_path): Update.
10275 (clear_counts_path): Update.
10276 (ssa_fix_duplicate_block_edges): Update.
10277 (duplicate_thread_path): Update.
10278 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10279 (case_bit_test_cmp): Update.
10280 (collect_switch_conv_info): Update.
10281 (gen_inbound_check): Update.
10282 (do_jump_if_equal): Update.
10283 (emit_cmp_and_jump_insns): Update.
10284 * tree-tailcall.c (decrease_profile): Update.
10285 (eliminate_tail_call): Update.
10286 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10287 (vect_do_peeling): Update.
10288 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10289 * ubsan.c (ubsan_expand_null_ifn): Update.
10290 (ubsan_expand_ptr_ifn): Update.
10291 * value-prof.c (gimple_divmod_fixed_value): Update.
10292 (gimple_mod_pow2): Update.
10293 (gimple_mod_subtract): Update.
10294 (gimple_ic): Update.
10295 (gimple_stringop_fixed_value): Update.
10297 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10300 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
10302 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
10304 PR rtl-optimization/82395
10305 * ira-color.c (allocno_priority_compare_func): Fix comparison step
10306 based on non_spilled_static_chain_regno_p.
10308 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10310 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
10311 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
10312 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
10314 2017-10-19 Martin Sebor <msebor@redhat.com>
10316 PR tree-optimization/82596
10317 * tree.c (array_at_struct_end_p): Handle STRING_CST.
10319 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10321 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
10322 (get_mem_refs_of_builtin_call): Likewise.
10323 * builtins.c (expand_builtin_apply): Adjust call to
10324 allocate_dynamic_stack_space.
10325 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
10326 the third argument to allocate_dynamic_stack_space, otherwise -1.
10327 (expand_builtin): Deal with all alloca variants.
10328 (is_inexpensive_builtin): Likewise.
10329 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
10330 * calls.c (special_function_p): Deal with all alloca variants.
10331 (initialize_argument_information): Adjust call to
10332 allocate_dynamic_stack_space.
10333 (expand_call): Likewise.
10334 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
10335 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
10336 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
10337 use it for the stack usage computation.
10338 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
10339 * function.c (gimplify_parameters): Call build_alloca_call_expr.
10340 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
10341 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
10342 (in_loop_p): Remove first argument and useless check.
10343 (pass_walloca::execute): Remove useless test and adjust call to above.
10344 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
10345 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
10346 (gimplify_call_expr): Deal with all alloca variants.
10347 * hsa-gen.c (gen_hsa_alloca): Likewise.
10348 (gen_hsa_insns_for_call): Likewise.
10349 * ipa-pure-const.c (special_builtin_state): Likewise.
10350 * tree-chkp.c (chkp_build_returned_bound): Likewise.
10351 * tree-object-size.c (alloc_object_size): Likewise.
10352 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
10353 (call_may_clobber_ref_p_1): Likewise.
10354 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
10355 (ccp_fold_stmt): Likewise.
10356 (optimize_stack_restore): Likewise.
10357 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
10358 (mark_all_reaching_defs_necessary_1): Likewise.
10359 (propagate_necessity): Likewise.
10360 (eliminate_unnecessary_stmts): Likewise.
10361 * tree.c (build_common_builtin_nodes): Build
10362 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
10363 (build_alloca_call_expr): New function.
10364 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
10365 (CASE_BUILT_IN_ALLOCA): Likewise.
10366 (build_alloca_call_expr): Declare.
10367 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
10369 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10372 * dwarf2out.c (new_die_raw): New static inline function.
10373 (new_die): Use it to create the DIE.
10374 (add_AT_external_die_ref): Likewise.
10375 (clone_die): Likewise.
10376 (clone_as_declaration): Likewise.
10377 (dwarf2out_vms_debug_main_pointer): Likewise.
10378 (base_type_die): Likewise. Remove early return for corner cases.
10379 Do not call add_pubtype on the DIE here.
10380 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
10381 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
10382 typedefs for base types with DW_AT_endianity. Make sure a DIE with
10383 native order exists for base types, attach the DIE manually and call
10384 add_pubtype on it. Do not equate a reverse order DIE to the type.
10386 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
10388 * config/arm/arm.c (align_ok_ldrd_strd): New function.
10389 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
10391 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
10393 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10395 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
10396 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
10397 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
10398 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
10399 * opts.c (sanitizer_opts): Add builtin.
10400 * ubsan.c (instrument_builtin): New function.
10401 (pass_ubsan::execute): Call it.
10402 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
10403 * doc/invoke.texi: Document -fsanitize=builtin.
10405 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
10406 builtins, store max (log2 (align), 0) into uchar field instead of
10407 align into uptr field.
10408 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
10409 store uchar 0 field instead of uptr 0 field.
10410 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
10411 instead of passing one address of struct with 2 locations pass
10412 two addresses of structs with 1 location each.
10413 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
10414 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
10415 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
10416 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
10417 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
10418 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
10419 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
10420 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
10422 2017-10-19 Martin Liska <mliska@suse.cz>
10425 * file-find.c (remove_prefix): Remove.
10426 * file-find.h (remove_prefix): Likewise.
10427 * gcc-ar.c: Remove smartness of lookup.
10429 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
10431 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
10432 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
10433 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
10435 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10438 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
10439 (setcc + and to xor + setcc): New peephole2.
10441 2017-10-19 Tom de Vries <tom@codesourcery.com>
10443 * doc/sourcebuild.texi (Test Directives, Variants of
10444 dg-require-support): Add dg-require-stack-size.
10446 2017-10-19 Martin Liska <mliska@suse.cz>
10449 * gimplify.c (gimplify_decl_expr): Do not instrument variables
10450 that have a large alignment.
10451 (gimplify_target_expr): Likewise.
10453 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
10455 PR rtl-optimization/82602
10456 * ira.c (rtx_moveable_p): Return false for volatile asm.
10458 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
10461 * config/i386/i386-modes.def (CCGZ): New CC mode.
10462 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
10463 * config/i386/predicates.md (ix86_comparison_operator):
10465 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10466 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
10467 with double-word subtraction.
10468 (put_condition_code): Handle CCGZmode.
10470 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
10472 * wide-int.cc (debug (const wide_int &)): New.
10473 (debug (const wide_int *)): New.
10474 (debug (const widest_int &)): New.
10475 (debug (const widest_int *)): New.
10477 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
10479 PR middle-end/82556
10480 * lra-constraints.c (curr_insn_transform): Use non-input operand
10481 instead of output one for matched reload.
10483 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10485 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
10486 (tree-ssa-loop-ivopts.h): New header file.
10487 (struct builtin_info): New fields.
10488 (classify_builtin_1): Compute and record base and offset parts for
10489 memset builtin partition by calling strip_offset.
10490 (offset_cmp, fuse_memset_builtins): New functions.
10491 (finalize_partitions): Fuse adjacent memset partitions by calling
10493 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
10494 Expose the interface.
10495 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
10497 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10499 PR tree-optimization/82574
10500 * tree-loop-distribution.c (find_single_drs): New parameter. Check
10501 that data reference must be executed exactly once per iteration
10502 against the outermost loop in nest.
10503 (classify_partition): Update call to above function.
10505 2017-10-18 Richard Biener <rguenther@suse.de>
10507 PR tree-optimization/82591
10508 * graphite.c (graphite_transform_loops): Move code gen message
10510 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10511 Here. Handle scop_to_isl_ast failing.
10512 (scop_to_isl_ast): Limit the number of ISL operations.
10514 2017-10-18 Richard Biener <rguenther@suse.de>
10516 * graphite-isl-ast-to-gimple.c
10517 (translate_isl_ast_to_gimple::set_rename): Simplify.
10518 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
10519 (graphite_copy_stmts_from_block): ... here.
10520 (copy_bb_and_scalar_dependences): Simplify.
10521 (add_parameters_to_ivs_params): Canonicalize.
10522 (generate_entry_out_of_ssa_copies): Simplify.
10523 * graphite-sese-to-poly.c (extract_affine_name): Simplify
10524 by passing in ISL dimension.
10525 (parameter_index_in_region_1): Rename to ...
10526 (parameter_index_in_region): ... this.
10527 (extract_affine): Adjust assert, pass down parameter index.
10528 (add_param_constraints): Use range-info when available.
10529 (build_scop_context): Adjust.
10530 * sese.c (new_sese_info): Adjust.
10531 (free_sese_info): Likewise.
10532 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
10533 Remove unused typedefs.
10534 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
10536 2017-10-18 Martin Liska <mliska@suse.cz>
10538 * combine.c (simplify_compare_const): Add gcc_fallthrough.
10540 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10542 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
10543 (s390_sched_init): Do not reset s390_sched_state if we entered the
10544 current basic block via a fallthru edge and all others are unlikely.
10546 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10548 * config/s390/s390.c (NUM_SIDES): New variable.
10549 (LONGRUNNING_THRESHOLD): New variable.
10550 (LATENCY_FACTOR): New variable.
10551 (s390_sched_score): Decrease score for long-running instructions on
10553 (s390_sched_variable_issue): Perform bookkeeping for long-running
10556 2017-10-18 Richard Biener <rguenther@suse.de>
10558 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
10559 Simplify with removal of the parameter rename map.
10560 (set_rename): Likewise.
10561 (should_copy_to_new_region): Likewise.
10562 (graphite_copy_stmts_from_block): Likewise.
10563 (copy_bb_and_scalar_dependences): Remove initialization of
10564 unused copied_bb_map.
10565 (copy_def): Remove.
10566 (copy_internal_parameters): Likewise.
10567 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
10568 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
10569 Use INTEGRAL_TYPE_P.
10570 (parameter_index_in_region_1): Rename to ...
10571 (assign_parameter_index_in_region): ... this. Assert we have
10572 a parameter we handle.
10573 (scan_tree_for_params): Adjust.
10574 * sese.h (parameter_rename_map_t): Remove.
10575 (struct sese_info_t): Remove unused parameter_rename_map and
10576 copied_bb_map members.
10577 * sese.c (new_sese_info): Adjust.
10578 (free_sese_info): Likewise.
10580 2017-10-18 Martin Liska <mliska@suse.cz>
10583 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
10584 on an abnormal edge.
10586 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
10588 * doc/invoke.texi (ffunction-sections and fdata-sections):
10591 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10593 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
10594 the use statement can throw internally.
10596 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10598 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
10599 any RTX present on the RHS of a SET.
10600 * compare-elim.c (try_eliminate_compare): Restore comment.
10602 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10604 * langhooks.h (struct lang_hooks): Document that tree_size langhook
10605 may be also called on tcc_type nodes.
10606 * langhooks.c (lhd_tree_size): Likewise.
10608 2017-10-17 David Malcolm <dmalcolm@redhat.com>
10610 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
10611 format_warning_at_substring.
10612 (maybe_warn): Convert source_range * param to a location_t. Pass
10613 UNKNOWN_LOCATION rather than NULL to fmtwarn.
10614 (format_directive): Remove code to extract source_ranges and
10615 source_range * in favor of just a location_t.
10616 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
10618 * substring-locations.c (format_warning_va): Convert
10619 source_range * param to a location_t.
10620 (format_warning_at_substring): Likewise.
10621 * substring-locations.h (format_warning_va): Likewise.
10622 (format_warning_at_substring): Likewise.
10624 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
10626 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
10628 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
10629 and scatter/gather ops.
10631 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
10632 vec_gather_load and vec_scatter_store.
10633 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
10634 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
10635 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
10636 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
10637 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
10638 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
10640 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
10642 * reg-stack.c (compare_for_stack_reg): Add bool argument.
10643 Detect FTST instruction and handle its register pops. Only pop
10644 second operand if can_pop_second_op is true.
10645 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
10646 set can_pop_second_op to false in the compare_for_stack_reg call.
10648 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
10649 output_fp_compare for stack register operands.
10650 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
10651 instructions here. Do not emit stack register pops here. Assert
10652 that FCOMPP pops next to top stack register. Rewrite function.
10654 2017-10-17 Nathan Sidwell <nathan@acm.org>
10656 PR middle-end/82577
10657 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
10658 use DECL_ASSEMBLER_NAME_RAW.
10660 PR middle-end/82546
10661 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
10664 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
10665 Wilco Dijkstra <wilco.dijkstra@arm.com>
10667 * builtins.c (expand_builtin_update_setjmp_buf): Add a
10668 converstion to Pmode from the buf_addr.
10670 2017-10-17 Richard Biener <rguenther@suse.de>
10672 * graphite-dependences.c (scop_get_reads_and_writes): Change
10673 output parameters to references.
10675 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
10677 PR 71026/tree-optimization
10678 * fold-const.c (distribute_real_division): Removed.
10679 (fold_binary_loc): Remove calls to distribute_real_divison.
10681 2017-10-17 Richard Biener <rguenther@suse.de>
10683 * graphite-scop-detection.c
10684 (scop_detection::stmt_has_simple_data_refs_p): Always use
10685 the full nest as region.
10686 (try_generate_gimple_bb): Likewise.
10687 * sese.c (scalar_evolution_in_region): Simplify now that
10688 SCEV can handle instantiation in regions.
10689 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
10690 in the non-loop part of a function if requested.
10692 2017-10-17 Richard Biener <rguenther@suse.de>
10694 PR tree-optimization/82563
10695 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
10697 (graphite_regenerate_ast_isl): Call it.
10698 * graphite-scop-detection.c (build_scops): Remove entry edge split.
10700 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10702 PR tree-optimization/82549
10703 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
10704 Formatting fixes. Instead of calling make_bit_field_ref with negative
10707 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
10709 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
10710 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
10711 _mm_maskz_reduce_ss): New.
10712 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
10713 __builtin_ia32_reducess_mask): Ditto..
10714 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
10715 * config/i386/sse.md (reduces<mode>): Renamed to ...
10716 (reduces<mode><mask_scalar_name>): ... this.
10717 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
10719 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
10720 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
10722 2017-10-16 David Malcolm <dmalcolm@redhat.com>
10724 * Makefile.in (OBJS): Add unique-ptr-tests.o.
10725 * selftest-run-tests.c (selftest::run_tests): Call
10726 selftest::unique_ptr_tests_cc_tests.
10727 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
10728 * unique-ptr-tests.cc: New file.
10730 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
10733 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10735 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10736 (make_hard_regno_born, make_hard_regno_dead): Update
10737 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
10739 2017-10-16 Jeff Law <law@redhat.com>
10741 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
10743 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
10745 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
10747 2017-10-16 Olivier Hainque <hainque@adacore.com>
10749 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
10750 with_cpu if we were configured for an e500v2 target cpu name.
10752 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
10754 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
10755 * doc/invoke.texi: Document +nodsp as a valid extension for
10758 2017-10-16 Martin Liska <mliska@suse.cz>
10760 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
10761 (test_set_range): Likewise.
10762 (test_range_functions): Rename to ...
10763 (test_bit_in_range): ... this.
10764 (sbitmap_c_tests): Add new test.
10766 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10768 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
10770 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
10771 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
10773 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10775 * config/aarch64/aarch64-builtins.c
10776 (aarch64_types_quadopu_lane_qualifiers): New.
10777 (TYPES_QUADOPU_LANE): New.
10778 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
10779 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
10780 (aarch64_<sur>dot_laneq<vsi2qi>): New.
10781 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
10782 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
10783 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
10784 (Vdottype, DOTPROD): New.
10785 (sur): Add SDOT and UDOT.
10787 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10789 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
10790 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
10791 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
10792 Add TARGET_DOTPROD.
10793 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
10794 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
10795 Enable TARGET_DOTPROD.
10796 (cortex-a75.cortex-a55): Likewise.
10797 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
10799 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10801 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
10802 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
10804 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
10806 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
10807 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
10808 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
10809 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
10810 * config/arm/types.md (neon_dot, neon_dot_q): New.
10811 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
10813 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10815 * config/arm/arm.h (TARGET_DOTPROD): New.
10816 * config/arm/arm.c (arm_arch_dotprod): New.
10817 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
10818 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
10819 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
10820 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
10821 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
10822 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
10823 * doc/invoke.texi (armv8.2-a): Document dotprod
10825 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10827 * i386.c (ix86_vec_cost): New function.
10828 (ix86_rtx_costs): Handle vector operations better.
10829 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
10830 * x86-tune-costs.h: Add new costs to all tables.
10832 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10834 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
10836 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
10837 divsd, sqrtss and sqrtsd
10838 * x86-tune-costs.h: Add new entries to all costs.
10839 (znver1_cost): Fix to match real instruction latencies.
10841 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10842 Michael Collison <michael.collison@arm.com>
10844 * compare-elim.c: Include emit-rtl.h.
10845 (can_merge_compare_into_arith): New function.
10846 (try_validate_parallel): Likewise.
10847 (try_merge_compare): Likewise.
10848 (try_eliminate_compare): Call the above when no previous clobber
10850 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
10853 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10855 PR middle-end/62263
10856 PR middle-end/82498
10857 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
10858 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
10860 PR middle-end/62263
10861 PR middle-end/82498
10862 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
10863 to be any operand_equal_p operands. For & (B - 1) require
10864 B to be power of 2. Recognize
10865 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
10867 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10870 * optabs.c (expand_memory_blockage): Fix call of
10871 targetm.have_memory_blockage.
10873 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10876 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
10877 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
10878 objects to extra_objs instead of overwriting it.
10880 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10882 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
10883 Use any_fp_register_operand as operand[3] predicate. Simplify
10884 equality test for operands[2] and operands[4] memory location.
10885 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
10886 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
10887 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
10888 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
10889 any_fp_register_operand as operand[1] predicate. Simplify
10890 equality test for operands[0] and operands[3] memory location.
10891 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
10892 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
10893 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
10895 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10897 * target-insns.def: Add memory_blockage.
10898 * optabs.c (expand_memory_blockage): New function.
10899 (expand_asm_memory_barrier): Rename ...
10900 (expand_asm_memory_blockage): ... to this.
10901 (expand_mem_thread_fence): Call expand_memory_blockage
10902 instead of expand_asm_memory_barrier.
10903 (expand_mem_singnal_fence): Ditto.
10904 (expand_atomic_load): Ditto.
10905 (expand_atomic_store): Ditto.
10906 * doc/md.texi (Standard Pattern Names For Generation):
10907 Document memory_blockage instruction pattern.
10909 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
10911 * config/rl78/rl78.c (rl78_emit_libcall): New function.
10912 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
10913 * config/rl78/rl78.md: New define_expand "adddi3".
10915 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
10917 * cfghooks.c (verify_flow_info): Disable check that all probabilities
10920 2017-10-13 Jeff Law <law@redhat.com>
10922 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
10924 2017-10-13 Jakub Jelinek <jakub@redhat.com>
10927 * internal-fn.c (expand_mul_overflow): If both operands have
10928 the same highpart of -1 or 0 and the topmost bit of lowpart
10929 is different, overflow is if res <= 0 rather than res < 0.
10931 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
10933 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
10934 TARGET_P9_VECTOR code for unaligned_load case.
10936 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
10938 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
10940 2017-10-13 Nathan Sidwell <nathan@acm.org>
10942 * tree-core.h (tree_contains_struct): Make bool.
10943 * tree.c (tree_contains_struct): Likewise.
10944 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
10945 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
10946 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
10947 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
10948 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
10950 2017-10-13 Richard Biener <rguenther@suse.de>
10952 * graphite-isl-ast-to-gimple.c
10953 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
10954 parameters and dominance check.
10955 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
10956 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
10957 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
10958 Do not update SSA form here or do intermediate IL verification.
10959 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
10960 (graphite_initialize): Remove check on the number of loops in
10961 the function and inline into graphite_transform_loops.
10962 (graphite_finalize): Inline into graphite_transform_loops.
10963 (graphite_transform_loops): Perform SSA update and IL verification
10965 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
10967 2017-10-13 Richard Biener <rguenther@suse.de>
10969 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
10970 graphite_expression_type_precision): Avoid global constructor
10972 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
10973 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
10974 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
10975 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
10976 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
10977 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
10979 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
10982 * config/i386/i386.h (ix86_red_zone_size): New.
10983 * config/i386/i386.md (push peephole2s): Replace
10984 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
10986 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10987 Alan Hayward <alan.hayward@arm.com>
10988 David Sherwood <david.sherwood@arm.com>
10990 * combine.c (can_change_dest_mode): Reject changes in
10991 REGMODE_NATURAL_SIZE.
10993 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10994 Alan Hayward <alan.hayward@arm.com>
10995 David Sherwood <david.sherwood@arm.com>
10997 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
10998 (expand_debug_source_expr): Likewise.
10999 * combine.c (combine_simplify_rtx): Likewise.
11000 * cse.c (fold_rtx): Likewise.
11001 * fwprop.c (canonicalize_address): Likewise.
11002 * targhooks.c (default_shift_truncation_mask): Likewise.
11004 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11005 Alan Hayward <alan.hayward@arm.com>
11006 David Sherwood <david.sherwood@arm.com>
11008 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
11009 (widened_mode): Likewise.
11010 (expand_unop): Likewise.
11011 * ree.c (transform_ifelse): Likewise.
11012 (merge_def_and_ext): Likewise.
11013 (combine_reaching_defs): Likewise.
11014 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11016 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11017 Alan Hayward <alan.hayward@arm.com>
11018 David Sherwood <david.sherwood@arm.com>
11020 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
11021 * combine.c (gen_lowpart_for_combine): Likewise.
11022 * dwarf2out.c (rtl_for_decl_location): Likewise.
11023 * final.c (alter_subreg): Likewise.
11024 * rtlhooks.c (gen_lowpart_general): Likewise.
11025 (gen_lowpart_if_possible): Likewise.
11027 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11028 Alan Hayward <alan.hayward@arm.com>
11029 David Sherwood <david.sherwood@arm.com>
11031 * calls.c (expand_call): Use subreg_lowpart_offset.
11032 * cse.c (cse_insn): Likewise.
11033 * regcprop.c (copy_value): Likewise.
11034 (copyprop_hardreg_forward_1): Likewise.
11036 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11039 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
11040 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
11041 =Q constraints to +Q and into insn condition add check
11042 that operands[0] and operands[1] are equal.
11043 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
11044 =Q constraints to +Q and into insn condition add check
11045 that operands[0] is equal to either operands[1] or operands[2].
11048 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
11049 instead of handling MINUS_EXPR twice (once for each argument),
11050 canonicalize operand order and handle just once, use rtype where
11051 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
11054 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
11055 any values of __C while still being pattern recognizable as a simple
11056 rotate instruction.
11058 2017-10-13 Richard Biener <rguenther@suse.de>
11060 PR tree-optimization/82451
11062 2017-10-02 Richard Biener <rguenther@suse.de>
11064 PR tree-optimization/82355
11065 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11066 a mapping for the enclosing loop but avoid generating one for
11067 the loop tree root.
11068 (copy_bb_and_scalar_dependences): Remove premature codegen
11069 error on PHIs in blocks duplicated into multiple places.
11070 * graphite-scop-detection.c
11071 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11072 in the region use it as loop and nest to analyze the DR in.
11073 (try_generate_gimple_bb): Likewise.
11074 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11075 (add_loop_constraints): For blocks in a loop not in the region
11076 create a dimension with a single iteration.
11077 * sese.h (gbb_loop_at_index): Remove assert.
11079 * cfgloop.c (loop_preheader_edge): For the loop tree root
11080 return the single successor of the entry block.
11081 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11082 Reset the SCEV hashtable and niters.
11083 * graphite-scop-detection.c
11084 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
11085 assert that we only have POLYNOMIAL_CHREC that vary in loops
11086 contained in the region.
11087 (scop_detection::graphite_can_represent_expr): Adjust.
11088 (scop_detection::stmt_has_simple_data_refs_p): For loops
11089 not in the region set loop to NULL. The nest is now the
11090 entry edge to the region.
11091 (try_generate_gimple_bb): Likewise.
11092 * sese.c (scalar_evolution_in_region): Adjust for
11093 instantiate_scev change.
11094 * tree-data-ref.h (graphite_find_data_references_in_stmt):
11095 Make nest parameter the edge into the region.
11096 (create_data_ref): Likewise.
11097 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
11098 entry edge into a region and adjust instantiate_scev calls.
11099 (create_data_ref): Likewise.
11100 (graphite_find_data_references_in_stmt): Likewise.
11101 (find_data_references_in_stmt): Pass the loop preheader edge
11102 from the nest argument.
11103 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
11104 parameter the edge into the region.
11105 (instantiate_parameters): Use the loop preheader edge as entry.
11106 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
11108 (get_instantiated_value_entry): Make instantiate_below parameter
11109 the edge into the region.
11110 (instantiate_scev_name): Likewise. Adjust dominance checks,
11111 when we cannot use loop-based instantiation instantiate by
11112 walking use-def chains.
11113 (instantiate_scev_poly): Adjust.
11114 (instantiate_scev_binary): Likewise.
11115 (instantiate_scev_convert): Likewise.
11116 (instantiate_scev_not): Likewise.
11117 (instantiate_array_ref): Remove.
11118 (instantiate_scev_3): Likewise.
11119 (instantiate_scev_2): Likewise.
11120 (instantiate_scev_1): Likewise.
11121 (instantiate_scev_r): Do not blindly handle N-operand trees.
11122 Do not instantiate array-refs. Handle all constants and invariants.
11123 (instantiate_scev): Make instantiate_below parameter
11124 the edge into the region.
11125 (resolve_mixers): Use the loop preheader edge for the region
11126 parameter to instantiate_scev_r.
11127 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
11129 2017-10-13 Richard Biener <rguenther@suse.de>
11131 PR tree-optimization/82525
11132 * graphite-isl-ast-to-gimple.c
11133 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
11135 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
11136 Fail code generation when we cannot represent the isl integer.
11137 (binary_op_to_tree): Elide modulo operations that are no-ops
11138 in the type we code generate. Remove now superfluous code
11141 2017-10-13 Richard Biener <rguenther@suse.de>
11143 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
11144 (scop_detection::harmful_loop_in_region): Remove premature
11145 IV type restriction.
11146 (scop_detection::graphite_can_represent_scev): We can handle
11147 pointer IVs just fine.
11149 2017-10-13 Alan Modra <amodra@gmail.com>
11151 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
11152 "Clobbers and Scratch Registers". Add paragraph on
11153 alternative to clobbers for scratch registers and OpenBLAS
11156 2017-10-13 Alan Modra <amodra@gmail.com>
11158 * doc/extend.texi (Clobbers): Correct vax example. Delete old
11159 example of a memory input for a string of known length. Move
11160 commentary out of table. Add a number of new examples
11161 covering array memory inputs.
11163 2017-10-12 Martin Liska <mliska@suse.cz>
11165 PR tree-optimization/82493
11166 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
11167 (test_range_functions): New function.
11168 (sbitmap_c_tests): Likewise.
11169 * selftest-run-tests.c (selftest::run_tests): Run new tests.
11170 * selftest.h (sbitmap_c_tests): New function.
11172 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11174 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11176 * config/rs6000/amo.h: Fix spacing issue.
11178 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11181 * config/i386/i386.md (*ashl<mode>3_mask_1,
11182 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
11183 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
11186 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11188 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
11189 (profile_probability): Set max_probability
11190 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
11192 * profile-count.c (profile_probability::differs_from_p): Do not
11193 rely on max_probaiblity == 10000
11195 2017-10-12 Jeff Law <law@redhat.com>
11197 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
11200 2017-10-12 Martin Sebor <msebor@redhat.com>
11204 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
11205 (handle_alias_pairs): Call it.
11206 * common.opt (-Wattribute-alias): New option.
11207 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
11208 * doc/invoke.texi (-Wattribute-alias): Document.
11210 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
11213 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11215 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11217 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11218 (make_hard_regno_born, make_hard_regno_dead): Update
11219 bb_killed_pseudos and bb_gen_pseudos.
11221 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11223 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
11225 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
11227 * config/alpha/alpha.c (alpha_split_conditional_move):
11228 Use std::swap instead of manually swapping.
11229 (alpha_stdarg_optimize_hook): Ditto.
11230 (alpha_canonicalize_comparison): Ditto.
11232 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11234 * tree-loop-distribution.c (struct builtin_info): New struct.
11235 (struct partition): Refactor fields into struct builtin_info.
11236 (partition_free): Free struct builtin_info.
11237 (build_size_arg_loc, build_addr_arg_loc): Delete.
11238 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
11239 information from struct builtin_info.
11240 (find_single_drs): New function refactored from classify_partition.
11241 Also moved builtin validity checks to this function.
11242 (compute_access_range, alloc_builtin): New functions.
11243 (classify_builtin_st, classify_builtin_ldst): New functions.
11244 (classify_partition): Refactor code into functions find_single_drs,
11245 classify_builtin_st and classify_builtin_ldst.
11246 (distribute_loop): Don't do runtime alias check when distributing
11248 (find_seed_stmts_for_distribution): New function.
11249 (pass_loop_distribution::execute): Refactor code finding seed
11250 stmts into above function. Support distribution for the innermost
11251 two-level loop nest. Adjust dump information.
11253 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11255 * tree-loop-distribution.c: Adjust the general comment.
11256 (NUM_PARTITION_THRESHOLD): New macro.
11257 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
11258 (classify_partition): Skip builtin pattern of loop nest's inner loop.
11259 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
11260 in call to build_partition_graph.
11261 (finalize_partitions): New parameter. Make loop distribution more
11262 conservative by fusing more partitions.
11263 (distribute_loop): Don't do runtime alias check in case of loop nest
11265 (find_seed_stmts_for_distribution): New function.
11266 (prepare_perfect_loop_nest): New function.
11267 (pass_loop_distribution::execute): Refactor code finding seed stmts
11268 and loop nest into above functions. Support loop nest distribution.
11269 Adjust dump information accordingly.
11271 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11273 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
11274 and set PTYPE_SEQUENTIAL for merged partition.
11276 2017-10-12 Richard Biener <rguenther@suse.de>
11278 PR tree-optimization/69728
11280 2017-09-19 Richard Biener <rguenther@suse.de>
11282 PR tree-optimization/69728
11283 * graphite-sese-to-poly.c (schedule_error): New global.
11284 (add_loop_schedule): Handle empty domain by failing the
11286 (build_original_schedule): Handle schedule_error.
11288 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
11289 domain by returning an unchanged schedule.
11291 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11293 * genrecog.c (validate_pattern): For VEC_SELECT verify that
11294 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
11296 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
11298 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
11301 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11304 * expr.c (store_field): Don't optimize away bitsize == 0 store
11305 from CALL_EXPR with addressable return type.
11307 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11309 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
11310 * config/rs6000/rs6000.md (sel): Delete mode attribute.
11311 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
11312 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
11313 TARGET_ISEL instead of TARGET_ISEL<sel>.
11315 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
11317 * config/rs6000/rs6000.c
11318 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
11320 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11322 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
11324 (reg_or_cint_operand): Fix comment.
11325 (reg_or_zero_operand): New predicate.
11326 * config/rs6000/rs6000-protos.h (output_isel): Delete.
11327 * config/rs6000/rs6000.c (output_isel): Delete.
11328 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
11329 instead of reg_or_cint_operand. Output instruction directly (not via
11331 (isel_unsigned_<mode>): Ditto.
11332 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
11333 gpc_reg_operand. Add an instruction alternative for this. Output
11334 instruction directly.
11335 (*isel_reversed_unsigned_<mode>): Ditto.
11337 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11339 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
11340 (TARGET_CANONICALIZE_COMPARISON): Define.
11342 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
11345 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
11346 Check whether the dest is REG before adding REG_EQUIV note.
11348 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11351 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11353 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11354 (make_hard_regno_born, make_hard_regno_dead): Update
11355 bb_killed_pseudos and bb_gen_pseudos.
11357 2017-10-11 Nathan Sidwell <nathan@acm.org>
11359 * incpath.h (enum incpath_kind): Name enum, prefix values.
11360 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
11361 * incpath.c (heads, tails): Use INC_MAX.
11362 (add_env_var_paths, add_standard_paths): Use incpath_kind.
11363 (merge_include_chains, split_quote_chain,
11364 register_include_chains): Update incpath_kind names.
11365 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
11366 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
11368 (add_framework_path, darwin_register_objc_includes): Likewise.
11369 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
11371 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11373 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
11374 Do not use float_operator operator predicate.
11375 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
11376 * config/i386/predicates.md (float_operator): Remove predicate.
11378 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11380 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
11381 (*jccxf_i387): Ditto.
11382 (*jcc<mode>_i387): Ditto.
11383 (*jccu<mode>_i387): Ditto.
11384 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
11385 (*jcc_*_i387 splitters): Remove.
11386 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
11387 * config/i386/i386.c (ix86_split_fp_branch): Remove.
11388 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
11391 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11393 * profile-count.h (slow_safe_scale_64bit): New function.
11394 (safe_scale_64bit): New inline.
11395 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
11396 * profile-count.c: Include wide-int.h
11397 (slow_safe_scale_64bit): New.
11399 2017-10-11 Nathan Sidwell <nathan@acm.org>
11401 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
11402 HAS_DECL_ASSEMBLER_NAME_P.
11403 * gimple-expr.c (gimple_decl_printable_name: Check
11404 HAS_DECL_ASSEMBLER_NAME_P too.
11405 * ipa-utils.h (type_in_anonymous_namespace_p): Check
11406 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
11407 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
11408 * passes.c (rest_of_decl_compilation): Check
11409 HAS_DECL_ASSEMBLER_NAME_P too.
11410 * recog.c (verify_changes): Likewise.
11411 * tree-pretty-print.c (dump_decl_name): Likewise.
11412 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
11414 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
11415 (SET_DECL_ASSEMBLER_NAME): Use it.
11416 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
11417 (COPY_DECL_ASSEMBLER_NAME): Likewise.
11418 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
11420 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11422 * config.gcc (i386, x86_64): Add extra objects.
11423 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
11424 (ix86_min_insn_size): Declare.
11425 (ix86_issue_rate): Declare.
11426 (ix86_adjust_cost): Declare.
11427 (ia32_multipass_dfa_lookahead): Declare.
11428 (ix86_macro_fusion_p): Declare.
11429 (ix86_macro_fusion_pair_p): Declare.
11430 (ix86_bd_has_dispatch): Declare.
11431 (ix86_bd_do_dispatch): Declare.
11432 (ix86_core2i7_init_hooks): Declare.
11433 (ix86_atom_sched_reorder): Declare.
11434 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
11435 (COSTS_N_BYTES): Move to x86-tune-costs.h.
11436 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
11437 (rip_relative_addr_p): Rename to ...
11438 (ix86_rip_relative_addr_p): ... this one; export.
11439 (memory_address_length): Update.
11440 (ix86_issue_rate): Move to x86-tune-sched.c.
11441 (ix86_flags_dependent): Move to x86-tune-sched.c.
11442 (ix86_agi_dependent): Move to x86-tune-sched.c.
11443 (exact_dependency_1): Move to x86-tune-sched.c.
11444 (exact_store_load_dependency): Move to x86-tune-sched.c.
11445 (ix86_adjust_cost): Move to x86-tune-sched.c.
11446 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
11447 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
11448 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
11449 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
11450 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
11451 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
11452 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
11453 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
11454 (min_insn_size): Rename to ...
11455 (ix86_min_insn_size): ... this one; export.
11456 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
11457 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
11458 (core2i7_first_cycle_multipass_backtrack): Move to
11459 x86-tune-sched-core.c.
11460 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
11461 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
11462 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
11463 (ix86_avoid_jump_mispredicts): Update.
11464 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
11465 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
11466 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
11467 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
11468 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
11469 (MAX_INSN): Move to ix86-tune-sched-bd.c.
11470 (MAX_IMM): Move to ix86-tune-sched-bd.c.
11471 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
11472 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
11473 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
11474 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
11475 (MAX_STORE): Move to ix86-tune-sched-bd.c.
11476 (BIG): Move to ix86-tune-sched-bd.c.
11477 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
11478 (enum insn_path): Move to ix86-tune-sched-bd.c.
11479 (get_mem_group): Move to ix86-tune-sched-bd.c.
11480 (is_cmp): Move to ix86-tune-sched-bd.c.
11481 (dispatch_violation): Move to ix86-tune-sched-bd.c.
11482 (is_branch): Move to ix86-tune-sched-bd.c.
11483 (is_prefetch): Move to ix86-tune-sched-bd.c.
11484 (init_window): Move to ix86-tune-sched-bd.c.
11485 (allocate_window): Move to ix86-tune-sched-bd.c.
11486 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
11487 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
11488 (process_end_window): Move to ix86-tune-sched-bd.c.
11489 (allocate_next_window): Move to ix86-tune-sched-bd.c.
11490 (find_constant): Move to ix86-tune-sched-bd.c.
11491 (get_num_immediates): Move to ix86-tune-sched-bd.c.
11492 (has_immediate): Move to ix86-tune-sched-bd.c.
11493 (get_insn_path): Move to ix86-tune-sched-bd.c.
11494 (get_insn_group): Move to ix86-tune-sched-bd.c.
11495 (count_num_restricted): Move to ix86-tune-sched-bd.c.
11496 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
11497 (add_insn_window): Move to ix86-tune-sched-bd.c.
11498 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
11499 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
11500 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
11501 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
11502 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
11503 (do_dispatch): Move to ix86-tune-sched-bd.c.
11504 (has_dispatch): Move to ix86-tune-sched-bd.c.
11505 * config/i386/t-i386: Add new object files.
11506 * config/i386/x86-tune-costs.h: New file.
11507 * config/i386/x86-tune-sched-atom.c: New file.
11508 * config/i386/x86-tune-sched-bd.c: New file.
11509 * config/i386/x86-tune-sched-core.c: New file.
11510 * config/i386/x86-tune-sched.c: New file.
11512 2017-10-11 Liu Hao <lh_mouse@126.com>
11514 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
11515 the generic version below instead.
11516 (should_colorize): Recognize Windows consoles as terminals
11518 * pretty-print.c [__MINGW32__] (write_all): New function.
11519 [__MINGW32__] (find_esc_head): Likewise.
11520 [__MINGW32__] (find_esc_terminator): Likewise.
11521 [__MINGW32__] (eat_esc_sequence): Likewise.
11522 [__MINGW32__] (mingw_ansi_fputs): New function that handles
11524 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
11527 2017-10-11 Richard Biener <rguenther@suse.de>
11529 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
11530 Properly call analyze_scalar_evolution with the loop of the stmt.
11532 2017-10-11 Richard Biener <rguenther@suse.de>
11534 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
11535 * tree-core.h (tree_base): Add chrec_var union member.
11536 * tree.h (CHREC_VAR): Remove.
11537 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
11538 * tree-chrec.h (build_polynomial_chrec): Adjust.
11539 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
11540 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
11542 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
11544 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
11545 * match.pd: ... here.
11546 ((T) X == (T) Y): Relax condition.
11548 2017-10-11 Bin Cheng <bin.cheng@arm.com>
11550 PR tree-optimization/82472
11551 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
11553 (break_alias_scc_partitions): Update postorder number.
11555 2017-10-11 Martin Liska <mliska@suse.cz>
11558 * opts.c (parse_no_sanitize_attribute): Do not use error_value
11560 * opts.h (parse_no_sanitize_attribute): Remove last argument.
11562 2017-10-11 Martin Liska <mliska@suse.cz>
11564 * print-rtl.c (print_insn): Move declaration of idbuf
11565 to same scope as name.
11567 2017-10-11 Martin Liska <mliska@suse.cz>
11572 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11573 volatile arguments.
11575 2017-10-11 Martin Liska <mliska@suse.cz>
11578 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11579 volatile arguments.
11581 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
11583 * config.gcc (default_gnu_indirect_function): Default to yes for
11584 arm*-*-linux* with glibc.
11586 2017-10-11 Richard Biener <rguenther@suse.de>
11588 * tree-scalar-evolution.c (get_scalar_evolution): Handle
11589 default-defs and types we do not want to analyze.
11590 (interpret_loop_phi): Replace unreachable code with an assert.
11591 (compute_scalar_evolution_in_loop): Remove and inline ...
11592 (analyze_scalar_evolution_1): ... here, replacing condition with
11593 what makes the intent clearer. Remove handling of cases
11594 get_scalar_evolution now handles.
11596 2017-10-10 Jim Wilson <wilson@tuliptree.org>
11598 PR rtl-optimization/81434
11599 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
11600 comment for main loop. In sched_group_found if, also add checks for
11601 pass and min_cost_group.
11603 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
11605 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
11606 (rs6000_insn_cost): New function.
11607 * config/rs6000/rs6000.md (cost): New attribute.
11609 2017-10-10 Jakub Jelinek <jakub@redhat.com>
11610 H.J. Lu <hongjiu.lu@intel.com>
11614 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
11615 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
11616 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
11617 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
11618 to OPTION_MASK_ISA_AVX512VL - builtins that have both
11619 OPTION_MASK_ISA_MMX and some other bit set require both
11620 mmx and the ISAs without the mmx bit.
11621 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
11622 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
11623 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
11624 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
11625 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
11626 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
11627 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
11628 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
11629 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
11630 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
11631 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
11632 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
11633 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
11634 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
11635 Add OPTION_MASK_ISA_MMX.
11637 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
11639 * config.gcc (armv7*-*-freebsd*): New target.
11640 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
11642 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
11644 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
11645 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
11648 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
11650 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
11651 a template parameter.
11652 (WIDE_INT_REF_FOR): Update accordingly.
11653 * tree.h (wi::int_traits <const_tree>): Delete.
11654 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
11655 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
11656 (wi::tree_to_wide_ref): New typedef.
11657 (wi::to_wide): New function.
11658 * calls.c (get_size_range): Use wi::to_wide when operating on
11659 trees as wide_ints.
11660 * cgraph.c (cgraph_node::create_thunk): Likewise.
11661 * config/i386/i386.c (ix86_data_alignment): Likewise.
11662 (ix86_local_alignment): Likewise.
11663 * dbxout.c (stabstr_O): Likewise.
11664 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
11665 * expr.c (const_vector_from_tree): Likewise.
11666 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
11667 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
11668 (fold_negate_expr_1, int_const_binop_1, const_binop)
11669 (fold_convert_const_int_from_real, optimize_bit_field_compare)
11670 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
11671 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
11672 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
11673 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
11674 (fold_not_const, round_up_loc): Likewise.
11675 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
11676 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
11677 (alloca_call_type): Likewise.
11678 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
11679 * godump.c (go_output_typedef): Likewise.
11680 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
11681 * internal-fn.c (get_min_precision): Likewise.
11682 * ipa-cp.c (ipcp_store_vr_results): Likewise.
11683 * ipa-polymorphic-call.c
11684 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
11685 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
11686 (ipa_modify_call_arguments): Likewise.
11687 * match.pd: Likewise.
11688 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
11689 * print-tree.c (print_node_brief, print_node): Likewise.
11690 * stmt.c (expand_case): Likewise.
11691 * stor-layout.c (layout_type): Likewise.
11692 * tree-affine.c (tree_to_aff_combination): Likewise.
11693 * tree-cfg.c (group_case_labels_stmt): Likewise.
11694 * tree-data-ref.c (dr_analyze_indices): Likewise.
11695 (prune_runtime_alias_test_list): Likewise.
11696 * tree-dump.c (dequeue_and_dump): Likewise.
11697 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
11698 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
11699 * tree-pretty-print.c (dump_generic_node): Likewise.
11700 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
11701 (simple_iv_with_niters): Likewise.
11702 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
11703 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
11704 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
11705 * tree-ssa-loop-niter.c (split_to_var_and_offset)
11706 (refine_value_range_using_guard, number_of_iterations_ne_max)
11707 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
11708 (get_cst_init_from_scev, record_nonwrapping_iv)
11709 (scev_var_range_cant_overflow): Likewise.
11710 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
11711 * tree-ssa-pre.c (compute_avail): Likewise.
11712 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
11713 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
11714 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
11715 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
11716 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
11717 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
11718 (dump_case_nodes, try_switch_expansion): Likewise.
11719 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
11720 (vect_do_peeling): Likewise.
11721 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
11722 * tree-vect-stmts.c (vectorizable_load): Likewise.
11723 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
11724 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
11725 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
11726 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
11727 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
11728 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
11729 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
11730 (evrp_dom_walker::before_dom_children): Likewise.
11731 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
11732 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
11733 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
11734 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
11735 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
11736 (get_range_pos_neg): Likewise.
11737 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
11738 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
11739 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
11740 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
11741 * config/avr/avr.c (avr_fold_builtin): Likewise.
11742 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
11743 * config/msp430/msp430.c (msp430_attr): Likewise.
11744 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
11745 * config/powerpcspe/powerpcspe-c.c
11746 (altivec_resolve_overloaded_builtin): Likewise.
11747 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
11748 (rs6000_expand_ternop_builtin): Likewise.
11749 * config/rs6000/rs6000-c.c
11750 (altivec_resolve_overloaded_builtin): Likewise.
11751 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
11752 (rs6000_expand_ternop_builtin): Likewise.
11753 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
11755 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11757 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
11758 when copying loop nest with only one inner loop.
11760 2017-10-10 Richard Biener <rguenther@suse.de>
11762 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
11763 blocks if SCEV is active.
11764 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
11766 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
11767 (scev_initialize): Assert we are not yet initialized.
11769 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11771 * tree-loop-distribution.c (generate_loops_for_partition): Remove
11772 inner loop's exit stmt by making it always exit the loop, otherwise
11773 we would generate an infinite empty loop.
11775 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11777 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
11778 renaming variables in new preheader if it's deleted.
11780 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11782 * tree-loop-distribution.c (struct partition): Remove unused field
11783 loops of the structure.
11784 (partition_alloc, partition_free): Ditto.
11785 (build_rdg_partition_for_vertex): Ditto.
11787 2017-10-09 Jeff Law <law@redhat.com>
11789 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
11790 return type to match prototype and documentation.
11792 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11794 * config/rs6000/rs6000.c (processor_costs): Move to ...
11795 * config/rs6000/rs6000.h: ... here.
11796 (rs6000_cost): Declare.
11798 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
11800 * except.c (setjmp_fn): New global variable.
11801 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
11802 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
11803 if DONT_USE_BUILTIN_SETJMP is defined.
11805 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11807 * target.def (insn_cost): New hook.
11808 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
11809 * doc/tm.texi: Regenerate.
11810 * rtlanal.c (insn_cost): Use the new hook.
11812 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11814 * combine.c (combine_validate_cost): Compute the new insn_cost,
11815 not just pattern_cost.
11816 (try_combine): Adjust comment.
11818 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11820 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
11822 * combine.c (uid_insn_cost): Adjust comment.
11823 (combine_validate_cost): Adjust comment. Use pattern_cost instead
11825 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
11826 * dse.c (find_shift_sequence): Ditto.
11827 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
11828 (bb_valid_for_noce_process_p): Use pattern_cost.
11829 * rtl.h (insn_rtx_cost): Delete.
11830 (pattern_cost): New prototype.
11831 (insn_cost): New prototype.
11832 * rtlanal.c (insn_rtx_cost): Rename to...
11833 (pattern_cost): ... this.
11836 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
11838 * config/i386/i386.md (*jcc_2): Remove insn pattern.
11839 (*jcc<mode>_0_r_i387): Ditto.
11840 (*jccxf_r_i387): Ditto.
11841 (*jcc<mode>_r_i387): Ditto.
11842 (*jccu<mode>_r_i387): Ditto.
11843 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
11844 (*jcc): Rename from *jcc_1.
11846 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11848 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
11849 deferred rescans after the lvx/stvx recombination pre-pass.
11851 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
11853 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
11854 memory operation instruction support.
11855 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
11856 (rs6000-ibm-aix[789]*): Likewise.
11857 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
11858 Document new functions.
11860 2017-10-09 Richard Biener <rguenther@suse.de>
11862 PR tree-optimization/82397
11863 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
11864 equality only for semantically equal trees.
11866 2017-10-09 Richard Biener <rguenther@suse.de>
11868 PR tree-optimization/82449
11869 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
11870 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
11871 allow constant addresses.
11872 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
11875 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11877 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
11880 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11883 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
11886 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11889 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
11891 2017-10-09 Jakub Jelinek <jakub@redhat.com>
11894 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
11895 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
11897 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
11899 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
11900 (WI_BINARY_PREDICATE_RESULT): Likewise.
11901 (wi::binary_traits::operator_result): New type.
11902 (wi::binary_traits::predicate_result): Likewise.
11903 (generic_wide_int::operator~, unary generic_wide_int::operator-)
11904 (generic_wide_int::operator==, generic_wide_int::operator!=)
11905 (generic_wide_int::operator&, generic_wide_int::and_not)
11906 (generic_wide_int::operator|, generic_wide_int::or_not)
11907 (generic_wide_int::operator^, generic_wide_int::operator+
11908 (binary generic_wide_int::operator-, generic_wide_int::operator*):
11910 (operator~, unary operator-, operator==, operator!=, operator&)
11911 (operator|, operator^, operator+, binary operator-, operator*): New
11913 * expr.c (get_inner_reference): Use wi::bit_and_not.
11914 * fold-const.c (fold_binary_loc): Likewise.
11915 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
11916 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
11917 (bit_value_binop): Likewise.
11918 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
11919 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
11920 (extract_range_from_binary_expr_1): Likewise.
11921 (masked_increment): Likewise.
11922 (simplify_bit_ops_using_ranges): Likewise.
11924 2017-10-09 Martin Jambor <mjambor@suse.cz>
11927 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
11928 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
11929 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
11931 (hsa_fixup_mov_insn_type): New function.
11932 (hsa_op_with_type::get_in_type): Use it.
11933 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
11934 immediates in an assert.
11935 (hsa_op_with_type::extend_int_to_32bit): New method.
11936 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
11937 types. Convert to dest type if necessary.
11938 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
11939 (reg_for_gimple_ssa): Pass false as min32int to
11940 hsa_type_for_scalar_tree_type.
11941 (gen_hsa_addr): Fixup type when creating addresable temporary.
11942 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
11943 (gen_hsa_unary_operation): Extend operands and convert to dest type if
11944 necessary. Call hsa_fixup_mov_insn_type.
11945 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
11946 extend operands and convert to dest type if necessary.
11947 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
11948 to dest type if necessary.
11949 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
11950 if conversion nt necessary and size matches.
11951 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
11952 to dest type if necessary.
11953 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
11954 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
11956 (gen_hsa_clrsb): Likewise.
11957 (gen_hsa_ffs): Likewise.
11958 (gen_hsa_divmod): Extend operands and convert to dest type if
11960 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
11962 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
11964 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
11965 Remove empty default arguments. Use a brace block as output
11967 (conditional return): Ditto.
11969 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
11970 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
11972 (group_ending_nop): Ditto.
11973 (doloop_end): Ditto.
11974 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
11975 (splitters for those): Ditto.
11977 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
11979 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
11980 a conditional jump (and the compare for it) so that pc_rtx is the
11982 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
11983 for the deleted and renamed ctr<mode>_internal[234] patterns.
11984 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
11985 Delete second conditional return pattern.
11986 (ctr<mode>_internal2): Delete this second bdnz pattern.
11987 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
11988 (ctr<mode>_internal4): Delete this second bdz pattern.
11990 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
11992 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
11993 (always_initialized_rtx_for_ssa_name_p): New predicate.
11994 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
11995 (finish_out_of_ssa): Free new field of SA.
11996 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
11997 * tree-ssa-coalesce.c: Include tree-ssa.h.
11998 (get_parm_default_def_partitions): Remove extern keyword.
11999 (get_undefined_value_partitions): New function.
12000 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
12001 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
12002 uninitialized bits.
12003 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
12005 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12007 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
12009 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12011 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
12012 for targets that preffer 128bit.
12014 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12016 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
12018 2017-10-08 Olivier Hainque <hainque@adacore.com>
12020 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
12021 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
12023 (thumb_set_return_address): Likewise.
12025 2017-10-08 Olivier Hainque <hainque@adacore.com>
12027 * common/config/arm/arm-common.c (arm_except_unwind_info):
12028 Handle DWARF2_UNWIND_INFO.
12030 2017-10-07 Michael Collison <michael.collison@arm.com>
12032 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
12035 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
12037 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
12038 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
12039 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
12040 defined, force the creation of a new block for a dispatch label.
12042 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
12044 * invoke.texi (Wsuggest-attribute=cold): Document.
12045 * common.opt (Wsuggest-attribute=cold): New
12046 * ipa-pure-const.c (warn_function_cold): New function.
12047 * predict.c (compute_function_frequency): Use it.
12048 * predict.h (warn_function_cold): Declare.
12050 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
12052 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
12055 2017-10-06 Martin Liska <mliska@suse.cz>
12057 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
12058 keyword for member functions.
12059 (struct sanopt_tree_couple): New struct.
12060 (struct sanopt_tree_couple_hash): New function.
12061 (struct sanopt_ctx): Add new hash_map.
12062 (has_dominating_ubsan_ptr_check): New function.
12063 (record_ubsan_ptr_check_stmt): Likewise.
12064 (maybe_optimize_ubsan_ptr_ifn): Likewise.
12065 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
12066 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
12068 2017-10-06 Sudakshina Das <sudi.das@arm.com>
12071 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
12072 aarch64_simd_valid_immediate on CONST_VECTORs.
12073 (aarch64_reg_or_bic_imm): Likewise.
12075 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
12077 PR rtl-optimization/82396
12078 * haifa-sched.c (ready_sort_real): Disable qsort checking.
12080 2017-10-06 Sebastian Pop <sebpop@gmail.com>
12082 * graphite-dependences.c (scop_get_reads): Move code to...
12083 (scop_get_must_writes): Move code to...
12084 (scop_get_may_writes): Move code to...
12085 (scop_get_reads_and_writes): ... here.
12086 (scop_get_dependences): Call scop_get_reads_and_writes.
12088 2017-10-06 Jakub Jelinek <jakub@redhat.com>
12090 PR tree-optimization/82434
12091 * fold-const.h (can_native_encode_type_p,
12092 can_native_encode_string_p): Remove.
12093 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
12094 don't encode anything, just return what would be otherwise returned.
12095 (native_encode_fixed, native_encode_complex, native_encode_vector):
12097 (native_encode_string): Likewise. Inline by hand
12098 can_native_encode_string_p.
12099 (can_native_encode_type_p): Remove.
12100 (can_native_encode_string_p): Remove.
12101 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
12102 STRING_CSTs using can_native_encode_string_p, test all
12103 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
12104 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
12105 argument from native_encode_expr.
12106 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
12107 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
12110 2017-10-06 Richard Biener <rguenther@suse.de>
12112 PR tree-optimization/82397
12113 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
12114 operand_equal_p but rely on data_ref_compare_tree for detecting
12116 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
12117 to match up with dr_group_sort_cmp.
12119 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12122 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
12124 * config/s390/s390-builtin-types.def: Regenerate.
12126 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12129 * config/s390/s390-builtin-types.def: Regenerate.
12130 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
12131 Change flag from B_VXE to B_VX.
12132 (s390_vec_min_dbl): Remove B_VXE flag.
12134 2017-10-06 Richard Biener <rguenther@suse.de>
12136 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
12137 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
12138 translate_isl_ast_to_gimple::is_valid_rename,
12139 translate_isl_ast_to_gimple::get_rename,
12140 translate_isl_ast_to_gimple::get_def_bb_for_const,
12141 translate_isl_ast_to_gimple::get_new_name,
12142 translate_isl_ast_to_gimple::collect_all_ssa_names,
12143 translate_isl_ast_to_gimple::copy_loop_phi_args,
12144 translate_isl_ast_to_gimple::collect_all_ssa_names,
12145 translate_isl_ast_to_gimple::copy_loop_phi_args,
12146 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
12147 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
12148 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
12149 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
12150 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
12151 translate_isl_ast_to_gimple::copy_cond_phi_args,
12152 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
12153 translate_isl_ast_to_gimple::edge_for_new_close_phis,
12154 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
12155 translate_isl_ast_to_gimple::rename_uses,
12156 translate_isl_ast_to_gimple::rename_all_uses): Remove.
12157 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
12158 (set_rename_for_each_def): Likewise.
12159 (graphite_copy_stmts_from_block): Handle debug stmt resetting
12160 here. Handle rewriting SCEV analyzable uses here.
12161 (copy_bb_and_scalar_dependences): Generate code for PHI
12163 (graphite_regenerate_ast_isl): Adjust.
12164 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
12165 (add_write, add_read): New functions.
12166 (build_cross_bb_scalars_def): Use it and simplify.
12167 (build_cross_bb_scalars_use): Likewise.
12168 (graphite_find_cross_bb_scalar_vars): Inline into...
12169 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
12170 simulating out-of-SSA. Compute liveout and add dependencies.
12171 (build_scops): Force an empty entry block.
12172 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
12174 (sese_build_liveouts): Declare.
12175 (sese_trivially_empty_bb_p): Likewise.
12176 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
12177 compute liveout and debug_liveout.
12178 (sese_bad_liveouts_use): Remove.
12179 (sese_reset_debug_liveouts_bb): Likewise.
12180 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
12181 (sese_build_liveouts): Build liveout and debug_liveout and store
12183 (new_sese_info): Adjust.
12184 (free_sese_info): Likewise.
12185 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
12186 do not build liveout here.
12187 (move_sese_in_condition): Adjust region entry.
12188 (scev_analyzable_p): Match up with chrec_apply requirements.
12189 (sese_trivially_empty_bb_p): New.
12190 * tree-into-ssa.c (get_reaching_def): Properly support generating
12191 default-defs for incremental rewrite of anonymous names.
12193 2017-10-06 Richard Biener <rguenther@suse.de>
12195 * graphite-sese-to-poly.c (extract_affine): For casts increasing
12196 precision do not perform modulo reduction.
12198 2017-10-06 Richard Biener <rguenther@suse.de>
12200 PR tree-optimization/82436
12201 * tree-vect-slp.c (vect_supported_load_permutation_p): More
12202 conservatively choose the vectorization factor when checking
12203 whether we can perform the required load permutation.
12204 (vect_transform_slp_perm_load): Assert when we may not fail.
12206 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
12208 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
12209 message for incompatible -msdata=* and -mcall-* options.
12211 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12213 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
12214 rate for post-reload scheduling.
12216 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12218 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
12220 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12222 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
12223 to improve monte carlo in scimark.
12225 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12227 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
12228 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
12229 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
12230 pentium4_cost, nocona_cost): Set reassociation width to 1.
12231 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
12232 width to 2 for fp operations and 1 otherwise.
12233 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
12235 (atom_cost): Set reassociation width to 2.
12236 (slm_cost, generic_cost): Set fp reassociation width
12237 to 2 and 1 otherwise.
12238 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
12239 (core_cost): Set fp reassociation width to 4 and vector to 2.
12240 (ix86_reassociation_width): Rewrite using cost table; special case
12241 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
12242 and TARGET_AVX128_OPTIMAL.
12243 * config/i386/i386.h (processor_costs): Add
12244 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
12245 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
12246 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
12247 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
12248 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
12249 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
12251 2017-10-05 Nathan Sidwell <nathan@acm.org>
12253 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
12255 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12257 * config/arm/arm.c (arm_test_fpu_data): New.
12258 (arm_run_selftests): Call arm_test_fpu_data.
12260 2017-10-04 Nathan Sidwell <nathan@acm.org>
12262 * toplev.c (toplev::main): Remove excess parens on pretty_printer
12264 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
12266 2017-10-04 Sudakshina Das <sudi.das@arm.com>
12268 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
12269 check type for aarch64_simd_valid_immediate.
12270 (aarch64_output_simd_mov_immediate): Update prototype.
12271 (aarch64_simd_valid_immediate): Update prototype.
12272 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
12273 support for ORR-immediate.
12274 (and<mode>3): modified pattern to add support for BIC-immediate.
12275 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
12276 now checks for valid immediate for BIC and ORR based on new enum
12278 (aarch64_output_simd_mov_immediate): Function now used to output
12279 BIC/ORR imm as well based on new enum argument.
12280 * config/aarch64/constraints.md (Do): New vector immediate constraint.
12282 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
12283 (aarch64_reg_or_bic_imm): Likewise.
12285 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12287 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
12288 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
12290 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12294 PR rtl-optimization/82396
12295 * haifa-sched.c (autopref_multipass_init): Simplify
12297 (autopref_rank_data): Simplify sort order.
12298 * sched-int.h (autopref_multipass_data_): Remove
12299 multi_mem_insn_p, min_offset and max_offset.
12301 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12303 * doc/sourcebuild.texi: Document vect_peeling_profitable.
12305 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12307 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
12308 vect_doubleint_cvt.
12310 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12312 * doc/sourcebuild.texi: Document vect_long_mult.
12314 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
12316 PR tree-optimization/82413
12317 * fold-const.c (build_range_check): Use widest_int when comparing
12318 the maximum ETYPE value with HIGH.
12320 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12322 PR rtl-optimization/82396
12323 * haifa-sched.c (autopref_multipass_init): Simplify
12325 (autopref_rank_data): Simplify sort order.
12326 * sched-int.h (autopref_multipass_data_): Remove
12327 multi_mem_insn_p, min_offset and max_offset.
12329 2017-10-04 Jakub Jelinek <jakub@redhat.com>
12331 PR tree-optimization/82381
12332 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
12333 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
12336 PR tree-optimization/82374
12337 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
12338 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
12339 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
12340 current_function_decl to the new decl.
12342 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
12344 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
12345 helper macro for IEEE float128 hardware built-in functions.
12346 (SQRTF128_ODD): Add built-in functions with the round-to-odd
12348 (TRUNCF128_ODD): Likewise.
12349 (ADDF128_ODD): Likewise.
12350 (SUBF128_ODD): Likewise.
12351 (MULF128_ODD): Likewise.
12352 (DIVF128_ODD): Likewise.
12353 (FMAF128_ODD): Likewise.
12354 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
12355 UNSPEC_TRUNC_ROUND_TO_ODD.
12356 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
12357 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
12358 floating point round to odd instructions.
12359 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
12360 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
12361 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
12362 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
12363 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
12364 (trunc<mode>sf2_hw): Change the truncate with round to odd
12365 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
12366 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
12367 to odd hardware instructions.
12368 (sub<mode>3_odd): Likewise.
12369 (mul<mode>3_odd): Likewise.
12370 (div<mode>3_odd): Likewise.
12371 (sqrt<mode>2_odd): Likewise.
12372 (fma<mode>4_odd): Likewise.
12373 (fms<mode>4_odd): Likewise.
12374 (nfma<mode>4_odd): Likewise.
12375 (nfms<mode>4_odd): Likewise.
12376 (trunc<mode>df2_odd): Change the truncate with round to odd
12377 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
12379 * doc/extend.texi (PowerPC built-in functions): Update documentation
12380 for existing IEEE float128-bit built-in functions. Add built-in
12381 functions that generate the IEEE 128-bit floating point round to
12384 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
12386 PR rtl-optimization/77729
12387 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
12388 to (X&(C1&~C2))|C2 transformations.
12390 2017-10-03 Martin Jambor <mjambor@suse.cz>
12392 PR tree-optimization/82363
12393 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
12394 mismatch, mark lacc written regardless of racc.
12396 2017-10-03 Jakub Jelinek <jakub@redhat.com>
12398 PR tree-optimization/82381
12399 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
12400 stmt_to_insert nor wheather SSA_NAMEs are default defs.
12401 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
12402 fallthrough into reassoc_stmt_dominates_stmt_p.
12405 * combine.c (combine_instructions): Don't combine in unreachable
12408 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
12411 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
12412 function to not use the have_cpu variable. Do not set cpu_index,
12413 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
12414 or the default cpu.
12415 (rs6000_valid_attribute_p): Remove duplicate initializations of
12416 old_optimize and func_optimize.
12417 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
12418 (rs6000_activate_target_options): Make global.
12419 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
12422 2017-10-02 Jakub Jelinek <jakub@redhat.com>
12424 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
12425 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
12426 Set *poffset to 0 and *psize and *pmax_size to -1 if
12427 *poffset + *psize overflows in HOST_WIDE_INT.
12429 PR tree-optimization/82387
12430 PR tree-optimization/82388
12431 PR tree-optimization/82389
12432 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
12433 instead of live_bytes non-NULL.
12435 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
12438 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
12441 2017-10-02 Richard Biener <rguenther@suse.de>
12443 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
12444 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
12445 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
12447 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12449 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
12450 requested precision matches the type's.
12451 * calls.c (alloc_max_size): Calculate the new candidate size as
12452 a widest_int and use wi::to_widest when comparing it with the
12453 current candidate size.
12454 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
12455 zero rather than integer_zero_node.
12456 * match.pd: Check for a no-op conversion before using wi::add
12457 rather than after. Use tree_to_uhwi when summing small shift
12458 counts into an unsigned int.
12460 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12461 Alan Hayward <alan.hayward@arm.com>
12462 David Sherwood <david.sherwood@arm.com>
12465 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
12466 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
12467 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
12468 POINTER_AND_FP_REGS.
12470 2017-10-02 Richard Biener <rguenther@suse.de>
12472 PR tree-optimization/82355
12473 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12474 a mapping for the enclosing loop but avoid generating one for
12475 the loop tree root.
12476 (copy_bb_and_scalar_dependences): Remove premature codegen
12477 error on PHIs in blocks duplicated into multiple places.
12478 * graphite-scop-detection.c
12479 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12480 in the region use it as loop and nest to analyze the DR in.
12481 (try_generate_gimple_bb): Likewise.
12482 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12483 (add_loop_constraints): For blocks in a loop not in the region
12484 create a dimension with a single iteration.
12485 * sese.h (gbb_loop_at_index): Remove assert.
12487 2017-10-01 Kevin Buettner <kevinb@redhat.com>
12489 * omp-expand.c (adjust_context_scope): New function.
12490 (expand_parallel_call): Call adjust_context_scope.
12492 2017-10-01 Jeff Law <law@redhat.com>
12494 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
12495 dom_opt_dom_walker class with direct access to private members.
12496 Add comments. Call test_for_singularity.
12497 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
12498 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
12499 m_dummy_cond anymore.
12500 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
12502 (pass_dominator:execute): Build the dummy_cond here and pass it
12503 to the dom_opt_dom_walker ctor.
12504 (test_for_singularity): New function.
12506 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
12507 Maya Rashish <coypu@sdf.org>
12509 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
12510 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
12511 (alpha*-*-netbsd*) Use nbsd_tm_file.
12512 (arm*-*-netbsdelf*) Likewise.
12513 (i[34567]86-*-netbsdelf*) Likewise.
12514 (x86_64-*-netbsd*) Likewise.
12515 (mips*-*-netbsd*) Likewise.
12516 (powerpc-*-netbsd*) Likewise.
12517 (sh*-*-netbsd*) Likewise.
12518 (sparc-*-netbsdelf*) Likewise.
12519 (sparc64-*-netbsd*) Likewise.
12520 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
12522 (vax-*-netbsdelf*) Likewise.
12523 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
12524 (UINT_FAST8_TYPE) Likewise.
12525 (INT_FAST16_TYPE) Check CHAR_FAST16.
12526 (UINT_FAST16_TYPE) Likewise.
12528 2017-09-30 Jakub Jelinek <jakub@redhat.com>
12531 * config/i386/i386.md
12532 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
12533 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
12534 *divmodsi4_zext_2): New define_insn_and_split.
12535 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
12536 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
12537 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
12538 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
12539 New define_insn_and_split.
12540 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
12541 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
12542 operands[1] having DImode when mode is SImode.
12544 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
12545 always SImode for DIV and MOD in REG_EQUAL notes.
12547 2017-09-29 Yury Gribov <tetra2005@gmail.com>
12549 PR middle-end/82319
12550 * match.pd: Fix handling of NaNs in pattern.
12552 2017-09-29 Jeff Law <law@redhat.com>
12554 * sbitmap.c (bitmap_bit_in_range_p): New function.
12555 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
12556 * tree-ssa-dse.c (live_bytes_read): New function.
12557 (dse_classify_store): Ignore reads of dead bytes.
12559 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
12560 typos and whitespace errors.
12561 * config/i386/predicates.md (address_no_seg_operand): Likewise.
12562 * config/s390/s390.c (s390_emit_prologue): Likewise.
12564 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12567 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
12568 with a symbol for LRA.
12570 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12572 PR rtl-optimization/82338
12573 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
12575 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
12577 * genmodes.c (calc_wider_mode): Suppress qsort macro.
12578 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
12579 (qsort_chk): Declare.
12580 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
12581 (qsort_chk): New function.
12583 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12585 PR tree-optimization/82337
12586 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12587 phi definition if the PHI result appears in an abnormal PHI.
12588 (find_basis_for_base_expr): Don't record a basis if the LHS of the
12589 basis appears in an abnormal PHI.
12591 2017-09-29 Richard Biener <rguenther@suse.de>
12593 * graphite-isl-ast-to-gimple.c
12594 (translate_isl_ast_to_gimple::set_codegen_error): New function.
12595 (binary_op_to_tree): Use it.
12596 (get_rename_from_scev): Likewise.
12597 (copy_loop_phi_nodes): Likewise.
12598 (copy_bb_and_scalar_dependences): Likewise.
12599 (translate_pending_phi_nodes): Likewise.
12601 2017-09-29 Jakub Jelinek <jakub@redhat.com>
12604 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
12605 for movabsq $(i32 << shift), r64.
12607 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
12609 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
12610 index when encoding %esp as %rsp to avoid 0x67 prefix.
12612 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
12614 * config/i386/i386.md (*movsf_internal, *movdf_internal):
12615 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
12617 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
12619 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
12620 Extensions with more than 16 double VFP registers.
12621 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
12622 to_clear_mask and all code related to it. Replace the remaining
12623 entry by a sbitmap and adapt code accordingly.
12625 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
12627 * brig-builtins.def: Change pure attributes to const.
12629 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12631 * config.gcc (default_gnu_indirect_function): Default to yes for
12632 sparc*-*-linux* with glibc.
12634 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12636 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
12637 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
12638 when creating .init_array and .fini_array sections with priority
12641 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
12644 * config/aarch64/aarch64.c
12645 (aarch64_builtin_support_vector_misalignment): Always return false
12646 when misalignment is unknown.
12648 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12650 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
12651 this function to return false if the definition used by the swap
12652 instruction is artificial, or if the memory address from which the
12653 constant value is loaded is not represented by a base address held
12654 in a register or if the base address register is a frame or stack
12655 pointer. Additionally, return false if the base address of the
12656 loaded constant is a SYMBOL_REF but is not considered to be a
12658 (replace_swapped_load_constant): New function.
12659 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
12660 loaded constant vector with a load of a swapped constant vector.
12662 2017-09-27 Carl Love <cel@us.ibm.com>
12664 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
12665 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
12666 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
12669 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
12671 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
12672 first, always call autopref_rank_data otherwise.
12674 2017-09-27 Richard Biener <rguenther@suse.de>
12676 * graphite-scop-detection.c (find_scop_parameters): Move
12677 loop bound handling ...
12678 (gather_bbs::before_dom_children): ... here, avoiding the need
12679 to build scop_info->loop_nest.
12680 (record_loop_in_sese): Remove.
12681 * sese.h (sese_info_t::loop_nest): Remove.
12682 * sese.c (new_sese_info): Do not allocate loop_nest.
12683 (free_sese_info): Do not free loop_nest.
12685 2017-09-27 Jakub Jelinek <jakub@redhat.com>
12688 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
12689 lhs from calls if the lhs has addressable type.
12691 2017-09-27 Richard Biener <rguenther@suse.de>
12693 * graphite.h (scop::max_alias_set): New member.
12694 * graphite-scop-detection.c: Remove references to non-existing
12695 --param in comments.
12696 (build_alias_sets): Record the maximum alias set used for drs.
12697 (build_scops): Support zero as unlimited for
12698 --param graphite-max-arrays-per-scop.
12699 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
12700 and inline into ...
12701 (build_poly_sr_1): ... here. Compute alias set based on the
12702 maximum alias set used for drs rather than
12703 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
12705 2017-09-27 Richard Biener <rguenther@suse.de>
12707 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
12708 --param loop-block-tile-size=0 to disable tiling.
12710 2017-09-27 Richard Biener <rguenther@suse.de>
12712 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
12713 (graphite-max-nb-scop-params): Document special value zero.
12714 * domwalk.h (dom_walker::STOP): New symbolical constant.
12715 (dom_walker::dom_walker): Add optional parameter for bb to
12717 (dom_walker::~dom_walker): Declare.
12718 (dom_walker::before_dom_children): Document STOP return value.
12719 (dom_walker::m_user_bb_to_rpo): New member.
12720 (dom_walker::m_bb_to_rpo): Likewise.
12721 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
12722 mapping here if not provided by the user.
12723 (dom_walker::~dom_walker): Free bb to RPO mapping if not
12724 provided by the user.
12725 (dom_walker::STOP): Define.
12726 (dom_walker::walk): Do not compute bb to RPO mapping here.
12727 Support STOP return value from before_dom_children to stop
12729 * graphite-optimize-isl.c (optimize_isl): If the schedule
12730 is the same still generate code if -fgraphite-identity
12731 or -floop-parallelize-all are given.
12732 * graphite-scop-detection.c: Include cfganal.h.
12733 (gather_bbs::gather_bbs): Get and pass through bb to RPO
12735 (gather_bbs::before_dom_children): Return STOP for BBs
12737 (build_scops): Compute bb to RPO mapping and pass it to
12738 the domwalk. Treat --param graphite-max-nb-scop-params=0
12739 as not limiting the number of params.
12740 * graphite.c (graphite_initialize): Remove limit on the
12741 number of basic-blocks in a function.
12742 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
12743 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
12744 default value of 10.
12746 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12748 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
12749 Adjust code to eliminate needing to do the shift right 32-bits
12750 operation after XSCVDPSPN.
12752 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12754 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
12755 ((X / Y) != 0 -> X >= Y): Likewise.
12757 2017-09-26 Carl Love <cel@us.ibm.com>
12759 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
12760 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
12761 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
12762 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
12763 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
12764 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
12765 definitions and overloading.
12766 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
12767 statement for P9V_BUILTIN_XST_LEN_R.
12768 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
12769 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
12770 define_expand and define_insn for the instructions and builtins.
12771 * doc/extend.texi: Update the built-in documentation file for the new
12772 built-in functions.
12773 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
12774 define_insn for the instructions
12776 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
12779 * gcc/config/netbsd-protos.h: New file.
12780 * gcc/config/netbsd.c: New file.
12781 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
12782 * gcc/config/t-netbsd: New file.
12783 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
12784 (tmake_file) Add t-netbsd.
12785 (extra_objs) Add netbsd.o.
12787 2017-09-26 Janus Weil <janus@gcc.gnu.org>
12791 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
12793 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12795 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
12796 sign extension from a vector register to a GPR by doing a 32-bit
12797 direct move and then an EXTSW.
12798 (extendsi<mode>2 splitter): Likewise.
12799 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
12800 right or vector extract after doing XSCVDPSPN. Use
12801 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
12803 (movdi_from_sf_zero_ext): Likewise.
12804 (reload_gpr_from_vsxsf): Likewise.
12805 (p8_mfvsrd_4_disf): Delete, no longer used.
12806 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
12807 then needing to move the SFmode to a GPR to use the XSCVDPSP
12808 instruction instead of FRSP and XSCVDPSPN.
12809 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
12810 it is adjacent to the other XSCVSPDP insns.
12811 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
12812 SFmode to be in traditional Altivec registers.
12813 (vsx_xscvdpspn): Eliminate useless alternative constraint.
12814 (vsx_xscvspdpn): Likewise.
12815 (vsx_xscvspdpn_scalar): Likewise.
12817 2017-09-26 Martin Jambor <mjambor@suse.cz>
12819 * tree-sra.c (compare_access_positions): Put integral types first,
12820 stabilize sorting of integral types, remove conditions putting
12821 non-full-precision integers last.
12822 (sort_and_splice_var_accesses): Disable scalarization if a
12823 non-integert would be represented by a non-full-precision integer.
12825 2017-09-26 Joseph Myers <joseph@codesourcery.com>
12827 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
12828 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
12829 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
12830 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
12831 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
12832 conditionals inside the function instead of around it. Call
12833 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
12834 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
12836 2017-09-26 Richard Biener <rguenther@suse.de>
12838 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
12840 (scop_detection::build_scop_breadth): ... this. Removed.
12841 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
12842 (scop_detection::harmful_stmt_in_bb): Likewise.
12843 (scop_detection::graphite_can_represent_stmt): Likewise.
12844 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
12845 (scop_detection::can_represent_loop): Remove recursion, fold in ...
12846 (scop_detection::can_represent_loop_1): ... this. Removed.
12847 (scop_detection::harmful_loop_in_region): Simplify after inlining
12848 the above and remove more quadraticness.
12849 (build_scops): Adjust.
12850 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
12853 2017-09-26 Jakub Jelinek <jakub@redhat.com>
12856 * config/i386/i386.c (ix86_print_operand_address_as): Only test
12857 REGNO (base) == SP_REG if base is a REG.
12859 PR middle-end/35691
12860 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
12861 if it is different SSA_NAME.
12862 (optimize_range_tests_cmp_bitwise): New function.
12863 (optimize_range_tests): Call it.
12865 2017-09-26 Richard Biener <rguenther@suse.de>
12867 PR tree-optimization/82321
12868 * graphite.c (canonicalize_loop_closed_ssa): Properly check
12869 for the def being inside the loop.
12871 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12873 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
12875 * config/s390/s390-builtins.def: Fix constraint on op4.
12877 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12879 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
12880 independent expanders.
12881 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
12882 ("vec_ordered", "vec_unordered"): New expanders.
12884 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12886 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
12889 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12891 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
12892 vec_unpacks_lo_v16qi.
12893 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
12895 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12897 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
12898 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
12899 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
12901 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12903 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
12905 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
12907 ("*vec_slb<mode>"): New insn pattern.
12908 ("vec_shr_<mode>"): New expander.
12909 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
12910 and force the shift count operand to V16QImode.
12911 ("vec_srb<mode>"): Set shift count mode to V16QI.
12913 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12915 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
12916 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
12917 ("vec_widen_smult_hi_<mode>"): New expander definitions.
12919 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
12922 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
12924 2017-09-26 Richard Biener <rguenther@suse.de>
12926 PR tree-optimization/82320
12927 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
12930 2017-09-25 Jeff Law <law@redhat.com>
12932 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
12933 prototype for new argument.
12934 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
12935 mostly extracted from rs6000_emit_allocate_stack.
12936 (rs6000_emit_probe_stack_range_stack_clash): New function.
12937 (rs6000_emit_allocate_stack): Call
12938 rs6000_emit_probe_stack_range_stack_clash as needed.
12939 (rs6000_emit_probe_stack_range): Add additional argument
12940 to call to gen_probe_stack_range{si,di}.
12941 (output_probe_stack_range): New.
12942 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
12943 (output_probe_stack_range_stack_clash): New.
12944 (rs6000_emit_prologue): Emit notes into dump file as requested.
12945 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
12946 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
12947 Add additional operand and pass it to output_probe_stack_range.
12949 2017-09-25 Bin Cheng <bin.cheng@arm.com>
12951 PR tree-optimization/82163
12952 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
12953 (checking_verify_loop_closed_ssa): New parameter.
12954 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
12955 (check_loop_closed_ssa_stmt): Delete.
12956 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
12957 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
12958 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
12961 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
12963 * brig-builtins.def: Treat HSAIL barrier builtins as
12964 setjmp/longjump style functions.
12966 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
12968 * target.def (constant_alignment): New hook.
12969 * defaults.h (CONSTANT_ALIGNMENT): Delete.
12970 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
12971 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
12972 * doc/tm.texi: Regenerate.
12973 * targhooks.h (default_constant_alignment): Declare.
12974 (constant_alignment_word_strings): Likewise.
12975 * targhooks.c (default_constant_alignment): New function.
12976 (constant_alignment_word_strings): Likewise.
12977 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
12978 instead of CONSTANT_ALIGNMENT.
12979 * varasm.c (align_variable, get_variable_align, build_constant_desc)
12980 (force_const_mem): Likewise.
12981 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
12982 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
12983 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
12984 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12985 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
12987 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
12988 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12989 constant_alignment_word_strings.
12990 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
12991 (CONSTANT_ALIGNMENT): Likewise.
12992 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12993 (arm_constant_alignment): New function.
12994 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
12995 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12996 constant_alignment_word_strings.
12997 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
12998 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12999 constant_alignment_word_strings.
13000 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
13001 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13002 (cris_constant_alignment): New function.
13003 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
13004 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13005 (epiphany_constant_alignment): New function.
13006 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
13007 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13008 constant_alignment_word_strings.
13009 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
13010 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13011 constant_alignment_word_strings.
13012 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
13013 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13014 constant_alignment_word_strings.
13015 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
13016 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
13017 * config/i386/i386.c (ix86_constant_alignment): Make static.
13018 Use the same interface as the target hook.
13019 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13020 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
13021 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13022 constant_alignment_word_strings.
13023 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
13024 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
13025 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13026 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
13027 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13028 constant_alignment_word_strings.
13029 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
13030 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13031 constant_alignment_word_strings.
13032 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
13033 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13034 constant_alignment_word_strings.
13035 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
13036 * config/microblaze/microblaze.c (microblaze_constant_alignment):
13038 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13039 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
13040 * config/mips/mips.c (mips_constant_alignment): New function.
13041 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13042 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
13043 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
13044 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13045 (mmix_constant_alignment): Make static. Use the same interface
13046 as the target hook.
13047 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
13048 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13049 constant_alignment_word_strings.
13050 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
13051 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13052 constant_alignment_word_strings.
13053 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
13054 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13055 constant_alignment_word_strings.
13056 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
13057 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13058 (rs6000_constant_alignment): New function.
13059 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
13060 * config/riscv/riscv.c (riscv_constant_alignment): New function.
13061 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13062 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
13063 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13064 (rs6000_constant_alignment): New function.
13065 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
13066 * config/s390/s390.c (s390_constant_alignment): New function.
13067 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13068 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
13069 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13070 constant_alignment_word_strings.
13071 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
13072 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13073 (sparc_constant_alignment): New function.
13074 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
13075 * config/spu/spu.c (spu_constant_alignment): New function.
13076 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13077 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
13078 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13079 constant_alignment_word_strings.
13080 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
13081 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13082 constant_alignment_word_strings.
13083 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
13084 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13085 constant_alignment_word_strings.
13086 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
13087 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13088 (visium_constant_alignment): New function.
13089 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
13090 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13091 (xtensa_constant_alignment): New function.
13092 * system.h (CONSTANT_ALIGNMENT): Poison.
13094 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
13096 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13097 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
13098 (rs6000_builtin_valid_without_lhs): New helper function.
13099 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13100 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
13102 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13104 * target.h (vec_perm_indices): Use unsigned short rather than
13106 (auto_vec_perm_indices): Likewise.
13107 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
13108 Use unsigned int rather than unsigned char.
13109 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
13111 2017-09-25 Richard Biener <rguenther@suse.de>
13113 * cfgloop.h (sort_sibling_loops): Declare.
13114 * cfgloop.c (sort_sibling_loops_cmp): New helper.
13115 (sort_sibling_loops): New function sorting the sibling loop list
13117 * graphite.c (graphite_transform_loops): Sort sibling loops.
13119 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
13121 * target.def (vec_perm_const_ok): Change sel parameter to
13123 * optabs-query.c (can_vec_perm_p): Update accordingly.
13124 * doc/tm.texi: Regenerate.
13125 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
13126 auto_vec_perm_indices and remove separate nelt field.
13127 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
13128 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
13129 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
13130 (aarch64_expand_vec_perm_const): Update accordingly.
13131 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
13132 to vec_perm_indices.
13133 * config/arm/arm.c (expand_vec_perm_d): Change perm to
13134 auto_vec_perm_indices and remove separate nelt field.
13135 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
13136 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
13137 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
13139 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
13140 to vec_perm_indices.
13141 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
13142 sel to vec_perm_indices.
13143 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
13144 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
13145 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
13147 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
13150 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
13153 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
13154 on the FUNCTION_DECL function context if it has a DIE that is a
13157 2017-09-25 Richard Biener <rguenther@suse.de>
13159 PR tree-optimization/82285
13160 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
13163 2017-09-25 Tom de Vries <tom@codesourcery.com>
13167 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
13170 2017-09-25 Richard Biener <rguenther@suse.de>
13172 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
13173 ISL errors other than isl_error_quota happen. Dump if the
13174 schedule is the same.
13175 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
13176 errors instead of aborting inside ISL.
13178 2017-09-25 Iain Sandoe <iain@codesourcery.com>
13181 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
13182 of libgcc_eh for m64.
13183 * config/i386/darwin64.h: Likewise.
13185 2017-09-25 Richard Biener <rguenther@suse.de>
13187 PR middle-end/82144
13188 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
13189 attribute for incomplete types nor twice for complete ones.
13191 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
13194 * config/i386/i386.c (ix86_print_operand_address_as): Encode
13195 %esp as %rsp to avoid 0x67 prefix if there is no index or base
13198 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
13201 * config/i386/i386.opt (mprefer-avx256): Use
13202 ix86_target_flags variable.
13203 * config/i386/i386.c (ix86_target_string): Move
13204 -mprefer-avx256 to flag2_opts.
13206 2017-09-22 Jakub Jelinek <jakub@redhat.com>
13208 PR middle-end/35691
13209 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
13210 and x != -1 | y != -1 into (x & y) != -1.
13212 2017-09-22 Steve Ellcey <sellcey@cavium.com>
13214 * config.gcc: Add new case statement to set
13215 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
13216 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
13217 s390x-*-linux* case statements. Added aarch64 to the list of
13218 supported architectures.
13220 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13222 PR tree-optimization/82289
13223 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
13224 STMT_VINFO_RELEVANT_P.
13226 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13227 Alan Hayward <alan.hayward@arm.com>
13228 David Sherwood <david.sherwood@arm.com>
13230 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
13231 for VR_RANGE only; don't allow VR_ANTI_RANGE.
13232 (extract_range_from_binary_expr_1): Don't call
13233 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
13235 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13236 Alan Hayward <alan.hayward@arm.com>
13237 David Sherwood <david.sherwood@arm.com>
13239 * target.def (preferred_vector_alignment): New hook.
13240 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
13242 * doc/tm.texi: Regenerate.
13243 * targhooks.h (default_preferred_vector_alignment): Declare.
13244 * targhooks.c (default_preferred_vector_alignment): New function.
13245 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
13247 (DR_TARGET_ALIGNMENT): New macro.
13248 (aligned_access_p): Update commentary.
13249 (vect_known_alignment_in_bytes): New function.
13250 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
13252 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
13253 Calculate the misalignment based on the target alignment rather than
13255 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
13256 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
13257 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
13258 the target alignment, rather than masking the element misalignment
13259 with the number of elements in a vector. Also use the target
13260 alignment when calculating the maximum number of peels.
13261 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
13262 instead of TYPE_ALIGN_UNIT.
13263 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
13264 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
13265 (vect_create_addr_base_for_vector_ref): Update call accordingly.
13266 (vect_create_data_ref_ptr): Likewise.
13267 (vect_setup_realignment): Realign by ANDing with
13268 -DR_TARGET_MISALIGNMENT.
13269 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
13270 the number of peels based on DR_TARGET_ALIGNMENT.
13271 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
13272 with the guaranteed alignment boundary when deciding whether
13274 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
13275 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
13276 (ensure_base_align): Remove stmt_info parameter. Get the
13277 target base alignment from DR_TARGET_ALIGNMENT.
13278 (vectorizable_store): Update call accordingly. Interpret
13279 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
13281 (vectorizable_load): Likewise.
13283 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13284 Alan Hayward <alan.hayward@arm.com>
13285 David Sherwood <david.sherwood@arm.com>
13287 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
13288 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
13289 (vect_enhance_data_refs_alignment): Likewise.
13291 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13293 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13294 error. Only quit immediately if parsing is complete.
13295 (BEGIN): Initialize fatal_err and parse_done.
13296 (begin fpu, end fpu): Check number of arguments.
13297 (begin arch, end arch): Likewise.
13298 (begin cpu, end cpu): Likewise.
13299 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13300 (optalias): Likewise.
13302 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13304 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13305 * config/arm/arm-isa.h: Delete. Move definitions to ...
13306 * arm-cpus.in: ... here. Use new feature and fgroup values.
13307 * config/arm/arm.c (arm_option_override): Use lower case for feature
13309 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13310 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13311 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13313 (print_isa_bits_for): New function.
13314 (gen_isa): New function.
13315 (gen_comm_data): Use print_isa_bits_for.
13316 (define feature): New keyword.
13317 (define fgroup): New keyword.
13318 * config/arm/t-arm (TM_H): Remove.
13319 (GTM_H): Add arm-isa.h.
13320 (arm-isa.h): Add rule to generate file.
13321 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13322 case for feature bit names.
13324 2017-09-22 Richard Biener <rguenther@suse.de>
13326 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
13328 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
13329 print of no dependency loops ...
13330 * graphite.c (graphite_transform_loops): ... here.
13331 (canonicalize_loop_closed_ssa_form): Work from inner to outer
13333 (same_close_phi_node, remove_duplicate_close_phi,
13334 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
13335 (canonicalize_loop_closed_ssa): ... here and simplify.
13336 * graphite-optimize-isl.c: Include tree-vectorizer.h.
13337 (optimize_isl): Use dump_printf_loc to tell when we stopped
13338 optimizing because of an ISL timeout.
13340 2017-09-22 Richard Biener <rguenther@suse.de>
13342 PR tree-optimization/82291
13343 * tree-if-conv.c (predicate_mem_writes): Make sure to
13344 remove writes in blocks predicated with false.
13346 2017-09-22 Richard Biener <rguenther@suse.de>
13348 * sese.c: Include cfganal.h.
13349 (if_region_set_false_region): Remove.
13350 (create_if_region_on_edge): Likewise.
13351 (move_sese_in_condition): Re-implement without destroying
13354 2017-09-22 Richard Biener <rguenther@suse.de>
13356 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
13357 Verify both BBs contain loop PHI nodes before dispatching to
13358 copy_loop_phi_args.
13359 (graphite_regenerate_ast_isl): Do not recompute dominators,
13360 do not verify three times. Restructure for clarity.
13361 * graphite-scop-detection.c (same_close_phi_node,
13362 remove_duplicate_close_phi, make_close_phi_nodes_unique,
13363 defined_in_loop_p, canonicalize_loop_closed_ssa,
13364 canonicalize_loop_closed_ssa_form): Simplify, remove excess
13365 checking and SSA rewrite, move to ...
13366 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
13367 (graphite_initialize): Do not pass in ctx, do not reset the
13368 SCEV cache, compute only dominators.
13369 (graphite_transform_loops): Allocate ISL ctx after
13370 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
13371 Maintain post-dominators only around build_scops.
13372 * sese.c (if_region_set_false_region): Make static. Free
13373 and recompute dominators.
13374 (move_sese_in_condition): Assert we don't get called with
13375 post-dominators computed.
13376 * sese.h (if_region_set_false_region): Remove.
13378 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
13380 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
13381 mode attribute for TARGET_AVX512VL.
13383 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
13385 * config/i386/i386.opt (mprefer-avx256): New option.
13386 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
13388 (ix86_preferred_simd_mode): Return 256-bit AVX modes
13389 for TARGET_PREFER_AVX256.
13390 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
13392 2017-09-21 Jeff Law <law@redhat.com>
13394 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
13395 Fix dump output if the only stack space is for pushed registers.
13397 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13399 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
13402 2017-09-21 Martin Sebor <msebor@redhat.com>
13405 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
13406 code (in C++) or code that triggers warnings.
13408 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
13410 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
13412 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
13414 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
13415 * sched-rgn.c: Ditto.
13416 * sel-sched-ir.c: Ditto.
13418 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
13420 * toplev.h (set_random_seed): Adjust return type.
13421 * toplev.c (init_local_tick): Move eager initialization of random_seed
13422 to get_random_seed. Adjust comment.
13423 (init_random_seed): Inline to get_random_seed, delete.
13424 (get_random_seed): Initialize random_seed lazily.
13425 (set_random_seed): Do not return previous value.
13426 (print_switch_value): Do not call get_random_seed.
13428 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
13430 * cgraph.c (delete_function_version): New, broken out from...
13431 (cgraph_node::delete_function_version): ...here. Rename to
13432 cgraph_node::delete_function_version_by_decl. Update all uses.
13433 (cgraph_node::remove): Call delete_function_version.
13435 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13438 * tree-inline.c (expand_call_inline): Emit clobber stmts for
13439 VAR_DECLs to which addressable non-volatile parameters are mapped
13440 and for id->retvar after the return value assignment. Clear
13441 id->retval and id->retbnd after inlining.
13443 2017-09-21 Richard Biener <rguenther@suse.de>
13445 PR tree-optimization/82276
13446 PR tree-optimization/82244
13447 * tree-vrp.c (build_assert_expr_for): Set
13448 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
13450 (remove_range_assertions): Revert earlier change.
13452 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
13455 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
13457 2017-09-21 Richard Biener <rguenther@suse.de>
13459 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
13460 Restore valid IL after code generation errors.
13461 * graphite.c (graphite_transform_loops): Diagnose code
13462 generation issues as MSG_MISSED_OPTIMIZATION and continue
13463 with processing SCOPs.
13465 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13466 Alan Hayward <alan.hayward@arm.com>
13467 David Sherwood <david.sherwood@arm.com>
13469 * calls.c (compute_argument_addresses): Use simplify_gen_binary
13470 rather than choosing between plus_constant and gen_rtx_<CODE>.
13471 * expr.c (emit_push_insn): Likewise.
13472 (expand_expr_real_2): Likewise.
13474 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13475 Alan Hayward <alan.hayward@arm.com>
13476 David Sherwood <david.sherwood@arm.com>
13478 * loop-unroll.c (split_iv): Call copy_rtx on the step.
13480 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13481 Alan Hayward <alan.hayward@arm.com>
13482 David Sherwood <david.sherwood@arm.com>
13484 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
13485 calling tree_to_uhwi.
13487 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13488 Alan Hayward <alan.hayward@arm.com>
13489 David Sherwood <david.sherwood@arm.com>
13491 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
13492 INTEGER_CST rather than a negative test for ADDR_EXPR.
13494 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13495 Alan Hayward <alan.hayward@arm.com>
13496 David Sherwood <david.sherwood@arm.com>
13498 * tree-vrp.c (extract_range_from_binary_expr_1): Check
13499 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
13501 2017-09-21 Richard Biener <rguenther@suse.de>
13503 PR tree-optimization/71351
13504 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
13505 graphite_create_new_loop_guard): Remove, fold remaining parts
13507 (translate_isl_ast_node_for): ... here and simplify.
13509 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13512 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
13513 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
13514 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
13515 alternative always use QI mode, for -Os imov (=R,R) alternative
13516 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
13519 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13520 Jeff Law <law@redhat.com>
13522 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
13523 (allocate_stack_space): New function, partially extracted from
13524 s390_emit_prologue.
13525 (s390_emit_prologue): Track offset to most recent stack probe.
13526 Code to allocate space moved into allocate_stack_space.
13527 Dump actions when no stack is allocated.
13528 (s390_prologue_plus_offset): New function.
13529 (s390_emit_stack_probe): Likewise.
13531 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
13533 * common.opt (Wa, Wl, Wp, g, gz=): Add
13535 (gno-column-info): Remove.
13536 (gcolumn-info): Drop RejectNegative.
13537 (gno-): New prefix.
13538 (gno-record-gcc-switches): Remove.
13539 (grecord-gcc-switches): Drop RejectNegative.
13540 (gno-split-dwarf): Remove.
13541 (gsplit-dwarf): Drop RejectNegative.
13542 (gno-strict-dwarf): Remove.
13543 (gstrict-dwarf): Drop RejectNegative.
13544 * config/darwin.opt (gfull, gused): Add RejectNegative.
13545 * dwarf2out.c (gen_producer_string): Drop
13546 gno-record-gcc-switches handler.
13547 * optc-gen.awk: Add g to prefixes with negative forms.
13548 * opts-common.c (remapping_prefix_p): New.
13549 (find_opt): Check it.
13550 (generate_canonical_option): Test g prefix.
13551 (option_map): Add -gno- mapping.
13552 (add_misspelling_candidates): Check remapping_prefix_p.
13554 2017-09-20 Jeff Law <law@redhat.com>
13556 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
13557 thinko in stack clash protection support.
13559 * explow.c (compute_stack_clash_protection_loop_data): Use
13560 CONST_INT_P instead of explicit test. Verify object is a
13561 CONST_INT_P before looking at INTVAL.
13562 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
13563 instead of explicit test.
13565 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
13568 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
13569 address instead of to r1 and r11.
13571 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
13573 * config.gcc: Support "knm".
13574 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
13575 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13577 * config/i386/i386.c (m_KNM): Define.
13578 (processor_target_table): Add "knm".
13580 (ix86_option_override_internal): Add "knm".
13581 (ix86_issue_rate): Add PROCESSOR_KNM.
13582 (ix86_adjust_cost): Ditto.
13583 (ia32_multipass_dfa_lookahead): Ditto.
13584 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
13585 (fold_builtin_cpu): Add M_INTEL_KNM.
13586 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
13587 (processor_type): Add PROCESSOR_KNM.
13588 * config/i386/x86-tune.def: Add m_KNM.
13589 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
13591 2017-09-20 Richard Biener <rguenther@suse.de>
13593 PR tree-optimization/80213
13594 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
13595 are allowed in empty BBs as well.
13596 (canonicalize_loop_closed_ssa): Also look for other complex
13598 (scop_detection::get_sese): Include the loop-closed PHI block
13600 (scop_detection::merge_sese): Remove code adding extra blocks.
13601 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
13602 (build_scops): Assert the final returned scop is invalid.
13604 2017-09-20 Richard Biener <rguenther@suse.de>
13606 PR tree-optimization/82264
13607 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
13609 (vn_phi_lookup): Likewise.
13610 (vn_phi_insert): Likewise.
13612 2017-09-20 Jakub Jelinek <jakub@redhat.com>
13614 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
13615 that fits into uhwi or shwi, add DW_AT_const_value regardless
13616 of early_dwarf without going through RTL, using add_AT_unsigned
13619 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
13620 (DEBUG_LTO_ABBREV_SECTION): Likewise.
13621 (DEBUG_LTO_MACINFO_SECTION): Likewise.
13622 (DEBUG_MACRO_SECTION): Likewise.
13623 (DEBUG_LTO_MACRO_SECTION): Likewise.
13624 (DEBUG_STR_DWO_SECTION): Likewise.
13625 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
13626 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
13627 (DEBUG_LTO_DWO_LINE_SECTION): Define.
13628 (DEBUG_LTO_LINE_STR_SECTION): Define.
13629 (init_sections_and_labels): Initialize debug_line_str_section
13630 variable. Initialize debug_loc_section for -gdwarf-5 to
13631 DEBUG_LOCLISTS_SECTION. Formatting fixes.
13633 2017-09-20 Richard Biener <rguenther@suse.de>
13635 * graphite-sese-to-poly.c (extract_affine): Properly handle
13636 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
13638 2017-09-20 Richard Biener <rguenther@suse.de>
13640 PR tree-optimization/81373
13641 * graphite-scop-detection.c (build_cross_bb_scalars_def):
13642 Force SESE live-out defs to be handled even if they are
13645 2017-09-19 Jeff Law <law@redhat.com>
13647 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
13648 nothing for stack adjustments with REG_STACK_CHECK.
13649 * sched-deps.c (parse_add_or_inc): Reject insns with
13650 REG_STACK_CHECK from dependency breaking.
13651 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
13652 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
13653 * reg-notes.def (STACK_CHECK): New note.
13655 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
13656 (ix86_expand_prologue): Dump stack clash info as needed.
13657 Call ix86_adjust_stack_and_probe_stack_clash as needed.
13659 * function.c (dump_stack_clash_frame_info): New function.
13660 * function.h (dump_stack_clash_frame_info): Prototype.
13661 (enum stack_clash_probes): New enum.
13663 * config/alpha/alpha.c (alpha_expand_prologue): Also check
13664 flag_stack_clash_protection.
13665 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
13666 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
13667 (arm_frame_pointer_required): Likewise.
13668 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
13669 (ia64_expand_prologue): Likewise.
13670 * config/mips/mips.c (mips_expand_prologue): Likewise.
13671 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
13672 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13673 (sparc_flat_expand_prologue): Likewise.
13674 * config/spu/spu.c (spu_expand_prologue): Likewise.
13676 * explow.c: Include "params.h".
13677 (anti_adjust_stack_and_probe_stack_clash): New function.
13678 (get_stack_check_protect): Likewise.
13679 (compute_stack_clash_protection_loop_data): Likewise.
13680 (emit_stack_clash_protection_loop_start): Likewise.
13681 (emit_stack_clash_protection_loop_end): Likewise.
13682 (allocate_dynamic_stack_space): Use get_stack_check_protect.
13683 Use anti_adjust_stack_and_probe_stack_clash.
13684 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
13685 (emit_stack_clash_protection_loop_start): Likewise.
13686 (emit_stack_clash_protection_loop_end): Likewise.
13687 * rtl.h (get_stack_check_protect): Prototype.
13688 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
13689 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
13690 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
13692 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
13694 * doc/tm.texi: Rebuilt.
13695 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
13696 get_stack_check_protect.
13697 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
13698 * config/arm/arm.c (arm_expand_prologue): Likewise.
13699 (arm_frame_pointer_required): Likewise.
13700 * config/i386/i386.c (ix86_expand_prologue): Likewise.
13701 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
13702 * config/mips/mips.c (mips_expand_prologue): Likewise.
13703 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
13704 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
13705 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13706 (sparc_flat_expand_prologue): Likewise.
13708 * common.opt (-fstack-clash-protection): New option.
13709 * flag-types.h (enum stack_check_type): Note difference between
13710 -fstack-check= and -fstack-clash-protection.
13711 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
13712 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
13713 * toplev.c (process_options): Issue warnings/errors for cases
13714 not handled with -fstack-clash-protection.
13715 * doc/invoke.texi (-fstack-clash-protection): Document new option.
13716 (-fstack-check): Note additional problem with -fstack-check=generic.
13717 Note that -fstack-check is primarily for Ada and refer users
13718 to -fstack-clash-protection for stack-clash-protection.
13719 Document new params for stack clash protection.
13721 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
13723 * config/i386/i386.c (ix86_split_long_move): Do not handle
13724 address used for LEA in a special way.
13726 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
13728 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
13730 2017-09-19 Martin Sebor <msebor@redhat.com>
13733 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
13734 of incompatible types.
13736 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
13738 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13739 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
13740 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13741 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
13743 2017-09-19 Richard Biener <rguenther@suse.de>
13745 PR tree-optimization/82244
13746 * tree-vrp.c (remove_range_assertions): Do not propagate
13747 a constant to abnormals but replace the assert with a copy.
13749 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
13751 PR rtl-optimization/57878
13752 PR rtl-optimization/68988
13753 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
13754 avoidance test involving non_reload_pseudos. Move frequency test
13755 below the general fragmentation avoidance test.
13757 2017-09-19 Richard Biener <rguenther@suse.de>
13759 PR tree-optimization/69728
13760 * graphite-sese-to-poly.c (schedule_error): New global.
13761 (add_loop_schedule): Handle empty domain by failing the
13763 (build_original_schedule): Handle schedule_error.
13765 2017-09-19 Richard Biener <rguenther@suse.de>
13767 * graphite-scop-detection.c (scop_detection::can_represent_loop):
13768 Do not iterate to sibling loops but only to siblings of inner
13771 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
13774 * config/m68k/m68k.md (moveq feeding equality comparison): Check
13775 that the registers are different.
13777 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
13779 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
13780 to processor_model and "amdfam17h" to arch_names_table.
13781 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
13783 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13786 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
13788 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13789 Alan Hayward <alan.hayward@arm.com>
13790 David Sherwood <david.sherwood@arm.com>
13792 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
13794 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
13796 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
13797 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
13798 vect_schedule_slp_instance.
13799 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
13800 Update call to vect_slp_analyze_node_operations. Simplify return
13802 (vect_slp_analyze_bb_1): Update call accordingly.
13803 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
13804 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
13805 (vect_schedule_slp): Update call accordingly.
13807 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13808 Alan Hayward <alan.hayward@arm.com>
13809 David Sherwood <david.sherwood@arm.com>
13811 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
13812 with types that aren't in fact scalar.
13814 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13816 * tree-vect-slp.c (vect_record_max_nunits): New function,
13818 (vect_build_slp_tree_1): ...here.
13819 (vect_build_slp_tree_2): Call it for phis too.
13821 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13823 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
13824 to vect_get_vec_def_for_operand when getting the mask operand.
13826 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13827 Alan Hayward <alan.hayward@arm.com>
13828 David Sherwood <david.sherwood@arm.com>
13830 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
13833 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13834 Alan Hayward <alan.hayward@arm.com>
13835 David Sherwood <david.sherwood@arm.com>
13837 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
13838 calculation for vector booleans.
13840 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13841 Alan Hayward <alan.hayward@arm.com>
13842 David Sherwood <david.sherwood@arm.com>
13844 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
13846 (vect_transform_stmt): ...here.
13847 (vect_analyze_stmt): Use it instead of calling
13848 vectorizable_live_operation directly.
13850 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
13852 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
13853 non-SIMT targets in acc vector loops.
13855 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
13857 * configure.ac: Add arc and check if assembler supports gdwarf2.
13858 * configure: Regenerate.
13860 2017-09-18 Richard Biener <rguenther@suse.de>
13862 PR tree-optimization/82220
13863 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
13864 epilogue niters from the min_profitable_iters compute.
13866 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13869 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
13870 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
13871 (ix86_init_pic_reg): Revert 2017-09-01 changes.
13873 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
13876 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
13877 switching to a new text section.
13879 2017-09-18 Richard Biener <rguenther@suse.de>
13881 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
13883 (build_alias_set): Reject aliases with no access function.
13885 2017-09-18 Richard Biener <rguenther@suse.de>
13887 PR tree-optimization/79622
13888 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
13890 (build_cross_bb_scalars_use): Likewise.
13892 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
13894 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
13896 2017-09-18 Alan Modra <amodra@gmail.com>
13899 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
13900 stack_pointer_rtx for count 0. Update comments. Break up
13901 large rtl expression.
13903 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
13905 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
13906 Increase to 20 bytes.
13907 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
13908 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
13909 or avx version of the stub.
13911 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
13914 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
13915 compute the minimum stack alignment. Also update preferred stack
13916 boundary for leaf functions.
13918 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
13920 PR tree-optimization/82228
13921 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
13924 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
13926 * common/config/nds32/nds32-common.c
13927 (nds32_option_optimization_table): Refine formatting.
13928 (nds32_option_optimization_table): Use -fsched-pressure and
13929 -fomit-frame-pointer for specific optimization level.
13931 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
13933 * config/nds32/nds32.c: Refine formatting and comments.
13934 * config/nds32/nds32.h: Likewise.
13935 * config/nds32/nds32.md: Likewise.
13936 * config/nds32/nds32-cost.c: Likewise.
13937 * config/nds32/nds32-isr.c: Likewise.
13938 * config/nds32/nds32-md-auxiliary.c: Likewise.
13939 * config/nds32/nds32-multiple.md: Likewise.
13940 * config/nds32/nds32-predicates.c: Likewise.
13942 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
13943 Jakub Jelinek <jakub@redhat.com>
13945 Add support for -std=c++2a.
13946 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
13948 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
13950 2017-09-15 Steve Ellcey <sellcey@cavium.com>
13953 * doc/extend.texi (Common Function Attributes): Add
13954 references to ARM, AArch64, and S/390 specific attributes.
13955 (Function Specific Option Pragmas): Add AArch64 and S/390
13956 to list of back ends that support the target pragma.
13958 2017-09-15 Nathan Sidwell <nathan@acm.org>
13960 * doc/standards.texi: Fix C++17 description. Update URLs for
13963 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
13965 * common.opt (Wcast-align=strict): New warning option.
13966 * doc/invoke.texi: Document -Wcast-align=strict.
13968 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
13970 * cgraph.h (cgraph_thunk_info): Add comments.
13971 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
13972 assert for VIRTUAL_* arguments stricter.
13974 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
13976 PR tree-optimization/71026
13977 * match.pd: Move RDIV patterns from fold-const.c
13978 * fold-const.c (distribute_real_division): Removed.
13979 (fold_binary_loc): Remove calls to distribute_real_divison.
13981 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13983 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
13984 c++1z and gnu++1z as deprecated. Change other references to
13985 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
13986 Change -Wc++1z-compat to -Wc++17-compat.
13987 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
13988 * dwarf2out.c (highest_c_language): Handle C++17.
13989 (gen_compile_unit_die): Likewise.
13991 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13993 PR rtl-optimization/82192
13994 * combine.c (make_extraction): Don't look through non-paradoxical
13995 SUBREGs or TRUNCATE if pos + len is or might be bigger than
13998 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13999 Alan Hayward <alan.hayward@arm.com>
14000 David Sherwood <david.sherwood@arm.com>
14002 * target.def (function_arg_offset): New hook.
14003 * targhooks.h (default_function_arg_offset): Declare.
14004 * targhooks.c (default_function_arg_offset): New function.
14005 * function.c (locate_and_pad_parm): Use
14006 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
14007 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
14008 (TARGET_FUNCTION_ARG_OFFSET): ...this.
14009 * doc/tm.texi: Regenerate.
14010 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
14011 * config/spu/spu.c (spu_function_arg_offset): New function.
14012 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
14013 * system.h (FUNCTION_ARG_OFFSET): Poison.
14015 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14016 Alan Hayard <alan.hayward@arm.com>
14017 David Sherwood <david.sherwood@arm.com>
14019 * target.def (truly_noop_truncation): New hook.
14020 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
14021 than TRULY_NOOP_TRUNCATION.
14022 * hooks.h (hook_bool_uint_uint_true): Declare.
14023 * hooks.c (hook_bool_uint_uint_true): New function.
14024 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
14025 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
14026 * doc/tm.texi: Regenerate.
14027 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
14028 rather than TRULY_NOOP_TRUNCATION in comments.
14029 (simplify_comparison): Likewise.
14030 (record_truncated_value): Likewise.
14031 * expmed.c (extract_bit_field_1): Likewise.
14032 (extract_split_bit_field): Likewise.
14033 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
14034 instead of TRULY_NOOP_TRUNCATION.
14035 * function.c (assign_parm_setup_block): Likewise.
14036 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
14037 * rtlhooks.c: Include target.h.
14038 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
14039 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
14040 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
14041 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
14042 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
14043 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
14044 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
14045 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
14046 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
14047 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
14048 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
14049 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
14050 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
14051 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
14052 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
14053 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
14054 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
14055 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
14056 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
14057 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
14058 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
14059 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
14060 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
14061 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
14062 * config/mips/mips.c (mips_truly_noop_truncation): New function.
14063 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14064 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14065 rather than TRULY_NOOP_TRUNCATION in comments.
14066 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
14067 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
14068 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
14069 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
14070 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
14071 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
14072 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
14073 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
14074 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
14075 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
14076 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
14077 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14078 rather than TRULY_NOOP_TRUNCATION in comments.
14079 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
14080 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
14081 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
14082 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
14083 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
14084 TRULY_NOOP_TRUNCATION condition.
14085 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
14086 (TRULY_NOOP_TRUNCATION): Delete.
14087 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
14088 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
14089 * config/spu/spu.c (spu_truly_noop_truncation): New function.
14090 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14091 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
14092 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
14093 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
14094 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14095 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14096 rather than TRULY_NOOP_TRUNCATION in comments.
14097 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
14098 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
14099 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
14100 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
14101 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
14102 * system.h (TRULY_NOOP_TRUNCATION): Poison.
14104 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
14107 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
14108 (*cmp_ior): Likewise.
14109 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
14110 (*ior_scc_scc_cmp): Likewise.
14111 (*and_scc_scc): Likewise.
14112 (*and_scc_scc_cmp): Likewise.
14114 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14115 Alan Hayard <alan.hayward@arm.com>
14116 David Sherwood <david.sherwood@arm.com>
14118 * target.def (can_change_mode_class): New hook.
14119 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
14120 (hard_regno_nregs): Likewise.
14121 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
14122 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
14123 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
14124 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
14125 (LOAD_EXTEND_OP): Update accordingly.
14126 * doc/tm.texi: Regenerate.
14127 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
14128 CANNOT_CHANGE_MODE_CLASS.
14129 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
14130 (REG_CAN_CHANGE_MODE_P): ...this new macro.
14131 * combine.c (simplify_set): Update accordingly.
14132 * emit-rtl.c (validate_subreg): Likewise.
14133 * recog.c (general_operand): Likewise.
14134 * regcprop.c (mode_change_ok): Likewise.
14135 * reload1.c (choose_reload_regs): Likewise.
14136 (inherit_piecemeal_p): Likewise.
14137 * rtlanal.c (simplify_subreg_regno): Likewise.
14138 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
14139 instead of CANNOT_CHANGE_MODE_CLASS.
14140 (reload_cse_simplify_operands): Likewise.
14141 * reload.c (push_reload): Use targetm.can_change_mode_class
14142 instead of CANNOT_CHANGE_MODE_CLASS.
14143 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
14144 REG_CANNOT_CHANGE_MODE_P.
14145 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14146 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
14147 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14148 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14149 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14150 (arm_can_change_mode_class): New function.
14151 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14152 than CANNOT_CHANGE_MODE_CLASS in comments.
14153 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14154 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
14155 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
14156 (ix86_can_change_mode_class): ...this new function, inverting the
14157 sense of the return value.
14158 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14159 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14160 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14161 (ia64_can_change_mode_class): New function.
14162 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14163 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
14164 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
14165 (m32c_can_change_mode_class): ...this new function, inverting the
14166 sense of the return value.
14167 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14168 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14169 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
14170 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
14171 (mips_can_change_mode_class): ...this new function, inverting the
14172 sense of the return value.
14173 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14174 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14175 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14176 (msp430_can_change_mode_class): New function.
14177 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14178 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
14179 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14180 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14181 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14182 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
14183 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14184 (pa_cannot_change_mode_class): Replace with...
14185 (pa_can_change_mode_class): ...this new function, inverting the
14186 sense of the return value.
14187 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14188 than CANNOT_CHANGE_MODE_CLASS in comments.
14189 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14190 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
14191 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14192 (pdp11_cannot_change_mode_class): Replace with...
14193 (pdp11_can_change_mode_class): ...this new function, inverting the
14194 sense of the return value.
14195 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14196 * config/powerpcspe/powerpcspe-protos.h
14197 (rs6000_cannot_change_mode_class_ptr): Delete.
14198 * config/powerpcspe/powerpcspe.c
14199 (rs6000_cannot_change_mode_class_ptr): Delete.
14200 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14201 (rs6000_option_override_internal): Assign to
14202 targetm.can_change_mode_class instead of
14203 rs6000_cannot_change_mode_class_ptr.
14204 (rs6000_cannot_change_mode_class): Replace with...
14205 (rs6000_can_change_mode_class): ...this new function, inverting the
14206 sense of the return value.
14207 (rs6000_debug_cannot_change_mode_class): Replace with...
14208 (rs6000_debug_can_change_mode_class): ...this new function.
14209 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14210 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
14211 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14212 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14213 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
14215 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
14216 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14217 (rs6000_option_override_internal): Assign to
14218 targetm.can_change_mode_class instead of
14219 rs6000_cannot_change_mode_class_ptr.
14220 (rs6000_cannot_change_mode_class): Replace with...
14221 (rs6000_can_change_mode_class): ...this new function, inverting the
14222 sense of the return value.
14223 (rs6000_debug_cannot_change_mode_class): Replace with...
14224 (rs6000_debug_can_change_mode_class): ...this new function.
14225 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14226 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
14227 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
14228 (s390_can_change_mode_class): ...this new function, inverting the
14229 sense of the return value.
14230 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14231 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14232 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
14233 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14234 (sh_cannot_change_mode_class): Replace with...
14235 (sh_can_change_mode_class): ...this new function, inverting the
14236 sense of the return value.
14237 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14238 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14239 (sparc_can_change_mode_class): New function.
14240 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14241 * config/spu/spu.c (spu_can_change_mode_class): New function.
14242 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14243 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14244 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14245 (visium_can_change_mode_class): New function.
14246 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
14248 2017-09-15 Richard Biener <rguenther@suse.de>
14250 PR tree-optimization/82217
14251 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
14252 but not undefined case.
14254 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14257 * postreload.c (reload_cse_simplify_operands): Skip
14258 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
14260 2017-09-15 Richard Biener <rguenther@suse.de>
14262 PR tree-optimization/68823
14263 * graphite-scop-detection.c (build_alias_set): If we have a
14264 possible dependence check whether we can handle them by just
14265 looking at the DRs DR_ACCESS_FNs.
14266 (build_scops): If build_alias_set fails, fail the SCOP.
14268 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
14270 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
14271 to support float128 built-in functions that require the ISA 3.0
14273 (BU_FLOAT128_3_HW): Likewise.
14274 (SQRTF128): Add support for the IEEE 128-bit square root and fma
14275 built-in functions.
14276 (FMAF128): Likewise.
14278 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14279 support for built-in functions that need the ISA 3.0 IEEE 128-bit
14280 floating point instructions.
14281 (rs6000_invalid_builtin): Likewise.
14282 (rs6000_builtin_mask_names): Likewise.
14283 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
14284 (RS6000_BTM_FLOAT128_HW): Likewise.
14285 (RS6000_BTM_COMMON): Likewise.
14286 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
14288 * doc/extend.texi (RS/6000 built-in functions): Document the
14289 IEEE 128-bit floating point square root and fused multiply-add
14290 built-in functions.
14292 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
14294 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
14295 reg (r2) isn't in the set of registers defined in the prologue.
14297 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14298 Alan Hayward <alan.hayward@arm.com>
14299 David Sherwood <david.sherwood@arm.com>
14301 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
14302 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
14303 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
14304 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
14305 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
14307 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
14308 max_vectorization_factor.
14309 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
14311 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14312 Alan Hayward <alan.hayward@arm.com>
14313 David Sherwood <david.sherwood@arm.com>
14315 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
14316 (vect_worthwhile_without_simd_p): Declare.
14317 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
14318 (vectorizable_reduction): Use it.
14319 * tree-vect-stmts.c (vectorizable_shift): Likewise.
14320 (vectorizable_operation): Likewise.
14322 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14323 Alan Hayward <alan.hayward@arm.com>
14324 David Sherwood <david.sherwood@arm.com>
14326 * tree-vectorizer.h (vect_get_num_copies): New function.
14327 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
14328 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14329 (vectorizable_induction): Likewise.
14330 (vectorizable_live_operation): Likewise.
14331 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14332 (vectorizable_bswap): Likewise.
14333 (vectorizable_call): Likewise.
14334 (vectorizable_conversion): Likewise.
14335 (vectorizable_assignment): Likewise.
14336 (vectorizable_shift): Likewise.
14337 (vectorizable_operation): Likewise.
14338 (vectorizable_store): Likewise.
14339 (vectorizable_load): Likewise.
14340 (vectorizable_condition): Likewise.
14341 (vectorizable_comparison): Likewise.
14342 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
14344 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14345 Alan Hayward <alan.hayward@arm.com>
14346 David Sherwood <david.sherwood@arm.com>
14348 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
14349 of vect_init_vector.
14351 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14352 Alan Hayward <alan.hayward@arm.com>
14353 David Sherwood <david.sherwood@arm.com>
14355 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
14356 an inline wrapper that provides a location.
14357 (gimple_build_vector): Likewise.
14358 * gimple-fold.c (gimple_build_vector_from_val): New function.
14359 (gimple_build_vector): Likewise.
14360 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
14361 functions to build the initial value. Always return a gimple value.
14362 (get_initial_defs_for_reduction): Likewise. Only compute
14364 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
14365 vect_init_vector on the results from get_initial_def(s)_for_reduction.
14366 (vectorizable_induction): Use gimple_build_vector rather than
14369 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14370 Alan Hayward <alan.hayward@arm.com>
14371 David Sherwood <david.sherwood@arm.com>
14373 * target.h (vec_perm_indices): New typedef.
14374 (auto_vec_perm_indices): Likewise.
14375 * optabs-query.h: Include target.h
14376 (can_vec_perm_p): Take a vec_perm_indices *.
14377 * optabs-query.c (can_vec_perm_p): Likewise.
14378 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
14379 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14380 * tree-vect-generic.c (lower_vec_perm): Likewise.
14381 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
14382 (vect_grouped_load_supported): Likewise.
14383 (vect_shift_permute_load_chain): Likewise.
14384 (vect_permute_store_chain): Use auto_vec_perm_indices.
14385 (vect_permute_load_chain): Likewise.
14386 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
14387 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
14388 Update uses of can_vec_perm_p.
14389 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
14390 mode with a number of elements. Take a vec_perm_indices *.
14391 (vect_create_epilog_for_reduction): Update accordingly.
14392 Use auto_vec_perm_indices.
14393 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
14394 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
14395 (vect_transform_slp_perm_load): Likewise.
14396 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
14397 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
14398 (vect_gen_perm_mask_checked): Likewise.
14399 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
14400 (vect_gen_perm_mask_checked): Likewise.
14401 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
14402 (vectorizable_store): Likewise.
14403 (vectorizable_load): Likewise.
14404 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
14405 (vectorizable_bswap): Likewise.
14407 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14408 Alan Hayward <alan.hayward@arm.com>
14409 David Sherwood <david.sherwood@arm.com>
14411 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
14412 * tree.c (build_vector): Likewise.
14413 (build_vector_from_ctor): Update accordingly.
14414 (build_vector_from_val): Likewise.
14415 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14416 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14417 * tree-vect-generic.c (add_rshift): Likewise.
14418 (expand_vector_divmod): Likewise.
14419 (optimize_vector_constructor): Likewise.
14420 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
14421 (vect_transform_slp_perm_load): Likewise.
14422 (vect_schedule_slp_instance): Likewise.
14423 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
14424 (vectorizable_call): Likewise.
14425 (vect_gen_perm_mask_any): Likewise. Add elements in order.
14426 * expmed.c (make_tree): Likewise.
14427 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
14428 a vector passed to build_vector.
14429 (fold_convert_const): Likewise.
14430 (exact_inverse): Likewise.
14431 (fold_ternary_loc): Likewise.
14432 (fold_relational_const): Likewise.
14433 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
14434 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
14435 (const_unop): Likewise. Store the reduction accumulator in a
14436 variable rather than an array.
14437 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
14438 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
14439 the new vector, rather than constructing it after the input arrays.
14440 (native_interpret_vector): Use auto_vec<tree> when building
14441 a vector passed to build_vector. Add elements in order.
14442 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
14443 auto_vec<tree> when building a vector passed to build_vector.
14444 (vect_create_epilog_for_reduction): Likewise.
14445 (vectorizable_induction): Likewise.
14446 (get_initial_def_for_reduction): Likewise. Fix indentation of
14448 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
14450 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
14451 passed to build_vector.
14453 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14454 Alan Hayward <alan.hayward@arm.com>
14455 David Sherwood <david.sherwood@arm.com>
14457 * tree-core.h (tree_base::u): Add an "nelts" field.
14458 (tree_vector): Use VECTOR_CST_NELTS as the length.
14459 * tree.c (tree_size): Likewise.
14460 (make_vector): Initialize VECTOR_CST_NELTS.
14461 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
14462 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
14463 TYPE_VECTOR_SUBPARTS.
14464 * expr.c (const_vector_mask_from_tree): Consistently use "units"
14465 as the number of units, setting it from VECTOR_CST_NELTS.
14466 (const_vector_from_tree): Likewise.
14467 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
14468 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
14469 (fold_negate_expr_1): Likewise.
14470 (fold_convert_const): Likewise.
14471 (const_binop): Likewise. Differentiate the number of output and
14473 (const_unop): Likewise.
14474 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
14475 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
14476 in cases that did the opposite.
14478 2017-09-14 Richard Biener <rguenther@suse.de>
14480 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
14483 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
14485 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
14487 2017-09-14 Jakub Jelinek <jakub@redhat.com>
14490 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
14491 if and where to split a bb, except for splitting before debug insn
14492 sequences followed by non-label real insn. Delete debug insns
14493 in between basic blocks.
14495 * combine.c (make_compound_operation_int): Formatting fixes.
14497 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
14498 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14499 * config/netbsd.h (LINK_EH_SPEC): Likewise.
14500 * config/sol2.h (LINK_EH_SPEC): Likewise.
14501 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14502 * config/s390/linux.h (LINK_SPEC): Likewise.
14503 * config/freebsd.h (LINK_EH_SPEC): Likewise.
14504 * config/openbsd.h (LINK_EH_SPEC): Likewise.
14505 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14506 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
14507 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
14508 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14509 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
14510 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
14512 2017-09-13 Jakub Jelinek <jakub@redhat.com>
14514 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
14516 (ENDFILE_LINUX_SPEC): Likewise.
14517 (LINK_EH_SPEC): Likewise.
14518 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
14519 (LINK_OS_LINUX_SPEC32): Likewise.
14520 (LINK_OS_LINUX_SPEC64): Likewise.
14521 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
14522 (LINK_OS_LINUX_SPEC): Likewise.
14524 2017-09-13 Martin Liska <mliska@suse.cz>
14526 PR middle-end/82154
14527 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
14528 CASE_HIGH is NULL_TREE.
14530 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14531 Alan Hayward <alan.hayward@arm.com>
14532 David Sherwood <david.sherwood@arm.com>
14534 * target.def (secondary_memory_needed): New hook.
14535 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
14536 instead of SECONDARY_MEMORY_NEEDED.
14537 (secondary_memory_needed_mode): Likewise.
14538 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
14539 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
14540 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
14541 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
14542 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
14543 * doc/tm.texi: Regenerate.
14544 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
14545 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
14546 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14547 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
14548 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
14549 * config/i386/i386.c (inline_secondary_memory_needed): Put the
14550 mode argument first and change the reg_class arguments to reg_class_t.
14551 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
14552 Make static. Update the call to inline_secondary_memory_needed.
14553 (ix86_register_move_cost): Update the call to
14554 inline_secondary_memory_needed.
14555 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14556 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
14558 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
14559 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14561 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
14562 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
14563 * config/mips/mips.c (mips_secondary_memory_needed): Make static
14564 and match hook interface. Add comment from mips.h.
14565 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14566 * config/mmix/mmix.md (truncdfsf2): Refer to
14567 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14569 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
14570 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
14571 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14572 (pa_secondary_memory_needed): New function.
14573 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
14574 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
14575 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14576 (pdp11_secondary_memory_needed): Make static and match hook interface.
14577 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
14578 * config/powerpcspe/powerpcspe-protos.h
14579 (rs6000_secondary_memory_needed_ptr): Delete.
14580 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
14582 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14583 (rs6000_option_override_internal): Assign to
14584 targetm.secondary_memory_needed rather than
14585 rs6000_secondary_memory_needed_ptr.
14586 (rs6000_secondary_memory_needed): Match hook interface.
14587 (rs6000_debug_secondary_memory_needed): Likewise.
14588 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
14589 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
14590 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
14591 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14592 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
14593 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
14595 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
14596 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14597 (rs6000_option_override_internal): Assign to
14598 targetm.secondary_memory_needed rather than
14599 rs6000_secondary_memory_needed_ptr.
14600 (rs6000_secondary_memory_needed): Match hook interface.
14601 (rs6000_debug_secondary_memory_needed): Likewise.
14602 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
14603 * config/s390/s390.c (s390_secondary_memory_needed): New function.
14604 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14605 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
14606 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14607 (sparc_secondary_memory_needed): New function.
14608 * lra-constraints.c (check_and_process_move): Refer to
14609 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14611 (curr_insn_transform): Likewise.
14612 (process_alt_operands): Use targetm.secondary_memory_needed
14613 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14614 (check_secondary_memory_needed_p): Likewise.
14615 (choose_split_class): Likewise.
14616 * reload.c: Unconditionally include code that was previously
14617 conditional on SECONDARY_MEMORY_NEEDED.
14618 (push_secondary_reload): Use targetm.secondary_memory_needed
14619 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14620 (push_reload): Likewise.
14621 * reload1.c: Unconditionally include code that was previously
14622 conditional on SECONDARY_MEMORY_NEEDED.
14623 (choose_reload_regs): Use targetm.secondary_memory_needed
14624 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14625 (gen_reload): Likewise.
14626 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
14628 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14629 Alan Hayward <alan.hayward@arm.com>
14630 David Sherwood <david.sherwood@arm.com>
14632 * target.def (secondary_memory_needed_mode): New hook:
14633 * targhooks.c (default_secondary_memory_needed_mode): Declare.
14634 * targhooks.h (default_secondary_memory_needed_mode): New function.
14635 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
14636 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
14637 * doc/tm.texi: Regenerate.
14638 * lra-constraints.c (check_and_process_move): Use
14639 targetm.secondary_memory_needed_mode instead of
14640 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
14641 (curr_insn_transform): Likewise.
14642 * reload.c (get_secondary_mem): Likewise.
14643 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14644 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
14646 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14647 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14648 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
14649 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14650 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
14652 * config/powerpcspe/powerpcspe-protos.h
14653 (rs6000_secondary_memory_needed_mode): Delete.
14654 * config/powerpcspe/powerpcspe.c
14655 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14656 (rs6000_secondary_memory_needed_mode): Make static.
14657 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14658 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
14660 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14662 (rs6000_secondary_memory_needed_mode): Make static.
14663 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14664 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
14665 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14666 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14667 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14669 (sparc_secondary_memory_needed_mode): New function.
14670 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
14672 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
14674 * config/aarch64/constraints.md (Umq): New constraint.
14675 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
14677 (mov<mode>): Update condition.
14679 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14681 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
14682 when bitposition is the same.
14684 2017-09-13 Richard Biener <rguenther@suse.de>
14686 * dwarf2out.c (output_die_symbol): Remove.
14687 (output_die): Do not output a DIEs symbol.
14689 2017-09-13 Richard Biener <rguenther@suse.de>
14691 PR middle-end/82128
14692 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
14693 default-def to avoid breaking iterator update with the weird
14694 interaction with cgraph_update_edges_for_call_stmt_node.
14696 2017-09-13 Richard Biener <rguenther@suse.de>
14698 * tree-cfg.c (verify_gimple_assign_binary): Add verification
14699 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
14700 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
14701 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
14703 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
14705 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
14706 Disable pc relative literal load irrespective of
14707 TARGET_FIX_ERR_A53_84341 for default.
14709 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
14711 * config/sparc/sparc.c (output_return): Output the source location of
14712 the insn in the delay slot, if any.
14713 (output_sibcall): Likewise.
14715 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
14718 * common.opt (-static-pie): New alias.
14719 (shared): Negate static-pie.
14720 (-no-pie): Update help text.
14722 (static-pie): New option.
14723 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
14724 -static-pie support.
14725 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
14726 (LINK_EH_SPEC): Likewise.
14727 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14728 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14729 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14730 * gcc.c (LINK_COMMAND_SPEC): Likewise.
14731 (init_gcc_specs): Likewise.
14732 (init_spec): Likewise.
14733 (display_help): Update help message for -pie.
14734 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
14737 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
14739 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
14740 (movdi_aarch64): Likewise.
14741 (movti_aarch64): Likewise.
14743 2017-09-12 Simon Wright <simon@pushface.org>
14746 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
14747 calculation of the minor version, always output as 0.
14749 2017-09-12 Jakub Jelinek <jakub@redhat.com>
14752 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14753 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
14754 on it early, rather than manual conversion late. For
14755 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
14756 instead of performing manual conversion.
14758 2017-09-12 Carl Love <cel@us.ibm.com>
14760 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
14761 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
14763 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14764 VMULOSW): Add definitions.
14765 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14766 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14767 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14768 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14769 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14771 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14773 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
14775 (movti_aarch64): Likewise.
14776 (movdf_aarch64): Likewise.
14777 (movtf_aarch64): Likewise.
14778 (load_pairdi): Likewise.
14779 (store_pairdi): Likewise.
14780 (load_pairdf): Likewise.
14781 (store_pairdf): Likewise.
14782 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
14783 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
14784 (ldr_got_small_<mode>): Likewise.
14785 (ldr_got_small_28k_<mode>): Likewise.
14786 (ldr_got_tiny): Likewise.
14787 * config/aarch64/iterators.md (ldst_sz): New.
14788 (ldpstp_sz): Likewise.
14789 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
14791 (thunderx_load): Split load_8 to load_16.
14792 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
14794 (thunderx2t99_storepair_basic): Split store_8 to store_16.
14795 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
14796 (xgene1_store_pair): Split store_8 to store_16.
14797 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
14798 (falkor_st_0_st_sd): Split store_8 to store_16.
14800 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14802 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
14803 and store1/2/3/4 to store_4/8/12/16.
14804 * config/aarch64/aarch64.md: Update for rename.
14805 * config/arm/arm.md: Likewise.: Likewise.
14806 * config/arm/arm.c: Likewise.
14807 * config/arm/thumb1.md: Likewise.
14808 * config/arm/thumb2.md: Likewise.
14809 * config/arm/vfp.md: Likewise.
14810 * config/arm/arm-generic.md: Likewise.
14811 * config/arm/arm1020e.md: Likewise.
14812 * config/arm/arm1026ejs.md: Likewise.
14813 * config/arm/arm1136jfs.md: Likewise.
14814 * config/arm/arm926ejs.md: Likewise.
14815 * config/arm/cortex-a15.md: Likewise.
14816 * config/arm/cortex-a17.md: Likewise.
14817 * config/arm/cortex-a5.md: Likewise.
14818 * config/arm/cortex-a53.md: Likewise.
14819 * config/arm/cortex-a57.md: Likewise.
14820 * config/arm/cortex-a7.md: Likewise.
14821 * config/arm/cortex-a8.md: Likewise.
14822 * config/arm/cortex-a9.md: Likewise.
14823 * config/arm/cortex-m4.md: Likewise.
14824 * config/arm/cortex-m7.md: Likewise.
14825 * config/arm/cortex-r4.md: Likewise.
14826 * config/arm/exynos-m1.md: Likewise.
14827 * config/arm/fa526.md: Likewise.
14828 * config/arm/fa606te.md: Likewise.
14829 * config/arm/fa626te.md: Likewise.
14830 * config/arm/fa726te.md: Likewise.
14831 * config/arm/fmp626.md: Likewise.
14832 * config/arm/iwmmxt.md: Likewise.
14833 * config/arm/ldmstm.md: Likewise.
14834 * config/arm/marvell-pj4.md: Likewise.
14835 * config/arm/xgene1.md: Likewise.
14836 * config/aarch64/thunderx.md: Likewise.
14837 * config/aarch64/thunderx2t99.md: Likewise.
14838 * config/aarch64/falkor.md: Likewise.
14840 2017-09-12 Martin Liska <mliska@suse.cz>
14842 * attribs.c (private_lookup_attribute): New function.
14843 * attribs.h (private_lookup_attribute): Declared here.
14844 (lookup_attribute): Called from this place.
14846 2017-09-12 Richard Biener <rguenther@suse.de>
14848 PR tree-optimization/82157
14849 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
14850 stmts with side-effects.
14852 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14853 Alan Hayward <alan.hayward@arm.com>
14854 David Sherwood <david.sherwood@arm.com>
14856 * target.def (hard_regno_nregs): New hook.
14857 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
14858 * targhooks.h (default_hard_regno_nregs): Declare.
14859 * targhooks.c (default_hard_regno_nregs): New function.
14860 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
14861 (TARGET_HARD_REGNO_NREGS): ...this hook.
14862 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
14863 (CLASS_MAX_NREGS): Likewise.
14864 * doc/tm.texi: Regenerate.
14865 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
14866 instead of HARD_REGNO_NREGS.
14867 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
14868 HARD_REGNO_NREGS in the comment.
14869 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
14870 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
14871 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
14872 Return an unsigned int.
14873 (TARGET_HARD_REGNO_NREGS): Redefine.
14874 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
14875 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
14876 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
14877 (arc_hard_regno_nregs): New function.
14878 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
14879 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
14880 (arm_hard_regno_nregs): New function.
14881 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
14882 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
14883 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
14884 (TARGET_HARD_REGNO_NREGS): Redefine.
14885 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
14886 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
14887 (HARD_REGNO_NREGS): Delete.
14888 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
14889 (cr16_hard_regno_nregs): New function.
14890 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
14891 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
14892 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
14893 (cris_hard_regno_nregs): New function.
14894 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
14895 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
14896 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
14897 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
14898 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
14899 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
14900 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
14901 (frv_hard_regno_nregs): Make static. Take and return an
14903 (frv_class_max_nregs): Remove outdated copy of documentation.
14904 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
14905 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
14906 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
14907 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
14908 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
14909 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
14910 (TARGET_HARD_REGNO_NREGS): Redefine.
14911 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
14912 (CLASS_MAX_NREGS): Update comment.
14913 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
14914 (ia64_hard_regno_nregs): New function.
14915 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
14916 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
14917 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
14918 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
14919 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
14921 (m32c_hard_regno_nregs): Likewise. Make static.
14922 (TARGET_HARD_REGNO_NREGS): Redefine.
14923 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
14924 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
14925 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
14926 (m68k_hard_regno_nregs): New function.
14927 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
14928 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
14929 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
14930 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
14931 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
14932 Take and return an unsigned int.
14933 (TARGET_HARD_REGNO_NREGS): Redefine.
14934 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
14935 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
14936 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
14937 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
14938 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
14939 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
14940 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
14941 (msp430_hard_regno_nregs): Make static. Take and return an
14943 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
14944 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
14945 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
14946 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
14947 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
14948 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
14949 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
14950 (TARGET_HARD_REGNO_NREGS): Redefine.
14951 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
14952 (PA_HARD_REGNO_NREGS): ...this.
14953 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
14954 (PA_HARD_REGNO_NREGS): ...this.
14955 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
14956 (pa_hard_regno_nregs): New function.
14957 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
14958 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
14959 (pdp11_hard_regno_nregs): New function.
14960 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
14961 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
14962 (rs6000_hard_regno_nregs_hook): New function.
14963 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
14964 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
14965 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
14966 Take and return an unsigned int. Move earlier in file.
14967 (TARGET_HARD_REGNO_NREGS): Redefine.
14968 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
14969 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
14970 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
14971 (rl78_hard_regno_nregs): Make static. Take and return an
14973 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
14974 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
14975 (rs6000_hard_regno_nregs_hook): New function.
14976 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
14977 * config/rx/rx.c (rx_hard_regno_nregs): New function.
14978 (TARGET_HARD_REGNO_NREGS): Redefine.
14979 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
14980 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
14981 instead of HARD_REGNO_NREGS.
14982 (s390_hard_regno_nregs): New function.
14983 (s390_hard_regno_mode_ok): Add comment from s390.h.
14984 (TARGET_HARD_REGNO_NREGS): Redefine.
14985 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
14986 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
14987 (sh_hard_regno_nregs): New function.
14988 (sh_pass_in_reg_p): Use it.
14989 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
14990 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
14991 (sparc_hard_regno_nregs): New function.
14992 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
14993 * config/spu/spu.c (spu_hard_regno_nregs): New function.
14994 (spu_function_arg_advance): Use it, supplying a valid register number.
14995 (TARGET_HARD_REGNO_NREGS): Redefine.
14996 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
14997 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
14998 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
14999 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
15000 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
15001 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
15002 (CLASS_MAX_NREGS): Remove copy of old documentation.
15003 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
15004 (visium_hard_regno_nregs): New function.
15005 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
15006 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
15007 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15008 xtensa_hard_regno_nregs): New function.
15009 * system.h (HARD_REGNO_NREGS): Poison.
15011 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15013 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
15014 hard_regno_nregs instead of HARD_REGNO_NREGS.
15015 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
15016 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
15017 (c6x_expand_epilogue): Likewise.
15018 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
15019 (frv_read_iacc_argument): Likewise.
15020 * config/sh/sh.c: Include regs.h.
15021 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15022 (regs_used): Likewise.
15023 (output_stack_adjust): Likewise.
15024 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
15025 * expmed.c: Include regs.h.
15026 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15027 * ree.c: Include regs.h.
15028 (combine_reaching_defs): Use hard_regno_nregs instead of
15030 (add_removable_extension): Likewise.
15032 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15034 * regs.h (hard_regno_nregs): Turn into a function.
15035 (end_hard_regno): Update accordingly.
15036 * caller-save.c (setup_save_areas): Likewise.
15037 (save_call_clobbered_regs): Likewise.
15038 (replace_reg_with_saved_mem): Likewise.
15039 (insert_restore): Likewise.
15040 (insert_save): Likewise.
15041 * combine.c (can_change_dest_mode): Likewise.
15042 (move_deaths): Likewise.
15043 (distribute_notes): Likewise.
15044 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
15045 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
15046 (rs6000_split_multireg_move): Likewise.
15047 (rs6000_register_move_cost): Likewise.
15048 (rs6000_memory_move_cost): Likewise.
15049 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
15050 (rs6000_split_multireg_move): Likewise.
15051 (rs6000_register_move_cost): Likewise.
15052 (rs6000_memory_move_cost): Likewise.
15053 * cselib.c (cselib_reset_table): Likewise.
15054 (cselib_lookup_1): Likewise.
15055 * emit-rtl.c (set_mode_and_regno): Likewise.
15056 * function.c (aggregate_value_p): Likewise.
15057 * ira-color.c (setup_profitable_hard_regs): Likewise.
15058 (check_hard_reg_p): Likewise.
15059 (calculate_saved_nregs): Likewise.
15060 (assign_hard_reg): Likewise.
15061 (improve_allocation): Likewise.
15062 (calculate_spill_cost): Likewise.
15063 * ira-emit.c (modify_move_list): Likewise.
15064 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
15065 (ira_hard_reg_in_set_p): Likewise.
15066 * ira.c (setup_reg_mode_hard_regset): Likewise.
15067 (clarify_prohibited_class_mode_regs): Likewise.
15068 (check_allocation): Likewise.
15069 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15070 (lra_setup_reg_renumber): Likewise.
15071 (setup_try_hard_regno_pseudos): Likewise.
15072 (spill_for): Likewise.
15073 (assign_hard_regno): Likewise.
15074 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
15075 * lra-constraints.c (in_class_p): Likewise.
15076 (lra_constraint_offset): Likewise.
15077 (simplify_operand_subreg): Likewise.
15078 (lra_constraints): Likewise.
15079 (split_reg): Likewise.
15080 (split_if_necessary): Likewise.
15081 (invariant_p): Likewise.
15082 (inherit_in_ebb): Likewise.
15083 * lra-lives.c (process_bb_lives): Likewise.
15084 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
15085 (get_hard_regs): Likewise.
15086 (do_remat): Likewise.
15087 * lra-spills.c (assign_spill_hard_regs): Likewise.
15088 * mode-switching.c (create_pre_exit): Likewise.
15089 * postreload.c (reload_combine_recognize_pattern): Likewise.
15090 * recog.c (peep2_find_free_register): Likewise.
15091 * regcprop.c (kill_value_regno): Likewise.
15092 (set_value_regno): Likewise.
15093 (copy_value): Likewise.
15094 (maybe_mode_change): Likewise.
15095 (find_oldest_value_reg): Likewise.
15096 (copyprop_hardreg_forward_1): Likewise.
15097 * regrename.c (check_new_reg_p): Likewise.
15098 (regrename_do_replace): Likewise.
15099 * reload.c (push_reload): Likewise.
15100 (combine_reloads): Likewise.
15101 (find_dummy_reload): Likewise.
15102 (operands_match_p): Likewise.
15103 (find_reloads): Likewise.
15104 (find_equiv_reg): Likewise.
15105 (reload_adjust_reg_for_mode): Likewise.
15106 * reload1.c (count_pseudo): Likewise.
15107 (count_spilled_pseudo): Likewise.
15108 (find_reg): Likewise.
15109 (clear_reload_reg_in_use): Likewise.
15110 (free_for_value_p): Likewise.
15111 (allocate_reload_reg): Likewise.
15112 (choose_reload_regs): Likewise.
15113 (reload_adjust_reg_for_temp): Likewise.
15114 (emit_reload_insns): Likewise.
15115 (delete_output_reload): Likewise.
15116 * rtlanal.c (subreg_get_info): Likewise.
15117 * sched-deps.c (sched_analyze_reg): Likewise.
15118 * sel-sched.c (init_regs_for_mode): Likewise.
15119 (mark_unavailable_hard_regs): Likewise.
15120 (choose_best_reg_1): Likewise.
15121 (verify_target_availability): Likewise.
15122 * valtrack.c (dead_debug_insert_temp): Likewise.
15123 * var-tracking.c (track_loc_p): Likewise.
15124 (emit_note_insn_var_location): Likewise.
15125 * varasm.c (make_decl_rtl): Likewise.
15126 * reginfo.c (choose_hard_reg_mode): Likewise.
15127 (init_reg_modes_target): Refer directly to
15128 this_target_regs->x_hard_regno_nregs.
15130 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15132 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
15133 instead of hard_regno_nregs.
15135 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15137 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
15138 end_hard_regno instead of hard_regno_nregs.
15139 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
15140 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
15141 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
15142 * ira-color.c (improve_allocation): Likewise.
15143 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15144 * lra-lives.c (mark_regno_live): Likewise.
15145 (mark_regno_dead): Likewise.
15146 * lra-remat.c (operand_to_remat): Likewise.
15147 * lra.c (collect_non_operand_hard_regs): Likewise.
15148 * postreload.c (reload_combine_note_store): Likewise.
15149 (move2add_valid_value_p): Likewise.
15150 * reload.c (regno_clobbered_p): Likewise.
15152 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15154 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
15156 * config/v850/v850.c (v850_reorg): Likewise.
15157 * reload.c (refers_to_regno_for_reload_p): Likewise.
15158 (find_equiv_reg): Likewise.
15159 * reload1.c (reload_reg_reaches_end_p): Likewise.
15161 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15163 * caller-save.c (add_used_regs): Use REG_NREGS instead of
15165 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
15166 * config/arm/arm.c (output_move_neon): Likewise.
15167 (arm_attr_length_move_neon): Likewise.
15168 (neon_split_vcombine): Likewise.
15169 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
15170 (c6x_mark_reg_written): Likewise.
15171 (c6x_dwarf_register_span): Likewise.
15172 * config/i386/i386.c (ix86_save_reg): Likewise.
15173 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
15174 (rws_access_reg): Likewise.
15175 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
15176 * mode-switching.c (create_pre_exit): Likewise.
15177 * ree.c (combine_reaching_defs): Likewise.
15178 (add_removable_extension): Likewise.
15179 * regcprop.c (find_oldest_value_reg): Likewise.
15180 (copyprop_hardreg_forward_1): Likewise.
15181 * reload.c (reload_inner_reg_of_subreg): Likewise.
15182 (push_reload): Likewise.
15183 (combine_reloads): Likewise.
15184 (find_dummy_reload): Likewise.
15185 (reload_adjust_reg_for_mode): Likewise.
15186 * reload1.c (find_reload_regs): Likewise.
15187 (forget_old_reloads_1): Likewise.
15188 (reload_reg_free_for_value_p): Likewise.
15189 (reload_adjust_reg_for_temp): Likewise.
15190 (emit_reload_insns): Likewise.
15191 (delete_output_reload): Likewise.
15192 * sel-sched.c (choose_best_reg_1): Likewise.
15193 (choose_best_pseudo_reg): Likewise.
15195 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15196 Alan Hayward <alan.hayward@arm.com>
15197 David Sherwood <david.sherwood@arm.com>
15199 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
15200 * target.def (slow_unaligned_access): New hook.
15201 * targhooks.h (default_slow_unaligned_access): Declare.
15202 * targhooks.c (default_slow_unaligned_access): New function.
15203 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
15204 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
15205 * doc/tm.texi: Regenerate.
15206 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
15207 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
15208 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
15210 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
15211 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
15213 (rs6000_slow_unaligned_access): New function.
15214 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15215 (expand_block_compare): Likewise.
15216 (expand_strn_compare): Likewise.
15217 (rs6000_rtx_costs): Likewise.
15218 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
15219 (riscv_slow_unaligned_access): Likewise.
15220 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
15221 (riscv_slow_unaligned_access_p): ...this and make static.
15222 (riscv_option_override): Update accordingly.
15223 (riscv_slow_unaligned_access): New function.
15224 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15225 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
15226 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15227 (rs6000_slow_unaligned_access): New function.
15228 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15229 (rs6000_rtx_costs): Likewise.
15230 * config/rs6000/rs6000-string.c (expand_block_compare)
15231 (expand_strn_compare): Use targetm.slow_unaligned_access instead
15232 of SLOW_UNALIGNED_ACCESS.
15233 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
15234 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
15235 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
15236 of SLOW_UNALIGNED_ACCESS.
15237 * expmed.c (simple_mem_bitfield_p): Likewise.
15238 * expr.c (alignment_for_piecewise_move): Likewise.
15239 (emit_group_load_1): Likewise.
15240 (emit_group_store): Likewise.
15241 (copy_blkmode_from_reg): Likewise.
15242 (emit_push_insn): Likewise.
15243 (expand_assignment): Likewise.
15244 (store_field): Likewise.
15245 (expand_expr_real_1): Likewise.
15246 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
15247 * lra-constraints.c (simplify_operand_subreg): Likewise.
15248 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15249 * gimple-ssa-store-merging.c: Likewise in block comment at start
15251 * tree-ssa-strlen.c: Include target.h.
15252 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
15253 of SLOW_UNALIGNED_ACCESS.
15254 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
15256 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15258 PR rtl-optimization/82185
15259 * expmed.c (emit_store_flag_int): Only test tem if it has been
15262 2017-09-12 Richard Biener <rguenther@suse.de>
15264 PR middle-end/82149
15265 * match.pd ((FTYPE) N CMP CST): Fix typo.
15267 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
15269 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
15271 (mips_near_type_p): Add 'short_call' attribute as a synonym
15273 * doc/extend.texi (short_call): Document new function attribute.
15275 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15278 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
15279 assertion check that in the condition.
15280 (get_atomic_generic_size): Likewise. Before testing if parameter
15281 has pointer type, if it has array type, call for C++
15282 default_conversion to perform array-to-pointer conversion.
15284 2017-09-12 Richard Biener <rguenther@suse.de>
15286 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
15287 for operations we cannot scalarize.
15289 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
15291 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
15292 vectors heap vectors. Clean up comments.
15293 Make visited_bbs a reference.
15294 (profitable_jump_thread_path): Make GC
15295 vectors heap vectors. Clean up comments.
15297 (convert_and_register_jump_thread_path): Make GC vectors heap
15299 (check_subpath_and_update_thread_path): Same. Clean up comments.
15300 Make visited_bbs a reference.
15301 (handle_phi): Abstract common code to to
15302 register_jump_thread_path_if_profitable.
15303 Rename VAR_BB to DEF_BB.
15305 Make GC vectors heap vectors.
15306 Make visited_bbs a reference.
15307 (handle_assignment): Same.
15308 (register_jump_thread_path_if_profitable): New.
15309 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
15311 Make GC vectors heap vectors. Clean up comments.
15312 Make visited_bbs a reference.
15313 (find_jump_threads_backwards): Make visited_bbs live in the stack.
15314 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
15317 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
15320 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
15321 words of E_DImode object are reachable by xtensa_uimm8x4 access.
15323 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
15325 Revert r251800 and r251799.
15327 2017-09-11 Martin Jambor <mjambor@suse.cz>
15330 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
15331 arguments in advance.
15332 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
15333 use it to find predecessor edges.
15334 (naive_outof_ssa): Collect vector of predecessors.
15336 2017-09-08 Jason Merrill <jason@redhat.com>
15338 PR c++/70029 - ICE with ref-qualifier and -flto
15339 * langhooks.h (struct lang_hooks_for_types): Add
15340 copy_lang_qualifiers.
15341 * attribs.c (build_type_attribute_qual_variant): Use it.
15342 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
15344 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
15345 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
15347 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
15350 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
15351 (*mulsi3_sp64): New instruction.
15352 (mulsi3): New expander.
15354 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
15356 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
15358 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15360 * sancov.c: Include memmodel.h.
15362 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
15365 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
15368 2017-09-07 Carl Love <cel@us.ibm.com>
15370 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
15371 the sldi instruction.
15373 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
15375 * sancov.c: Include tm_p.h.
15377 2017-09-07 Jakub Jelinek <jakub@redhat.com>
15380 * output.h (switch_to_other_text_partition): New declaration.
15381 * varasm.c (switch_to_other_text_partition): New function.
15382 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
15383 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
15384 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
15385 to the other text partition before emitting LCL label and switch back
15386 after emitting the word after it.
15388 2017-09-07 Richard Biener <rguenther@suse.de>
15390 * passes.def (pass_split_crit_edges): Remove instance before PRE.
15391 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
15392 critical edges here, after loop init.
15393 (pass_data_pre): Remove PROP_no_crit_edges flags.
15394 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
15395 for valueization of call args to avoid leaking VN_TOP.
15396 (visit_use): Assert we do not visit default defs.
15397 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
15398 Use error_mark_node to more easily detect leaking VN_TOP.
15399 All default-defs are varying, not VN_TOP. Mark them visited.
15400 (run_scc_vn): Make code match comment.
15402 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15404 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
15405 OPTION_MASK_FLOAT128_KEYWORD.
15406 (POWERPC_MASKS): Likewise.
15407 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
15408 support for the -mfloat128-type option, and make -mfloat128
15409 default on PowerPC Linux systems. Define or undefine
15410 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
15411 Define __float128 to be __ieee128 if IEEE 128-bit support is
15412 enabled, or undefine it.
15413 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
15414 Delete defining __FLOAT128_TYPE__.
15415 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
15416 -mfloat128-type option and make -mfloat128 default on PowerPC
15418 (TARGET_FLOAT128_TYPE): Likewise.
15419 (-mfloat128-type): Likewise.
15420 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15421 Delete the -mfloat128-type option and make -mfloat128 default on
15422 PowerPC Linux systems. Always use __ieee128 to be the keyword for
15423 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
15424 128-bit floating point is enabled. Change tests from using
15425 -mfloat128-type to -mfloat128.
15426 (rs6000_mangle_type): Use the correct mangling for the __float128
15427 type even if normal long double is restricted to 64-bits.
15428 (floatn_mode): Enable the _Float128 type by default on VSX Linux
15430 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
15431 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
15432 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
15434 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
15435 documentation for -mfloat128.
15437 2017-09-06 Olivier Hainque <hainque@adacore.com>
15439 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
15441 2017-09-06 Wish Wu <wishwu007@gmail.com>
15442 Jakub Jelinek <jakub@redhat.com>
15444 * asan.c (initialize_sanitizer_builtins): Add
15445 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
15446 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
15447 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
15448 BT_FN_VOID_UINT64_PTR variables.
15449 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
15450 (BT_FN_VOID_UINT16_UINT16): Likewise.
15451 (BT_FN_VOID_UINT32_UINT32): Likewise.
15452 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
15453 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
15454 (BT_FN_VOID_UINT64_PTR): Likewise.
15455 * common.opt (flag_sanitize_coverage): New variable.
15456 (fsanitize-coverage=trace-pc): Remove.
15457 (fsanitize-coverage=): Add.
15458 * flag-types.h (enum sanitize_coverage_code): New enum.
15459 * fold-const.c (fold_range_test): Disable non-short-circuit
15460 optimization if flag_sanitize_coverage.
15461 (fold_truth_andor): Likewise.
15462 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15463 * opts.c (COVERAGE_SANITIZER_OPT): Define.
15464 (coverage_sanitizer_opts): New array.
15465 (get_closest_sanitizer_option): Add OPTS argument, handle also
15466 OPT_fsanitize_coverage_.
15467 (parse_sanitizer_options): Adjusted to also handle
15468 OPT_fsanitize_coverage_.
15469 (common_handle_option): Add OPT_fsanitize_coverage_.
15470 * sancov.c (instrument_comparison, instrument_switch): New function.
15471 (sancov_pass): Add trace-cmp support.
15472 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
15473 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
15474 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
15475 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
15476 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
15477 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
15478 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
15479 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
15480 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
15481 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
15483 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15485 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
15486 error. Only quit immediately if parsing is complete.
15487 (BEGIN): Initialize fatal_err and parse_done.
15488 (begin fpu, end fpu): Check number of arguments.
15489 (begin arch, end arch): Likewise.
15490 (begin cpu, end cpu): Likewise.
15491 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
15492 (optalias): Likewise.
15494 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15496 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
15497 * config/arm/arm-isa.h: Delete. Move definitions to ...
15498 * arm-cpus.in: ... here. Use new feature and fgroup values.
15499 * config/arm/arm.c (arm_option_override): Use lower case for feature
15501 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
15502 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
15503 * config/arm/parsecpu.awk (END): Add new command 'isa'.
15505 (print_isa_bits_for): New function.
15506 (gen_isa): New function.
15507 (gen_comm_data): Use print_isa_bits_for.
15508 (define feature): New keyword.
15509 (define fgroup): New keyword.
15510 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
15511 (arm-isa.h): Add rule to generate file.
15512 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
15513 case for feature bit names.
15515 2017-09-06 Richard Biener <rguenther@suse.de>
15517 * tree-ssa-pre.c (NECESSARY): Remove.
15518 (create_expression_by_pieces): Do not touch pass-local flags.
15519 (insert_into_preds_of_block): Likewise.
15520 (do_pre_regular_insertion): Likewise.
15521 (eliminate_insert): Likewise.
15522 (eliminate_dom_walker::before_dom_children): Likewise.
15523 (fini_eliminate): Do not look at inserted_exprs.
15524 (mark_operand_necessary): Remove.
15525 (remove_dead_inserted_code): Replace with simple work-list
15526 algorithm based on inserted_exprs and SSA uses.
15527 (pass_pre::execute): Re-order fini_eliminate and
15528 remove_dead_inserted_code.
15530 2017-09-06 Olivier Hainque <hainque@adacore.com>
15532 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
15533 for VxWorks 7. Adjust surrounding comments.
15535 2017-09-06 Richard Biener <rguenther@suse.de>
15537 * gimple-ssa-strength-reduction.c
15538 (find_candidates_dom_walker::before_dom_children): Also allow
15541 2017-09-06 Richard Biener <rguenther@suse.de>
15543 PR tree-optimization/82108
15544 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
15545 for gap in the non-permutation SLP case.
15547 2017-09-06 Martin Jambor <mjambor@suse.cz>
15549 PR tree-optimization/82078
15550 * tree-sra.c (sort_and_splice_var_accesses): Move call to
15551 add_access_to_work_queue...
15552 (build_accesses_from_assign): ...here.
15553 (propagate_all_subaccesses): Make sure racc is the group
15554 representative, if there is one.
15556 2017-09-06 Jakub Jelinek <jakub@redhat.com>
15558 PR middle-end/82095
15559 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
15562 2017-09-06 Richard Biener <rguenther@suse.de>
15564 * gimple-ssa-strength-reduction.c
15565 (find_candidates_dom_walker::before_doom_children): Use a
15566 type and not a mode check.
15568 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
15571 * config/arm/predicates.md (arm_general_adddi_operand): Create new
15573 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
15575 2017-09-05 Jeff Law <law@redhat.com>
15577 PR tree-optimization/64910
15578 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
15579 cases where we have 3 or more operands.
15581 2017-09-05 Jakub Jelinek <jakub@redhat.com>
15583 PR middle-end/81768
15584 * omp-low.c (lower_omp_for): Recompute tree invariant if
15585 gimple_omp_for_initial/final is ADDR_EXPR.
15587 PR middle-end/81768
15588 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
15589 into gimple val before gimplification fo the COND_EXPR.
15591 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
15593 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
15594 REGION_COPY argument.
15595 (thread_through_all_blocks): Remove unused argument to
15596 duplicate_thread_path.
15598 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15599 Alan Hayward <alan.hayward@arm.com>
15600 David Sherwood <david.sherwood@arm.com>
15602 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
15603 Take a scalar_mode rather than a machine_mode.
15604 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15605 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
15606 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15607 (aarch64_gen_adjusted_ldpstp): Likewise.
15608 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
15610 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15611 Alan Hayward <alan.hayward@arm.com>
15612 David Sherwood <david.sherwood@arm.com>
15614 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
15615 Take a scalar_int_mode instead of a machine_mode.
15616 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15617 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15618 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15619 (aarch64_simd_attr_length_rglist): Delete.
15620 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
15621 a scalar_int_mode instead of a machine_mode.
15622 (aarch64_add_offset): Likewise.
15623 (aarch64_internal_mov_immediate): Likewise
15624 (aarch64_add_constant_internal): Likewise.
15625 (aarch64_add_constant): Likewise.
15626 (aarch64_movw_imm): Likewise.
15627 (aarch64_rtx_arith_op_extract_p): Likewise.
15628 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15629 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15630 Remove assert that the mode isn't a vector.
15631 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15632 (aarch64_expand_mov_immediate): Update calls after above changes.
15633 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
15634 (aarch64_and_bitmask_imm): Check for scalar integer modes.
15635 (aarch64_move_imm): Likewise.
15636 (aarch64_can_const_movi_rtx_p): Likewise.
15637 (aarch64_strip_extend): Likewise.
15638 (aarch64_extr_rtx_p): Likewise.
15639 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
15640 a CONST_INT when the mode parameter is VOIDmode.
15641 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
15643 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15645 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
15646 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15647 (bitwise_type_for_mode): Update accordingly.
15649 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15651 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
15652 * stor-layout.c (mode_for_size_tree): Likewise.
15653 (mode_for_array): Update accordingly.
15654 (layout_decl): Likewise.
15655 (compute_record_mode): Likewise. Only set the mode once.
15657 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15659 * target.def (get_mask_mode): Change return type to opt_mode.
15661 * doc/tm.texi: Regenerate.
15662 * targhooks.h (default_get_mask_mode): Return an opt_mode.
15663 * targhooks.c (default_get_mask_mode): Likewise.
15664 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15665 * optabs-query.c (can_vec_mask_load_store_p): Update use of
15666 targetm.get_mask_mode.
15667 * tree.c (build_truth_vector_type): Likewise.
15669 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15671 * machmode.h (mode_for_vector): Return an opt_mode.
15672 * stor-layout.c (mode_for_vector): Likewise.
15673 (mode_for_int_vector): Update accordingly.
15674 (layout_type): Likewise.
15675 * config/i386/i386.c (emit_memmov): Likewise.
15676 (ix86_expand_set_or_movmem): Likewise.
15677 (ix86_expand_vector_init): Likewise.
15678 (ix86_get_mask_mode): Likewise.
15679 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
15681 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
15682 * expmed.c (extract_bit_field_1): Likewise.
15683 * expr.c (expand_expr_real_2): Likewise.
15684 * optabs-query.c (can_vec_perm_p): Likewise.
15685 (can_vec_mask_load_store_p): Likewise.
15686 * optabs.c (expand_vec_perm): Likewise.
15687 * targhooks.c (default_get_mask_mode): Likewise.
15688 * tree-vect-stmts.c (vectorizable_store): Likewise.
15689 (vectorizable_load): Likewise.
15690 (get_vectype_for_scalar_type_and_size): Likewise.
15692 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15694 * machmode.h (mode_for_int_vector): New function.
15695 * stor-layout.c (mode_for_int_vector): Likewise.
15696 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
15697 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
15698 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15699 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15700 (s390_expand_vcond): Likewise.
15702 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15704 * machmode.h (opt_machine_mode): New type.
15705 (opt_mode<T>): Allow construction from anything that can be
15707 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
15708 (mode_for_size): Return an opt_machine_mode.
15709 * stor-layout.c (mode_for_size): Likewise.
15710 (mode_for_size_tree): Update call accordingly.
15711 (bitwise_mode_for_mode): Likewise.
15712 (make_fract_type): Likewise.
15713 (make_accum_type): Likewise.
15714 * caller-save.c (replace_reg_with_saved_mem): Update call
15716 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15717 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15718 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15719 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15720 * expmed.c (extract_bit_field_1): Likewise.
15721 * reload.c (get_secondary_mem): Likewise.
15722 * varasm.c (assemble_integer): Likewise.
15723 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
15726 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15728 * machmode.h (decimal_float_mode_for_size): New function.
15729 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
15730 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
15731 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
15732 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
15733 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
15734 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
15736 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15738 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
15739 (get_builtin_sync_mode): Likewise.
15740 (expand_ifn_atomic_compare_exchange): Likewise.
15741 (expand_builtin_atomic_clear): Likewise.
15742 (expand_builtin_atomic_test_and_set): Likewise.
15743 (fold_builtin_atomic_always_lock_free): Likewise.
15744 * calls.c (compute_argument_addresses): Likewise.
15745 (emit_library_call_value_1): Likewise.
15746 (store_one_arg): Likewise.
15747 * combine.c (combine_instructions): Likewise.
15748 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
15749 * config/arm/arm.c (arm_function_value): Likewise.
15750 (aapcs_allocate_return_reg): Likewise.
15751 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
15752 * config/i386/i386.c (construct_container): Likewise.
15753 (ix86_gimplify_va_arg): Likewise.
15754 (ix86_expand_sse_cmp): Likewise.
15755 (emit_memmov): Likewise.
15756 (emit_memset): Likewise.
15757 (expand_small_movmem_or_setmem): Likewise.
15758 (ix86_expand_pextr): Likewise.
15759 (ix86_expand_pinsr): Likewise.
15760 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
15761 * config/microblaze/microblaze.c (microblaze_block_move_straight):
15763 * config/mips/mips.c (mips_function_value_1) Likewise.
15764 (mips_block_move_straight): Likewise.
15765 (mips_expand_ins_as_unaligned_store): Likewise.
15766 * config/powerpcspe/powerpcspe.c
15767 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15768 (rs6000_darwin64_record_arg_flush): Likewise.
15769 * config/rs6000/rs6000.c
15770 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15771 (rs6000_darwin64_record_arg_flush): Likewise.
15772 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
15773 (sparc_function_value_1): Likewise.
15774 * config/spu/spu.c (adjust_operand): Likewise.
15775 (spu_emit_branch_or_set): Likewise.
15776 (arith_immediate_p): Likewise.
15777 * emit-rtl.c (gen_lowpart_common): Likewise.
15778 * expr.c (expand_expr_real_1): Likewise.
15779 * function.c (assign_parm_setup_block): Likewise.
15780 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
15781 * reload1.c (alter_reg): Likewise.
15782 * stor-layout.c (mode_for_vector): Likewise.
15783 (layout_type): Likewise.
15785 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15787 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
15788 (spu_convert_move): Likewise.
15789 * lower-subreg.c (resolve_simple_move): Likewise.
15791 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15794 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
15795 define_insn to a define_expand.
15796 (altivec_vsum2sws_direct): New define_insn.
15797 (altivec_vsumsws): Convert from a define_insn to a define_expand.
15799 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15801 * config/arm/arm.c (arm_option_params_internal): Improve setting of
15804 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
15809 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
15810 realign stack if stack alignment needed is less than incoming
15813 2017-09-05 Marek Polacek <polacek@redhat.com>
15816 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
15819 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15821 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
15823 2017-09-05 Richard Biener <rguenther@suse.de>
15825 PR tree-optimization/82084
15826 * fold-const.c (can_native_encode_string_p): Handle wide characters.
15828 2017-09-05 Richard Biener <rguenther@suse.de>
15830 PR tree-optimization/82102
15831 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
15833 2017-09-05 Martin Liska <mliska@suse.cz>
15835 PR tree-optimization/82032
15836 * tree-cfg.c (generate_range_test): New function.
15837 * tree-cfg.h (generate_range_test): Declared here.
15838 * tree-cfgcleanup.c (convert_single_case_switch): New function.
15839 (cleanup_control_expr_graph): Use it.
15840 * tree-switch-conversion.c (try_switch_expansion): Remove
15842 (emit_case_nodes): Use generate_range_test.
15844 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15847 * config/i386/i386.md (*<btsc><mode>_mask): Add
15848 TARGET_USE_BT to insn constraint.
15849 (*btr<mode>_mask): Ditto.
15851 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
15853 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
15854 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
15856 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
15859 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
15860 TARGET_NEON and TARGET_IWMMXT.
15861 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
15862 TARGET_NEON and TARGET_IWMMXT.
15863 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
15865 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15867 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
15868 (ix86_rewrite_tls_address): Ditto.
15869 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
15870 (ix86_rewrite_tls_address_1): Ditto.
15871 (ix86_rewrite_tls_address): Ditto.
15872 * config/i386/predicates.md (tls_address_pattern): New predicate.
15873 * config/i386/i386.md (TLS address splitter): New splitter.
15875 2017-09-04 Richard Biener <rguenther@suse.de>
15877 PR tree-optimization/82084
15878 * fold-const.h (can_native_encode_string_p): Declare.
15879 * fold-const.c (can_native_encode_string_p): Factor out from ...
15880 (native_encode_string): ... here.
15881 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
15882 vectorizing stores from constants we later cannot handle.
15884 2017-09-04 Marek Polacek <polacek@redhat.com>
15887 * doc/invoke.texi: Update -Wtautological-compare documentation.
15889 2017-09-04 Jeff Law <law@redhat.com>
15891 PR tree-optimization/64910
15892 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
15893 swap the first and last operand if the last is a constant.
15895 2017-09-04 Marek Polacek <polacek@redhat.com>
15898 * convert.c (do_narrow): When sanitizing signed integer overflows,
15899 bail out for signed types.
15900 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
15902 2017-09-04 Richard Biener <rguenther@suse.de>
15904 PR tree-optimization/82060
15905 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15906 Move devirtualization after stmt folding and before EH/AB/noreturn
15907 cleanup to get the stmt refs canonicalized. Use a bool instead
15908 of gimple_modified_p since that doesn't work for NOPs. Schedule
15909 NOPs generated by folding for removal.
15911 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15912 Alan Hayward <alan.hayward@arm.com>
15913 David Sherwood <david.sherwood@arm.com>
15915 * coretypes.h (pad_direction): New enum.
15916 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
15917 (FUNCTION_ARG_PADDING): Likewise.
15918 * target.def (function_arg_padding): New hook.
15919 * targhooks.h (default_function_arg_padding): Declare.
15920 * targhooks.c (default_function_arg_padding): New function.
15921 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
15922 (TARGET_FUNCTION_ARG_PADDING): ...this.
15923 * doc/tm.texi: Regenerate.
15924 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
15925 instead of direction.
15926 (compute_argument_addresses): Likewise.
15927 (load_register_parameters): Likewise.
15928 (emit_library_call_value_1): Likewise.
15929 (store_one_arg): Use targetm.calls.function_arg_padding instead
15930 of FUNCTION_ARG_PADDING.
15931 (must_pass_in_stack_var_size_or_pad): Likewise.
15932 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
15933 (emit_group_store): Likewise.
15934 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
15935 instead of FUNCTION_ARG_PADDING.
15936 (emit_push_insn): Likewise, and propagate enum change throughout
15938 * function.h (direction): Delete.
15939 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
15941 * function.c (assign_parm_find_stack_rtl): Likewise.
15942 (assign_parm_setup_block_p): Likewise.
15943 (assign_parm_setup_block): Likewise.
15944 (gimplify_parameters): Likewise.
15945 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
15946 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
15948 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
15949 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15950 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
15951 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
15952 (aarch64_function_arg_padding): ...this new function.
15953 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
15954 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15955 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
15956 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15957 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
15958 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15959 (arm_pad_arg_upward): Replace with...
15960 (arm_function_arg_padding): ...this new function.
15961 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
15963 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
15964 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
15965 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15966 (ia64_hpux_function_arg_padding): Replace with...
15967 (ia64_function_arg_padding): ...this new function. Use pad_direction
15968 instead of direction. Check for TARGET_HPUX.
15969 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
15970 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15971 (iq2000_function_arg_padding): New function.
15972 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
15973 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
15974 (mips_function_arg_padding): ...this new function.
15975 (mips_pad_reg_upward): Update accordingly.
15976 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15977 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
15978 targetm.calls.function_arg_padding.
15979 (FUNCTION_ARG_PADDING): Delete.
15980 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15981 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
15982 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15983 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
15984 (nios2_block_reg_padding): Return pad_direction instead of direction.
15985 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
15986 instead of direction.
15987 (nios2_function_arg_padding): Likewise. Make static.
15988 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15989 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
15990 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
15991 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
15992 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15993 (pa_function_arg_padding): Make static. Return pad_direction instead
15995 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
15996 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15997 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
15998 instead of direction. Use targetm.calls.function_arg_padding.
15999 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
16000 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16001 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
16002 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
16003 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
16005 (function_arg_padding): Rename to...
16006 (rs6000_function_arg_padding): ...this. Make static. Return
16007 pad_direction instead of direction.
16008 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16009 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
16010 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16011 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
16012 instead of direction. Use targetm.calls.function_arg_padding.
16013 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
16014 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16015 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
16016 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
16017 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16018 (function_arg_padding): Rename to...
16019 (rs6000_function_arg_padding): ...this. Make static. Return
16020 pad_direction instead of direction.
16021 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16022 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
16023 * config/s390/s390.c (s390_function_arg_padding): New function.
16024 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16025 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
16026 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
16027 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16028 (function_arg_padding): Rename to...
16029 (sparc_function_arg_padding): ...this. Make static. Return
16030 pad_direction instead of direction.
16031 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
16032 * config/spu/spu.c (spu_function_arg_padding): New function.
16033 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16034 * system.h (FUNCTION_ARG_PADDING): Poison.
16036 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16037 Alan Hayward <alan.hayward@arm.com>
16038 David Sherwood <david.sherwood@arm.com>
16040 * target.def (modes_tieable_p): New hook.
16041 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
16042 (TARGET_MODES_TIEABLE_P): ...this.
16043 * doc/tm.texi.in: Regenerate.
16044 * hooks.h (hook_bool_mode_mode_true): Declare.
16045 * hooks.c (hook_bool_mode_mode_true): New function.
16046 * combine.c (subst): Use targetm.modes_tieable_p instead of
16048 * dse.c (find_shift_sequence): Likewise.
16049 * expmed.c (extract_low_bits): Likewise.
16050 * lower-subreg.c: Include target.h.
16051 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
16053 * rtlanal.c (rtx_cost): Likewise.
16054 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
16055 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
16056 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
16057 (TARGET_MODES_TIEABLE_P): Redefine.
16058 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
16059 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
16060 (TARGET_MODES_TIEABLE_P): Redefine.
16061 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
16062 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
16063 (arc_modes_tieable_p): New function.
16064 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
16065 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
16066 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
16067 (arm_modes_tieable_p): Make static.
16068 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
16069 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
16070 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
16071 (TARGET_MODES_TIEABLE_P): Redefine.
16072 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
16073 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
16074 (TARGET_MODES_TIEABLE_P): Redefine.
16075 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
16076 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
16077 (cr16_modes_tieable_p): New function.
16078 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
16079 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
16080 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
16081 (TRULY_NOOP_TRUNCATION): Update comment.
16082 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
16083 (TRULY_NOOP_TRUNCATION): Update comment.
16084 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
16085 (frv_modes_tieable_p): New function.
16086 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
16087 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
16088 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
16089 (TARGET_MODES_TIEABLE_P): Redefine.
16090 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
16091 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
16092 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
16093 (TARGET_MODES_TIEABLE_P): Redefine.
16094 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
16095 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
16096 (ia64_modes_tieable_p): New function.
16097 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
16098 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
16099 (iq2000_modes_tieable_p): New function.
16100 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
16101 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
16102 (lm32_modes_tieable_p): New function.
16103 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
16104 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
16105 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
16106 (TARGET_MODES_TIEABLE_P): Redefine.
16107 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
16108 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
16109 (m32r_modes_tieable_p): New function.
16110 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
16111 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
16112 (m68k_modes_tieable_p): New function.
16113 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
16114 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
16115 (mcore_modes_tieable_p): New function.
16116 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
16117 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
16119 (TARGET_MODES_TIEABLE_P): Redefine.
16120 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
16121 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
16122 * config/mips/mips.c (mips_modes_tieable_p): Make static.
16123 (TARGET_MODES_TIEABLE_P): Redefine.
16124 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
16125 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
16126 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
16127 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
16128 (mn10300_modes_tieable_p): ...this and make static.
16129 (TARGET_MODES_TIEABLE_P): Redefine.
16130 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
16131 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
16132 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
16133 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
16134 (msp430_modes_tieable_p): Make static.
16135 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
16136 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
16137 (TARGET_MODES_TIEABLE_P): Redefine.
16138 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
16139 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
16140 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
16141 (TARGET_MODES_TIEABLE_P): Redefine.
16142 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
16143 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
16144 * config/pa/pa.c (pa_modes_tieable_p): Make static.
16145 (TARGET_MODES_TIEABLE_P): Redefine.
16146 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
16147 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
16148 (pdp11_modes_tieable_p): New function.
16149 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
16150 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
16151 (rs6000_modes_tieable_p): New function.
16152 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16153 * config/powerpcspe/powerpcspe.md: Update comment.
16154 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
16155 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
16156 (TARGET_MODES_TIEABLE_P): Redefine.
16157 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
16158 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
16159 (rl78_modes_tieable_p): New function.
16160 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
16161 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
16162 (rs6000_modes_tieable_p): New function.
16163 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16164 * config/rs6000/rs6000.md: Update comment.
16165 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
16166 * config/rx/rx.c (rx_modes_tieable_p): New function.
16167 (TARGET_MODES_TIEABLE_P): Redefine.
16168 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
16169 * config/s390/s390.c (s390_modes_tieable_p): New function.
16170 (TARGET_MODES_TIEABLE_P): Redefine.
16171 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
16172 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
16173 (sh_modes_tieable_p): New function.
16174 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
16175 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
16176 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
16177 (sparc_modes_tieable_p): Make static.
16178 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
16179 * config/spu/spu.c (spu_modes_tieable_p): New function.
16180 (TARGET_MODES_TIEABLE_P): Redefine.
16181 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
16182 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
16183 (TARGET_MODES_TIEABLE_P): Redefine.
16184 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
16185 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
16186 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
16187 * config/v850/v850.c (v850_modes_tieable_p): New function.
16188 (TARGET_MODES_TIEABLE_P): Redefine.
16189 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
16190 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
16191 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
16192 (visium_modes_tieable_p): New function.
16193 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
16194 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
16195 (xtensa_modes_tieable_p): New function.
16196 * system.h (MODES_TIEABLE_P): Poison.
16198 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16199 Alan Hayward <alan.hayward@arm.com>
16200 David Sherwood <david.sherwood@arm.com>
16202 * target.def (hard_regno_mode_ok): New hook.
16203 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
16204 (TARGET_HARD_REGNO_MODE_OK): ...this.
16205 * doc/tm.texi.in: Regenerate.
16206 * hooks.h (hook_bool_uint_mode_true): Declare.
16207 * hooks.c (hook_bool_uint_mode_true): New function.
16208 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
16209 HARD_REGNO_MODE_OK.
16210 * genpreds.c (write_insn_preds_c): Add an include of target.h.
16211 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
16212 instead of HARD_REGNO_MODE_OK.
16213 * caller-save.c: Include target.h.
16214 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
16215 HARD_REGNO_MODE_OK.
16216 * combine.c (can_combine_p): Likewise.
16217 (combinable_i3pat): Likewise.
16218 (can_change_dest_mode): Likewise.
16219 * expr.c (init_expr_target): Likewise.
16220 (convert_move): Likewise.
16221 (convert_modes): Likewise.
16222 * ira.c (setup_prohibited_class_mode_regs): Likewise.
16223 (setup_prohibited_mode_move_regs): Likewise.
16224 * ira.h (target_ira): Likewise.
16225 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16226 * lra-constraints.c (process_alt_operands): Likewise.
16227 (split_reg): Likewise.
16228 * recog.c (peep2_find_free_register): Likewise.
16229 * ree.c (combine_reaching_defs): Likewise.
16230 * regcprop.c (maybe_mode_change): Likewise.
16231 * reginfo.c (init_reg_sets_1): Likewise.
16232 (choose_hard_reg_mode): Likewise.
16233 (simplifiable_subregs): Likewise.
16234 * regrename.c (check_new_reg_p): Likewise.
16235 * reload.c (find_valid_class): Likewise.
16236 (find_valid_class_1): Likewise.
16237 (reload_inner_reg_of_subreg): Likewise.
16238 (push_reload): Likewise.
16239 (combine_reloads): Likewise.
16240 (find_dummy_reload): Likewise.
16241 (find_reloads): Likewise.
16242 * reload1.c (find_reg): Likewise.
16243 (set_reload_reg): Likewise.
16244 (allocate_reload_reg): Likewise.
16245 (choose_reload_regs): Likewise.
16246 (reload_adjust_reg_for_temp): Likewise.
16247 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
16248 (simplify_subreg_regno): Likewise.
16249 * sel-sched.c (init_regs_for_mode): Likewise.
16250 * varasm.c (make_decl_rtl): Likewise.
16251 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
16252 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
16253 HARD_REGNO_MODE_OK.
16254 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
16255 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
16256 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16257 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
16258 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
16259 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16260 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
16261 (arc_mode_class): Delete.
16262 (HARD_REGNO_MODE_OK): Delete.
16263 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16264 (arc_hard_regno_mode_ok): Rename old array to...
16265 (arc_hard_regno_mode_ok_modes): ...this.
16266 (arc_conditional_register_usage): Update accordingly.
16267 (arc_mode_class): Make static.
16268 (arc_hard_regno_mode_ok): New function.
16269 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
16270 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
16271 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16272 (arm_hard_regno_mode_ok): Make static.
16273 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
16274 HARD_REGNO_MODE_OK.
16275 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
16276 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
16277 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
16279 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16280 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
16281 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
16282 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
16283 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
16284 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16285 * config/bfin/predicates.md (valid_reg_operand): Use
16286 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
16287 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
16288 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
16289 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16290 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
16291 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
16292 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16293 (cr16_hard_regno_mode_ok): Make static and return a bool.
16294 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
16295 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16296 (cris_hard_regno_mode_ok): New function.
16297 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
16298 (epiphany_mode_class): Delete.
16299 (HARD_REGNO_MODE_OK): Delete.
16300 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
16301 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16302 (hard_regno_mode_ok): Rename to...
16303 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
16304 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
16305 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
16306 HARD_REGNO_MODE_OK.
16307 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
16308 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
16309 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16310 (frv_hard_regno_mode_ok): Make static and return a bool.
16311 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
16312 HARD_REGNO_MODE_OK.
16313 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
16314 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
16315 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
16316 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
16318 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16319 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
16320 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
16321 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
16323 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16324 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
16325 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16326 (ia64_hard_regno_mode_ok): New function.
16327 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
16328 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16329 (iq2000_hard_regno_mode_ok): New function.
16330 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
16331 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16332 (lm32_hard_regno_mode_ok): New function.
16333 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
16334 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
16335 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
16336 instead of HARD_REGNO_MODE_OK.
16337 (m32c_hard_regno_ok): Rename to...
16338 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
16339 (m32c_cannot_change_mode_class): Update accordingly.
16340 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16341 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
16342 (m32r_mode_class): Delete.
16343 (HARD_REGNO_MODE_OK): Delete.
16344 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16345 (m32r_hard_regno_mode_ok): Rename to...
16346 (m32r_hard_regno_modes): ...this.
16347 (m32r_mode_class): Make static.
16348 (m32r_hard_regno_mode_ok): New function.
16349 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
16350 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
16351 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16352 (m68k_hard_regno_mode_ok): Make static.
16353 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
16354 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16355 (mcore_hard_regno_mode_ok): New function.
16356 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
16357 (HARD_REGNO_MODE_OK): Delete.
16358 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
16360 (microblaze_hard_regno_mode_ok_p): ...this and make static.
16361 (microblaze_hard_regno_mode_ok): New function.
16362 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16363 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
16364 (mips_hard_regno_mode_ok): Delete.
16365 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
16366 (mips_hard_regno_mode_ok_p): ...this and make static.
16367 (mips_hard_regno_mode_ok_p): Rename to...
16368 (mips_hard_regno_mode_ok_uncached): ...this.
16369 (mips_hard_regno_mode_ok): New function.
16370 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
16371 of HARD_REGNO_MODE_OK.
16372 (mips_option_override): Update after above name changes.
16373 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16374 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
16375 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
16376 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
16377 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
16378 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16379 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
16380 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
16381 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
16382 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16383 (msp430_hard_regno_mode_ok): Make static and return a bool.
16384 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
16385 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
16386 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
16388 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16389 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
16390 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
16391 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
16392 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
16393 (PA_HARD_REGNO_MODE_OK): ...this
16394 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
16395 (PA_HARD_REGNO_MODE_OK): ...this.
16396 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16397 (pa_hard_regno_mode_ok): New function.
16398 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
16399 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16400 (pdp11_hard_regno_mode_ok): New function.
16401 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
16402 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
16404 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
16406 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16407 (rs6000_hard_regno_mode_ok): Rename to...
16408 (rs6000_hard_regno_mode_ok_uncached): ...this.
16409 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16410 (rs6000_hard_regno_mode_ok): New function.
16411 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
16412 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
16413 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
16414 (riscv_hard_regno_mode_ok): ...this and make static.
16415 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16416 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
16417 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
16418 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16419 (rl78_hard_regno_mode_ok): Make static and return bool.
16420 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
16421 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
16423 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
16424 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16425 (rs6000_hard_regno_mode_ok): Rename to...
16426 (rs6000_hard_regno_mode_ok_uncached): ...this.
16427 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16428 (rs6000_hard_regno_mode_ok): New function.
16429 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
16430 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
16431 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16432 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
16433 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
16434 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
16435 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16436 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
16437 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
16438 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16439 (sh_hard_regno_mode_ok): Make static.
16440 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
16441 instead of HARD_REGNO_MODE_OK.
16442 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
16443 (sparc_mode_class): Delete.
16444 (HARD_REGNO_MODE_OK): Delete.
16445 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16446 (hard_regno_mode_classes): Make static.
16447 (sparc_mode_class): Likewise.
16448 (sparc_hard_regno_mode_ok): New function.
16449 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
16450 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
16451 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
16453 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16454 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
16455 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
16456 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
16457 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
16458 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16459 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
16460 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
16461 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16462 (visium_hard_regno_mode_ok): New function.
16463 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
16464 instead of HARD_REGNO_MODE_OK.
16465 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
16466 (HARD_REGNO_MODE_OK): Delete.
16467 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
16468 (xtensa_hard_regno_mode_ok_p): ...this and make static.
16469 (xtensa_option_override): Update accordingly.
16470 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16471 (xtensa_hard_regno_mode_ok): New function.
16472 * system.h (HARD_REGNO_MODE_OK): Poison.
16474 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16475 Alan Hayward <alan.hayward@arm.com>
16476 David Sherwood <david.sherwood@arm.com>
16478 * target.def (hard_regno_call_part_clobbered): New hook.
16479 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
16480 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
16481 * doc/tm.texi: Regenerate.
16482 * hooks.h (hook_bool_uint_mode_false): Declare.
16483 * hooks.c (hook_bool_uint_mode_false): New function.
16484 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16485 * cselib.c (cselib_process_insn): Use
16486 targetm.hard_regno_call_part_clobbered instead of
16487 HARD_REGNO_CALL_PART_CLOBBERED.
16488 * ira-conflicts.c (ira_build_conflicts): Likewise.
16489 * ira-costs.c (ira_tune_allocno_costs): Likewise.
16490 * lra-constraints.c (need_for_call_save_p): Likewise.
16491 * lra-lives.c: Include target.h.
16492 (check_pseudos_live_through_calls): Use
16493 targetm.hard_regno_call_part_clobbered instead of
16494 HARD_REGNO_CALL_PART_CLOBBERED.
16495 * regcprop.c: Include target.h.
16496 (copyprop_hardreg_forward_1): Use
16497 targetm.hard_regno_call_part_clobbered instead of
16498 HARD_REGNO_CALL_PART_CLOBBERED.
16499 * reginfo.c (choose_hard_reg_mode): Likewise.
16500 * regrename.c (check_new_reg_p): Likewise.
16501 * reload.c (find_equiv_reg): Likewise.
16502 * reload1.c (emit_reload_insns): Likewise.
16503 * sched-deps.c (deps_analyze_insn): Likewise.
16504 * sel-sched.c (init_regs_for_mode): Likewise.
16505 (mark_unavailable_hard_regs): Likewise.
16506 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
16507 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16508 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
16510 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16511 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16512 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
16514 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
16516 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16517 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16518 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
16520 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16521 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16522 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
16524 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16525 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
16527 * config/powerpcspe/powerpcspe.c
16528 (rs6000_hard_regno_call_part_clobbered): New function.
16529 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16530 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16531 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
16533 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16534 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16535 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
16537 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16538 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16539 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
16541 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16542 Alan Hayward <alan.hayward@arm.com>
16543 David Sherwood <david.sherwood@arm.com>
16545 * rtl.h (subreg_memory_offset): Declare.
16546 * emit-rtl.c (subreg_memory_offset): New function.
16547 * expmed.c (store_bit_field_1): Use it.
16548 * expr.c (undefined_operand_subword_p): Likewise.
16549 * simplify-rtx.c (simplify_subreg): Likewise.
16551 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
16553 PR rtl-optimization/57448
16556 * optabs.c (expand_atomic_load): Place compiler memory barriers if
16557 using atomic_load pattern.
16558 (expand_atomic_store): Likewise.
16560 2017-09-04 Jakub Jelinek <jakub@redhat.com>
16563 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
16564 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
16565 handling. Use replace_call_with_value with NULL instead of
16566 gsi_replace, unlink_stmt_vdef and release_defs.
16568 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
16571 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
16573 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16576 * rtl.h (emit_library_call_value_1): Declare.
16577 (emit_library_call): Replace declaration with a series of overloads.
16578 Remove the parameter count argument.
16579 (emit_library_call_value): Likewise.
16580 * calls.c (emit_library_call_value_1): Make global. Replace varargs
16581 with an "rtx_mode_t *".
16582 (emit_library_call_value): Delete.
16583 (emit_library_call): Likewise.
16584 * asan.c (asan_emit_stack_protection): Update calls accordingly.
16585 (asan_emit_allocas_unpoison): Likewise.
16586 * builtins.c (expand_builtin_powi): Likewise.
16587 (expand_asan_emit_allocas_unpoison): Likewise.
16588 * cfgexpand.c (expand_main_function): Likewise.
16589 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
16590 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
16591 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
16592 * config/arm/arm.c (arm_trampoline_init): Likewise.
16593 (arm_call_tls_get_addr): Likewise.
16594 (arm_expand_divmod_libfunc): Likewise.
16595 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
16596 (smulsi3_highpart): Likewise.
16597 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
16598 (c6x_expand_compare): Likewise.
16599 (c6x_expand_movmem): Likewise.
16600 * config/frv/frv.c (frv_trampoline_init): Likewise.
16601 * config/i386/i386.c (ix86_trampoline_init): Likewise.
16602 (ix86_expand_divmod_libfunc): Likewise.
16603 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
16604 (ia64_expand_compare): Likewise.
16605 (ia64_profile_hook): Likewise.
16606 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
16607 (nonlocal_goto): Likewise.
16608 (restore_stack_nonlocal): Likewise.
16609 * config/m32r/m32r.c (block_move_call): Likewise.
16610 (m32r_trampoline_init): Likewise.
16611 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
16612 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
16613 (m68k_call_m68k_read_tp): Likewise.
16614 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
16615 (microblaze_expand_divide): Likewise.
16616 * config/mips/mips.h (mips_args): Likewise.
16617 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
16618 (MIPS_ICACHE_SYNC): Likewise.
16619 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
16620 (nios2_trampoline_init): Likewise.
16621 * config/pa/pa.c (hppa_tls_call): Likewise.
16622 (pa_trampoline_init): Likewise.
16623 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
16624 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
16625 (expand_strn_compare): Likewise.
16626 (rs6000_generate_compare): Likewise.
16627 (rs6000_expand_float128_convert): Likewise.
16628 (output_profile_hook): Likewise.
16629 (rs6000_trampoline_init): Likewise.
16630 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
16631 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
16632 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
16633 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
16634 (rs6000_generate_compare): Likewise.
16635 (rs6000_expand_float128_convert): Likewise.
16636 (output_profile_hook): Likewise.
16637 (rs6000_trampoline_init): Likewise.
16638 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
16639 * config/sh/sh.c (sh_trampoline_init): Likewise.
16640 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
16641 (sparc_emit_float_lib_cmp): Likewise.
16642 (sparc32_initialize_trampoline): Likewise.
16643 (sparc64_initialize_trampoline): Likewise.
16644 (sparc_profile_hook): Likewise.
16645 * config/spu/spu.c (ea_load_store): Likewise.
16646 * config/spu/spu.md (floatunssidf2): Likewise.
16647 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
16648 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
16649 * config/visium/visium.c (expand_block_move_4): Likewise.
16650 (expand_block_move_2): Likewise.
16651 (expand_block_move_1): Likewise.
16652 (expand_block_set_4): Likewise.
16653 (expand_block_set_2): Likewise.
16654 (expand_block_set_1): Likewise.
16655 (visium_trampoline_init): Likewise.
16656 (visium_profile_hook): Likewise.
16657 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
16658 (xtensa_setup_frame_addresses): Likewise.
16659 (xtensa_trampoline_init): Likewise.
16660 * except.c (sjlj_emit_function_enter): Likewise.
16661 (sjlj_emit_function_exit): Likewise.
16662 * explow.c (allocate_dynamic_stack_space): Likewise.
16663 (probe_stack_range): Likewise.
16664 * expr.c (convert_mode_scalar): Likewise.
16665 * optabs.c (expand_binop): Likewise.
16666 (expand_twoval_binop_libfunc): Likewise.
16667 (expand_unop): Likewise.
16668 (prepare_cmp_insn): Likewise.
16669 (prepare_float_lib_cmp): Likewise.
16670 (expand_float): Likewise.
16671 (expand_fix): Likewise.
16672 (expand_fixed_convert): Likewise.
16673 (maybe_emit_sync_lock_test_and_set): Likewise.
16674 (expand_atomic_compare_and_swap): Likewise.
16675 (expand_mem_thread_fence): Likewise.
16676 (expand_atomic_fetch_op): Likewise.
16678 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
16680 * doc/generic.texi (OpenACC): Adjust URL.
16681 * doc/invoke.texi (C Dialect Options): Ditto.
16683 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
16685 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
16686 predicate for operand 1. Add (m,<S>) constraint.
16687 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
16688 Prevent memory operand 1 with register operand 2.
16690 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
16692 PR rtl-optimization/82024
16693 * combine.c (try_combine): If the combination result is a PARALLEL,
16694 and we only need to retain the SET in there that would be placed
16695 at I2, check that we can place that at I3 instead, before doing so.
16697 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16700 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
16702 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
16703 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
16706 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
16707 Jeff Law <law@redhat.com>
16709 * varasm.c (bss_initializer_p): Do not put constants into .bss
16710 (categorize_decl_for_section): Handle bss_initializer_p returning
16711 false when DECL_INITIAL is NULL.
16713 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16716 * config/s390/s390.c (s390_can_inline_p): New function.
16718 2017-09-01 Jeff Law <law@redhat.com>
16720 PR tree-optimization/82052
16721 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16722 Always initialize the returned slot after a hash table miss
16723 when INSERT is true.
16725 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
16727 * config/s390/s390.md (mem_signal_fence): Remove.
16728 * doc/md.texi (mem_signal_fence): Remove.
16729 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
16731 * target-insns.def (mem_signal_fence): Remove.
16733 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16736 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
16739 * asan.c (create_odr_indicator): Strip name encoding from assembler
16740 name before appending it after __odr_asan_.
16742 2017-09-01 Martin Liska <mliska@suse.cz>
16744 PR tree-optimization/82059
16745 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
16746 frequency only when an edge is redirected.
16748 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16750 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
16751 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
16752 (arc_conditional_register_usage): Remove ARC600 lp_count
16754 (arc_file_start): Emit Tag_ARC_CPU_variation.
16755 (arc_can_use_doloop_p): New conditions to use ZOLs.
16756 (hwloop_fail): New function.
16757 (hwloop_optimize): Likewise.
16758 (hwloop_pattern_reg): Likewise.
16759 (arc_doloop_hooks): New struct, to be used with reorg_loops.
16760 (arc_reorg_loops): New function, calls reorg_loops.
16761 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
16762 (arc600_corereg_hazard): Remove ZOL checking, case handled by
16764 (arc_loop_hazard): Remove function, functionality moved into
16766 (arc_hazard): Remove arc_loop_hazard call.
16767 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
16768 into hwloop_optimize.
16769 (arc_label_align): Remove ZOL handling.
16770 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
16771 * config/arc/arc.md (doloop_begin): Remove pattern.
16772 (doloop_begin_i): Likewise.
16773 (doloop_end_i): Likewise.
16774 (doloop_fallback): Likewise.
16775 (doloop_fallback_m): Likewise.
16776 (doloop_end): Reimplement expand.
16777 (arc_lp): New pattern for LP instruction.
16778 (loop_end): New pattern.
16779 (loop_fail): Likewise.
16780 (decrement_and_branch_until_zero): Likewise.
16781 * config/arc/arc.opt (mlpc-width): New option.
16782 * doc/invoke.texi (mlpc-width): Document option.
16784 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16786 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
16787 (arc_ccfsm_advance): Fix checking for delay slots.
16788 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
16790 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16792 * config/arc/arc.md (movqi_insn): Add stores to save constant long
16794 (movhi_insn): Update store instruction constraint which are saving
16795 6-bit short immediates.
16796 (movsi_insn): Consider also short scaled load operations.
16797 (zero_extendhisi2_i): Use Usd constraint instead of T.
16798 (extendhisi2_i): Add q constraint.
16799 (arc_clzsi2): Add type and length attributes.
16800 (arc_ctzsi2): Likewise.
16801 * config/arc/constraints.md (Usc): Update constraint, the
16802 assembler can parse two relocations for a single instruction.
16804 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16806 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
16807 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
16809 2017-08-31 Olivier Hainque <hainque@adacore.com>
16811 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
16812 match as powerpc-wrs-vxworks*.
16814 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
16816 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
16817 register constraint for by-element operand.
16818 (aarch64_mls_elt_merge<mode>): Likewise.
16820 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16822 * config/arc/arc.c (arc_can_follow_jump): Check for short
16825 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16827 * config.gcc: Use g.opt for arc.
16828 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
16829 functionality moved to ...
16830 (legitimate_scaled_address_p): New function, ...here.
16831 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
16832 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
16833 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
16835 (arc_override_options): Handle G option.
16836 (arc_output_pic_addr_const): Correct function definition.
16837 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
16838 (arc_decl_anon_ns_mem_p): Delete.
16839 (arc_in_small_data_p): Overhaul this function to take into
16840 consideration the value given via G option.
16841 (arc_rewrite_small_data_1): Renamed and corrected old
16842 arc_rewrite_small_data function.
16843 (arc_rewrite_small_data): New function.
16844 (small_data_pattern): Don't use pic_offset_table_rtx.
16845 * config/arc/arc.h (CC1_SPEC): Recognize G option.
16846 * config/arc/simdext.md (movmisalignv2hi): Use
16847 prepare_move_operands function.
16849 (movmisalign*): Likewise.
16850 * doc/invoke.texi (ARC options): Document -G option.
16852 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16854 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
16856 * config/arc/arc.c (arc_print_operand): Output scalled address for
16857 sdata whenever is possible.
16858 (arc_in_small_data_p): Allow sdata for 64bit datum when double
16859 load/stores are available.
16860 (compact_sda_memory_operand): Check for the alignment required by
16861 code density instructions.
16862 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
16864 * config/arc/constraints.md (Usd): Update constraint.
16865 (Us0): New constraint.
16866 (Usc): Update constraint.
16868 2017-08-31 Richard Biener <rguenther@suse.de>
16870 PR middle-end/82054
16871 * dwarf2out.c (dwarf2out_early_global_decl): Process each
16872 function only once.
16874 2017-08-31 Tamar Christina <tamar.christina@arm.com>
16876 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
16877 Resize type_signature.
16879 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16880 Alan Hayward <alan.hayward@arm.com>
16881 David Sherwood <david.sherwood@arm.com>
16883 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
16884 subregs whose inner modes can be stored in GPRs.
16885 (aarch64_classify_index): Likewise.
16887 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16888 Alan Hayward <alan.hayward@arm.com>
16889 David Sherwood <david.sherwood@arm.com>
16891 * config/aarch64/iterators.md (V_cmp_result): Rename to...
16892 (V_INT_EQUIV): ...this.
16893 (v_cmp_result): Rename to...
16894 (v_int_equiv): ...this.
16895 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
16896 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
16897 (copysign<mode>3): Likewise.
16898 (aarch64_simd_bsl<mode>_internal): Likewise.
16899 (aarch64_simd_bsl<mode>): Likewise.
16900 (vec_cmp<mode><mode>): Likewise.
16901 (vcond<mode><mode>): Likewise.
16902 (vcond<v_cmp_mixed><mode>): Likewise.
16903 (vcondu<mode><v_cmp_mixed>): Likewise.
16904 (aarch64_cm<optab><mode>): Likewise.
16905 (aarch64_cmtst<mode>): Likewise.
16906 (aarch64_fac<optab><mode>): Likewise.
16907 (vec_perm_const<mode>): Likewise.
16908 (vcond_mask_<mode><v_cmp_result>): Rename to...
16909 (vcond_mask_<mode><v_int_equiv>): ...this.
16910 (vec_cmp<mode><v_cmp_result>): Rename to...
16911 (vec_cmp<mode><v_int_equiv>): ...this.
16913 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16914 Alan Hayward <alan.hayward@arm.com>
16915 David Sherwood <david.sherwood@arm.com>
16917 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
16919 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
16920 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
16921 (UNSPEC_LD4_DREG): New unspecs.
16922 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
16923 (aarch64_ld2<mode>_dreg_be): Replace with...
16924 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
16926 (aarch64_ld3<mode>_dreg_le)
16927 (aarch64_ld3<mode>_dreg_be): Replace with...
16928 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
16930 (aarch64_ld4<mode>_dreg_le)
16931 (aarch64_ld4<mode>_dreg_be): Replace with...
16932 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
16935 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16937 PR tree-optimization/81987
16938 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
16939 insert an initializer in a location not dominated by the stride
16942 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
16944 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
16945 on the entire header of the finally block in the fallthru case.
16947 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
16949 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
16951 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
16953 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
16954 rs6000_emit_move_from_cr and call renamed function.
16955 (rs6000_emit_prologue): Call renamed functions.
16956 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
16957 movesi_from_cr, remove volatile CRs.
16959 2017-08-30 Jon Beniston <jon@beniston.com>
16960 Richard Biener <rguenther@suse.de>
16962 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
16963 of VECTOR_MODE_P check.
16964 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
16965 element vector types.
16967 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16969 * df.h (df_read_modify_subreg_p): Remove in favor of...
16970 * rtl.h (read_modify_subreg_p): ...this new function. Take a
16971 const_rtx instead of an rtx.
16972 * cprop.c (local_cprop_find_used_regs): Update accordingly.
16973 * df-problems.c (df_word_lr_mark_ref): Likewise.
16974 * ira-lives.c (mark_pseudo_reg_live): Likewise.
16975 (mark_pseudo_reg_dead): Likewise.
16976 (mark_ref_dead): Likewise.
16977 * reginfo.c (init_subregs_of_mode): Likewise.
16978 * sched-deps.c (sched_analyze_1): Likewise.
16979 * df-scan.c (df_def_record_1): Likewise.
16980 (df_uses_record): Likewise.
16981 (df_read_modify_subreg_p): Remove in favor of...
16982 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
16983 const_rtx instead of an rtx.
16985 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16986 Alan Hayward <alan.hayward@arm.com>
16987 David Sherwood <david.sherwood@arm.com>
16989 * rtl.h (partial_subreg_p): New function.
16990 * caller-save.c (save_call_clobbered_regs): Use it.
16991 * calls.c (expand_call): Likewise.
16992 * combine.c (combinable_i3pat): Likewise.
16993 (simplify_set): Likewise.
16994 (make_extraction): Likewise.
16995 (make_compound_operation_int): Likewise.
16996 (gen_lowpart_or_truncate): Likewise.
16997 (force_to_mode): Likewise.
16998 (make_field_assignment): Likewise.
16999 (reg_truncated_to_mode): Likewise.
17000 (record_truncated_value): Likewise.
17001 (move_deaths): Likewise.
17002 * cse.c (record_jump_cond): Likewise.
17003 (cse_insn): Likewise.
17004 * cselib.c (cselib_lookup_1): Likewise.
17005 * expmed.c (extract_bit_field_using_extv): Likewise.
17006 * function.c (assign_parm_setup_reg): Likewise.
17007 * ifcvt.c (noce_convert_multiple_sets): Likewise.
17008 * ira-build.c (create_insn_allocnos): Likewise.
17009 * lra-coalesce.c (merge_pseudos): Likewise.
17010 * lra-constraints.c (match_reload): Likewise.
17011 (simplify_operand_subreg): Likewise.
17012 (curr_insn_transform): Likewise.
17013 * lra-lives.c (process_bb_lives): Likewise.
17014 * lra.c (new_insn_reg): Likewise.
17015 (lra_substitute_pseudo): Likewise.
17016 * regcprop.c (mode_change_ok): Likewise.
17017 (maybe_mode_change): Likewise.
17018 (copyprop_hardreg_forward_1): Likewise.
17019 * reload.c (push_reload): Likewise.
17020 (find_reloads): Likewise.
17021 (find_reloads_subreg_address): Likewise.
17022 * reload1.c (alter_reg): Likewise.
17023 (eliminate_regs_1): Likewise.
17024 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17026 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
17028 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
17029 back to if statements, including unpack.
17031 2017-08-30 Martin Liska <mliska@suse.cz>
17033 PR inline-asm/82001
17034 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
17036 (func_checker::compare_asm_inputs_outputs): ... this function.
17037 (func_checker::compare_gimple_asm): Use the function to compare
17038 also ASM constrains.
17039 * ipa-icf-gimple.h: Rename the function.
17041 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17042 Alan Hayward <alan.hayward@arm.com>
17043 David Sherwood <david.sherwood@arm.com>
17045 * coretypes.h (complex_mode): New type.
17046 * gdbhooks.py (build_pretty_printer): Handle it.
17047 * machmode.h (complex_mode): New class.
17048 (complex_mode::includes_p): New function.
17049 (is_complex_int_mode): Likewise.
17050 (is_complex_float_mode): Likewise.
17051 * genmodes.c (get_mode_class): Handle complex mode classes.
17052 * function.c (expand_function_end): Use is_complex_int_mode.
17054 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17055 Alan Hayward <alan.hayward@arm.com>
17056 David Sherwood <david.sherwood@arm.com>
17058 * coretypes.h (scalar_mode_pod): New typedef.
17059 * gdbhooks.py (build_pretty_printer): Handle it.
17060 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
17061 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
17062 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
17063 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
17064 as_a <scalar_mode>.
17066 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17067 Alan Hayward <alan.hayward@arm.com>
17068 David Sherwood <david.sherwood@arm.com>
17070 * machmode.h (mode_for_vector): Take a scalar_mode instead
17072 * stor-layout.c (mode_for_vector): Likewise.
17073 * explow.c (promote_mode): Use as_a <scalar_mode>.
17074 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
17076 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17077 Alan Hayward <alan.hayward@arm.com>
17078 David Sherwood <david.sherwood@arm.com>
17080 * target.def (preferred_simd_mode): Take a scalar_mode
17081 instead of a machine_mode.
17082 * targhooks.h (default_preferred_simd_mode): Likewise.
17083 * targhooks.c (default_preferred_simd_mode): Likewise.
17084 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
17085 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
17086 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
17087 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
17088 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
17089 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
17090 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
17091 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
17093 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
17094 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
17095 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
17096 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
17097 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
17098 * doc/tm.texi: Regenerate.
17099 * optabs-query.c (can_vec_mask_load_store_p): Return false for
17102 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17103 Alan Hayward <alan.hayward@arm.com>
17104 David Sherwood <david.sherwood@arm.com>
17106 * target.def (scalar_mode_supported_p): Take a scalar_mode
17107 instead of a machine_mode.
17108 * targhooks.h (default_scalar_mode_supported_p): Likewise.
17109 * targhooks.c (default_scalar_mode_supported_p): Likewise.
17110 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
17111 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17112 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
17113 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
17114 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
17115 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
17116 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
17117 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
17118 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
17119 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
17120 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
17121 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
17123 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
17124 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
17125 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17126 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17127 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
17129 * doc/tm.texi: Regenerate.
17131 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17132 Alan Hayward <alan.hayward@arm.com>
17133 David Sherwood <david.sherwood@arm.com>
17135 * coretypes.h (opt_scalar_mode): New typedef.
17136 * gdbhooks.py (build_pretty_printers): Handle it.
17137 * machmode.h (mode_iterator::get_2xwider): Add overload for
17139 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
17141 * expr.c (convert_mode_scalar): Likewise.
17142 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17143 * optabs.c (expand_float): Likewise.
17144 (expand_fix): Likewise.
17145 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17147 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17148 Alan Hayward <alan.hayward@arm.com>
17149 David Sherwood <david.sherwood@arm.com>
17151 * optabs.c (expand_float): Explicitly check for scalars before
17152 using a branching expansion.
17153 (expand_fix): Likewise.
17155 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17156 Alan Hayward <alan.hayward@arm.com>
17157 David Sherwood <david.sherwood@arm.com>
17159 * expr.c (convert_mode): Split scalar handling out into...
17160 (convert_mode_scalar): ...this new function. Treat the modes
17163 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17164 Alan Hayward <alan.hayward@arm.com>
17165 David Sherwood <david.sherwood@arm.com>
17167 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
17169 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
17171 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17172 Alan Hayward <alan.hayward@arm.com>
17173 David Sherwood <david.sherwood@arm.com>
17175 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
17176 rather than a machine_mode.
17177 (fixed_from_string): Likewise.
17178 (fixed_convert): Likewise.
17179 (fixed_convert_from_int): Likewise.
17180 (fixed_convert_from_real): Likewise.
17181 (real_convert_from_fixed): Likewise.
17182 * fixed-value.c (fixed_from_double_int): Likewise.
17183 (fixed_from_string): Likewise.
17184 (fixed_convert): Likewise.
17185 (fixed_convert_from_int): Likewise.
17186 (fixed_convert_from_real): Likewise.
17187 (real_convert_from_fixed): Likewise.
17188 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
17190 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17191 Alan Hayward <alan.hayward@arm.com>
17192 David Sherwood <david.sherwood@arm.com>
17194 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
17195 of separate mode class checks. Do not allow vector modes here.
17196 (immed_wide_int_const): Use as_a <scalar_mode>.
17197 * explow.c (trunc_int_for_mode): Likewise.
17198 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
17199 (wi::shwi): Likewise.
17200 (wi::min_value): Likewise.
17201 (wi::max_value): Likewise.
17202 * dwarf2out.c (loc_descriptor): Likewise.
17203 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
17204 for CONST_WIDE_INT.
17206 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17207 Alan Hayward <alan.hayward@arm.com>
17208 David Sherwood <david.sherwood@arm.com>
17210 * tree.h (SCALAR_TYPE_MODE): New macro.
17211 * expr.c (expand_expr_addr_expr_1): Use it.
17212 (expand_expr_real_2): Likewise.
17213 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
17214 (fold_convert_const_fixed_from_int): Likewise.
17215 (fold_convert_const_fixed_from_real): Likewise.
17216 (native_encode_fixed): Likewise
17217 (native_encode_complex): Likewise
17218 (native_encode_vector): Likewise.
17219 (native_interpret_fixed): Likewise.
17220 (native_interpret_real): Likewise.
17221 (native_interpret_complex): Likewise.
17222 (native_interpret_vector): Likewise.
17223 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
17224 (simd_clone_adjust_argument_types): Likewise.
17225 (simd_clone_init_simd_arrays): Likewise.
17226 (simd_clone_adjust): Likewise.
17227 * stor-layout.c (layout_type): Likewise.
17228 * tree.c (build_minus_one_cst): Likewise.
17229 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17230 * tree-inline.c (estimate_move_cost): Likewise.
17231 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
17232 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
17233 (vectorizable_reduction): Likewise.
17234 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
17235 (vect_recog_mixed_size_cond_pattern): Likewise.
17236 (check_bool_pattern): Likewise.
17237 (adjust_bool_pattern): Likewise.
17238 (search_type_for_mask_1): Likewise.
17239 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
17240 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17241 (vectorizable_load): Likewise.
17242 (vectorizable_store): Likewise.
17243 * ubsan.c (ubsan_encode_value): Likewise.
17244 * varasm.c (output_constant): Likewise.
17246 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17247 Alan Hayward <alan.hayward@arm.com>
17248 David Sherwood <david.sherwood@arm.com>
17250 * coretypes.h (scalar_mode): New class.
17251 * machmode.h (scalar_mode): Likewise.
17252 (scalar_mode::includes_p): New function.
17253 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
17254 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
17255 * genmodes.c (get_mode_class): Handle remaining scalar modes.
17256 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
17257 * expmed.c (store_bit_field_1): Likewise.
17258 (extract_bit_field_1): Likewise.
17259 * expr.c (write_complex_part): Likewise.
17260 (read_complex_part): Likewise.
17261 (emit_move_complex_push): Likewise.
17262 (expand_expr_real_2): Likewise.
17263 * function.c (assign_parm_setup_reg): Likewise.
17264 (assign_parms_unsplit_complex): Likewise.
17265 * optabs.c (expand_binop): Likewise.
17266 * rtlanal.c (subreg_get_info): Likewise.
17267 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17268 * varasm.c (output_constant_pool_2): Likewise.
17270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17271 Alan Hayward <alan.hayward@arm.com>
17272 David Sherwood <david.sherwood@arm.com>
17274 * expmed.c (extract_high_half): Use scalar_int_mode and remove
17276 (expmed_mult_highpart_optab): Likewise.
17277 (expmed_mult_highpart): Likewise.
17279 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17280 Alan Hayward <alan.hayward@arm.com>
17281 David Sherwood <david.sherwood@arm.com>
17283 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
17284 instead of a machine_mode.
17285 (builtin_memset_read_str): Likewise.
17286 * builtins.c (c_readstr): Likewise.
17287 (builtin_memcpy_read_str): Likewise.
17288 (builtin_strncpy_read_str): Likewise.
17289 (builtin_memset_read_str): Likewise.
17290 (builtin_memset_gen_str): Likewise.
17291 (expand_builtin_signbit): Use scalar_int_mode for local variables.
17292 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
17293 instead of a machine_mode.
17294 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
17296 (make_extraction): Likewise.
17297 (try_widen_shift_mode): Take and return scalar_int_modes instead
17299 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
17300 a scalar_int_mode instead of a machine_mode.
17301 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
17302 (avr_addr_space_pointer_mode): Likewise.
17303 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
17304 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
17305 (msp430_unwind_word_mode): Likewise.
17306 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
17307 (spu_addr_space_pointer_mode): Likewise.
17308 (spu_addr_space_address_mode): Likewise.
17309 (spu_libgcc_cmp_return_mode): Likewise.
17310 (spu_libgcc_shift_count_mode): Likewise.
17311 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
17312 (rl78_addr_space_pointer_mode): Likewise.
17313 (fl78_unwind_word_mode): Likewise.
17314 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
17316 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
17317 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
17318 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
17319 (mips_valid_pointer_mode): Likewise.
17320 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
17321 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
17322 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
17324 (ft32_addr_space_address_mode): Likewise.
17325 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
17326 scalar_int_mode instead of a machine_mode.
17327 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
17329 (m32c_addr_space_address_mode): Likewise.
17330 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
17331 (rs6000_eh_return_filter_mode): Likewise.
17332 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
17333 (rs6000_eh_return_filter_mode): Likewise.
17334 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
17335 (s390_libgcc_shift_count_mode): Likewise.
17336 (s390_unwind_word_mode): Likewise.
17337 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
17339 * target.def (mode_rep_extended): Likewise.
17340 (valid_pointer_mode): Likewise.
17341 (addr_space.valid_pointer_mode): Likewise.
17342 (eh_return_filter_mode): Return a scalar_int_mode rather than
17344 (libgcc_cmp_return_mode): Likewise.
17345 (libgcc_shift_count_mode): Likewise.
17346 (unwind_word_mode): Likewise.
17347 (addr_space.pointer_mode): Likewise.
17348 (addr_space.address_mode): Likewise.
17349 * doc/tm.texi: Regenerate.
17350 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
17352 (do_jump): Use scalar_int_mode for local variables.
17353 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
17354 rather than a machine_mode.
17355 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
17356 (scompare_loc_descriptor_wide): Likewise.
17357 (scompare_loc_descriptor_narrow): Likewise.
17358 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
17360 * except.c (sjlj_emit_dispatch_table): Likewise.
17361 (expand_builtin_eh_copy_values): Likewise.
17362 * explow.c (convert_memory_address_addr_space_1): Likewise.
17363 Take a scalar_int_mode rather than a machine_mode.
17364 (convert_memory_address_addr_space): Take a scalar_int_mode rather
17365 than a machine_mode.
17366 (memory_address_addr_space): Use scalar_int_mode for local variables.
17367 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
17368 rather than a machine_mode.
17369 * expmed.c (mask_rtx): Likewise.
17370 (init_expmed_one_conv): Likewise.
17371 (expand_mult_highpart_adjust): Likewise.
17372 (extract_high_half): Likewise.
17373 (expmed_mult_highpart_optab): Likewise.
17374 (expmed_mult_highpart): Likewise.
17375 (expand_smod_pow2): Likewise.
17376 (expand_sdiv_pow2): Likewise.
17377 (emit_store_flag_int): Likewise.
17378 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
17380 (extract_low_bits): Likewise.
17381 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
17383 * expr.c (pieces_addr::adjust): Likewise.
17384 (can_store_by_pieces): Likewise.
17385 (store_by_pieces): Likewise.
17386 (clear_by_pieces_1): Likewise.
17387 (expand_expr_addr_expr_1): Likewise.
17388 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
17389 (expand_expr_real_1): Likewise.
17390 (try_casesi): Likewise.
17391 * final.c (shorten_branches): Likewise.
17392 * fold-const.c (fold_convert_const_int_from_fixed): Change the
17393 type of "mode" to machine_mode.
17394 * internal-fn.c (expand_arith_overflow_result_store): Take a
17395 scalar_int_mode rather than a machine_mode.
17396 (expand_mul_overflow): Use scalar_int_mode for local variables.
17397 * loop-doloop.c (doloop_modify): Likewise.
17398 (doloop_optimize): Likewise.
17399 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
17400 than a machine_mode.
17401 (expand_doubleword_shift_condmove): Likewise.
17402 (expand_doubleword_shift): Likewise.
17403 (expand_doubleword_clz): Likewise.
17404 (expand_doubleword_popcount): Likewise.
17405 (expand_doubleword_parity): Likewise.
17406 (expand_absneg_bit): Use scalar_int_mode for local variables.
17407 (prepare_float_lib_cmp): Likewise.
17408 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
17409 rather than a machine_mode.
17410 (convert_memory_address_addr_space): Likewise.
17411 (get_mode_bounds): Likewise.
17412 (get_address_mode): Return a scalar_int_mode rather than a
17414 * rtlanal.c (get_address_mode): Likewise.
17415 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
17416 than a machine_mode.
17417 * targhooks.c (default_mode_rep_extended): Likewise.
17418 (default_valid_pointer_mode): Likewise.
17419 (default_addr_space_valid_pointer_mode): Likewise.
17420 (default_eh_return_filter_mode): Return a scalar_int_mode rather
17421 than a machine_mode.
17422 (default_libgcc_cmp_return_mode): Likewise.
17423 (default_libgcc_shift_count_mode): Likewise.
17424 (default_unwind_word_mode): Likewise.
17425 (default_addr_space_pointer_mode): Likewise.
17426 (default_addr_space_address_mode): Likewise.
17427 * targhooks.h (default_eh_return_filter_mode): Likewise.
17428 (default_libgcc_cmp_return_mode): Likewise.
17429 (default_libgcc_shift_count_mode): Likewise.
17430 (default_unwind_word_mode): Likewise.
17431 (default_addr_space_pointer_mode): Likewise.
17432 (default_addr_space_address_mode): Likewise.
17433 (default_mode_rep_extended): Take a scalar_int_mode rather than
17435 (default_valid_pointer_mode): Likewise.
17436 (default_addr_space_valid_pointer_mode): Likewise.
17437 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
17439 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
17440 rather than a machine_mode.
17441 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
17442 for local variables.
17443 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17444 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
17445 than a machine_mode.
17447 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17448 Alan Hayward <alan.hayward@arm.com>
17449 David Sherwood <david.sherwood@arm.com>
17451 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
17452 the mode argument to scalar_int_mode.
17453 (do_jump_by_parts_zero_rtx): Likewise.
17454 (do_jump_by_parts_equality_rtx): Likewise.
17455 (do_jump_by_parts_greater): Take a mode argument.
17456 (do_jump_by_parts_equality): Likewise.
17457 (do_jump_1): Update calls accordingly.
17459 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17460 Alan Hayward <alan.hayward@arm.com>
17461 David Sherwood <david.sherwood@arm.com>
17463 * is-a.h (safe_dyn_cast): New function.
17464 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
17465 (jump_table_for_label): Likewise.
17466 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
17467 instead of an rtx_insn *.
17468 (shorten_branches): Use dyn_cast instead of LABEL_P and
17469 JUMP_TABLE_DATA_P. Use jump_table_for_label and
17470 rtx_jump_table_data::get_data_mode.
17471 (final_scan_insn): Likewise.
17473 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17474 Alan Hayward <alan.hayward@arm.com>
17475 David Sherwood <david.sherwood@arm.com>
17477 * combine.c (try_combine): Use is_a <scalar_int_mode> when
17478 trying to combine a full-register integer set with a subreg
17481 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17482 Alan Hayward <alan.hayward@arm.com>
17483 David Sherwood <david.sherwood@arm.com>
17485 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
17486 that is always either address_mode or pointer_mode.
17488 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17489 Alan Hayward <alan.hayward@arm.com>
17490 David Sherwood <david.sherwood@arm.com>
17492 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
17493 when the two are known to be equal.
17495 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17496 Alan Hayward <alan.hayward@arm.com>
17497 David Sherwood <david.sherwood@arm.com>
17499 * simplify-rtx.c (simplify_const_unary_operation): Use
17500 is_a <scalar_int_mode> instead of checking for a nonzero
17501 precision. Forcibly convert op_mode to a scalar_int_mode
17502 in that case. More clearly differentiate the operand and
17503 result modes and use the former when deciding what the value
17504 of a count-bits operation should be. Use is_int_mode instead
17505 of checking for a MODE_INT. Remove redundant check for whether
17506 this mode has a zero precision.
17508 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17509 Alan Hayward <alan.hayward@arm.com>
17510 David Sherwood <david.sherwood@arm.com>
17512 * optabs.c (widen_leading): Change the type of the mode argument
17513 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
17514 (widen_bswap): Likewise.
17515 (expand_parity): Likewise.
17516 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
17517 (expand_ffs): Likewise.
17518 (epand_unop): Check for scalar integer modes before calling the
17521 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17522 Alan Hayward <alan.hayward@arm.com>
17523 David Sherwood <david.sherwood@arm.com>
17525 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
17527 (expand_expr_real_1): Update call accordingly.
17529 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17530 Alan Hayward <alan.hayward@arm.com>
17531 David Sherwood <david.sherwood@arm.com>
17533 * expmed.c (store_bit_field_using_insv): Add op0_mode and
17534 value_mode arguments. Use scalar_int_mode internally.
17535 (store_bit_field_1): Rename the new integer mode from imode
17536 to op0_mode and use it instead of GET_MODE (op0). Update calls
17537 to store_split_bit_field, store_bit_field_using_insv and
17538 store_fixed_bit_field.
17539 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
17540 Use scalar_int_mode internally. Use a bit count rather than a mode
17541 when calculating the largest bit size for get_best_mode.
17542 Update calls to store_split_bit_field and store_fixed_bit_field_1.
17543 (store_fixed_bit_field_1): Add mode and value_mode arguments.
17544 Remove assertion that OP0 has a scalar integer mode.
17545 (store_split_bit_field): Add op0_mode and value_mode arguments.
17546 Update calls to extract_fixed_bit_field.
17547 (extract_bit_field_using_extv): Add an op0_mode argument.
17548 Use scalar_int_mode internally.
17549 (extract_bit_field_1): Rename the new integer mode from imode to
17550 op0_mode and use it instead of GET_MODE (op0). Update calls to
17551 extract_split_bit_field, extract_bit_field_using_extv and
17552 extract_fixed_bit_field.
17553 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
17554 to extract_split_bit_field and extract_fixed_bit_field_1.
17555 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
17556 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
17557 on the target mode.
17558 (extract_split_bit_field): Add an op0_mode argument. Update call
17559 to extract_fixed_bit_field.
17561 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17562 Alan Hayward <alan.hayward@arm.com>
17563 David Sherwood <david.sherwood@arm.com>
17565 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
17566 * explow.c (hard_function_value): Likewise.
17567 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
17568 convert_to_mode call outside the loop.
17569 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
17570 for the mode iterator. Require the mode specified by max_pieces
17572 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
17574 (copy_blkmode_to_reg): Likewise.
17575 (set_storage_via_setmem): Likewise.
17576 * optabs.c (prepare_cmp_insn): Likewise.
17577 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17578 * stor-layout.c (finish_bitfield_representative): Likewise.
17580 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17581 Alan Hayward <alan.hayward@arm.com>
17582 David Sherwood <david.sherwood@arm.com>
17584 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
17585 * expr.c (convert_move): Use them.
17586 (convert_modes): Likewise.
17587 (store_expr_with_bounds): Likewise.
17589 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17590 Alan Hayward <alan.hayward@arm.com>
17591 David Sherwood <david.sherwood@arm.com>
17593 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
17594 parameter for the mode of "x". Remove the "known_x", "known_mode"
17595 and "known_ret" arguments. Change the type of the mode argument
17596 to scalar_int_mode.
17597 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
17598 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
17599 (reg_num_sign_bit_copies_for_combine): Likewise.
17600 * rtlanal.c (nonzero_bits1): Likewise.
17601 (num_sign_bit_copies1): Likewise.
17602 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
17603 (reg_num_sign_bit_copies_general): Likewise.
17604 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
17605 (reg_nonzero_bits_general): Likewise.
17607 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17608 Alan Hayward <alan.hayward@arm.com>
17609 David Sherwood <david.sherwood@arm.com>
17611 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
17612 than in subroutines. Return 1 for non-integer modes.
17613 (cached_num_sign_bit_copies): Change the type of the mode parameter
17614 to scalar_int_mode.
17615 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
17616 classes. Handle CONST_INT_P first and then check whether X also
17617 has a scalar integer mode. Check the same thing for inner registers
17618 of a SUBREG and for values that are being extended or truncated.
17620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17621 Alan Hayward <alan.hayward@arm.com>
17622 David Sherwood <david.sherwood@arm.com>
17624 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
17625 in subroutines. Return the mode mask for non-integer modes.
17626 (cached_nonzero_bits): Change the type of the mode parameter
17627 to scalar_int_mode.
17628 (nonzero_bits1): Likewise. Remove early exit for other mode
17629 classes. Handle CONST_INT_P first and then check whether X
17630 also has a scalar integer mode.
17632 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17633 Alan Hayward <alan.hayward@arm.com>
17634 David Sherwood <david.sherwood@arm.com>
17636 * expr.c (widest_int_mode_for_size): Make the comment match the code.
17637 Return a scalar_int_mode and assert that the size is greater than
17639 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
17640 (op_by_pieces_d::op_by_pieces_d): Likewise.
17641 (op_by_pieces_d::run): Likewise.
17642 (can_store_by_pieces): Likewise.
17644 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17645 Alan Hayward <alan.hayward@arm.com>
17646 David Sherwood <david.sherwood@arm.com>
17648 * combine.c (extract_left_shift): Add a mode argument and update
17650 (make_compound_operation_int): Change the type of the mode parameter
17651 to scalar_int_mode and update the call to extract_left_shift.
17653 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17654 Alan Hayward <alan.hayward@arm.com>
17655 David Sherwood <david.sherwood@arm.com>
17657 * combine.c (simplify_and_const_int): Change the type of the mode
17658 parameter to scalar_int_mode.
17659 (simplify_and_const_int_1): Likewise. Update recursive call.
17661 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17662 Alan Hayward <alan.hayward@arm.com>
17663 David Sherwood <david.sherwood@arm.com>
17665 * combine.c (simplify_compare_const): Check that the mode is a
17666 scalar_int_mode (rather than VOIDmode) before testing its
17668 (simplify_comparison): Move COMPARISON_P handling out of the
17669 loop and restrict the latter part of the loop to scalar_int_modes.
17670 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
17671 and when considering SUBREG_REGs. Use is_int_mode instead of
17672 checking GET_MODE_CLASS against MODE_INT.
17674 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17675 Alan Hayward <alan.hayward@arm.com>
17676 David Sherwood <david.sherwood@arm.com>
17678 * combine.c (try_widen_shift_mode): Move check for equal modes to...
17679 (simplify_shift_const_1): ...here. Use scalar_int_mode for
17680 shift_unit_mode and for modes involved in scalar shifts.
17682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17683 Alan Hayward <alan.hayward@arm.com>
17684 David Sherwood <david.sherwood@arm.com>
17686 * combine.c (force_int_to_mode): New function, split out from...
17687 (force_to_mode): ...here. Keep xmode up-to-date and use it
17688 instead of GET_MODE (x).
17690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17691 Alan Hayward <alan.hayward@arm.com>
17692 David Sherwood <david.sherwood@arm.com>
17694 * optabs-query.h (extraction_insn::struct_mode): Change type to
17695 opt_scalar_int_mode and update comment.
17696 (extraction_insn::field_mode): Change type to scalar_int_mode.
17697 (extraction_insn::pos_mode): Likewise.
17698 * combine.c (make_extraction): Update accordingly.
17699 * optabs-query.c (get_traditional_extraction_insn): Likewise.
17700 (get_optab_extraction_insn): Likewise.
17701 * recog.c (simplify_while_replacing): Likewise.
17702 * expmed.c (narrow_bit_field_mem): Change the type of the mode
17703 parameter to opt_scalar_int_mode.
17705 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17706 Alan Hayward <alan.hayward@arm.com>
17707 David Sherwood <david.sherwood@arm.com>
17709 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
17710 to a scalar_int_mode instead of a machine_mode.
17711 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
17712 (get_best_mode): Return a boolean and use a pointer argument to store
17713 the selected mode. Replace the limit mode parameter with a bit limit.
17714 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
17715 for the values returned by bit_field_mode_iterator::next_mode.
17716 (store_bit_field): Update call to get_best_mode.
17717 (store_fixed_bit_field): Likewise.
17718 (extract_fixed_bit_field): Likewise.
17719 * expr.c (optimize_bitfield_assignment_op): Likewise.
17720 * fold-const.c (optimize_bit_field_compare): Likewise.
17721 (fold_truth_andor_1): Likewise.
17722 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
17723 Update for new type of m_mode.
17724 (get_best_mode): As above.
17726 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17727 Alan Hayward <alan.hayward@arm.com>
17728 David Sherwood <david.sherwood@arm.com>
17730 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
17731 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
17732 (store_bit_field): Check is_a <scalar_int_mode> before calling
17733 strict_volatile_bitfield_p.
17734 (extract_bit_field): Likewise.
17736 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17737 Alan Hayward <alan.hayward@arm.com>
17738 David Sherwood <david.sherwood@arm.com>
17740 * target.def (cstore_mode): Return a scalar_int_mode.
17741 * doc/tm.texi: Regenerate.
17742 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
17743 * targhooks.h (default_cstore_mode): Likewise.
17744 * targhooks.c (default_cstore_mode): Likewise, using a forced
17746 * expmed.c (emit_cstore): Expect the target of the cstore to be
17749 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17750 Alan Hayward <alan.hayward@arm.com>
17751 David Sherwood <david.sherwood@arm.com>
17753 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
17755 (niter_desc): Likewise mode.
17756 (iv_analyze): Add a mode parameter.
17758 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
17759 and change its type to scalar_int_mode.
17760 * loop-iv.c: Update commentary at head of file.
17761 (iv_constant): Pass the mode paraeter before the rtx it describes
17762 and change its type to scalar_int_mode. Remove VOIDmode handling.
17763 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
17764 (iv_extend): Likewise.
17765 (shorten_into_mode): Likewise.
17766 (iv_add): Use scalar_int_mode.
17767 (iv_mult): Likewise.
17768 (iv_shift): Likewise.
17769 (canonicalize_iv_subregs): Likewise.
17770 (get_biv_step_1): Pass the outer_mode parameter before the rtx
17771 it describes and change its mode to scalar_int_mode. Also change
17772 the type of the returned inner_mode to scalar_int_mode.
17773 (get_biv_step): Likewise, turning outer_mode from a pointer
17774 into a direct parameter. Update call to get_biv_step_1.
17775 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
17776 iv_constant and get_biv_step.
17777 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
17778 and change its type to scalar_int_mode. Don't initialise iv->mode
17779 to VOIDmode and remove later checks for its still being VOIDmode.
17780 Update calls to iv_analyze_op and iv_analyze_expr. Check
17781 is_a <scalar_int_mode> when changing the mode under consideration.
17782 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
17783 Update call to iv_analyze_expr.
17784 (iv_analyze_op): Add a mode parameter. Reject subregs whose
17785 inner register is not also a scalar_int_mode. Update call to
17787 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
17788 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
17789 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
17790 separate mode class checks. Update calls to iv_analyze. Remove
17791 fix-up of VOIDmodes after iv_analyze_biv.
17792 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
17793 don't have a scalar_int_mode. Update call to biv_p.
17795 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17796 Alan Hayward <alan.hayward@arm.com>
17797 David Sherwood <david.sherwood@arm.com>
17799 * cfgexpand.c (convert_debug_memory_address): Use
17800 as_a <scalar_int_mode>.
17801 * combine.c (expand_compound_operation): Likewise.
17802 (make_extraction): Likewise.
17803 (change_zero_ext): Likewise.
17804 (simplify_comparison): Likewise.
17805 * cse.c (cse_insn): Likewise.
17806 * dwarf2out.c (minmax_loc_descriptor): Likewise.
17807 (mem_loc_descriptor): Likewise.
17808 (loc_descriptor): Likewise.
17809 * expmed.c (init_expmed_one_mode): Likewise.
17810 (synth_mult): Likewise.
17811 (emit_store_flag_1): Likewise.
17812 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
17813 of a comparison with size.
17814 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
17815 (reduce_to_bit_field_precision): Likewise.
17816 * function.c (expand_function_end): Likewise.
17817 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
17818 * loop-doloop.c (doloop_modify): Likewise.
17819 * optabs.c (expand_binop): Likewise.
17820 (expand_unop): Likewise.
17821 (expand_copysign_absneg): Likewise.
17822 (prepare_cmp_insn): Likewise.
17823 (maybe_legitimize_operand): Likewise.
17824 * recog.c (const_scalar_int_operand): Likewise.
17825 * rtlanal.c (get_address_mode): Likewise.
17826 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17827 (simplify_cond_clz_ctz): Likewise.
17828 * tree-nested.c (get_nl_goto_field): Likewise.
17829 * tree.c (build_vector_type_for_mode): Likewise.
17830 * var-tracking.c (use_narrower_mode): Likewise.
17832 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17833 Alan Hayward <alan.hayward@arm.com>
17834 David Sherwood <david.sherwood@arm.com>
17836 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
17837 * builtins.c (expand_builtin_signbit): Use it.
17838 * cfgexpand.c (expand_debug_expr): Likewise.
17839 * dojump.c (do_jump): Likewise.
17840 (do_compare_and_jump): Likewise.
17841 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
17842 * expmed.c (make_tree): Likewise.
17843 * expr.c (expand_expr_real_2): Likewise.
17844 (expand_expr_real_1): Likewise.
17845 (try_casesi): Likewise.
17846 * fold-const-call.c (fold_const_call_ss): Likewise.
17847 * fold-const.c (unextend): Likewise.
17848 (extract_muldiv_1): Likewise.
17849 (fold_single_bit_test): Likewise.
17850 (native_encode_int): Likewise.
17851 (native_encode_string): Likewise.
17852 (native_interpret_int): Likewise.
17853 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
17854 * internal-fn.c (expand_addsub_overflow): Likewise.
17855 (expand_neg_overflow): Likewise.
17856 (expand_mul_overflow): Likewise.
17857 (expand_arith_overflow): Likewise.
17858 * match.pd: Likewise.
17859 * stor-layout.c (layout_type): Likewise.
17860 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17861 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
17862 * tree-ssanames.c (get_range_info): Likewise.
17863 * tree-switch-conversion.c (array_value_type) Likewise.
17864 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
17865 (vect_recog_divmod_pattern): Likewise.
17866 (vect_recog_mixed_size_cond_pattern): Likewise.
17867 * tree-vrp.c (extract_range_basic): Likewise.
17868 (simplify_float_conversion_using_ranges): Likewise.
17869 * tree.c (int_fits_type_p): Likewise.
17870 * ubsan.c (instrument_bool_enum_load): Likewise.
17871 * varasm.c (mergeable_string_section): Likewise.
17872 (narrowing_initializer_constant_valid_p): Likewise.
17873 (output_constant): Likewise.
17875 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17876 Alan Hayward <alan.hayward@arm.com>
17877 David Sherwood <david.sherwood@arm.com>
17879 * machmode.h (NARROWEST_INT_MODE): New macro.
17880 * expr.c (alignment_for_piecewise_move): Use it instead of
17881 GET_CLASS_NARROWEST_MODE (MODE_INT).
17882 (push_block): Likewise.
17883 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
17885 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17887 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17888 Alan Hayward <alan.hayward@arm.com>
17889 David Sherwood <david.sherwood@arm.com>
17891 * postreload.c (move2add_valid_value_p): Change the type of the
17892 mode parameter to scalar_int_mode.
17893 (move2add_use_add2_insn): Add a mode parameter and use it instead
17895 (move2add_use_add3_insn): Likewise.
17896 (reload_cse_move2add): Update accordingly.
17898 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17899 Alan Hayward <alan.hayward@arm.com>
17900 David Sherwood <david.sherwood@arm.com>
17902 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
17904 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
17905 * optabs.c (expand_unop): Likewise.
17907 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17908 Alan Hayward <alan.hayward@arm.com>
17909 David Sherwood <david.sherwood@arm.com>
17911 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
17912 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
17913 (popcount_loc_descriptor): Likewise.
17914 (bswap_loc_descriptor): Likewise.
17915 (rotate_loc_descriptor): Likewise.
17916 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
17917 calling the functions above.
17919 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17920 Alan Hayward <alan.hayward@arm.com>
17921 David Sherwood <david.sherwood@arm.com>
17923 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
17925 (try_combine): Likewise.
17926 (simplify_if_then_else): Likewise.
17927 * cse.c (cse_insn): Likewise.
17928 * dwarf2out.c (mem_loc_descriptor): Likewise.
17929 * emit-rtl.c (gen_lowpart_common): Likewise.
17930 * simplify-rtx.c (simplify_truncation): Likewise.
17931 (simplify_binary_operation_1): Likewise.
17932 (simplify_const_relational_operation): Likewise.
17933 (simplify_ternary_operation): Likewise.
17934 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
17936 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17937 Alan Hayward <alan.hayward@arm.com>
17938 David Sherwood <david.sherwood@arm.com>
17940 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
17941 * reload.c (push_reload): Likewise.
17942 (find_reloads): Likewise.
17944 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17945 Alan Hayward <alan.hayward@arm.com>
17946 David Sherwood <david.sherwood@arm.com>
17948 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
17949 (make_compound_operation_int): Likewise.
17950 (change_zero_ext): Likewise.
17951 * expr.c (convert_move): Likewise.
17952 (convert_modes): Likewise.
17953 * fwprop.c (forward_propagate_subreg): Likewise.
17954 * loop-iv.c (get_biv_step_1): Likewise.
17955 * optabs.c (widen_operand): Likewise.
17956 * postreload.c (move2add_valid_value_p): Likewise.
17957 * recog.c (simplify_while_replacing): Likewise.
17958 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17959 (simplify_binary_operation_1): Likewise. Remove redundant
17960 mode equality check.
17962 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17963 Alan Hayward <alan.hayward@arm.com>
17964 David Sherwood <david.sherwood@arm.com>
17966 * combine.c (combine_simplify_rtx): Add checks for
17967 is_a <scalar_int_mode>.
17968 (simplify_if_then_else): Likewise.
17969 (make_field_assignment): Likewise.
17970 (simplify_comparison): Likewise.
17971 * ifcvt.c (noce_try_bitop): Likewise.
17972 * loop-invariant.c (canonicalize_address_mult): Likewise.
17973 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17975 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17976 Alan Hayward <alan.hayward@arm.com>
17977 David Sherwood <david.sherwood@arm.com>
17979 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
17980 is_a <scalar_int_mode> instead of != BLKmode.
17982 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17983 Alan Hayward <alan.hayward@arm.com>
17984 David Sherwood <david.sherwood@arm.com>
17986 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
17987 instead of != VOIDmode.
17988 * combine.c (if_then_else_cond): Likewise.
17989 (change_zero_ext): Likewise.
17990 * dwarf2out.c (mem_loc_descriptor): Likewise.
17991 (loc_descriptor): Likewise.
17992 * rtlanal.c (canonicalize_condition): Likewise.
17993 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
17995 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17996 Alan Hayward <alan.hayward@arm.com>
17997 David Sherwood <david.sherwood@arm.com>
17999 * simplify-rtx.c (simplify_binary_operation_1): Use
18000 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
18002 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18003 Alan Hayward <alan.hayward@arm.com>
18004 David Sherwood <david.sherwood@arm.com>
18006 * wide-int.h (int_traits<unsigned char>) New class.
18007 (int_traits<unsigned short>) Likewise.
18008 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
18009 Use GET_MODE_UNIT_PRECISION and remove redundant test for
18011 * combine.c (set_nonzero_bits_and_sign_copies): Use
18012 is_a <scalar_int_mode>.
18013 (find_split_point): Likewise.
18014 (combine_simplify_rtx): Likewise.
18015 (simplify_logical): Likewise.
18016 (expand_compound_operation): Likewise.
18017 (expand_field_assignment): Likewise.
18018 (make_compound_operation): Likewise.
18019 (extended_count): Likewise.
18020 (change_zero_ext): Likewise.
18021 (simplify_comparison): Likewise.
18022 * dwarf2out.c (scompare_loc_descriptor): Likewise.
18023 (ucompare_loc_descriptor): Likewise.
18024 (minmax_loc_descriptor): Likewise.
18025 (mem_loc_descriptor): Likewise.
18026 (loc_descriptor): Likewise.
18027 * expmed.c (init_expmed_one_mode): Likewise.
18028 * lra-constraints.c (lra_constraint_offset): Likewise.
18029 * optabs.c (prepare_libcall_arg): Likewise.
18030 * postreload.c (move2add_note_store): Likewise.
18031 * reload.c (operands_match_p): Likewise.
18032 * rtl.h (load_extend_op): Likewise.
18033 * rtlhooks.c (gen_lowpart_general): Likewise.
18034 * simplify-rtx.c (simplify_truncation): Likewise.
18035 (simplify_unary_operation_1): Likewise.
18036 (simplify_binary_operation_1): Likewise.
18037 (simplify_const_binary_operation): Likewise.
18038 (simplify_const_relational_operation): Likewise.
18039 (simplify_subreg): Likewise.
18040 * stor-layout.c (bitwise_mode_for_mode): Likewise.
18041 * var-tracking.c (adjust_mems): Likewise.
18042 (prepare_call_arguments): Likewise.
18044 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18045 Alan Hayward <alan.hayward@arm.com>
18046 David Sherwood <david.sherwood@arm.com>
18048 * machmode.h (is_int_mode): New fuction.
18049 * combine.c (find_split_point): Use it.
18050 (combine_simplify_rtx): Likewise.
18051 (simplify_if_then_else): Likewise.
18052 (simplify_set): Likewise.
18053 (simplify_shift_const_1): Likewise.
18054 (simplify_comparison): Likewise.
18055 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
18056 * cse.c (notreg_cost): Likewise.
18057 (cse_insn): Likewise.
18058 * cselib.c (cselib_lookup_1): Likewise.
18059 * dojump.c (do_jump_1): Likewise.
18060 (do_compare_rtx_and_jump): Likewise.
18061 * dse.c (get_call_args): Likewise.
18062 * dwarf2out.c (rtl_for_decl_init): Likewise.
18063 (native_encode_initializer): Likewise.
18064 * expmed.c (emit_store_flag_1): Likewise.
18065 (emit_store_flag): Likewise.
18066 * expr.c (convert_modes): Likewise.
18067 (store_field): Likewise.
18068 (expand_expr_real_1): Likewise.
18069 * fold-const.c (fold_read_from_constant_string): Likewise.
18070 * gimple-ssa-sprintf.c (get_format_string): Likewise.
18071 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
18072 * optabs.c (expand_binop): Likewise.
18073 (expand_unop): Likewise.
18074 (expand_abs_nojump): Likewise.
18075 (expand_one_cmpl_abs_nojump): Likewise.
18076 * simplify-rtx.c (mode_signbit_p): Likewise.
18077 (val_signbit_p): Likewise.
18078 (val_signbit_known_set_p): Likewise.
18079 (val_signbit_known_clear_p): Likewise.
18080 (simplify_relational_operation_1): Likewise.
18081 * tree.c (vector_type_mode): Likewise.
18083 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18084 Alan Hayward <alan.hayward@arm.com>
18085 David Sherwood <david.sherwood@arm.com>
18087 * machmode.h (smallest_mode_for_size): Fix formatting.
18088 (smallest_int_mode_for_size): New function.
18089 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
18090 instead of smallest_mode_for_size.
18091 * combine.c (make_extraction): Likewise.
18092 * config/arc/arc.c (arc_expand_movmem): Likewise.
18093 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
18094 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
18095 * config/s390/s390.c (s390_expand_insv): Likewise.
18096 * config/sparc/sparc.c (assign_int_registers): Likewise.
18097 * config/spu/spu.c (spu_function_value): Likewise.
18098 (spu_function_arg): Likewise.
18099 * coverage.c (get_gcov_type): Likewise.
18100 (get_gcov_unsigned_t): Likewise.
18101 * dse.c (find_shift_sequence): Likewise.
18102 * expmed.c (store_bit_field_1): Likewise.
18103 * expr.c (convert_move): Likewise.
18104 (store_field): Likewise.
18105 * internal-fn.c (expand_arith_overflow): Likewise.
18106 * optabs-query.c (get_best_extraction_insn): Likewise.
18107 * optabs.c (expand_twoval_binop_libfunc): Likewise.
18108 * stor-layout.c (layout_type): Likewise.
18109 (initialize_sizetypes): Likewise.
18110 * targhooks.c (default_get_mask_mode): Likewise.
18111 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
18113 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18114 Alan Hayward <alan.hayward@arm.com>
18115 David Sherwood <david.sherwood@arm.com>
18117 * machmode.h (opt_mode::else_blk): New function.
18118 (int_mode_for_mode): Declare.
18119 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
18120 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
18122 * cfgexpand.c (expand_debug_expr): Likewise.
18123 * combine.c (gen_lowpart_or_truncate): Likewise.
18124 (gen_lowpart_for_combine): Likewise.
18125 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
18126 * config/avr/avr.c (avr_to_int_mode): Likewise.
18127 (avr_out_plus_1): Likewise.
18128 (avr_out_plus): Likewise.
18129 (avr_out_round): Likewise.
18130 * config/i386/i386.c (ix86_split_to_parts): Likewise.
18131 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
18132 (s390_expand_vcond): Likewise.
18133 * config/spu/spu.c (spu_split_immediate): Likewise.
18134 (spu_expand_mov): Likewise.
18135 * dse.c (get_stored_val): Likewise.
18136 * expmed.c (store_bit_field_1): Likewise.
18137 (convert_extracted_bit_field): Use int_mode_for_mode instead of
18139 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
18140 (extract_low_bits): Likewise.
18141 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
18142 handling rather than repeating the check.
18143 (emit_group_store): Likewise.
18144 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
18145 * optabs.c (expand_absneg_bit): Likewise.
18146 (expand_copysign_absneg): Likewise.
18147 (expand_copysign_bit): Likewise.
18148 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
18149 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18150 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18151 * var-tracking.c (prepare_call_arguments): Likewise.
18152 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
18153 int_mode_for_mode instead of mode_for_size.
18154 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
18156 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18157 Alan Hayward <alan.hayward@arm.com>
18158 David Sherwood <david.sherwood@arm.com>
18160 * machmode.h (int_mode_for_size): New function.
18161 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
18162 instead of mode_for_size.
18163 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
18165 * combine.c (expand_field_assignment): Use int_mode_for_size
18166 instead of mode_for_size.
18167 (make_extraction): Likewise.
18168 (simplify_shift_const_1): Likewise.
18169 (simplify_comparison): Likewise.
18170 * dojump.c (do_jump): Likewise.
18171 * dwarf2out.c (mem_loc_descriptor): Likewise.
18172 * emit-rtl.c (init_derived_machine_modes): Likewise.
18173 * expmed.c (flip_storage_order): Likewise.
18174 (convert_extracted_bit_field): Likewise.
18175 * expr.c (copy_blkmode_from_reg): Likewise.
18176 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
18177 * internal-fn.c (expand_mul_overflow): Likewise.
18178 * lower-subreg.c (simple_move): Likewise.
18179 * optabs-libfuncs.c (init_optabs): Likewise.
18180 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18181 * tree.c (vector_type_mode): Likewise.
18182 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18183 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
18184 * tree-vect-generic.c (expand_vector_parallel): Likewise.
18185 * tree-vect-stmts.c (vectorizable_load): Likewise.
18186 (vectorizable_store): Likewise.
18188 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18189 Alan Hayward <alan.hayward@arm.com>
18190 David Sherwood <david.sherwood@arm.com>
18192 * coretypes.h (pod_mode): New type.
18193 (scalar_int_mode_pod): New typedef.
18194 * machmode.h (pod_mode): New class.
18195 (int_n_data_t::m): Change type to scalar_int_mode_pod.
18196 * genmodes.c (emit_mode_int_n): Update accordingly.
18197 * lower-subreg.h (target_lower_subreg): Change type to
18198 scalar_int_mode_pod.
18199 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
18200 scalar_int_mode_pod.
18202 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18203 Alan Hayward <alan.hayward@arm.com>
18204 David Sherwood <david.sherwood@arm.com>
18206 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
18207 machine_mode to scalar_int_mode.
18208 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
18209 (rs6000_option_override_internal): Remove cast to int.
18210 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
18211 machine_mode to scalar_int_mode.
18212 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
18213 (rs6000_option_override_internal): Remove cast to int.
18214 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
18215 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
18217 * config/s390/s390.c (s390_expand_builtin): Likewise.
18218 * coretypes.h (scalar_int_mode): New type.
18219 (opt_scalar_int_mode): New typedef.
18220 * machmode.h (scalar_int_mode): New class.
18221 (scalar_int_mode::includes_p): New function.
18222 (byte_mode): Change type to scalar_int_mode.
18223 (word_mode): Likewise.
18224 (ptr_mode): Likewise.
18225 * emit-rtl.c (byte_mode): Likewise.
18226 (word_mode): Likewise.
18227 (ptr_mode): Likewise.
18228 (init_derived_machine_modes): Update accordingly.
18229 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
18230 and MODE_PARTIAL_INT.
18231 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
18232 opt_scalar_int_mode.
18234 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18235 Alan Hayward <alan.hayward@arm.com>
18236 David Sherwood <david.sherwood@arm.com>
18238 * target.def (libgcc_floating_mode_supported_p): Take a
18240 * doc/tm.texi: Regenerate.
18241 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
18243 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18244 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
18247 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18248 Alan Hayward <alan.hayward@arm.com>
18249 David Sherwood <david.sherwood@arm.com>
18251 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
18252 * doc/tm.texi: Regenerate.
18253 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
18254 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
18255 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
18256 * targhooks.h (default_floatn_mode): Likewise.
18257 * targhooks.c (default_floatn_mode): Likewise.
18258 * tree.c (build_common_tree_nodes): Update accordingly.
18260 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18261 Alan Hayward <alan.hayward@arm.com>
18262 David Sherwood <david.sherwood@arm.com>
18264 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
18265 (mode_iterator::iterate_p): Likewise.
18266 (mode_iterator::get_wider): Likewise.
18267 * expr.c (init_expr_target): Use opt_scalar_float_mode.
18269 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18270 Alan Hayward <alan.hayward@arm.com>
18271 David Sherwood <david.sherwood@arm.com>
18273 * coretypes.h (opt_scalar_float_mode): New typedef.
18274 * machmode.h (float_mode_for_size): New function.
18275 * emit-rtl.c (double_mode): Delete.
18276 (init_emit_once): Use float_mode_for_size.
18277 * stor-layout.c (layout_type): Likewise.
18278 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
18280 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18281 Alan Hayward <alan.hayward@arm.com>
18282 David Sherwood <david.sherwood@arm.com>
18284 * output.h (assemble_real): Take a scalar_float_mode.
18285 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
18286 * config/arm/arm.md (consttable_4): Likewise.
18287 (consttable_8): Likewise.
18288 (consttable_16): Likewise.
18289 * config/mips/mips.md (consttable_float): Likewise.
18290 * config/s390/s390.c (s390_output_pool_entry): Likewise.
18291 * varasm.c (assemble_real): Take a scalar_float_mode.
18292 (output_constant_pool_2): Update accordingly.
18293 (output_constant): Likewise.
18295 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18296 Alan Hayward <alan.hayward@arm.com>
18297 David Sherwood <david.sherwood@arm.com>
18299 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
18300 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
18301 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
18302 (native_encode_real): Likewise.
18303 (native_interpret_real): Likewise.
18304 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
18305 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18307 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18308 Alan Hayward <alan.hayward@arm.com>
18309 David Sherwood <david.sherwood@arm.com>
18311 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
18312 <scalar_float_mode>. Simplify.
18313 (gen_extend_conv_libfunc): Likewise.
18315 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18316 Alan Hayward <alan.hayward@arm.com>
18317 David Sherwood <david.sherwood@arm.com>
18319 * coretypes.h (scalar_float_mode): New type.
18320 * machmode.h (mode_traits::from_int): Use machine_mode if
18321 USE_ENUM_MODES is defined.
18322 (is_a): New function.
18324 (dyn_cast): Likewise.
18325 (scalar_float_mode): New class.
18326 (scalar_float_mode::includes_p): New function.
18327 (is_float_mode): Likewise.
18328 * gdbhooks.py (MachineModePrinter): New class.
18329 (build_pretty_printer): Use it for scalar_float_mode.
18330 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
18331 (format_helper::format_helper): Turn into a template.
18332 * genmodes.c (get_mode_class): New function.
18333 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
18334 or machine_mode if none.
18335 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
18336 as_a <scalar_float_mode>.
18337 * dwarf2out.c (mem_loc_descriptor): Likewise.
18338 (insert_float): Likewise.
18339 (add_const_value_attribute): Likewise.
18340 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18341 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
18342 (expand_unop): Update accordingly.
18343 (expand_abs_nojump): Likewise.
18344 (expand_copysign_absneg): Take a scalar_float_mode.
18345 (expand_copysign_bit): Likewise.
18346 (expand_copysign): Update accordingly.
18348 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18349 Alan Hayward <alan.hayward@arm.com>
18350 David Sherwood <david.sherwood@arm.com>
18352 * coretypes.h (opt_mode): New class.
18353 * machmode.h (opt_mode): Likewise.
18354 (opt_mode::else_void): New function.
18355 (opt_mode::require): Likewise.
18356 (opt_mode::exists): Likewise.
18357 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
18358 (GET_MODE_2XWIDER_MODE): Likewise.
18359 (mode_iterator::get_wider): Update accordingly.
18360 (mode_iterator::get_2xwider): Likewise.
18361 (mode_iterator::get_known_wider): Likewise, turning into a template.
18362 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
18363 forcing a wider mode to exist.
18364 * config/cr16/cr16.h (LONG_REG_P): Likewise.
18365 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18366 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
18367 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
18368 * lower-subreg.c (init_lower_subreg): Likewise.
18369 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
18370 on the final iteration.
18371 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
18372 a wider mode exists before asking for a move pattern.
18373 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
18374 forcing a wider mode to exist.
18375 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
18376 returning false if no such mode exists.
18377 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
18378 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
18379 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
18380 Avoid checking for a MODE_INT if we already know the mode is not a
18382 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
18383 forcing a wider mode to exist.
18384 (expmed_mult_highpart_optab): Likewise.
18385 (expmed_mult_highpart): Likewise.
18386 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
18388 * lto-streamer-in.c (lto_input_mode_table): Likewise.
18389 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
18390 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
18391 * internal-fn.c (expand_mul_overflow): Update use of
18392 GET_MODE_2XWIDER_MODE.
18393 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18394 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
18395 GET_MODE_WIDER_MODE.
18396 (convert_plusminus_to_widen): Likewise.
18397 * tree-switch-conversion.c (array_value_type): Likewise.
18398 * var-tracking.c (emit_note_insn_var_location): Likewise.
18399 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18400 Return false inside rather than outside the loop if no wider mode
18402 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
18403 and GET_MODE_2XWIDER_MODE
18404 (can_compare_p): Use else_void.
18405 * gdbhooks.py (OptMachineModePrinter): New class.
18406 (build_pretty_printer): Use it for opt_mode.
18408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18409 Alan Hayward <alan.hayward@arm.com>
18410 David Sherwood <david.sherwood@arm.com>
18412 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
18413 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
18415 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18416 Alan Hayward <alan.hayward@arm.com>
18417 David Sherwood <david.sherwood@arm.com>
18419 * machmode.h (mode_traits): New structure.
18420 (get_narrowest_mode): New function.
18421 (mode_iterator::start): Likewise.
18422 (mode_iterator::iterate_p): Likewise.
18423 (mode_iterator::get_wider): Likewise.
18424 (mode_iterator::get_known_wider): Likewise.
18425 (mode_iterator::get_2xwider): Likewise.
18426 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
18427 (FOR_EACH_MODE): Likewise.
18428 (FOR_EACH_MODE_FROM): Likewise.
18429 (FOR_EACH_MODE_UNTIL): Likewise.
18430 (FOR_EACH_WIDER_MODE): Likewise.
18431 (FOR_EACH_2XWIDER_MODE): Likewise.
18432 * builtins.c (expand_builtin_strlen): Use new mode iterators.
18433 * combine.c (simplify_comparison): Likewise
18434 * config/i386/i386.c (type_natural_mode): Likewise.
18435 * cse.c (cse_insn): Likewise.
18436 * dse.c (find_shift_sequence): Likewise.
18437 * emit-rtl.c (init_derived_machine_modes): Likewise.
18438 (init_emit_once): Likewise.
18439 * explow.c (hard_function_value): Likewise.
18440 * expmed.c (extract_fixed_bit_field_1): Likewise.
18441 (extract_bit_field_1): Likewise.
18442 (expand_divmod): Likewise.
18443 (emit_store_flag_1): Likewise.
18444 * expr.c (init_expr_target): Likewise.
18445 (convert_move): Likewise.
18446 (alignment_for_piecewise_move): Likewise.
18447 (widest_int_mode_for_size): Likewise.
18448 (emit_block_move_via_movmem): Likewise.
18449 (copy_blkmode_to_reg): Likewise.
18450 (set_storage_via_setmem): Likewise.
18451 (compress_float_constant): Likewise.
18452 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18453 * optabs-query.c (get_best_extraction_insn): Likewise.
18454 * optabs.c (expand_binop): Likewise.
18455 (expand_twoval_unop): Likewise.
18456 (expand_twoval_binop): Likewise.
18457 (widen_leading): Likewise.
18458 (widen_bswap): Likewise.
18459 (expand_parity): Likewise.
18460 (expand_unop): Likewise.
18461 (prepare_cmp_insn): Likewise.
18462 (prepare_float_lib_cmp): Likewise.
18463 (expand_float): Likewise.
18464 (expand_fix): Likewise.
18465 (expand_sfix_optab): Likewise.
18466 * postreload.c (move2add_use_add2_insn): Likewise.
18467 * reg-stack.c (reg_to_stack): Likewise.
18468 * reginfo.c (choose_hard_reg_mode): Likewise.
18469 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18470 * stor-layout.c (mode_for_size): Likewise.
18471 (smallest_mode_for_size): Likewise.
18472 (mode_for_vector): Likewise.
18473 (finish_bitfield_representative): Likewise.
18474 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
18475 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
18476 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18477 * var-tracking.c (prepare_call_arguments): Likewise.
18479 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18480 Alan Hayward <alan.hayward@arm.com>
18481 David Sherwood <david.sherwood@arm.com>
18483 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
18484 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
18485 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
18486 * machmode.h (mode_size): Move earlier in file.
18487 (mode_precision): Likewise.
18488 (mode_inner): Likewise.
18489 (mode_nunits): Likewise.
18490 (mode_unit_size): Likewise.
18491 (unit_unit_precision): Likewise.
18492 (mode_wider): Likewise.
18493 (mode_2xwider): Likewise.
18494 (machine_mode): New class.
18495 (mode_to_bytes): New function.
18496 (mode_to_bits): Likewise.
18497 (mode_to_precision): Likewise.
18498 (mode_to_inner): Likewise.
18499 (mode_to_unit_size): Likewise.
18500 (mode_to_unit_precision): Likewise.
18501 (mode_to_nunits): Likewise.
18502 (GET_MODE_SIZE): Use mode_to_bytes.
18503 (GET_MODE_BITSIZE): Use mode_to_bits.
18504 (GET_MODE_PRECISION): Use mode_to_precision.
18505 (GET_MODE_INNER): Use mode_to_inner.
18506 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
18507 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
18508 (GET_MODE_NUNITS): Use mode_to_nunits.
18509 * system.h (ALWAYS_INLINE): New macro.
18510 * config/powerpcspe/powerpcspe-c.c
18511 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
18512 int for arg1_mode and arg2_mode.
18514 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18515 Alan Hayward <alan.hayward@arm.com>
18516 David Sherwood <david.sherwood@arm.com>
18518 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18519 Prefix mode names with E_ in case statements.
18520 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18521 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
18522 (aarch64_split_simd_move): Likewise.
18523 (aarch64_gen_storewb_pair): Likewise.
18524 (aarch64_gen_loadwb_pair): Likewise.
18525 (aarch64_gen_store_pair): Likewise.
18526 (aarch64_gen_load_pair): Likewise.
18527 (aarch64_get_condition_code_1): Likewise.
18528 (aarch64_constant_pool_reload_icode): Likewise.
18529 (get_rsqrte_type): Likewise.
18530 (get_rsqrts_type): Likewise.
18531 (get_recpe_type): Likewise.
18532 (get_recps_type): Likewise.
18533 (aarch64_gimplify_va_arg_expr): Likewise.
18534 (aarch64_simd_container_mode): Likewise.
18535 (aarch64_emit_load_exclusive): Likewise.
18536 (aarch64_emit_store_exclusive): Likewise.
18537 (aarch64_expand_compare_and_swap): Likewise.
18538 (aarch64_gen_atomic_cas): Likewise.
18539 (aarch64_emit_bic): Likewise.
18540 (aarch64_emit_atomic_swap): Likewise.
18541 (aarch64_emit_atomic_load_op): Likewise.
18542 (aarch64_evpc_trn): Likewise.
18543 (aarch64_evpc_uzp): Likewise.
18544 (aarch64_evpc_zip): Likewise.
18545 (aarch64_evpc_ext): Likewise.
18546 (aarch64_evpc_rev): Likewise.
18547 (aarch64_evpc_dup): Likewise.
18548 (aarch64_gen_ccmp_first): Likewise.
18549 (aarch64_gen_ccmp_next): Likewise.
18550 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18551 (alpha_emit_xfloating_libcall): Likewise.
18552 (emit_insxl): Likewise.
18553 (alpha_arg_type): Likewise.
18554 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
18555 (arc_preferred_simd_mode): Likewise.
18556 (arc_secondary_reload): Likewise.
18557 (get_arc_condition_code): Likewise.
18558 (arc_print_operand): Likewise.
18559 (arc_legitimate_constant_p): Likewise.
18560 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18561 * config/arc/arc.md (casesi_load): Likewise.
18562 (casesi_compact_jump): Likewise.
18563 * config/arc/predicates.md (proper_comparison_operator): Likewise.
18564 (cc_use_register): Likewise.
18565 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18566 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18567 (arm_init_iwmmxt_builtins): Likewise.
18568 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
18569 (neon_expand_vector_init): Likewise.
18570 (arm_attr_length_move_neon): Likewise.
18571 (maybe_get_arm_condition_code): Likewise.
18572 (arm_emit_vector_const): Likewise.
18573 (arm_preferred_simd_mode): Likewise.
18574 (arm_output_iwmmxt_tinsr): Likewise.
18575 (thumb1_output_casesi): Likewise.
18576 (thumb2_output_casesi): Likewise.
18577 (arm_emit_load_exclusive): Likewise.
18578 (arm_emit_store_exclusive): Likewise.
18579 (arm_expand_compare_and_swap): Likewise.
18580 (arm_evpc_neon_vuzp): Likewise.
18581 (arm_evpc_neon_vzip): Likewise.
18582 (arm_evpc_neon_vrev): Likewise.
18583 (arm_evpc_neon_vtrn): Likewise.
18584 (arm_evpc_neon_vext): Likewise.
18585 (arm_validize_comparison): Likewise.
18586 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
18587 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
18588 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
18589 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
18590 (c6x_preferred_simd_mode): Likewise.
18591 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
18592 (epiphany_rtx_costs): Likewise.
18593 * config/epiphany/predicates.md (proper_comparison_operator):
18595 * config/frv/frv.c (condexec_memory_operand): Likewise.
18596 (frv_emit_move): Likewise.
18597 (output_move_single): Likewise.
18598 (output_condmove_single): Likewise.
18599 (frv_hard_regno_mode_ok): Likewise.
18600 (frv_matching_accg_mode): Likewise.
18601 * config/h8300/h8300.c (split_adds_subs): Likewise.
18602 (h8300_rtx_costs): Likewise.
18603 (h8300_print_operand): Likewise.
18604 (compute_mov_length): Likewise.
18605 (output_logical_op): Likewise.
18606 (compute_logical_op_length): Likewise.
18607 (compute_logical_op_cc): Likewise.
18608 (h8300_shift_needs_scratch_p): Likewise.
18609 (output_a_shift): Likewise.
18610 (compute_a_shift_length): Likewise.
18611 (compute_a_shift_cc): Likewise.
18612 (expand_a_rotate): Likewise.
18613 (output_a_rotate): Likewise.
18614 * config/i386/i386.c (classify_argument): Likewise.
18615 (function_arg_advance_32): Likewise.
18616 (function_arg_32): Likewise.
18617 (function_arg_64): Likewise.
18618 (function_value_64): Likewise.
18619 (ix86_gimplify_va_arg): Likewise.
18620 (ix86_legitimate_constant_p): Likewise.
18621 (put_condition_code): Likewise.
18622 (split_double_mode): Likewise.
18623 (ix86_avx256_split_vector_move_misalign): Likewise.
18624 (ix86_expand_vector_logical_operator): Likewise.
18625 (ix86_split_idivmod): Likewise.
18626 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
18627 (ix86_build_const_vector): Likewise.
18628 (ix86_build_signbit_mask): Likewise.
18629 (ix86_match_ccmode): Likewise.
18630 (ix86_cc_modes_compatible): Likewise.
18631 (ix86_expand_branch): Likewise.
18632 (ix86_expand_sse_cmp): Likewise.
18633 (ix86_expand_sse_movcc): Likewise.
18634 (ix86_expand_int_sse_cmp): Likewise.
18635 (ix86_expand_vec_perm_vpermi2): Likewise.
18636 (ix86_expand_vec_perm): Likewise.
18637 (ix86_expand_sse_unpack): Likewise.
18638 (ix86_expand_int_addcc): Likewise.
18639 (ix86_split_to_parts): Likewise.
18640 (ix86_vectorize_builtin_gather): Likewise.
18641 (ix86_vectorize_builtin_scatter): Likewise.
18642 (avx_vpermilp_parallel): Likewise.
18643 (inline_memory_move_cost): Likewise.
18644 (ix86_tieable_integer_mode_p): Likewise.
18645 (x86_maybe_negate_const_int): Likewise.
18646 (ix86_expand_vector_init_duplicate): Likewise.
18647 (ix86_expand_vector_init_one_nonzero): Likewise.
18648 (ix86_expand_vector_init_one_var): Likewise.
18649 (ix86_expand_vector_init_concat): Likewise.
18650 (ix86_expand_vector_init_interleave): Likewise.
18651 (ix86_expand_vector_init_general): Likewise.
18652 (ix86_expand_vector_set): Likewise.
18653 (ix86_expand_vector_extract): Likewise.
18654 (emit_reduc_half): Likewise.
18655 (ix86_emit_i387_round): Likewise.
18656 (ix86_mangle_type): Likewise.
18657 (ix86_expand_round_sse4): Likewise.
18658 (expand_vec_perm_blend): Likewise.
18659 (canonicalize_vector_int_perm): Likewise.
18660 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
18661 (expand_vec_perm_1): Likewise.
18662 (expand_vec_perm_interleave3): Likewise.
18663 (expand_vec_perm_even_odd_pack): Likewise.
18664 (expand_vec_perm_even_odd_1): Likewise.
18665 (expand_vec_perm_broadcast_1): Likewise.
18666 (ix86_vectorize_vec_perm_const_ok): Likewise.
18667 (ix86_expand_vecop_qihi): Likewise.
18668 (ix86_expand_mul_widen_hilo): Likewise.
18669 (ix86_expand_sse2_abs): Likewise.
18670 (ix86_expand_pextr): Likewise.
18671 (ix86_expand_pinsr): Likewise.
18672 (ix86_preferred_simd_mode): Likewise.
18673 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
18674 * config/i386/sse.md (*andnot<mode>3): Likewise.
18675 (<mask_codefor><code><mode>3<mask_name>): Likewise.
18676 (*<code><mode>3): Likewise.
18677 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
18678 (ia64_expand_atomic_op): Likewise.
18679 (ia64_arg_type): Likewise.
18680 (ia64_mode_to_int): Likewise.
18681 (ia64_scalar_mode_supported_p): Likewise.
18682 (ia64_vector_mode_supported_p): Likewise.
18683 (expand_vec_perm_broadcast): Likewise.
18684 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
18685 (iq2000_function_arg_advance): Likewise.
18686 (iq2000_function_arg): Likewise.
18687 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
18688 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
18689 (m68k_libcall_value): Likewise.
18690 (m68k_function_value): Likewise.
18691 (sched_attr_op_type): Likewise.
18692 * config/mcore/mcore.c (mcore_output_move): Likewise.
18693 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
18695 (microblaze_function_arg): Likewise.
18696 * config/mips/mips.c (mips16_build_call_stub): Likewise.
18697 (mips_print_operand): Likewise.
18698 (mips_mode_ok_for_mov_fmt_p): Likewise.
18699 (mips_vector_mode_supported_p): Likewise.
18700 (mips_preferred_simd_mode): Likewise.
18701 (mips_expand_vpc_loongson_even_odd): Likewise.
18702 (mips_expand_vec_unpack): Likewise.
18703 (mips_expand_vi_broadcast): Likewise.
18704 (mips_expand_vector_init): Likewise.
18705 (mips_expand_vec_reduc): Likewise.
18706 (mips_expand_msa_cmp): Likewise.
18707 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
18708 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
18709 (cc_flags_for_mode): Likewise.
18710 * config/msp430/msp430.c (msp430_print_operand): Likewise.
18711 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
18712 (nds32_output_casesi_pc_relative): Likewise.
18713 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18714 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
18715 (nvptx_gen_unpack): Likewise.
18716 (nvptx_gen_pack): Likewise.
18717 (nvptx_gen_shuffle): Likewise.
18718 (nvptx_gen_wcast): Likewise.
18719 (nvptx_preferred_simd_mode): Likewise.
18720 * config/pa/pa.c (pa_secondary_reload): Likewise.
18721 * config/pa/predicates.md (base14_operand): Likewise.
18722 * config/powerpcspe/powerpcspe-c.c
18723 (altivec_resolve_overloaded_builtin): Likewise.
18724 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
18726 (rs6000_preferred_simd_mode): Likewise.
18727 (output_vec_const_move): Likewise.
18728 (rs6000_expand_vector_extract): Likewise.
18729 (rs6000_split_vec_extract_var): Likewise.
18730 (reg_offset_addressing_ok_p): Likewise.
18731 (rs6000_legitimate_offset_address_p): Likewise.
18732 (rs6000_legitimize_address): Likewise.
18733 (rs6000_emit_set_const): Likewise.
18734 (rs6000_const_vec): Likewise.
18735 (rs6000_emit_move): Likewise.
18736 (spe_build_register_parallel): Likewise.
18737 (rs6000_darwin64_record_arg_recurse): Likewise.
18738 (swap_selector_for_mode): Likewise.
18739 (spe_init_builtins): Likewise.
18740 (paired_init_builtins): Likewise.
18741 (altivec_init_builtins): Likewise.
18742 (do_load_for_compare): Likewise.
18743 (rs6000_generate_compare): Likewise.
18744 (rs6000_expand_float128_convert): Likewise.
18745 (emit_load_locked): Likewise.
18746 (emit_store_conditional): Likewise.
18747 (rs6000_output_function_epilogue): Likewise.
18748 (rs6000_handle_altivec_attribute): Likewise.
18749 (rs6000_function_value): Likewise.
18750 (emit_fusion_gpr_load): Likewise.
18751 (emit_fusion_p9_load): Likewise.
18752 (emit_fusion_p9_store): Likewise.
18753 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
18754 (fusion_gpr_mem_load): Likewise.
18755 (fusion_addis_mem_combo_load): Likewise.
18756 (fusion_addis_mem_combo_store): Likewise.
18757 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
18758 (fusion_gpr_mem_load): Likewise.
18759 (fusion_addis_mem_combo_load): Likewise.
18760 (fusion_addis_mem_combo_store): Likewise.
18761 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18763 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
18764 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
18765 (rs6000_preferred_simd_mode): Likewise.
18766 (output_vec_const_move): Likewise.
18767 (rs6000_expand_vector_extract): Likewise.
18768 (rs6000_split_vec_extract_var): Likewise.
18769 (reg_offset_addressing_ok_p): Likewise.
18770 (rs6000_legitimate_offset_address_p): Likewise.
18771 (rs6000_legitimize_address): Likewise.
18772 (rs6000_emit_set_const): Likewise.
18773 (rs6000_const_vec): Likewise.
18774 (rs6000_emit_move): Likewise.
18775 (rs6000_darwin64_record_arg_recurse): Likewise.
18776 (swap_selector_for_mode): Likewise.
18777 (paired_init_builtins): Likewise.
18778 (altivec_init_builtins): Likewise.
18779 (rs6000_expand_float128_convert): Likewise.
18780 (emit_load_locked): Likewise.
18781 (emit_store_conditional): Likewise.
18782 (rs6000_output_function_epilogue): Likewise.
18783 (rs6000_handle_altivec_attribute): Likewise.
18784 (rs6000_function_value): Likewise.
18785 (emit_fusion_gpr_load): Likewise.
18786 (emit_fusion_p9_load): Likewise.
18787 (emit_fusion_p9_store): Likewise.
18788 * config/rx/rx.c (rx_gen_move_template): Likewise.
18789 (flags_from_mode): Likewise.
18790 * config/s390/predicates.md (s390_alc_comparison): Likewise.
18791 (s390_slb_comparison): Likewise.
18792 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
18793 (s390_vector_mode_supported_p): Likewise.
18794 (s390_cc_modes_compatible): Likewise.
18795 (s390_match_ccmode_set): Likewise.
18796 (s390_canonicalize_comparison): Likewise.
18797 (s390_emit_compare_and_swap): Likewise.
18798 (s390_branch_condition_mask): Likewise.
18799 (s390_rtx_costs): Likewise.
18800 (s390_secondary_reload): Likewise.
18801 (__SECONDARY_RELOAD_CASE): Likewise.
18802 (s390_expand_cs): Likewise.
18803 (s390_preferred_simd_mode): Likewise.
18804 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
18805 * config/sh/sh.c (sh_print_operand): Likewise.
18806 (dump_table): Likewise.
18807 (sh_secondary_reload): Likewise.
18808 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18809 * config/sh/sh.md (casesi_worker_1): Likewise.
18810 (casesi_worker_2): Likewise.
18811 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
18812 (fcc_comparison_operator): Likewise.
18813 * config/sparc/sparc.c (sparc_expand_move): Likewise.
18814 (emit_soft_tfmode_cvt): Likewise.
18815 (sparc_preferred_simd_mode): Likewise.
18816 (output_cbranch): Likewise.
18817 (sparc_print_operand): Likewise.
18818 (sparc_expand_vec_perm_bmask): Likewise.
18819 (vector_init_bshuffle): Likewise.
18820 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18821 (spu_vector_mode_supported_p): Likewise.
18822 (spu_expand_insv): Likewise.
18823 (spu_emit_branch_or_set): Likewise.
18824 (spu_handle_vector_attribute): Likewise.
18825 (spu_builtin_splats): Likewise.
18826 (spu_builtin_extract): Likewise.
18827 (spu_builtin_promote): Likewise.
18828 (spu_expand_sign_extend): Likewise.
18829 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18830 (tilegx_simd_int): Likewise.
18831 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
18832 (tilepro_simd_int): Likewise.
18833 * config/v850/v850.c (const_double_split): Likewise.
18834 (v850_print_operand): Likewise.
18835 (ep_memory_offset): Likewise.
18836 * config/vax/vax.c (vax_rtx_costs): Likewise.
18837 (vax_output_int_move): Likewise.
18838 (vax_output_int_add): Likewise.
18839 (vax_output_int_subtract): Likewise.
18840 * config/visium/predicates.md (visium_branch_operator): Likewise.
18841 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
18842 (visium_print_operand_address): Likewise.
18843 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18844 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
18845 (xtensa_expand_conditional_branch): Likewise.
18846 (xtensa_copy_incoming_a7): Likewise.
18847 (xtensa_output_literal): Likewise.
18848 * dfp.c (decimal_real_maxval): Likewise.
18849 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18851 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18852 Alan Hayward <alan.hayward@arm.com>
18853 David Sherwood <david.sherwood@arm.com>
18855 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
18856 (mode_nunits_inline): Likewise.
18857 (mode_inner_inline): Likewise.
18858 (mode_unit_size_inline): Likewise.
18859 (mode_unit_precision_inline): Likewise.
18860 (emit_insn_modes_h): Likewise. Also emit a #define of the
18862 (emit_mode_wider): Add an E_ prefix to mode names.
18863 (emit_mode_complex): Likewise.
18864 (emit_mode_inner): Likewise.
18865 (emit_mode_adjustments): Likewise.
18866 (emit_mode_int_n): Likewise.
18867 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
18868 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
18869 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
18870 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
18871 (CRC32_BUILTIN, ENTRY): Likewise.
18872 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
18873 (aarch64_pop_regs): Likewise.
18874 (aarch64_process_components): Likewise.
18875 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
18876 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
18877 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
18878 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
18879 * config/arm/arm.c (arm_init_libfuncs): Likewise.
18880 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
18882 * config/i386/i386-builtin.def (pcmpestr): Likewise.
18883 (pcmpistr): Likewise.
18884 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
18885 * config/mmix/mmix.c (mmix_output_condition): Likewise.
18886 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18888 * config/rl78/rl78.c (mduc_regs): Likewise.
18889 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18890 (htm_expand_builtin): Likewise.
18891 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
18892 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
18893 * config/xtensa/xtensa.c (print_operand): Likewise.
18894 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
18895 (NUM_MODE_VECTOR_INT): Likewise.
18896 * genoutput.c (null_operand): Likewise.
18897 (output_operand_data): Likewise.
18898 * genrecog.c (print_parameter_value): Likewise.
18899 * lra.c (debug_operand_data): Likewise.
18901 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18903 * dwarf2out.c (scompare_loc_descriptor_wide)
18904 (scompare_loc_descriptor_narrow): New functions, split out from...
18905 (scompare_loc_descriptor): ...here.
18906 * expmed.c (emit_store_flag_int): New function, split out from...
18907 (emit_store_flag): ...here.
18909 2017-08-30 Richard Biener <rguenther@suse.de>
18911 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
18912 (dwarf2out_early_finish): Move setting of AT_pubnames from
18913 early debug output to early finish.
18915 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18917 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
18918 and -mdata-region to the assembler.
18920 2017-08-30 Richard Biener <rguenther@suse.de>
18922 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
18924 (gen_subprogram_die): Add DW_AT_object_pointer only early.
18925 (dwarf2out_early_global_decl): Only generate a DIE for the
18926 abstract origin if it doesn't already exist or is a declaration DIE.
18927 (resolve_addr): Do not add the linkage name twice when
18928 generating a stub DIE for the DW_TAG_GNU_call_site target.
18930 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18932 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18933 Use machine_mode rather than int for arg1_mode.
18935 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
18938 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
18939 that the second argument of the built-in functions to unpack
18940 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
18941 switch statement instead a lot of if statements.
18942 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
18943 Allow 64-bit values to be in Altivec registers as well as
18944 traditional floating point registers.
18945 (pack<mode>, FMOVE128_VSX iterator): Likewise.
18947 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
18949 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
18950 MAX_REGS_PER_ADDRESS == 1.
18952 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
18954 * config/i386/i386.opt (flag_fentry): Do not init to -1.
18955 * config/i386/i386.c (ix86_option_override_internal): Simplify
18956 setting of opts->x_flag_entry.
18958 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18959 Jakub Jelinek <jakub@redhat.com>
18960 Richard Biener <rguenther@suse.de>
18962 PR tree-optimization/81503
18963 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
18964 folded constant fits in the target type; reorder tests for clarity.
18966 2017-08-29 Martin Liska <mliska@suse.cz>
18968 * passes.def: Include pass_lower_switch.
18969 * stmt.c (dump_case_nodes): Remove and move to
18970 tree-switch-conversion.
18971 (case_values_threshold): Likewise.
18972 (expand_switch_as_decision_tree_p): Likewise.
18973 (emit_case_decision_tree): Likewise.
18974 (expand_case): Likewise.
18975 (balance_case_nodes): Likewise.
18976 (node_has_low_bound): Likewise.
18977 (node_has_high_bound): Likewise.
18978 (node_is_bounded): Likewise.
18979 (emit_case_nodes): Likewise.
18980 (struct simple_case_node): New struct.
18981 (add_case_node): Remove.
18982 (emit_case_dispatch_table): Use vector instead of case_list.
18983 (reset_out_edges_aux): Remove.
18984 (compute_cases_per_edge): Likewise.
18985 (expand_case): Build list of simple_case_node.
18986 (expand_sjlj_dispatch_table): Use it.
18987 * tree-switch-conversion.c (struct case_node): Moved from
18988 stmt.c and adjusted.
18989 (emit_case_nodes): Likewise.
18990 (node_has_low_bound): Likewise.
18991 (node_has_high_bound): Likewise.
18992 (node_is_bounded): Likewise.
18993 (case_values_threshold): Likewise.
18994 (reset_out_edges_aux): Likewise.
18995 (compute_cases_per_edge): Likewise.
18996 (add_case_node): Likewise.
18997 (dump_case_nodes): Likewise.
18998 (balance_case_nodes): Likewise.
18999 (expand_switch_as_decision_tree_p): Likewise.
19000 (emit_jump): Likewise.
19001 (emit_case_decision_tree): Likewise.
19002 (try_switch_expansion): Likewise.
19003 (do_jump_if_equal): Likewise.
19004 (emit_cmp_and_jump_insns): Likewise.
19005 (fix_phi_operands_for_edge): New function.
19006 (record_phi_operand_mapping): Likewise.
19007 (class pass_lower_switch): New pass.
19008 (pass_lower_switch::execute): New function.
19009 (make_pass_lower_switch): Likewise.
19010 (conditional_probability):
19011 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
19012 * tree-pass.h: Add make_pass_lower_switch.
19014 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19017 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
19020 2017-08-29 Richard Biener <rguenther@suse.de>
19022 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
19023 we do not add a DW_AT_inline attribute twice.
19024 (gen_subprogram_die): Remove code setting DW_AT_inline on
19025 DECL_ABSTRACT_P nodes.
19027 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
19029 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
19030 calls to internal functions.
19031 (gimplify_modify_expr): Likewise.
19032 * tree-call-cdce.c (use_internal_fn): Likewise.
19033 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
19034 (convert_to_divmod): Set the nothrow flag.
19035 * tree-if-conv.c (predicate_mem_writes): Likewise.
19036 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19037 (vectorizable_call): Likewise.
19038 (vectorizable_store): Likewise.
19039 (vectorizable_load): Likewise.
19040 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
19041 (vect_recog_mask_conversion_pattern): Likewise.
19043 2017-08-29 Martin Liska <mliska@suse.cz>
19046 * gcc.c (driver_handle_option): Add new argument.
19047 * opts-common.c (handle_option): Pass
19048 target_option_override_hook.
19049 * opts-global.c (lang_handle_option): Add new option.
19050 (set_default_handlers): Add new argument.
19051 (decode_options): Likewise.
19052 * opts.c (target_handle_option): Likewise.
19053 (common_handle_option): Call target_option_override_hook.
19054 * opts.h (struct cl_option_handler_func): Add hook for
19055 target option override.
19056 (struct cl_option_handlers): Likewise.
19057 (set_default_handlers): Add new argument.
19058 (decode_options): Likewise.
19059 (common_handle_option): Likewise.
19060 (target_handle_option): Likewise.
19061 * toplev.c (toplev::main): Pass targetm.target_option.override
19064 2017-08-29 Richard Biener <rguenther@suse.de>
19065 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
19067 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
19068 life to the active subtree.
19070 2017-08-28 Jeff Law <law@redhat.com>
19072 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
19073 derive_equivalences.
19074 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
19075 Code moved into....
19076 (edge_info::derive_equivalences): New private member function
19078 * tree-ssa-dom.c (class edge_info): Changed from a struct
19079 to a class. Add ctor/dtor, methods and data members.
19080 (edge_info::edge_info): Renamed from allocate_edge_info.
19081 Initialize additional members.
19082 (edge_info::~edge_info): New.
19083 (free_dom_edge_info): Delete the edge info.
19084 (record_edge_info): Use new class & associated member functions.
19085 Tighten forms for testing for edge equivalences.
19086 (record_temporary_equivalences): Iterate over the simple
19087 equivalences rather than assuming there's only one per edge.
19088 (cprop_into_successor_phis): Iterate over the simple
19089 equivalences rather than assuming there's only one per edge.
19090 (optimize_stmt): Use operand_equal_p rather than pointer
19091 equality for mini-DSE code.
19093 2017-08-28 Nathan Sidwell <nathan@acm.org>
19095 * gcc.c (execute): Fold SIGPIPE handling into switch
19096 statement. Adjust internal error message.
19098 2017-08-28 Richard Biener <rguenther@suse.de>
19101 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
19102 Do nothing for removed DIEs.
19104 2017-08-28 Richard Biener <rguenther@suse.de>
19106 PR tree-optimization/81977
19107 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
19110 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
19113 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
19114 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
19115 using targetm.gen_mem_thread_fence.
19117 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
19120 * config/i386/i386.md (*<btsc><mode>): Change operand 2
19121 predicate to register_operand. Reorder operands.
19122 (*btr<mode>): Ditto.
19123 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
19124 (*btr<mode>_mask): Ditto.
19126 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
19128 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
19129 * config/rs6000/xmmintrin.h: New file.
19130 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
19132 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19135 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
19136 parameter and_insn and return it.
19137 (recombine_lvx_pattern): Insert a copy to ensure availability of
19138 the base register of the copied masking operation at the point of
19139 the instruction replacement.
19140 (recombine_stvx_pattern): Likewise.
19142 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
19144 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
19145 undocumented switches.
19146 (-mpower9-dform-vector): Likewise.
19147 (-mpower9-dform): Likewise.
19148 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19149 comments to delete references to -mpower9-dform* switches.
19150 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
19151 Delete reference to -mpower9-dform* switches, test for
19152 -mpower9-vector instead.
19153 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
19154 (OTHER_P9_VECTOR_MASKS): Likewise.
19155 (POWERPC_MASKS): Likewise.
19156 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
19157 tests against -mpower9-dform* to -mpower9-vector. Delete code
19158 that checked for -mpower9-dform* consistancy with other options.
19159 Add test for -mpower9-misc to enable other power9 switches.
19160 (rs6000_init_hard_regno_mode_ok): Likewise.
19161 (rs6000_option_override_internal): Likewise.
19162 (rs6000_emit_prologue): Likewise.
19163 (rs6000_emit_epilogue): Likewise.
19164 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
19165 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
19166 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
19168 (emit_fusion_p9_store): Likewise.
19169 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
19170 resetting these macros if the assembler does not support ISA 3.0
19172 (TARGET_P9_DFORM_VECTOR): Likewise.
19173 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
19174 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
19176 2017-08-25 Alan Modra <amodra@gmail.com>
19178 PR rtl-optimization/81747
19179 * cse.c (cse_extended_basic_block): Don't attempt to record
19180 equivalences for degenerate conditional branches that branch
19181 to their fall-through.
19183 2017-08-24 Martin Sebor <msebor@redhat.com>
19185 PR middle-end/81908
19186 * gimple-fold.c (size_must_be_zero_p): New function.
19187 (gimple_fold_builtin_memory_op): Call it.
19189 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
19191 * config/rs6000/mm_malloc.h: New file.
19193 2017-08-24 Bin Cheng <bin.cheng@arm.com>
19195 PR tree-optimization/81913
19196 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
19197 analysis when either IVs in condition can wrap.
19199 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
19201 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
19202 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
19204 2017-08-24 Richard Biener <rguenther@suse.de>
19207 * targhooks.c (default_target_can_inline_p): Properly
19208 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
19209 is present and always compare.
19210 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
19211 infer -mfpmath=sse from TARGET_SSE_P.
19212 (ix86_can_inline_p): Properly use target_option_default_node when
19213 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
19215 2017-08-24 Richard Biener <rguenther@suse.de>
19218 * dwarf2out.c (output_die): Handle flag_generate_offload like
19220 (output_comp_unit): Likewise.
19221 (gen_array_type_die): Likewise.
19222 (dwarf2out_early_finish): Likewise.
19223 (note_variable_value_in_expr): Likewise.
19224 (dwarf2out_finish): Likewise. Adjust assert.
19225 * cgraphunit.c (symbol_table::compile): Move setting of
19226 flag_generate_offload earlier ...
19227 (symbol_table::finalize_compilation_unit): ... here, before
19228 early debug finalization.
19230 2017-08-24 Richard Biener <rguenther@suse.de>
19232 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
19233 and ipa-fnsummary.h.
19234 (ix86_can_inline_p): When ix86_fpmath flags do not match
19235 check whether the callee uses FP math at all.
19237 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
19239 PR middle-end/81931
19240 * tree-ssanames.c (get_nonzero_bits): Use element_precision
19241 instead of TYPE_PRECISION.
19243 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
19244 Alan Hayward <alan.hayward@arm.com>
19245 David Sherwood <david.sherwood@arm.com>
19247 * combine.c (make_extraction): Use subreg_offset_from_lsb.
19249 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
19251 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
19253 (ix86_frame::stack_realign_allocate): New field.
19254 (struct machine_frame_state): Modify comments.
19255 (machine_frame_state::sp_realigned_fp_end): New field.
19256 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
19257 layout calculation.
19258 (sp_valid_at): Add assertion to assure no attempt to access invalid
19259 offset of a realigned stack.
19260 (fp_valid_at): Likewise.
19261 (choose_baseaddr): Modify comments.
19262 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
19263 ix86_expand_prologue.
19264 (ix86_expand_prologue): Modify stack realignment and allocation.
19265 (ix86_expand_epilogue): Modify comments.
19266 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
19267 avx2_runtime, avx512f, and avx512f_runtime.
19269 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
19271 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
19272 (mstackrealign): Do not init to -1.
19273 * config/i386/i386.c (ix86_option_override_internal):
19274 Check opts_set, not opts when setting default value of
19275 opts->x_ix86_force_align_arg_pointer.
19277 2017-08-23 Richard Biener <rguenther@suse.de>
19279 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
19280 lang_hooks.decl_printable_name.
19281 * print-rtl-function.c (print_rtx_function): Likewise.
19282 * tree-pretty-print.c (dump_function_header): Likewise.
19284 2017-08-23 Richard Biener <rguenther@suse.de>
19287 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
19288 -g0 at compile-time.
19290 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19292 PR middle-end/19706
19293 * doc/sourcebuild.texi (Other hardware attributes):
19296 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19298 PR middle-end/19706
19299 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
19300 Fix single-use check.
19302 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
19304 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
19306 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19308 * doc/install.texi: Modify to add more details on running selected
19311 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19313 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
19314 is combined with -mabi=ms.
19315 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
19318 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19320 PR tree-optimization/81488
19321 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
19322 and cached_basis fields.
19323 (MAX_SPREAD): New constant.
19324 (alloc_cand_and_find_basis): Initialize new fields.
19325 (clear_visited): New function.
19326 (create_phi_basis_1): Rename from create_phi_basis, set visited
19327 and cached_basis fields.
19328 (create_phi_basis): New wrapper function.
19329 (phi_add_costs_1): Rename from phi_add_costs, add spread
19330 parameter, set visited field, short-circuit when limits reached.
19331 (phi_add_costs): New wrapper function.
19332 (record_phi_increments_1): Rename from record_phi_increments, set
19334 (record_phi_increments): New wrapper function.
19335 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
19336 (phi_incr_cost): New wrapper function.
19337 (all_phi_incrs_profitable_1): Rename from
19338 all_phi_incrs_profitable, set visited field.
19339 (all_phi_incrs_profitable): New wrapper function.
19341 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19342 Alan Hayward <alan.hayward@arm.com>
19343 David Sherwood <david.sherwood@arm.com>
19345 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
19346 that takes the outer and inner modes.
19347 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
19348 comparison as the canonical test for a paradoxical subreg.
19349 * combine.c (simplify_set): Use paradoxical_subreg_p.
19350 (make_extraction): Likewise.
19351 (force_to_mode): Likewise.
19352 (rtx_equal_for_field_assignment_p): Likewise.
19353 (gen_lowpart_for_combine): Likewise.
19354 (simplify_comparison): Likewise.
19355 * cse.c (equiv_constant): Likewise.
19356 * expmed.c (store_bit_field_1): Likewise.
19357 * final.c (alter_subreg): Likewise.
19358 * fwprop.c (propagate_rtx): Likewise.
19359 (forward_propagate_subreg): Likewise.
19360 * ira-conflicts.c (ira_build_conflicts): Likewise.
19361 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
19362 * lra-constraints.c (curr_insn_transform): Likewise.
19363 (split_reg): Likewise.
19364 * lra-eliminations.c (move_plus_up): Likewise.
19365 (lra_eliminate_regs_1): Likewise.
19366 * recog.c (general_operand): Likewise.
19367 * ree.c (combine_reaching_defs): Likewise.
19368 * reload.c (push_reload): Likewise.
19369 (find_reloads): Likewise.
19370 * reload1.c (elimination_effects): Likewise.
19371 (compute_reload_subreg_offset): Likewise.
19372 (choose_reload_regs): Likewise.
19373 * rtlanal.c (subreg_lsb_1): Likewise.
19374 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19375 (simplify_subreg): Likewise.
19376 * var-tracking.c (track_loc_p): Likewise.
19377 * emit-rtl.c (byte_lowpart_offset): Likewise.
19378 (paradoxical_subreg_p): Delete out-of-line definition.
19380 2017-08-22 Jeff Law <law@redhat.com>
19382 PR tree-optimization/81741
19383 PR tree-optimization/71947
19384 * tree-ssa-dom.c: Include tree-inline.h.
19385 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
19386 equivalences if one is more expensive to compute than the other.
19387 * tree-ssa-scopedtables.h (class const_or_copies): Make
19388 record_const_or_copy_raw method private.
19389 (class avail_exprs_stack): New method simplify_binary_operation.
19390 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
19391 avail_exprs_stack::simplify_binary_operation as needed.
19392 (avail_exprs_stack::simplify_binary_operation): New function.
19394 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
19396 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
19397 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
19398 (DOT_SYMBOLS): Likewise.
19399 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
19400 (RELOCATABLE_NEEDS_FIXUP): Likewise.
19401 (RS6000_ABI_NAME): Likewise.
19402 (TARGET_CMODEL): Likewise.
19403 (TOC_SECTION_ASM_OP): Likewise.
19404 (SET_CMODEL): New macro.
19405 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
19407 2017-08-22 Richard Biener <rguenther@suse.de>
19409 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
19410 to allow for free-lang-data replacements similar to verify_type_variant.
19412 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19413 Alan Hayward <alan.hayward@arm.com>
19414 David Sherwood <david.sherwood@arm.com>
19416 * config/aarch64/aarch64.md (casesi): Use DImode rather than
19417 VOIDmode for the LABEL_REF.
19419 2017-08-22 Richard Biener <rguenther@suse.de>
19421 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
19423 2017-08-22 Richard Biener <rguenther@suse.de>
19425 * common.opt (feliminate-dwarf2-dups): Ignore.
19426 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
19427 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
19428 same_die_p_wrap, compute_section_prefix,
19429 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
19430 (comdat_symbol_id, comdat_symbol_number): Likewise.
19431 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
19433 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
19434 (output_die): Mark unreachable path unreachable.
19435 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
19436 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
19437 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
19438 (dwarf2out_early_finish): Likewise.
19440 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
19442 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
19444 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
19447 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
19448 not VAR_P. Filter attribute warnings with OPT_Wattributes.
19449 (avr_attribute_table) <io, io_low, address>: Initialize
19450 .decl_required with true.
19452 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
19454 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
19455 undocumented debugging options.
19456 (-mvsx-scalar-double): Likewise.
19457 (-mallow-df-permute): Likewise.
19458 (-mvectorize-builtins): Likewise.
19459 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19460 (rs6000_builtin_vectorized_function): Likewise.
19461 (rs6000_builtin_md_vectorized_function): Likewise.
19462 (rs6000_opt_vars): Likewise.
19464 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
19467 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
19468 (*btrq_imm): Rename from *btrq.
19469 (*btcq_imm): Rename from *btcq.
19470 (btsc): New code attribute.
19471 (*<btsc><mode>): New insn pattern.
19472 (*btr<mode>): Ditto.
19473 (*<btsc><mode>_mask): New insn_and_split pattern.
19474 (*btr<mode>_mask): Ditto.
19476 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19477 Alan Hayward <alan.hayward@arm.com>
19478 David Sherwood <david.sherwood@arm.com>
19480 * function.c (pad_below): Simplify padding calculation.
19482 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19483 Alan Hayward <alan.hayward@arm.com>
19484 David Sherwood <david.sherwood@arm.com>
19486 * target.def (function_prologue): Remove frame size argument.
19487 (function_epilogue): Likewise.
19488 * doc/tm.texi: Regenerate.
19489 * final.c (final_start_function): Update call to function_prologue.
19490 (final_end_function): Update call to function_epilogue.
19491 (default_function_pro_epilogue): Remove frame size argument.
19492 * output.h (default_function_pro_epilogue): Likewise.
19493 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
19494 (arm_output_function_prologue): Likewise.
19495 * config/frv/frv.c (frv_function_prologue): Likewise.
19496 (frv_function_epilogue): Likewise.
19497 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
19498 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
19499 (ia64_output_function_epilogue): Likewise.
19500 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
19501 (m32r_output_function_epilogue): Likewise.
19502 * config/microblaze/microblaze.c (microblaze_function_prologue)
19503 (microblaze_function_epilogue): Likewise.
19504 * config/mips/mips.c (mips_output_function_prologue): Likewise.
19505 (mips_output_function_epilogue): Likewise.
19506 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
19507 (mmix_target_asm_function_epilogue): Likewise.
19508 * config/msp430/msp430.c (msp430_start_function): Likewise.
19509 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
19510 (nds32_asm_function_epilogue): Likewise.
19511 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
19512 * config/pa/pa.c (pa_output_function_prologue): Likewise.
19513 (pa_output_function_epilogue): Likewise.
19514 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
19515 (rs6000_output_function_epilogue): Likewise.
19516 * config/rl78/rl78.c (rl78_start_function): Likewise.
19517 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
19518 (rs6000_output_function_epilogue): Likewise.
19519 * config/rx/rx.c (rx_output_function_prologue): Likewise.
19520 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
19521 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
19522 (sparc_asm_function_epilogue): Likewise.
19524 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19526 * tree.h (type_has_mode_precision_p): New function.
19527 * convert.c (convert_to_integer_1): Use it.
19528 * expr.c (expand_expr_real_2): Likewise.
19529 (expand_expr_real_1): Likewise.
19530 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
19531 * match.pd: Likewise.
19532 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
19533 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
19534 * tree-tailcall.c (process_assignment): Likewise.
19535 * tree-vect-loop.c (vectorizable_reduction): Likewise.
19536 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
19537 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
19538 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19539 (vectorizable_assignment): Likewise.
19540 (vectorizable_shift): Likewise.
19541 (vectorizable_operation): Likewise.
19542 * tree-vrp.c (register_edge_assert_for_2): Likewise.
19544 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
19546 * match.pd: Add pow (C, x) simplification.
19548 2017-08-21 Richard Biener <rguenther@suse.de>
19550 PR tree-optimization/81900
19551 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
19552 for blocks with abnormal predecessors.
19553 (compute_antic): Do not set visited flag prematurely.
19555 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
19558 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
19560 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19562 * stor-layout.h (vector_type_mode): Move to...
19563 * tree.h (vector_type_mode): ...here.
19564 * stor-layout.c (vector_type_mode): Move to...
19565 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
19567 2017-08-21 Richard Biener <rguenther@suse.de>
19569 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
19570 register_external_die hooks.
19571 (debug_false_tree_charstarstar_uhwistar): Declare.
19572 (debug_nothing_tree_charstar_uhwi): Likewise.
19573 * debug.c (do_nothing_debug_hooks): Adjust.
19574 (debug_false_tree_charstarstar_uhwistar): New do nothing.
19575 (debug_nothing_tree_charstar_uhwi): Likewise.
19576 * dbxout.c (dbx_debug_hooks): Adjust.
19577 (xcoff_debug_hooks): Likewise.
19578 * sdbout.c (sdb_debug_hooks): Likewise.
19579 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19580 * dwarf2out.c (macinfo_label_base): New global.
19581 (dwarf2out_register_external_die): New function for the
19582 register_external_die hook.
19583 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
19584 (dwarf2_debug_hooks): Use them.
19585 (dwarf2_lineno_debug_hooks): Adjust.
19586 (struct die_struct): Add with_offset flag.
19587 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
19588 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
19589 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
19590 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
19591 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
19592 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
19593 defining section names for the early LTO debug variants.
19594 (reset_indirect_string): New helper.
19595 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
19596 (print_dw_val): Add support for offsetted symbol references.
19597 (get_ultimate_context): Split out from is_cxx.
19598 (is_cxx): Use get_ultimate_context.
19599 (is_fortran): Add decl overload.
19600 (compute_comp_unit_symbol): Split out worker from
19601 compute_section_prefix.
19602 (compute_section_prefix): Call compute_comp_unit_symbol and
19603 set comdat_type_p here.
19604 (output_die): Skip DIE symbol output for the LTO added one.
19605 Handle DIE symbol references with offset.
19606 (output_comp_unit): Guard section name mangling properly.
19607 For LTO debug sections emit a symbol at the section beginning
19608 which we use to refer to its DIEs.
19609 (add_abstract_origin_attribute): For DIEs registered via
19610 dwarf2out_register_external_die directly refer to the early
19611 DIE rather than indirectly through the shadow one we created.
19612 Remove obsolete call to dwarf2out_abstract_function for
19613 non-function/block origins.
19614 (gen_array_type_die): When generating early LTO debug do
19615 not emit DW_AT_string_length.
19616 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
19617 late when in LTO. As suggested place a gcc_unreachable for
19618 the DECL_ABSTRACT_P case.
19619 (gen_subprogram_die): Avoid another specification DIE
19620 for early built declarations/definitions for the late LTO case.
19621 (gen_variable_die): Add type references for late duplicated VLA dies
19623 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
19624 we have the abstract instance already.
19625 (process_scope_var): Adjust decl DIE contexts in LTO which
19626 first puts them in limbo.
19627 (gen_decl_die): Do not generate type DIEs late apart from
19628 types for VLAs or for decls we do not yet have a DIE. Do not
19629 call dwarf2out_abstract_function late.
19630 (dwarf2out_early_global_decl): Make sure to create DIEs
19631 for abstract instances of a decl first.
19632 (dwarf2out_late_global_decl): Adjust comment.
19633 (output_macinfo_op): With multiple macro sections use
19634 macinfo_label_base to distinguish labels.
19635 (output_macinfo): Likewise. Update macinfo_label_base.
19636 Pass in the line info label.
19637 (note_variable_value_in_expr): When generating LTO resolve
19638 all variable values here by generating DIEs as needed.
19639 (init_sections_and_labels): Add early LTO debug flag parameter
19640 and generate different sections and names if set. Add generation
19641 counter for the labels so we can have multiple of them.
19642 (reset_dies): Helper to allow DIEs to be output multiple times.
19643 (dwarf2out_finish): When outputting DIEs to the fat part of an
19644 LTO object first reset DIEs.
19645 (dwarf2out_early_finish): Output early DIEs when generating LTO.
19646 (modified_type_die): Check for decl_ultimate_origin being self
19648 (gen_type_die_with_usage): Likewise.
19649 (gen_typedef_die): Allow decl_ultimate_origin being self.
19650 (set_decl_abstract_flags): Remove.
19651 (set_block_abstract_flags): Likewise.
19652 (dwarf2out_abstract_function): Treat the early generated DIEs
19653 as the abstract copy and only add DW_AT_inline and
19654 DW_AT_artificial here and call set_decl_origin_self.
19655 If the DIE has an abstract origin don't do anything.
19656 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
19657 if we have none yet (Go fails to build one, PR78628).
19658 (variably_modified_type_p): Prevent endless recursion for Ada
19659 cyclic pointer types.
19660 * lto-streamer-in.c: Include debug.h.
19661 (dref_queue): New global.
19662 (lto_read_tree_1): Stream in DIE references.
19663 (lto_input_tree): Register DIE references.
19664 (input_function): Stream DECL_DEBUG_ARGS.
19665 * lto-streamer-out.c: Include debug.h.
19666 (lto_write_tree_1): Output DIE references.
19667 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
19668 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
19669 (output_function): Stream DECL_DEBUG_ARGS.
19670 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
19671 Stream DECL_ABSTRACT_ORIGIN.
19672 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
19673 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
19674 DECL_CONTEXT for file-scope decls.
19675 * lto-streamer.h (struct dref_entry): Declare.
19676 (dref_queue): Likewise.
19677 * cfgexpand.c (pass_expand::execute): Do not call the
19678 outlining_inline_function hook here.
19679 * lto-wrapper.c (debug_obj): New global.
19680 (tool_cleanup): Unlink it if required.
19681 (debug_objcopy): New function.
19682 (run_gcc): Handle early debug sections in the IL files by
19683 extracting them to separate files, partially linkin them and
19684 feeding the result back as result to the linker.
19685 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
19686 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
19687 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
19688 sections into a separate segment.
19689 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
19691 (darwin_asm_dwarf_section): Likewise.
19692 (darwin_asm_output_dwarf_offset): Likewise.
19693 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
19695 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19696 Alan Hayward <alan.hayward@arm.com>
19697 David Sherwood <david.sherwood@arm.com>
19699 * read-md.h (md_reader::record_potential_iterator_use): Replace
19700 pointer argument with an rtx and an index.
19701 * read-rtl.c (iterator_group::apply_iterator): Likewise.
19702 (apply_mode_iterator): Likewise.
19703 (apply_code_iterator): Likewise.
19704 (apply_int_iterator): Likewise.
19705 (apply_subst_iterator): Likewise.
19706 (record_iterator_use): Likewise.
19707 (record_attribute_use): Likewise.
19708 (md_reader::record_potential_iterator_use): Likewise. Update calls
19709 to record_iterator_use and apply_iterator.
19710 (iterator_use): Replace ptr with x and index.
19711 (attribute_use): Likewise.
19712 (apply_attribute_uses): Update calls to apply_iterator.
19713 (apply_iterators): Likewise. Update initialization of iterator_use.
19714 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
19715 and record_potential_iterator_use.
19716 (rtx_reader::read_rtx_operand): Likewise.
19718 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19719 Alan Hayward <alan.hayward@arm.com>
19720 David Sherwood <david.sherwood@arm.com>
19722 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
19725 2017-08-21 Richard Biener <rguenther@suse.de>
19727 PR middle-end/81884
19728 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
19729 at struct end conservatively when comparing common bases.
19731 2017-08-21 Richard Biener <rguenther@suse.de>
19733 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
19734 (mem_ref_in_stmt): Remove.
19735 (determine_max_movement): Use ref index to get at the reference.
19736 (invariantness_dom_walker::before_dom_children): Deal with
19737 lim data already initialized.
19738 (gather_mem_refs_stmt): Initialize lim data and record ref index.
19740 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19742 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
19743 (TARGET_ISA_ROUND): Ditto.
19744 (TARGET_ROUND): Ditto.
19745 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
19746 * config/i386/i386.md: Ditto.
19747 * config/i386/sse.md: Ditto.
19748 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
19749 with OPTION_MASK_ISA_SSE4_1.
19751 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19754 * doc/extend.texi (x86 Built-in Functions): Correct the name of
19755 __builtin_ia32_lzcnt_u16.
19757 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
19760 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
19761 (rs6000_set_current_function): Rewrite function to use it.
19763 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
19766 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
19767 and TYPE_WARN_IF_NOT_ALIGN.
19768 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
19769 (handle_warn_if_not_align): New.
19770 (place_union_field): Call handle_warn_if_not_align.
19771 (place_field): Call handle_warn_if_not_align.
19772 Copy TYPE_WARN_IF_NOT_ALIGN.
19773 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
19774 (layout_type): Likewise.
19775 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
19777 (tree_decl_common): Add warn_if_not_align.
19778 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
19779 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
19780 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
19781 (DECL_WARN_IF_NOT_ALIGN): Likewise.
19782 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
19783 * doc/extend.texi: Document warn_if_not_aligned attribute.
19784 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
19786 2017-08-17 Martin Liska <mliska@suse.cz>
19789 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
19790 (get_data_dependence): Use it as pointer type.
19791 (distribute_loop): Likewise.
19793 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19795 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
19796 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
19797 (p8_vmrgow_<mode>_direct): New define_insn.
19798 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
19799 handle endianness for vmrgew and vmrgow permute patterns.
19801 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19803 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
19804 * config/rs6000/rs6000-cpus.def: Remove comment.
19805 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
19806 (POWERPC_MASKS): Likewise.
19807 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
19808 use of TARGET_VSX_TIMODE.
19809 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
19810 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
19811 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
19812 (rs6000_option_override_internal): Remove dead code.
19813 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
19814 (rs6000_legitimize_reload_address): Likewise.
19815 (rs6000_legitimate_address_p): Likewise.
19816 (rs6000_opt_masks): Delete "vsx-timode".
19817 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
19818 from function comment.
19819 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
19820 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
19821 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
19823 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
19824 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
19825 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
19826 (VSX_TI): Likewise.
19828 (define_peephole2): Likewise.
19830 2017-08-17 Martin Sebor <msebor@redhat.com>
19833 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
19834 past the end of an array.
19835 (test_pp_format): Add test cases.
19837 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
19839 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
19840 missing ECF_NOTHROW flags.
19842 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19845 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
19846 operands residing in integer registers.
19847 (*vsx_le_perm_load_<mode>): Likewise.
19848 (*vsx_le_perm_store_<mode>): Likewise.
19849 (define_peephole2): Add peepholes to optimize the above.
19851 2017-08-17 Marek Polacek <polacek@redhat.com>
19853 PR middle-end/81814
19854 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
19855 to mimic what shorten_compare did. Change the return type to bool.
19856 (fold_cond_expr_with_comparison): Update call to
19857 operand_equal_for_comparison_p.
19858 (fold_ternary_loc): Likewise.
19860 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
19862 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
19864 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
19866 2017-08-17 Richard Biener <rguenther@suse.de>
19868 * tree-ssa-structalias.c (solve_graph): When propagating
19869 to successors update the graphs succ edges and avoid duplicate work.
19871 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
19874 * config/i386/i386.c (ix86_option_override_internal): Save target
19875 specific options after ix86_stack_protector_guard_reg was changed.
19877 2017-08-17 Richard Biener <rguenther@suse.de>
19879 PR tree-optimization/81827
19880 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
19881 (new_var_info): Initialize it conservatively.
19882 (get_call_vi): Mark register vars.
19883 (new_scalar_tmp_constraint_exp): Likewise.
19884 (handle_rhs_call): Likewise.
19885 (handle_const_call): Likewise.
19886 (create_function_info_for): Likewise.
19887 (solve_constraints): Sort varinfos to separate register from
19888 non-register vars to pack points-to solution bitmaps during
19891 2017-08-17 Marek Polacek <polacek@redhat.com>
19893 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
19895 2017-08-17 Richard Biener <rguenther@suse.de>
19897 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
19898 to true when overflow is undefined and we saturated the result.
19900 2017-08-17 Alan Modra <amodra@gmail.com>
19903 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
19904 Don't use store multiple if only one reg needs saving.
19905 (interesting_frame_related_regno): New function.
19906 (rs6000_frame_related): Don't emit frame info for regs that
19908 (rs6000_emit_epilogue): Likewise.
19910 2017-08-16 Nathan Sidwell <nathan@acm.org>
19912 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
19913 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
19914 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
19915 * tree.c (free_lang_data_in_type): Use else-if chain. Always
19916 clear TYPE_LANG_1. Remove obsolete member-function stripping.
19917 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
19918 (verify_type): Adjust for TYPE_BINFO move.
19919 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
19920 process TYPE_BINFO directly.
19921 (hash_tree): Likewise.
19922 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
19924 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
19927 2017-08-16 David Malcolm <dmalcolm@redhat.com>
19929 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
19931 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
19934 * config/i386/i386.md (*anddi_1_btr): Change predicates of
19935 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
19936 Add ix86_binary_operator_ok to insn constraint.
19937 (*iordi_1_bts): Ditto.
19938 (*xordi_1_btc): Ditto.
19939 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
19940 Update corresponding peephole2 pattern.
19944 2017-08-16 Bin Cheng <bin.cheng@arm.com>
19946 PR tree-optimization/81832
19947 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
19948 copy loop header which has IFN_LOOP_DIST_ALIAS call.
19950 2017-08-16 Marek Polacek <polacek@redhat.com>
19953 * fold-const.c (fold_indirect_ref_1): Restore original behavior
19954 regarding size_zero_node.
19956 2017-08-16 Martin Liska <mliska@suse.cz>
19959 * config.gcc: Respect previously set extra_objs in case
19962 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
19964 PR tree-optimization/81835
19965 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
19966 the phi SSA_NAME. Check that the condition in a COND_EXPR does
19967 not depend on the phi.
19969 2017-08-16 Alan Modra <amodra@gmail.com>
19971 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
19974 2017-08-16 Alan Modra <amodra@gmail.com>
19976 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
19977 (save_reg_p): ..into this. Update all callers.
19978 (first_reg_to_save): Simplify.
19980 2017-08-16 Alan Modra <amodra@gmail.com>
19982 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
19985 2017-08-15 Joseph Myers <joseph@codesourcery.com>
19989 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
19990 constant count if that count is less than 32.
19992 2017-08-15 Nathan Sidwell <nathan@acm.org>
19994 * gcc.c (execute): Emit friendlier message if inferior is killed
19995 by an external cause.
19997 2017-08-15 Richard Biener <rguenther@suse.de>
19999 PR tree-optimization/81790
20000 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
20001 CONSTRUCTORs from simplifying and VN.
20003 2017-08-14 Martin Sebor <msebor@redhat.com>
20005 * builtin-attrs.def: Add comments.
20007 2017-08-14 Martin Sebor <msebor@redhat.com>
20010 * doc/extend.texi (attribute nonstring): Document new attribute.
20012 2017-08-14 Martin Sebor <msebor@redhat.com>
20015 * tree-diagnostic.c (default_tree_printer): Handle %G.
20016 * gimple-pretty-print.h (percent_G_format): Declare new function.
20017 * gimple-pretty-print.c (percent_G_format): Define.
20018 * tree-pretty-print.c (percent_K_format): Add argument.
20020 2017-08-14 Martin Sebor <msebor@redhat.com>
20022 PR translation/79998
20023 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
20024 Remove a stray space.
20026 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
20029 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
20030 (*iordi_1_bts): Ditto.
20031 (*xordi_1_btc): Ditto.
20033 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20036 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
20037 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20039 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
20040 quoted strings, and make more translator-friendly.
20041 (darwin_rs6000_override_options): Likewise.
20042 (rs6000_option_override_internal): Likewise.
20043 (rs6000_return_in_memory): Fix overlong line.
20044 (init_cmulative_args): Use quoted strings, and make more
20045 translator-friendly.
20046 (rs6000_pass_by_reference): Fix overlong line.
20047 (def_builtin): Use quoted strings.
20048 (altivec_expand_predicate_builtin): Use quoted strings, and make
20049 more translator-friendly.
20050 (htm_expand_builtin): Use quoted strings.
20051 (cpu_expand_builtin): Use quoted strings, and make more
20052 translator-friendly.
20053 (altivec_expand_builtin): Likewise.
20054 (paired_expand_predicate_builtin): Likewise.
20055 (rs6000_invalid_builtin): Likewise.
20056 (builtin_function_type): Use quoted strings.
20057 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
20058 more translator-friendly.
20059 (rs6000_trampoline_init): Likewise.
20060 (rs6000_handle_altivec_attribute): Likewise.
20061 (rs6000_inner_target_options): Use quoted strings.
20062 (rs6000_disable_incompatible_switches): Likewise.
20063 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
20064 strings, and make more translator-friendly.
20065 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
20067 2017-08-14 Bin Cheng <bin.cheng@arm.com>
20069 PR tree-optimization/81799
20070 * tree-loop-distribution.c (version_loop_by_alias_check): Force
20071 cond_expr to simple gimple operand.
20073 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
20075 PR middle-end/46932
20076 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
20078 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
20082 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
20083 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
20084 TARGET_GASISR_PROLOGUES.
20085 * config/avr/avr.c (avr_option_override): Same.
20086 (avr_pass_pre_proep::execute): Same.
20088 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
20091 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
20092 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
20094 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
20096 * config/i386/i386.md (*load_tp_<mode>): Redefine as
20097 define_insn_and_split. Split to a memory load from 0 in
20098 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
20099 using PTR mode iterator.
20100 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
20101 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
20102 (*add_tp_<mode>): Redefine as define_insn_and_split.
20103 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
20104 address space. Merge with *add_tp_x32 using PTR mode iterator.
20105 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
20106 Split to an add with a memory load from 0 in
20107 DEFAULT_TLS_SEG_REG address space.
20109 2017-08-12 Andrew Pinski <apinski@cavium.com>
20111 * config/aarch64/aarch64-option-extensions.def (rdma):
20112 Fix feature string to what Linux prints out in /proc/cpuinfo.
20114 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
20117 * dwarf2out.c (modified_type_die): For C typedef types that have
20118 an ultimate origin, process the ultimate origin instead of the
20120 (gen_typedef_die): Assert that input DECLs have no ultimate
20122 (gen_type_die_with_usage): For typedef variants that have an
20123 ultimate origin, just call gen_decl_die on the original DECL.
20124 (process_scope_var): Avoid creating DIEs for local typedefs and
20125 concrete static variables.
20127 2017-08-12 Alan Modra <amodra@gmail.com>
20131 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
20132 match gnu-user.h startfile.
20133 (ENDFILE_LINUX_SPEC): Similarly.
20135 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
20138 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
20140 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
20142 2017-08-11 Tamar Christina <tamar.christina@arm.com>
20143 * config/aarch64/aarch64.md (mov<mode>): Change.
20144 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20145 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
20146 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
20148 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
20150 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
20151 for storage order barriers.
20153 2017-08-11 Martin Liska <mliska@suse.cz>
20156 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
20157 variables of void type.
20159 2017-08-11 Martin Liska <mliska@suse.cz>
20161 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
20162 TARGET_SUPPORTS_ALIASES.
20163 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
20164 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
20165 (optimize_weakref): Likewise.
20166 * symtab.c (symtab_node::noninterposable_alias): Likewise.
20167 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
20168 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
20170 2017-08-11 Martin Liska <mliska@suse.cz>
20173 * config/i386/i386.c (make_resolver_func): Do complete
20174 refactoring of the function.
20176 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20179 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
20180 * config/i386/i386.c (ix86_stack_protect_guard): Use
20181 ix86_stack_protect_guard_symbol_str to generate varible declaration.
20182 * doc/invoke.texi (x86 Options): Document
20183 -mstack-protector-guard-symbol= option.
20185 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20187 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
20188 * config/i386/i386.c (ix86_split_stack_guard): New function.
20189 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
20190 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
20191 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
20192 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
20193 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
20194 (split_stack_space_check): Call ix86_split_stack_guard.
20196 2017-08-10 Martin Sebor <msebor@redhat.com>
20198 * print-tree.c (print_node): Print location using the established
20200 Replace spaces with colons.
20201 (debug_raw, debug): Ditto.
20203 2017-08-10 Martin Sebor <msebor@redhat.com>
20206 * pretty-print.c (pp_format): Correct the handling of %s precision.
20208 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
20211 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20213 (ix86_finalize_stack_frame_flags): This. Also clear
20214 frame_pointer_needed if -fno-omit-frame-pointer is used without
20216 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20217 with ix86_finalize_stack_frame_flags.
20218 (ix86_expand_epilogue): Likewise.
20219 (ix86_expand_split_stack_prologue): Likewise.
20220 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
20222 2017-08-10 Martin Liska <mliska@suse.cz>
20225 * c-attribs.c (handle_target_attribute):
20226 Report warning for an empty string argument of target attribute.
20228 2017-08-09 Jakub Jelinek <jakub@redhat.com>
20231 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
20233 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
20234 or DECL_NONLOCAL labels.
20235 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
20236 or DECL_NONLOCAL labels here.
20238 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
20240 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
20241 to indicate when early gimple folding has been disabled.
20242 (rs6000_gimple_fold_builtin): Add debug content.
20243 (rs6000_invalid_builtin): Fix whitespace.
20244 (rs6000_expand_builtin): Fix whitespace.
20245 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
20247 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
20250 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
20251 SAVE_MULTIPLE if not all the registers that saves, should be saved.
20253 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
20255 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
20256 (qdf24xx): Likewise.
20257 * config/aarch64/aarch64.md: Include falkor.md.
20258 * config/aarch64/falkor.md: New.
20260 2017-08-09 Marek Polacek <polacek@redhat.com>
20263 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
20264 * diagnostic.c (emit_diagnostic): Add a comment.
20265 (emit_diagnostic_valist): New function.
20267 2017-08-09 Marek Polacek <polacek@redhat.com>
20270 * input.c (make_location): New overload.
20271 * input.h (make_location): Declare.
20273 2017-08-08 Alan Modra <amodra@gmail.com>
20274 H.J. Lu <hongjiu.lu@intel.com>
20277 * gcc.c (NO_PIE_SPEC): Delete.
20278 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
20280 (LINK_PIE_SPEC): ..to here.
20281 (LINK_COMMAND_SPEC): Support -no-pie.
20282 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
20283 chain of crtbegin*.o selection, update for PIE_SPEC changes and
20285 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
20286 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
20287 (ENDFILE_CRTEND_SPEC): Similarly.
20289 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
20292 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
20293 (mstack-protector-guard-offset=): Ditto.
20294 * config/i386/i386.c (ix86_option_override): Handle
20295 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
20297 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
20298 ix86_stack_protect_guard_offset variables.
20299 (TARGET_STACK_PROTECT_GUARD): Always define.
20300 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
20301 and -mstack-protector-guard-offset= options.
20303 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20305 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
20306 boundary case for the last candidate.
20308 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20310 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
20311 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
20313 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20315 PR middle-end/19706
20316 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
20317 * config/aarch64/aarch64-builtins.c
20318 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
20319 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
20320 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
20322 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20323 Andrew Pinski <pinskia@gmail.com>
20325 PR middle-end/19706
20326 * internal-fn.def (XORSIGN): New.
20327 * optabs.def (xorsign_optab): New.
20328 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
20329 (convert_expand_mult_copysign): New.
20330 (pass_optimize_widening_mul::execute): Call
20331 convert_expand_mult_copysign.
20333 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20335 PR tree-optimization/81354
20336 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
20337 Insert on edges rather than explicitly creating landing pads.
20338 (analyze_candidates_and_replace): Commit edge inserts.
20340 2017-08-08 Richard Biener <rguenther@suse.de>
20342 PR middle-end/81719
20343 * tree-ssa-loop-niter.c: Include tree-dfa.h.
20344 (expand_simple_operations): Also look through ADDR_EXPRs with
20345 MEM_REF bases treating them as POINTER_PLUS_EXPR.
20347 2017-08-08 Richard Biener <rguenther@suse.de>
20349 PR tree-optimization/81723
20350 * tree-vect-slp.c (struct bst_traits): New hash traits.
20351 (bst_fail): New global.
20352 (vect_build_slp_tree_2): New worker, split out from ...
20353 (vect_build_slp_tree): ... this now wrapping it with using
20354 bst_fail set to cache SLP tree build fails. Properly handle
20356 (vect_analyze_slp_instance): Allocate and free bst_fail.
20358 2017-08-08 Martin Liska <mliska@suse.cz>
20361 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
20362 LABEL_DECLs that can be from a different function.
20364 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20366 PR tree-optimization/81744
20367 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
20368 loop's number of iterations.
20370 2017-08-08 Martin Liska <mliska@suse.cz>
20372 * asan.c: Include header files.
20373 * attribs.c (build_decl_attribute_variant): New function moved
20375 (build_type_attribute_qual_variant): Likewise.
20376 (cmp_attrib_identifiers): Likewise.
20377 (simple_cst_list_equal): Likewise.
20378 (omp_declare_simd_clauses_equal): Likewise.
20379 (attribute_value_equal): Likewise.
20380 (comp_type_attributes): Likewise.
20381 (build_type_attribute_variant): Likewise.
20382 (lookup_ident_attribute): Likewise.
20383 (remove_attribute): Likewise.
20384 (merge_attributes): Likewise.
20385 (merge_type_attributes): Likewise.
20386 (merge_decl_attributes): Likewise.
20387 (merge_dllimport_decl_attributes): Likewise.
20388 (handle_dll_attribute): Likewise.
20389 (attribute_list_equal): Likewise.
20390 (attribute_list_contained): Likewise.
20391 * attribs.h (lookup_attribute): New function moved from tree.[ch].
20392 (lookup_attribute_by_prefix): Likewise.
20393 * bb-reorder.c: Include header files.
20394 * builtins.c: Likewise.
20395 * calls.c: Likewise.
20396 * cfgexpand.c: Likewise.
20397 * cgraph.c: Likewise.
20398 * cgraphunit.c: Likewise.
20399 * convert.c: Likewise.
20400 * dwarf2out.c: Likewise.
20401 * final.c: Likewise.
20402 * fold-const.c: Likewise.
20403 * function.c: Likewise.
20404 * gimple-expr.c: Likewise.
20405 * gimple-fold.c: Likewise.
20406 * gimple-pretty-print.c: Likewise.
20407 * gimple.c: Likewise.
20408 * gimplify.c: Likewise.
20409 * hsa-common.c: Likewise.
20410 * hsa-gen.c: Likewise.
20411 * internal-fn.c: Likewise.
20412 * ipa-chkp.c: Likewise.
20413 * ipa-cp.c: Likewise.
20414 * ipa-devirt.c: Likewise.
20415 * ipa-fnsummary.c: Likewise.
20416 * ipa-inline.c: Likewise.
20417 * ipa-visibility.c: Likewise.
20419 * lto-cgraph.c: Likewise.
20420 * omp-expand.c: Likewise.
20421 * omp-general.c: Likewise.
20422 * omp-low.c: Likewise.
20423 * omp-offload.c: Likewise.
20424 * omp-simd-clone.c: Likewise.
20425 * opts-global.c: Likewise.
20426 * passes.c: Likewise.
20427 * predict.c: Likewise.
20428 * sancov.c: Likewise.
20429 * sanopt.c: Likewise.
20430 * symtab.c: Likewise.
20431 * toplev.c: Likewise.
20432 * trans-mem.c: Likewise.
20433 * tree-chkp.c: Likewise.
20434 * tree-eh.c: Likewise.
20435 * tree-into-ssa.c: Likewise.
20436 * tree-object-size.c: Likewise.
20437 * tree-parloops.c: Likewise.
20438 * tree-profile.c: Likewise.
20439 * tree-ssa-ccp.c: Likewise.
20440 * tree-ssa-live.c: Likewise.
20441 * tree-ssa-loop.c: Likewise.
20442 * tree-ssa-sccvn.c: Likewise.
20443 * tree-ssa-structalias.c: Likewise.
20444 * tree-ssa.c: Likewise.
20445 * tree-streamer-in.c: Likewise.
20446 * tree-vectorizer.c: Likewise.
20447 * tree-vrp.c: Likewise.
20448 * tsan.c: Likewise.
20449 * ubsan.c: Likewise.
20450 * varasm.c: Likewise.
20451 * varpool.c: Likewise.
20452 * tree.c: Remove functions moved to attribs.[ch].
20453 * tree.h: Likewise.
20454 * config/aarch64/aarch64.c: Add attrs.h header file.
20455 * config/alpha/alpha.c: Likewise.
20456 * config/arc/arc.c: Likewise.
20457 * config/arm/arm.c: Likewise.
20458 * config/avr/avr.c: Likewise.
20459 * config/bfin/bfin.c: Likewise.
20460 * config/c6x/c6x.c: Likewise.
20461 * config/cr16/cr16.c: Likewise.
20462 * config/cris/cris.c: Likewise.
20463 * config/darwin.c: Likewise.
20464 * config/epiphany/epiphany.c: Likewise.
20465 * config/fr30/fr30.c: Likewise.
20466 * config/frv/frv.c: Likewise.
20467 * config/ft32/ft32.c: Likewise.
20468 * config/h8300/h8300.c: Likewise.
20469 * config/i386/winnt.c: Likewise.
20470 * config/ia64/ia64.c: Likewise.
20471 * config/iq2000/iq2000.c: Likewise.
20472 * config/lm32/lm32.c: Likewise.
20473 * config/m32c/m32c.c: Likewise.
20474 * config/m32r/m32r.c: Likewise.
20475 * config/m68k/m68k.c: Likewise.
20476 * config/mcore/mcore.c: Likewise.
20477 * config/microblaze/microblaze.c: Likewise.
20478 * config/mips/mips.c: Likewise.
20479 * config/mmix/mmix.c: Likewise.
20480 * config/mn10300/mn10300.c: Likewise.
20481 * config/moxie/moxie.c: Likewise.
20482 * config/msp430/msp430.c: Likewise.
20483 * config/nds32/nds32-isr.c: Likewise.
20484 * config/nds32/nds32.c: Likewise.
20485 * config/nios2/nios2.c: Likewise.
20486 * config/nvptx/nvptx.c: Likewise.
20487 * config/pa/pa.c: Likewise.
20488 * config/pdp11/pdp11.c: Likewise.
20489 * config/powerpcspe/powerpcspe.c: Likewise.
20490 * config/riscv/riscv.c: Likewise.
20491 * config/rl78/rl78.c: Likewise.
20492 * config/rx/rx.c: Likewise.
20493 * config/s390/s390.c: Likewise.
20494 * config/sh/sh.c: Likewise.
20495 * config/sol2.c: Likewise.
20496 * config/sparc/sparc.c: Likewise.
20497 * config/spu/spu.c: Likewise.
20498 * config/stormy16/stormy16.c: Likewise.
20499 * config/tilegx/tilegx.c: Likewise.
20500 * config/tilepro/tilepro.c: Likewise.
20501 * config/v850/v850.c: Likewise.
20502 * config/vax/vax.c: Likewise.
20503 * config/visium/visium.c: Likewise.
20504 * config/xtensa/xtensa.c: Likewise.
20506 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20509 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
20510 constraints since the -mupper-regs-* switches have been
20512 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
20513 into a vector from a double word element that was extracted from
20514 another vector, and eliminate extra XXPERMDI instructions.
20515 (vsx_concat_<mode>_2): Likewise.
20516 (vsx_concat_<mode>_3): Likewise.
20517 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
20518 concat to allow optimizing inserts from previous extracts.
20520 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
20522 * config/i386/i386.c (ix86_stack_protect_guard): Generate
20523 memory reference to a SSP offset in TLS address space.
20524 (ix86_print_operand) <case '@'>: Remove.
20525 (ix86_print_operand_punct_valid_p): Remove '@' code.
20526 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
20527 UNSPEC_SP_TLS_TEST.
20528 (stack_tls_protect_set_<mode>): Remove.
20529 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
20530 (stack_tls_protect_test_<mode>): Remove.
20531 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
20533 2017-08-07 Olivier Hainque <hainque@adacore.com>
20536 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
20538 2017-08-07 Douglas Rupp <rupp@adacore.com>
20540 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
20541 variable was referenced as multidir in command.
20543 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20546 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
20549 2017-08-07 Martin Liska <mliska@suse.cz>
20551 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
20552 * config/rl78/rl78.c: Add include of attribs.h.
20553 * config/sh/sh.c: Likewise.
20554 * config/v850/v850.c: Likewise.
20556 2017-08-07 Tom de Vries <tom@codesourcery.com>
20558 PR middle-end/78266
20559 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
20561 2017-08-07 Martin Liska <mliska@suse.cz>
20563 * config/mips/mips.c: Include attribs.h.
20565 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
20568 * doc/invoke.texi: Document change in behvaior for -Ofast for
20571 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
20573 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
20575 (aarch64_pop_regs): Likewise.
20576 (aarch64_gen_load_pair): Likewise.
20577 (aarch64_save_callee_saves): Likewise.
20578 (aarch64_restore_callee_saves): Likewise.
20580 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20582 * config/i386/i386.c: Revert the last change.
20584 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20587 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20589 (ix86_finalize_stack_frame_flags): This. Also clear
20590 frame_pointer_needed if -fno-omit-frame-pointer is used without
20592 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20593 with ix86_finalize_stack_frame_flags.
20594 (ix86_expand_epilogue): Likewise.
20595 (ix86_expand_split_stack_prologue): Likewise.
20597 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20600 * config/i386/i386.c (get_builtin_code_for_version): Set priority
20601 to P_AES for Westmere.
20603 2017-08-07 Jonathan Yong <10walls@gmail.com>
20605 * config/i386/mingw.opt (fset-stack-executable): Removed.
20606 * config/i386/cygming.opt (fset-stack-executable): Moved
20608 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
20610 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
20612 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
20614 2017-08-07 Marek Polacek <polacek@redhat.com>
20616 PR middle-end/81737
20617 * fold-const.c (fold_indirect_ref_1): Check type_domain.
20619 2017-08-07 Martin Liska <mliska@suse.cz>
20621 * attribs.h (canonicalize_attr_name): New function.
20622 (cmp_attribs): Move from c-format.c and adjusted.
20623 (is_attribute_p): Moved from tree.h.
20624 * tree-inline.c: Add new includes.
20625 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
20626 (private_is_attribute_p): Remove.
20627 (private_lookup_attribute): Likewise.
20628 (private_lookup_attribute_by_prefix): Simplify.
20629 (remove_attribute): Use is_attribute_p.
20630 * tree.h: Remove removed declarations.
20632 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20634 PR middle-end/81698
20635 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
20636 instead of computing it in the function. Formatting fix.
20637 (expand_case): Don't rely on default_edge being the first edge,
20638 clear it if removing it, pass default_edge to
20639 emit_case_dispatch_table.
20640 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
20643 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
20645 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
20646 insn in the function, emit NOP after the insn.
20648 2017-08-06 Tom de Vries <tom@codesourcery.com>
20650 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
20653 2017-08-06 Tom de Vries <tom@codesourcery.com>
20655 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
20658 2017-08-04 Yury Gribov <tetra2005@gmail.com>
20660 PR tree-optimization/57371
20661 * match.pd: New pattern.
20663 2017-08-04 Marek Polacek <polacek@redhat.com>
20665 PR middle-end/81695
20666 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
20667 perform the computation in offset_int.
20669 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20671 PR tree-optimization/81136
20672 * tree-vectorizer.h: Include tree-hash-traits.h.
20673 (vec_base_alignments): New typedef.
20674 (vec_info): Add a base_alignments field.
20675 (vect_record_base_alignments): Declare.
20676 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
20678 (DR_IS_CONDITIONAL_IN_STMT): New macro.
20679 (create_data_ref): Add an is_conditional_in_stmt argument.
20680 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
20681 the is_conditional_in_stmt field.
20682 (data_ref_loc): Add an is_conditional_in_stmt field.
20683 (get_references_in_stmt): Set the is_conditional_in_stmt field.
20684 (find_data_references_in_stmt): Update call to create_data_ref.
20685 (graphite_find_data_references_in_stmt): Likewise.
20686 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
20687 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
20688 (vect_record_base_alignment): New function.
20689 (vect_record_base_alignments): Likewise.
20690 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
20691 for nested statements even if we fail to compute a misalignment.
20692 Use pooled base alignments for unconditional references.
20693 (vect_find_same_alignment_drs): Compare base addresses instead
20695 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
20696 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
20698 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20700 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
20701 Add an explicit name for the enum. Use auto_vec for slp_instances
20702 and grouped_stores.
20703 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
20705 (_bb_vec_info): Add a constructor and destructor.
20706 (vinfo_for_stmt): Return NULL for uids of -1 as well.
20707 (destroy_loop_vec_info): Delete.
20708 (vect_destroy_datarefs): Likewise.
20709 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
20710 (vec_info::vec_info): New function.
20711 (vec_info::~vec_info): Likewise.
20712 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
20713 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
20714 destroy_loop_vec_info.
20715 * tree-vect-loop.c (new_loop_vec_info): Replace with...
20716 (_loop_vec_info::_loop_vec_info): ...this.
20717 (destroy_loop_vec_info): Replace with...
20718 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
20719 the stmt_vec_infos. Leave handling of vec_info information to its
20720 destructor. Remove explicit vector releases.
20721 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
20722 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
20723 * tree-vect-slp.c (new_bb_vec_info): Replace with...
20724 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
20725 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
20726 (destroy_bb_vec_info): Replace with...
20727 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
20728 information to its destructor.
20729 (vect_slp_analyze_bb_1): Use new and delete instead of
20730 new_bb_vec_info and destroy_bb_vec_info.
20731 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
20734 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20736 * tree-data-ref.h (subscript): Add access_fn field.
20737 (data_dependence_relation): Add could_be_independent_p.
20738 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
20739 (same_access_functions): Move to tree-data-ref.c.
20740 * tree-data-ref.c (ref_contains_union_access_p): New function.
20741 (access_fn_component_p): Likewise.
20742 (access_fn_components_comparable_p): Likewise.
20743 (dr_analyze_indices): Add a reference to access_fn_component_p.
20744 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
20746 (constant_access_functions): Likewise.
20747 (add_other_self_distances): Likewise.
20748 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
20749 (initialize_data_dependence_relation): Use XCNEW and remove
20750 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
20751 of access functions that have the same type. Allow the
20752 subsequence to end with different bases in some circumstances.
20753 Record the chosen access functions in SUB_ACCESS_FN.
20754 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
20755 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
20756 (subscript_dependence_tester_1): Likewise dra and drb.
20757 (build_classic_dist_vector): Update calls accordingly.
20758 (subscript_dependence_tester): Likewise.
20759 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
20760 DDR_COULD_BE_INDEPENDENT_P.
20761 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
20762 comp_alias_ddrs instead of may_alias_ddrs.
20763 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
20765 (vect_analyze_data_ref_dependence): Use it if
20766 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
20767 distance vectors if that fails.
20768 (dependence_distance_ge_vf): New function.
20769 (vect_prune_runtime_alias_test_list): Use it. Don't clear
20770 LOOP_VINFO_MAY_ALIAS_DDRS.
20772 2017-08-04 Richard Biener <rguenther@suse.de>
20774 PR middle-end/81705
20775 * fold-const.c (fold_binary_loc): Properly restrict
20776 minus_var0 && minus_var1 case when associating undefined overflow
20779 2017-08-04 Tom de Vries <tom@codesourcery.com>
20781 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
20783 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20785 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20786 Don't start diagnostic messages with a capital letter.
20787 * config/rs6000/rs6000.c (rs6000_option_override_internal):
20789 (rs6000_invalid_builtin): Likewise.
20790 (rs6000_trampoline_init): Likewise.
20792 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20795 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
20796 after setting changeable df flags.
20798 2017-08-03 Richard Biener <rguenther@suse.de>
20800 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
20801 up if the use is in USE - X.
20803 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
20805 * toplev.c (dumpfile.h): New include.
20806 (internal_error_reentered): New static function. Use it...
20807 (internal_error_function): ...here to handle reentered internal_error.
20809 2017-08-03 Richard Biener <rguenther@suse.de>
20811 PR middle-end/81148
20812 * fold-const.c (split_tree): Add minus_var and minus_con
20813 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
20814 here but always use minus_*.
20815 (associate_trees): Assert we never associate with MINUS_EXPR
20816 and NULL first operand. Do not recurse for PLUS_EXPR operands
20817 when associating as MINUS_EXPR either.
20818 (fold_binary_loc): Track minus_var and minus_con.
20820 2017-08-03 Tom de Vries <tom@codesourcery.com>
20823 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
20824 ACCEL_COMPILER, apply finish_options on
20825 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
20827 2017-08-03 Tom de Vries <tom@codesourcery.com>
20830 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
20831 function_entry_patch_area_size > 0.
20833 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20836 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
20837 instead of 10??LU, perform unit multiplication in wide_int,
20838 don't change alloc_object_size_limit if the limit is larger
20841 PR tree-optimization/81655
20842 PR tree-optimization/81588
20843 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
20844 the case when ranges[i].low and high are 1 for unsigned type with
20847 PR middle-end/81052
20848 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
20849 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
20851 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20853 * tree-vrp.h: Add include guard.
20855 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
20858 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
20859 (ud2): New insn pattern.
20860 * config/i386/i386.c (ix86_expand_epilogue):
20861 For naked functions, generate ud2 instead of trap insn.
20863 2017-08-02 Marek Polacek <polacek@redhat.com>
20866 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
20868 2017-08-02 Tom de Vries <tom@codesourcery.com>
20869 Cesar Philippidis <cesar@codesourcery.com>
20871 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
20872 Add missing edge probabilities.
20874 2017-08-02 Tamar Christina <tamar.christina@arm.com>
20876 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
20877 Correct endianness.
20879 2017-08-02 Jakub Jelinek <jakub@redhat.com>
20881 PR middle-end/79499
20882 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
20883 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
20884 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
20885 prologue_seq sequences - if any.
20887 2017-08-02 Richard Biener <rguenther@suse.de>
20889 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
20890 via vectors if supported, integer extracts via punning if supported
20891 or otherwise vector extracts.
20893 2017-08-02 Richard Biener <rguenther@suse.de>
20895 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
20897 (bitmap_insert_into_set): ... this.
20899 2017-08-02 Richard Biener <rguenther@suse.de>
20901 PR tree-optimization/81633
20903 2015-08-17 Alan Hayward <alan.hayward@arm.com>
20905 PR tree-optimization/71752
20906 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
20908 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
20910 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
20911 (machine_function::call_ms2sysv_pad_out): Remove field.
20912 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
20913 (ix86_compute_frame_layout): Likewise.
20915 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
20918 * config/i386/i386.c (ix86_set_func_type): Disallow naked
20919 attribute with interrupt attribute.
20921 2017-08-01 Andrew Pinski <apinski@cavium.com>
20923 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
20924 BIT_INSERT_EXPR's operand 1
20925 to see if the types precision matches.
20927 2017-08-01 Martin Liska <mliska@suse.cz>
20929 PR middle-end/70140
20930 * builtins.c (expand_builtin_memcpy_args): Remove.
20931 (expand_builtin_memcpy): Call newly added function
20932 expand_builtin_memory_copy_args.
20933 (expand_builtin_memcpy_with_bounds): Likewise.
20934 (expand_builtin_mempcpy): Remove last argument.
20935 (expand_builtin_mempcpy_with_bounds): Likewise.
20936 (expand_builtin_memory_copy_args): New function created from
20937 expand_builtin_mempcpy_args with small modifications.
20938 (expand_builtin_mempcpy_args): Remove.
20939 (expand_builtin_stpcpy): Remove unused argument.
20940 (expand_builtin): Likewise.
20941 (expand_builtin_with_bounds): Likewise.
20943 2017-08-01 Martin Liska <mliska@suse.cz>
20946 Make mempcpy more optimal (PR middle-end/70140).
20948 2017-08-01 Jakub Jelinek <jakub@redhat.com>
20951 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
20952 __builtin_vec_cmpne verify both arguments are compatible vectors
20953 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
20954 verify arg1_type is a pointer or array type. For __builtin_vec_st,
20955 move computation of aligned to after checking the argument types.
20959 * config/rs6000/vsx.md (vextract_fp_from_shorth,
20960 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
20963 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
20964 Doug Rupp <rupp@adacore.com>
20965 Olivier Hainque <hainque@adacore.com>
20967 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
20968 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
20970 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
20971 for TARGET_OS_CPP_BUILTIN.
20972 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
20973 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
20975 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
20976 passing required abi options to the assembler for EABI configurations.
20977 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
20978 of .text.hot and .text.unlikely sections for kernel modules when
20979 using ARM style exceptions.
20980 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
20981 options. Add EXTRA_CC1_SPEC.
20982 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
20984 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
20986 (ARM_TARGET2_DWARF_FORMAT): Define.
20987 * config/arm/t-vxworks: Adjust multilib control to removal of the
20988 Diab command line options.
20990 2017-08-01 Martin Liska <mliska@suse.cz>
20992 PR gcov-profile/81561
20993 * gcov.c (unblock): Make unblocking safe as we need to preserve
20994 index correspondence of blocks and block_lists.
20996 2017-08-01 Richard Biener <rguenther@suse.de>
20998 PR tree-optimization/81181
20999 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
21000 (compute_antic): ... end of iteration here.
21002 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
21004 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
21005 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
21006 (ftree-slp-vectorize): Likewise.
21007 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
21008 can no longer be set independent of flag_tree_loop_vectorize.
21009 * omp-general.c (emp_max_vf): Likewise.
21010 * opts.c (enable_fdo_optimizations): Remove references to
21011 flag_tree_vectorize, these are now implicit.
21012 (common_handle_option): Remove handling for OPT_ftree_vectorize,
21013 and leave it for the options machinery.
21015 2017-08-01 Martin Liska <mliska@suse.cz>
21017 PR middle-end/70140
21018 * builtins.c (expand_builtin_memcpy_args): Remove.
21019 (expand_builtin_memcpy): Call newly added function
21020 expand_builtin_memory_copy_args.
21021 (expand_builtin_memcpy_with_bounds): Likewise.
21022 (expand_builtin_mempcpy): Remove last argument.
21023 (expand_builtin_mempcpy_with_bounds): Likewise.
21024 (expand_builtin_memory_copy_args): New function created from
21025 expand_builtin_mempcpy_args with small modifications.
21026 (expand_builtin_mempcpy_args): Remove.
21027 (expand_builtin_stpcpy): Remove unused argument.
21028 (expand_builtin): Likewise.
21029 (expand_builtin_with_bounds): Likewise.
21031 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21034 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
21035 print "ds:" only for immediates in generic address space.
21037 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21040 * config/i386/i386.c (ix86_funciton_naked): New prototype.
21041 (ix86_function_ok_for_sibcall): Return false for naked functions.
21043 2017-08-01 Richard Biener <rguenther@suse.de>
21045 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
21046 (compute_antic): Seed worklist with exit block predecessors.
21047 * cfganal.c (dfs_find_deadend): For a cycle return the source
21048 of the edge closing it.
21050 2017-08-01 Tamar Christina <tamar.christina@arm.com>
21052 * config/aarch64/aarch64.c
21053 (aarch64_can_const_movi_rtx_p): Move 0 check.
21055 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21057 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
21058 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
21060 * match.pd: Ditto in address comparison pattern.
21062 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21064 PR tree-optimization/81627
21065 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
21066 closed ssa form for store-store chain.
21068 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21070 PR tree-optimization/81620
21071 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
21072 for store-store chain.
21074 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21076 PR tree-optimization/81588
21077 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
21078 ranges[i].in_p, invert comparison code ccode. For >/>=,
21079 swap rhs1 and rhs2 and comparison code unconditionally,
21080 for </<= don't do that. Don't swap rhs1/rhs2 again if
21081 ranges[i].in_p, instead invert comparison code ccode if
21082 opcode or oe->rank is BIT_IOR_EXPR.
21085 * optabs.def (vec_extract_optab, vec_init_optab): Change from
21086 a direct optab to conversion optab.
21087 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
21088 with GET_MODE_INNER as last argument instead of optab_handler.
21089 * expmed.c (extract_bit_field_1): Likewise. Use vector from
21090 vector extraction if possible and optab is available.
21091 * expr.c (store_constructor): Use convert_optab_handler instead
21092 of optab_handler. Use vector initialization from smaller
21093 vectors if possible and optab is available.
21094 * tree-vect-stmts.c (vectorizable_load): Likewise.
21095 * doc/md.texi (vec_extract, vec_init): Document that the optabs
21096 now have two modes.
21097 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
21098 of vec_init from half-sized vectors with the same element mode.
21099 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
21100 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
21101 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
21102 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
21103 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
21104 after mode in gen_vec_extract* calls.
21105 (vec_extract<mode>): Renamed to ...
21106 (vec_extract<mode><ssescalarmodelower>): ... this.
21107 (vec_extract<mode><ssehalfvecmodelower>): New expander.
21108 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
21109 element mode after mode in gen_vec_init* calls.
21110 (VEC_INIT_HALF_MODE): New mode iterator.
21111 (vec_init<mode>): Renamed to ...
21112 (vec_init<mode><ssescalarmodelower>): ... this.
21113 (vec_init<mode><ssehalfvecmodelower>): New expander.
21114 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
21115 (vec_extractv2sfsf): ... this.
21116 (vec_initv2sf): Renamed to ...
21117 (vec_initv2sfsf): ... this.
21118 (vec_extractv2si): Renamed to ...
21119 (vec_extractv2sisi): ... this.
21120 (vec_initv2si): Renamed to ...
21121 (vec_initv2sisi): ... this.
21122 (vec_extractv4hi): Renamed to ...
21123 (vec_extractv4hihi): ... this.
21124 (vec_initv4hi): Renamed to ...
21125 (vec_initv4hihi): ... this.
21126 (vec_extractv8qi): Renamed to ...
21127 (vec_extractv8qiqi): ... this.
21128 (vec_initv8qi): Renamed to ...
21129 (vec_initv8qiqi): ... this.
21130 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
21131 (vec_init<mode>): Renamed to ...
21132 (vec_init<mode><VEC_base_l>): ... this.
21133 (vec_extract<mode>): Renamed to ...
21134 (vec_extract<mode><VEC_base_l>): ... this.
21135 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
21136 (vec_initv2sfsf): ... this.
21137 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
21138 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21139 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
21140 element mode after mode in gen_vec_init* calls.
21141 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
21142 (vec_init<mode><Vel>): ... this.
21143 (vec_extract<mode>): Renamed to ...
21144 (vec_extract<mode><Vel>): ... this.
21145 * config/aarch64/iterators.md (Vel): New mode attribute.
21146 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
21147 Add element mode after mode in gen_vec_extract* calls.
21148 * config/s390/vector.md (non_vec_l): New mode attribute.
21149 (vec_extract<mode>): Renamed to ...
21150 (vec_extract<mode><non_vec_l>): ... this.
21151 (vec_init<mode>): Renamed to ...
21152 (vec_init<mode><non_vec_l>): ... this.
21153 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
21154 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
21156 * config/arm/iterators.md (V_elem_l): New mode attribute.
21157 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
21158 (vec_extract<mode><V_elem_l>): ... this.
21159 (vec_extractv2di): Renamed to ...
21160 (vec_extractv2didi): ... this.
21161 (vec_init<mode>): Renamed to ...
21162 (vec_init<mode><V_elem_l>): ... this.
21163 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
21164 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
21165 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
21166 Add element mode after gen_vec_extract* calls.
21167 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
21168 (vec_init<mode><unitmode>): ... this.
21169 (vec_extract<mode>): Renamed to ...
21170 (vec_extract<mode><unitmode>): ... this.
21171 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
21172 (vec_init<mode><unitmode>): ... this.
21173 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
21174 (vec_initv2sfsf): ... this.
21175 (vec_extractv2sf): Renamed to ...
21176 (vec_extractv2sfsf): ... this.
21177 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
21178 Add element mode after gen_vec_extract* calls.
21179 * config/mips/mips.md (unitmode): New mode iterator.
21180 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
21181 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
21182 * config/spu/spu.md (inner_l): New mode attribute.
21183 (vec_init<mode>): Renamed to ...
21184 (vec_init<mode><inner_l>): ... this.
21185 (vec_extract<mode>): Renamed to ...
21186 (vec_extract<mode><inner_l>): ... this.
21187 * config/sparc/sparc.md (veltmode): New mode iterator.
21188 (vec_init<VMALL:mode>): Renamed to ...
21189 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
21190 * config/ia64/vect.md (vec_initv2si): Renamed to ...
21191 (vec_initv2sisi): ... this.
21192 (vec_initv2sf): Renamed to ...
21193 (vec_initv2sfsf): ... this.
21194 (vec_extractv2sf): Renamed to ...
21195 (vec_extractv2sfsf): ... this.
21196 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
21197 (vec_init<mode>): Renamed to ...
21198 (vec_init<mode><VEC_base_l>): ... this.
21199 (vec_extract<mode>): Renamed to ...
21200 (vec_extract<mode><VEC_base_l>): ... this.
21201 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
21202 (vec_initv2sfsf): ... this.
21203 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
21204 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21205 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
21206 gen_vec_init* calls.
21208 2017-08-01 Richard Biener <rguenther@suse.de>
21210 PR tree-optimization/81297
21211 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
21212 TREE_OVERFLOW from INTEGER_CSTs.
21214 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
21216 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
21218 2017-07-31 Carl Love <cel@us.ibm.com>
21220 * config/rs6000/rs6000-c: Add support for built-in functions
21221 vector signed char vec_xl_be (signed long long, signed char *);
21222 vector unsigned char vec_xl_be (signed long long, unsigned char *);
21223 vector signed int vec_xl_be (signed long long, signed int *);
21224 vector unsigned int vec_xl_be (signed long long, unsigned int *);
21225 vector signed long long vec_xl_be (signed long long, signed long long *);
21226 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
21227 vector signed short vec_xl_be (signed long long, signed short *);
21228 vector unsigned short vec_xl_be (signed long long, unsigned short *);
21229 vector double vec_xl_be (signed long long, double *);
21230 vector float vec_xl_be (signed long long, float *);
21231 * config/rs6000/altivec.h (vec_xl_be): Add #define.
21232 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
21233 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
21235 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
21236 (altivec_expand_builtin): Add switch statement to call
21237 altivec_expand_xl_be for each builtin.
21238 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
21239 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
21240 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
21241 __builtin_vsx_le_be_v16qi.
21242 * doc/extend.texi: Update the built-in documentation file for the
21243 new built-in functions.
21245 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21248 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
21250 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21252 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21254 * config.gcc: Add z14.
21255 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
21256 CPU model numbers for z13s and z14.
21257 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
21259 * config/s390/s390-opts.h (enum processor_type): Rename
21260 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21261 * config/s390/s390.c (processor_table): Add field for CPU name to
21262 be passed to Binutils.
21263 (s390_asm_output_machine_for_arch): Use the new field in
21264 processor_table for Binutils.
21265 (s390_expand_builtin): Replace arch12 with z14.
21266 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21267 (s390_get_sched_attrmask): Likewise.
21268 (s390_get_unit_mask): Likewise.
21269 * config/s390/s390.opt: Add z14 to processor_type enum.
21271 2017-07-31 Martin Jambor <mjambor@suse.cz>
21274 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
21275 regardless of optimization level.
21277 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
21278 Martin Liska <mliska@suse.cz>
21280 * predict.def: Remove old comment and adjust probability.
21281 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
21282 PREDICT statements.
21284 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21287 * config/i386/i386.c (ix86_function_naked): New function.
21288 (ix86_can_use_return_insn_p): Return false for naked functions.
21289 (ix86_expand_prologue): Skip prologue for naked functions.
21290 (ix86_expand_epilogue): Skip epilogue for naked functions
21291 and emit trap instruction.
21292 (ix86_warn_func_return): New function.
21293 (ix86_attribute_table): Add "naked" attribute specification.
21294 (TARGET_WARN_FUNC_RETURN): Define.
21295 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
21297 2017-07-31 Martin Liska <mliska@suse.cz>
21299 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
21300 (dump_gimple_bb_header): Always dump BB info.
21301 (pp_cfg_jump): Do not append info about BB when dumping a jump.
21303 2017-07-31 Martin Liska <mliska@suse.cz>
21306 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
21307 also with current_function_decl non-null equality.
21309 2017-07-31 Jakub Jelinek <jakub@redhat.com>
21312 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
21313 change type to the element type, instead add eltype variable and
21314 use it where we are interested in the element type.
21316 PR tree-optimization/81603
21317 * ipa-polymorphic-call.c
21318 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
21319 offset arithmetic in offset_int, bail out if the resulting bit offset
21320 doesn't fit into shwi.
21322 2017-07-31 Martin Liska <mliska@suse.cz>
21324 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
21325 (gimplify_save_expr): Fix comment.
21327 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
21330 * config/i386/i386.c (ix86_function_arg): Update arguments for
21332 (ix86_compute_frame_layout): Set the initial stack offset to
21333 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
21334 INCOMING_FRAME_SP_OFFSET.
21335 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
21336 stack before exception handler returns.
21337 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
21338 the 'ERROR_CODE' for exception handler.
21340 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
21342 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
21343 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
21344 (ASM_OUTPUT_REG_POP): Ditto.
21345 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
21346 instead of asm_fprintf to output pure string.
21348 2017-07-29 Jakub Jelinek <jakub@redhat.com>
21350 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
21351 to imported_module_or_decl hook.
21352 (debug_nothing_tree_tree_tree_bool): Remove.
21353 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
21354 * debug.c (do_nothing_debug_hooks): Use
21355 debug_nothing_tree_tree_tree_bool_bool instead of
21356 debug_nothing_tree_tree_tree_bool.
21357 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21358 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
21359 * sdbout.c (sdb_debug_hooks): Likewise.
21360 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
21361 (gen_namespace_die): Add DW_AT_export_symbols attribute if
21363 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
21364 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
21365 attribute, don't add anything.
21367 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21369 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
21370 (fold_build2_stat_loc): Likewise.
21371 (fold_build3_stat_loc): Likewise.
21372 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
21373 (fold_build1_loc): Remove macro.
21374 (fold_build2_loc): Likewise.
21375 (fold_build3_loc): Likewise.
21377 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21379 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
21380 (gimple_build_debug_bind_source_stat): Likewise.
21381 * gimple.h (gimple_build_debug_bind): Remove macro.
21382 (gimple_build_debug_bind_source): Likewise.
21384 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21386 * bitmap.c (bitmap_alloc): Adjust.
21387 (bitmap_gc_alloc): Likewise.
21388 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
21390 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21392 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
21393 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
21394 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
21395 (bitmap_gc_alloc_stat): Likewise.
21396 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
21398 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21400 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
21401 * rtl.h (shallow_copy_rtx): Remove macro.
21403 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21405 * emit-rtl.c (gen_raw_REG): Adjust.
21406 * gengenrtl.c (gendef): Likewise.
21407 * rtl.c (rtx_alloc_stat): Remove _stat from name.
21408 * rtl.h (rtx_alloc): Remove macro.
21410 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21412 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
21413 (build_tree_list_stat): Likewise.
21414 * tree.h (build_tree_list): Remove macro.
21415 (build_tree_list_vec): Likewise.
21417 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21419 * tree.c (make_vector_stat): Remove _stat from name.
21420 (build_vector_stat): Likewise.
21421 * tree.h (make_vector_stat): Remove macro.
21422 (build_vector_stat): Likewise.
21424 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21426 * tree.h (build_var_debug_value): Remove prototype.
21428 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21430 * tree.c (tree_cons_stat): Remove _stat from name.
21431 * tree.h (tree_cons): Remove macro.
21433 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21435 * tree.c (build_vl_exp_stat): Remove _stat from name.
21436 * tree.h (build_vl_exp): Remove macro.
21438 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21440 * tree.c (build_decl_stat): Remove _stat from name.
21441 * tree.h (build_decl): Remove macro.
21443 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21445 * gimple.c (gimple_build_with_ops_stat): Adjust.
21446 (gimple_alloc_stat): Remove _stat from name.
21447 * gimple.h (gimple_alloc): Remove macro.
21449 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21451 * tree.c (make_tree_vec_stat): Remove _stat from name.
21452 (grow_tree_vec_stat): Likewise.
21453 * tree.h (make_tree_vec_stat): Adjust prototype.
21454 (grow_tree_vec_stat): Likewise.
21455 (make_tree_vec): Remove macro.
21456 (grow_tree_vec): Likewise.
21458 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21460 * fold-const.c (fold_build1_stat_loc): Adjust.
21461 (fold_build2_stat_loc): Likewise.
21462 (fold_build3_stat_loc): Likewise.
21463 * tree.c (build0_stat): Remove _stat from name.
21464 (build1_stat): Likewise.
21465 (build2_stat): Likewise.
21466 (build3_stat): Likewise.
21467 (build4_stat): Likewise.
21468 (build5_stat): Likewise.
21469 * tree.h (build1_loc): Remove macro, and rename _stat function
21471 (build2_loc): Likewise.
21472 (build3_loc): Likewise.
21473 (build4_loc): Likewise.
21474 (build5_loc): Likewise.
21476 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21478 * tree.c (make_int_cst_stat): Remove _stat from name.
21479 * tree.h (make_int_cst_stat): Adjust prototype.
21480 (make_int_cst): Remove macro.
21482 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21484 * tree.c (make_tre_binfo_stat): Remove _stat from name.
21485 * tree.h (make_tree_binfo_stat): Adjust prototype.
21486 (make_tree_binfo): Remove.
21488 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21490 * tree.c (copy_node_stat): Rename to copy_node.
21491 (build_distinct_type_copy): Adjust.
21492 * tree.h (copy_node_stat): Adjust prototype.
21493 (copy_node): Remove macro.
21495 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21497 * tree.c (make_node_stat): rename to make_node.
21498 (build_tree_list_stat): Adjust.
21499 (build0_stat): Likewise.
21500 (build2_stat): Likewise.
21501 (build3_stat): Likewise.
21502 (build4_stat): Likewise.
21503 (build5_stat): Likewise.
21504 (build_decl_stat): Likewise.
21505 * tree.h (make_node_stat): Adjust prototype.
21506 (make_node): remove macro.
21508 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
21510 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
21511 (PPC_FEATURE2_SCV): Likewise.
21512 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
21514 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21516 * config/aarch64/aarch64.c
21517 (aarch64_internal_mov_immediate): Add new special pattern.
21518 * config/aarch64/aarch64.md (*movdi_aarch64):
21519 Add reg/32bit const mov case.
21521 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21522 Richard Sandiford <richard.sandiford@linaro.org>
21524 * config/aarch64/aarch64.md (mov<mode>): Generalize.
21525 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21526 Add integer and movi cases.
21527 (movi-split-hf-df-sf split, fp16): New.
21528 (enabled): Added TARGET_FP_F16INST.
21529 * config/aarch64/iterators.md (GPF_HF): New.
21530 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
21532 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21534 * config/aarch64/aarch64.c
21535 (aarch64_simd_container_mode): Add prototype.
21536 (aarch64_expand_mov_immediate): Add HI support.
21537 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
21538 (aarch64_can_const_movi_rtx_p): New.
21539 (aarch64_preferred_reload_class):
21540 Remove restrictions of using FP registers for certain SIMD operations.
21541 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
21542 (aarch64_valid_floating_const): Add integer move validation.
21543 (aarch64_simd_imm_scalar_p): Remove.
21544 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
21545 (aarch64_legitimate_constant_p): Expand list of supported cases.
21546 * config/aarch64/aarch64-protos.h
21547 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
21548 (aarch64_reinterpret_float_as_int): New.
21549 (aarch64_simd_imm_scalar_p): Remove.
21550 * config/aarch64/constraints.md (Uvi): New.
21551 (Dd): Split into Ds and new Dd.
21552 * config/aarch64/aarch64.md (*movsi_aarch64):
21554 (*movdi_aarch64): Add SIMD mov case.
21556 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21558 * tree-predcom.c: (struct chain): Handle store-store chain in which
21559 stores for elimination only store loop invariant values.
21560 (execute_pred_commoning_chain): Ditto.
21561 (prepare_initializers_chain_store_elim): Ditto.
21562 (prepare_finalizers): Ditto.
21563 (is_inv_store_elimination_chain): New function.
21564 (initialize_root_vars_store_elim_1): New function.
21566 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21568 * tree-predcom.c: Revise general description of the pass.
21569 (enum chain_type): New enum type for store elimination.
21570 (struct chain): New field supporting store elimination.
21571 (struct component): Ditto.
21572 (dump_chain): Dump store-stores chain.
21573 (release_chain): Release resources.
21574 (split_data_refs_to_components): Compute and create component
21575 contains only stores for elimination.
21576 (get_chain_last_ref_at): New function.
21577 (make_invariant_chain): Initialization.
21578 (make_rooted_chain): Specify chain type in parameter and record it.
21579 (add_looparound_copies): Skip for store-stores chain.
21580 (determine_roots_comp): Compute type of chain and pass it to
21582 (initialize_root_vars_store_elim_2): New function.
21583 (finalize_eliminated_stores): New function.
21584 (remove_stmt): Handle store for elimination.
21585 (execute_pred_commoning_chain): Execute predictive commoning on
21586 store-store chains.
21587 (determine_unroll_factor): Skip unroll for store-stores chain.
21588 (prepare_initializers_chain_store_elim): New function.
21589 (prepare_initializers_chain): Hanlde store-store chain.
21590 (prepare_finalizers_chain, prepare_finalizers): New function.
21591 (tree_predictive_commoning_loop): Return integer value indicating
21592 if loop is unrolled or lcssa form is corrupted.
21593 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
21595 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21597 * tree-predcom.c (initialize_root): Delete.
21598 (execute_pred_commoning_chain): Initialize root vars and replace
21599 reference of non-combined chain directly, rather than call above
21602 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21604 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
21605 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
21607 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21609 * tree-predcom.c (struct chain): New field init_seq.
21610 (release_chain): Release init_seq.
21611 (prepare_initializers_chain): Record intialization stmts in above
21613 (insert_init_seqs): New function.
21614 (tree_predictive_commoning_loop): Call insert_init_seqs.
21616 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21618 * tree-predcom.c (determine_roots_comp): Skip trivial components.
21620 2017-07-28 Richard Biener <rguenther@suse.de>
21622 * match.pd: Remove superfluous :c.
21623 * genmatch.c (simplify::id): Add member.
21624 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
21626 (current_id): New global.
21627 (dt_node::parent): Move from ...
21628 (dt_operand::parent): ... here. Add for_id member.
21629 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
21630 (decision_tree::find_node): Relax order requirement when
21631 merging DT_TRUE nodes to ones inbetween the current simplify
21632 and the one we try to merge with. Add diagnostic whenever
21633 we need to enforce pattern order by not merging.
21634 (decision_tree::insert): Set current_id.
21635 (decision_tree::print_node): Dump parent node and for_id.
21636 (parser::last_id): Add member.
21637 (parser::push_simplify): Assign unique id.
21638 (parser::parser): Initialize last_id.
21640 2017-07-28 Martin Liska <mliska@suse.cz>
21643 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
21644 gimple_build_debug_bind.
21646 2017-07-28 Richard Biener <rguenther@suse.de>
21648 PR tree-optimization/81502
21649 * match.pd: Add pattern combining BIT_INSERT_EXPR with
21651 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
21653 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
21654 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
21655 for BIT_FIELD_REF args.
21656 * fold-const.c (make_bit_field_ref): Likewise.
21657 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
21659 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21662 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
21663 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
21664 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
21665 Or it into SANITIZER_UNDEFINED.
21666 * ubsan.c: Include gimple-fold.h and varasm.h.
21667 (ubsan_expand_ptr_ifn): New function.
21668 (instrument_pointer_overflow): New function.
21669 (maybe_instrument_pointer_overflow): New function.
21670 (instrument_object_size): Formatting fix.
21671 (pass_ubsan::execute): Call instrument_pointer_overflow
21672 and maybe_instrument_pointer_overflow.
21673 * internal-fn.c (expand_UBSAN_PTR): New function.
21674 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
21675 * sanitizer.def (__ubsan_handle_pointer_overflow,
21676 __ubsan_handle_pointer_overflow_abort): New builtins.
21677 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
21678 * internal-fn.def (UBSAN_PTR): New internal function.
21679 * opts.c (sanitizer_opts): Add pointer-overflow.
21680 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
21681 * fold-const.c (build_range_check): Compute pointer range check in
21682 integral type if pointer arithmetics would be needed. Formatting
21685 2017-07-28 Martin Liska <mliska@suse.cz>
21688 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
21689 parameters that are of a variable-length.
21691 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
21693 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
21695 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
21696 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
21697 (CRT_CALL_STATIC_FUNCTION): Likewise.
21698 (ASM_DEFAULT_SPEC): New define.
21699 (ASM_SPEC32): Likewise.
21700 (ASM_SPEC64): Likewise.
21701 (ASM_SPEC_COMMON): Likewise.
21702 (ASM_SPEC): Likewise.
21703 (INVALID_64BIT): Likewise.
21704 (LINK_OS_DEFAULT_SPEC): Likewise.
21705 (LINK_OS_SPEC32): Likewise.
21706 (LINK_OS_SPEC64): Likewise.
21707 (POWERPC_LINUX): Likewise.
21708 (PTRDIFF_TYPE): Likewise.
21709 (RESTORE_FP_PREFIX): Likewise.
21710 (RESTORE_FP_SUFFIX): Likewise.
21711 (SAVE_FP_PREFIX): Likewise.
21712 (SAVE_FP_SUFFIX): Likewise.
21713 (SIZE_TYPE): Likewise.
21714 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
21715 (TARGET_64BIT): Likewise.
21716 (TARGET_64BIT): Likewise.
21717 (TARGET_AIX): Likewise.
21718 (WCHAR_TYPE_SIZE): Likewise.
21719 (WCHAR_TYPE): Undefine.
21720 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
21721 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
21722 (CPP_OS_RTEMS_SPEC): Delete.
21723 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
21724 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
21726 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
21728 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21730 PR tree-optimization/81578
21731 * tree-parloops.c (build_new_reduction): Bail out if
21732 reduction_code isn't one of the standard OpenMP reductions.
21733 Move the details printing after that decision.
21735 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21737 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
21738 related to reload_in_progress.
21739 (splat_input_operand): Likewise.
21740 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
21742 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
21744 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
21745 (TARGET_INSTANTIATE_DECLS): Likewise.
21746 (legitimate_indexed_address_p): Delete reload_in_progress code.
21747 (rs6000_debug_legitimate_address_p): Likewise.
21748 (rs6000_eliminate_indexed_memrefs): Likewise.
21749 (rs6000_emit_le_vsx_store): Likewise.
21750 (rs6000_emit_move_si_sf_subreg): Likewise.
21751 (rs6000_emit_move): Likewise.
21752 (register_to_reg_type): Likewise.
21753 (rs6000_pre_atomic_barrier): Likewise.
21754 (rs6000_machopic_legitimize_pic_address): Likewise.
21755 (rs6000_allocate_stack_temp): Likewise.
21756 (rs6000_address_for_fpconvert): Likewise.
21757 (rs6000_address_for_altivec): Likewise.
21758 (rs6000_secondary_memory_needed_rtx): Delete function.
21759 (rs6000_check_sdmode): Likewise.
21760 (rs6000_alloc_sdmode_stack_slot): Likewise.
21761 (rs6000_instantiate_decls): Likewise.
21762 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
21763 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
21764 Delete reload_in_progress.
21765 (*vec_reload_and_plus_<mptrsize>): Likewise.
21766 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
21767 (vsx_div_v2di): Likewise.
21768 (vsx_udiv_v2di): Likewise.
21770 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21772 * config/rs6000/rs6000.opt (mlra): Replace with stub.
21773 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
21774 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
21775 (rs6000_debug_reg_global): Delete print of LRA status.
21776 (rs6000_option_override_internal): Delete dead LRA related code.
21777 (rs6000_lra_p): Delete function.
21778 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
21780 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
21782 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
21783 * config/riscv/rtems.h: New file.
21785 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21786 Sudakshina Das <sudi.das@arm.com>
21788 * config/aarch64/aarch64.md
21789 (define_split for and<mode>3nr_compare): Move
21790 non aarch64_logical_operand to a register.
21791 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
21792 register immediate operand to a register.
21793 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
21795 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21797 PR middle-end/81564
21798 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
21800 2017-07-27 Richard Biener <rguenther@suse.de>
21802 PR tree-optimization/81573
21803 PR tree-optimization/81494
21804 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
21805 multi defuse cycle case.
21807 2017-07-27 Richard Biener <rguenther@suse.de>
21809 PR tree-optimization/81571
21810 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
21813 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
21815 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
21816 earlier and only if MASK_FPU is set. Adjust formatting.
21818 2017-07-27 Martin Liska <mliska@suse.cz>
21820 * opt-functions.awk: Add validation of value of Init.
21821 * optc-gen.awk: Pass new argument.
21823 2017-07-27 Martin Liska <mliska@suse.cz>
21825 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
21826 Fix wrong condition.
21828 2017-07-27 Martin Liska <mliska@suse.cz>
21830 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
21831 BBs and edges seen by autoFDO.
21833 2017-07-27 Richard Biener <rguenther@suse.de>
21835 PR tree-optimization/81502
21836 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
21837 with incompatible but same sized type.
21838 (execute_update_addresses_taken): Likewise.
21840 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
21842 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
21843 flag_tree_loop_vectorize rather than flag_tree_vectorize.
21845 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21848 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
21849 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
21850 Change s_operand to memory_operand.
21852 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
21854 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
21855 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
21856 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
21857 Emit instructions rather than returning an expression. Handle TFmode
21858 and KFmode by casting to TImode.
21859 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
21860 (rs6000_emit_le_vsx_store): Likewise.
21861 * config/rs6000/vsx.md (VSX_TI): New iterator.
21862 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
21863 (*vsx_le_undo_permute_<mode>): Likewise.
21864 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
21865 emit the split sequence.
21866 (*vsx_le_perm_store_<mode>): Likewise.
21868 2017-07-27 Jakub Jelinek <jakub@redhat.com>
21870 PR tree-optimization/81555
21871 PR tree-optimization/81556
21872 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
21873 if true, force CHANGED for the recursive invocation.
21874 (reassociate_bb): Remember original length of ops array, pass
21875 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
21877 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
21878 attributes for noipa attribute. For naked attribute use
21879 lookup_attribute first before lookup_attribute_spec.
21880 * final.c (rest_of_handle_final): Disable IPA RA for functions with
21882 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
21883 for functions with noipa attribute.
21884 (cgraph_externally_visible_p): Return true for functions with noipa
21886 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
21887 for functions with noipa attribute.
21888 * doc/extend.texi: Document noipa function attribute.
21889 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
21890 also for functions with noipa attribute.
21891 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
21893 2017-07-26 Andrew Pinski <apinski@cavium.com>
21895 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
21896 vec_unalign_load_cost and vec_unalign_store_cost.
21898 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
21900 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
21901 -mvsx-small-integer option.
21902 (ISA_3_0_MASKS_IEEE): Likewise.
21903 (OTHER_VSX_VECTOR_MASKS): Likewise.
21904 (POWERPC_MASKS): Likewise.
21905 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
21906 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
21907 code, only testing for DImode being allowed in non-VSX floating
21909 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
21910 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
21912 (rs6000_option_override_internal): Delete -mvsx-small-integer.
21913 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
21914 TARGET_P8_VECTOR test.
21915 (rs6000_secondary_reload_simple_move): Likewise.
21916 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
21917 since TARGET_P9_VECTOR was already tested.
21918 (rs6000_opt_masks): Remove -mvsx-small-integer.
21919 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
21920 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21922 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
21923 test for TARGET_VEXTRACTUB was used, and that uses
21925 (p9 extract splitter): Likewise.
21926 (vsx_extract_<mode>_di_p9): Likewise.
21927 (vsx_extract_<mode>_store_p9): Likewise.
21928 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
21929 for TARGET_P9_VECTOR was used. Delete code that is now dead with
21930 the elimination of TARGET_VSX_SMALL_INTEGER.
21931 (vsx_extract_<mode>_p8): Likewise.
21932 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
21933 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
21934 (vsx_set_<mode>_p9): Likewise.
21935 (vsx_set_v4sf_p9): Likewise.
21936 (vsx_set_v4sf_p9_zero): Likewise.
21937 (vsx_insert_extract_v4sf_p9): Likewise.
21938 (vsx_insert_extract_v4sf_p9_2): Likewise.
21939 * config/rs6000/rs6000.md (sign extend splitter): Change
21940 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
21941 (floatsi<mode>2_lfiwax_mem): Likewise.
21942 (floatunssi<mode>2_lfiwzx_mem): Likewise.
21943 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
21944 since a test for TARGET_P9_VECTOR was used.
21945 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21946 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
21947 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21948 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
21949 TARGET_P8_VECTOR test.
21950 (fix_trunc<mode>si2_stfiwx): Likewise.
21951 (fix_trunc<mode>si2_internal): Likewise.
21952 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
21953 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21955 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
21956 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
21957 TARGET_P8_VECTOR test.
21958 (fixuns_trunc<mode>si2_stfiwx): Likewise.
21959 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
21960 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21962 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
21963 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
21964 since a test for TARGET_P9_VECTOR was used.
21965 (splitter for loading small constants): Likewise.
21967 2017-07-26 Andrew Pinski <apinski@cavium.com>
21969 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
21972 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
21975 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
21976 (fp_valid_at): Likewise.
21978 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
21980 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
21981 (qdf24xx_addrcost_table): Likewise.
21982 (cortexa57_tunings): Update to use generic_branch_cost.
21983 (cortexa72_tunings): Likewise.
21984 (cortexa73_tunings): Likewise.
21985 (qdf24xx_tunings): Likewise.
21987 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
21989 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
21990 (thunderx2t99_branch_cost): Likewise.
21991 (cortexa35_tunings): Update to use generic_branch_cost.
21992 (cortexa53_tunings): Likewise.
21993 (cortexa57_tunings): Likewise.
21994 (cortexa72_tunings): Likewise.
21995 (cortexa73_tunings): Likewise.
21996 (thunderx2t99_tunings): Likewise.
21998 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22000 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
22001 (sparc_option_override): Honour MASK_FSMULD.
22002 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
22003 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
22004 * config/sparc/sparc.opt (mfsmuld): New option.
22005 * doc/invoke.texi (mfsmuld): Document option.
22007 2017-07-26 Marek Polacek <polacek@redhat.com>
22009 PR middle-end/70992
22010 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
22012 2017-07-26 Richard Biener <rguenther@suse.de>
22014 * gimple-match-head.c (do_valueize): Return OP if valueize
22016 (get_def): New helper to get at the def stmt of a SSA name
22017 if valueize allows.
22018 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
22019 do_valueize to get at the def stmt.
22020 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
22022 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
22024 PR middle-end/46932
22025 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
22027 2017-07-26 Martin Liska <mliska@suse.cz>
22030 * function.c (expand_function_start): Make expansion of
22031 nonlocal_goto_save_area after parm_birth_insn.
22033 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22035 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
22036 from all CPU target flags enable members.
22038 2017-07-26 Richard Biener <rguenther@suse.de>
22040 * genmatch.c (dt_simplify::gen): Make iterator vars const.
22041 (decision_tree::gen): Make 'type' const.
22042 (write_predicate): Likewise.
22044 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22046 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
22047 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
22048 (rs6000_option_override_internal): Likewise.
22049 (rs6000_expand_vector_set): Likewise.
22050 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
22051 (TARGET_UPPER_REGS_SF): Likewise.
22052 (TARGET_UPPER_REGS_DI): Likewise.
22053 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
22054 (TARGET_DIRECT_MOVE_64BIT): Likewise.
22055 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
22056 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22057 (Splitters for DI constants in Altivec registers): Likewise.
22058 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
22059 (vsx_set_v4sf_p9): Likewise.
22060 (vsx_set_v4sf_p9_zero): Likewise.
22061 (vsx_insert_extract_v4sf_p9): Likewise.
22062 (vsx_insert_extract_v4sf_p9_2): Likewise.
22064 2017-07-25 Carl Love <cel@us.ibm.com>
22066 * doc/extend.texi: Update the built-in documentation file for the
22067 existing built-in functions
22068 vector signed char vec_cnttz (vector signed char);
22069 vector unsigned char vec_cnttz (vector unsigned char);
22070 vector signed short vec_cnttz (vector signed short);
22071 vector unsigned short vec_cnttz (vector unsigned short);
22072 vector signed int vec_cnttz (vector signed int);
22073 vector unsigned int vec_cnttz (vector unsigned int);
22074 vector signed long long vec_cnttz (vector signed long long);
22075 vector unsigned long long vec_cnttz (vector unsigned long long);
22077 2017-07-25 Andrew Pinski <apinski@cavium.com>
22079 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
22080 accesses where the use is for the first operand of a BIT_INSERT.
22082 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22085 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22086 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22088 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22090 * config/i386/gstabs.h: Delete.
22091 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
22093 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
22095 * config/i386/i386.c (ix86_decompose_address): Do not check for
22096 register RTX when looking at index_reg or base_reg.
22097 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
22099 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
22101 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
22102 to update EH info here.
22104 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22106 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
22108 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22110 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
22112 2017-07-25 Torsten Duwe <duwe@suse.de>
22114 * common.opt: Introduce -fpatchable-function-entry
22115 command line option, and its variables function_entry_patch_area_size
22116 and function_entry_patch_area_start.
22117 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
22118 including a two-value parser.
22119 * target.def (print_patchable_function_entry): New target hook.
22120 * targhooks.h (default_print_patchable_function_entry): New function.
22121 * targhooks.c (default_print_patchable_function_entry): Likewise.
22122 * toplev.c (process_options): Switch off IPA-RA if
22123 patchable function entries are being generated.
22124 * varasm.c (assemble_start_function): Look at the
22125 patchable-function-entry command line switch and current
22126 function attributes and maybe generate NOP instructions by
22127 calling the print_patchable_function_entry hook.
22128 * doc/extend.texi: Document patchable_function_entry attribute.
22129 * doc/invoke.texi: Document -fpatchable_function_entry
22130 command line option.
22131 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
22133 * doc/tm.texi: Re-generate.
22135 2017-07-25 Jakub Jelinek <jakub@redhat.com>
22138 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
22139 TARGET_AVX512DQ rather than TARGET_AVX512BW.
22141 2017-07-25 Tamar Christina <tamar.christina@arm.com>
22143 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
22145 2017-07-25 Richard Biener <rguenther@suse.de>
22147 PR tree-optimization/81455
22148 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
22149 not walk in cycles when looking for guards.
22151 2017-07-25 Richard Biener <rguenther@suse.de>
22153 PR tree-optimization/81529
22154 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
22155 when optimizing backedge uses.
22157 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
22159 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
22161 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
22162 to dl_section_ref. On AIX, append an expression to subtract
22163 the size of the section length to dl_section_ref.
22165 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
22167 * configure.ac: If any of the config.* scripts fail, exit 1.
22168 * configure: Regenerate.
22170 2017-07-25 Richard Biener <rguenther@suse.de>
22172 PR middle-end/81546
22173 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
22174 of immediate uses, be more verbose on errors.
22176 2017-07-25 Richard Biener <rguenther@suse.de>
22178 PR tree-optimization/81510
22179 * tree-vect-loop.c (vect_is_simple_reduction): When the
22180 reduction stmt is not inside the loop bail out.
22182 2017-07-25 Richard Biener <rguenther@suse.de>
22184 PR tree-optimization/81303
22185 * tree-vect-loop-manip.c (vect_loop_versioning): Build
22186 profitability check against LOOP_VINFO_NITERSM1.
22188 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22190 * domwalk.c (cmp_bb_postorder): Simplify.
22191 (sort_bbs_postorder): New function. Use it...
22192 (dom_walker::walk): ...here to optimize common cases.
22194 2017-07-25 Martin Liska <mliska@suse.cz>
22197 * ipa-visibility.c (function_and_variable_visibility): Make the
22198 redirection just on target that supports aliasing.
22199 Fix GNU coding style.
22201 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
22204 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
22206 * config/arm/rtems.h: Likewise.
22207 * config/bfin/rtems.h: Likewise.
22208 * config/i386/rtemself.h: Likewise.
22209 * config/lm32/rtems.h: Likewise.
22210 * config/m32c/rtems.h: Likewise.
22211 * config/m68k/rtemself.h: Likewise.
22212 * config/microblaze/rtems.h: Likewise.
22213 * config/mips/rtems.h: Likewise.
22214 * config/moxie/rtems.h: Likewise.
22215 * config/nios2/rtems.h: Likewise.
22216 * config/powerpcspe/rtems.h: Likewise.
22217 * config/rs6000/rtems.h: Likewise.
22218 * config/rtems.h: Likewise.
22219 * config/sh/rtems.h: Likewise.
22220 * config/sh/rtemself.h: Likewise.
22221 * config/sparc/rtemself.h: Likewise.
22223 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22226 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
22227 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
22228 * tree-ssa-structalias.c (alias_get_name): Same.
22230 2017-07-25 Bin Cheng <bin.cheng@arm.com>
22233 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
22234 instructions if no du chain is found.
22236 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22238 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
22240 2017-07-25 Richard Biener <rguenther@suse.de>
22242 PR middle-end/81505
22243 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
22246 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22248 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
22249 upper-regs options.
22250 (ISA_2_7_MASKS_SERVER): Likewise.
22251 (ISA_3_0_MASKS_IEEE): Likewise.
22252 (OTHER_P8_VECTOR_MASKS): Likewise.
22253 (OTHER_VSX_VECTOR_MASKS): Likewise.
22254 (POWERPC_MASKS): Likewise.
22255 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
22256 duplicate list of options.
22257 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
22258 explicit -mupper-regs options.
22259 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
22260 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
22261 alias for -mupper-regs-df.
22262 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
22263 (rs6000_init_hard_regno_mode_ok): Likewise.
22264 (rs6000_option_override_internal): Likewise.
22265 (rs6000_opt_masks): Likewise.
22266 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
22267 options in terms of whether -mvsx or -mpower8-vector was used.
22268 (TARGET_UPPER_REGS_DI): Likewise.
22269 (TARGET_UPPER_REGS_SF): Likewise.
22270 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
22271 -mupper-regs-* options.
22273 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22275 * passes.c (emergency_dump_function): Print some empty lines and a
22276 header before the RTL dump.
22278 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22280 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
22282 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
22285 * config/aarch64/aarch64.c (aarch64_classify_symbol):
22286 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
22288 2017-07-24 Carl Love <cel@us.ibm.com>
22290 * config/rs6000/rs6000-c.c: Add support for built-in functions
22291 vector float vec_extract_fp32_from_shorth (vector unsigned short);
22292 vector float vec_extract_fp32_from_shortl (vector unsigned short);
22293 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
22294 vec_extract_fp_from_shortl): Add defines for the two builtins.
22295 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
22296 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
22298 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
22299 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
22300 * doc/extend.texi: Update the built-in documentation file for the
22301 new built-in function.
22303 2017-07-24 Jakub Jelinek <jakub@redhat.com>
22306 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
22308 * doc/generic.texi: Likewise.
22309 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22310 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22312 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
22314 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
22315 (aarch64_mls_elt_merge<mode>): Likewise.
22317 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
22319 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22320 having __cxa_atexit.
22322 2017-07-23 Michael Collison <michael.collison@arm.com>
22324 * config/arm/arm.c (arm_option_override): Deprecate
22325 use of -mstructure-size-boundary.
22326 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
22327 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
22329 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22332 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22333 Reduce cost estimate for direct moves.
22335 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
22338 * config/i386/i386.c (ix86_option_override_internal): Disable
22339 BMI, BMI2 and TBM instructions for -m16.
22341 2017-07-21 Carl Love <cel@us.ibm.com>
22343 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22344 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22345 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22346 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22347 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22348 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22349 VMULOSW): New enum "unspec" values.
22350 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22351 altivec_vmulosw): New patterns.
22352 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22353 VMULOSW): Add definitions.
22355 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
22357 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
22358 (qdf24xx): Likewise.
22359 * config/aarch64/aarch64-options-extensions.def (rdma); New.
22360 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
22361 (AARCH64_FL_V8_1): Renumber.
22362 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
22363 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
22364 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
22365 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
22366 rdma to feature modifiers list.
22368 2017-07-21 Yury Gribov <tetra2005@gmail.com>
22370 PR middle-end/56727
22371 * ipa-visibility (function_and_variable_visibility): Convert
22372 recursive PLT call to direct call if appropriate.
22374 2017-07-21 Andrew Pinski <apinski@cavium.com>
22376 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
22377 operand 1 to see if the types precision matches.
22378 * fold-const.c (operand_equal_p): Likewise.
22380 2017-07-21 Richard Biener <rguenther@suse.de>
22382 PR tree-optimization/81303
22383 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
22384 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
22385 (vect_peeling_hash_get_lowest_cost): Adjust.
22386 (vect_enhance_data_refs_alignment): Likewise. Use
22387 vect_get_peeling_costs_all_drs to compute the penalty for no
22388 peeling to match up costs.
22390 2017-07-21 Richard Biener <rguenther@suse.de>
22392 PR tree-optimization/81500
22393 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
22394 we didn't identify a reduction path.
22396 2017-07-21 Tom de Vries <tom@codesourcery.com>
22397 Cesar Philippidis <cesar@codesourcery.com>
22399 PR gcov-profile/81442
22400 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
22403 2017-07-21 Tom de Vries <tom@codesourcery.com>
22406 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
22408 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
22409 nvptx_override_options_after_change.
22411 2017-07-21 Ulrich Drepper <drepper@redhat.com>
22413 * dwarf2out.c (output_file_names): Avoid double testing for
22414 dwarf_version >= 5.
22416 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
22418 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
22420 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22422 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
22424 (try_crossjump_to_edge): Do not punt on partitioned functions.
22426 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22428 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
22429 Put all BBs reachable only via paths crossing cold region to cold
22431 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
22433 2016-07-21 Richard Biener <rguenther@suse.de>
22435 PR tree-optimization/81303
22436 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
22437 into account prologue and epilogue iterations when raising
22438 min_profitable_iters to sth at least covering one vector iteration.
22440 2017-07-21 Tamar Christina <tamar.christina@arm.com>
22442 * config/arm/arm.c (arm_test_cpu_arch_dat):
22445 2017-07-20 Nathan Sidwell <nathan@acm.org>
22447 Remove TYPE_METHODS.
22448 * tree.h (TYPE_METHODS): Delete.
22449 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
22450 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
22451 (dbxout_type_methods): Scan TYPE_FIELDS.
22452 (dbxout_type): Don't check TYPE_METHODS here.
22453 * function.c (use_register_for_decl): Always ignore register for
22454 class types when not optimizing.
22455 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
22456 * tree.c (free_lang_data_in_type): Stitch out member functions and
22457 templates from TYPE_FIELDS.
22458 (build_distinct_type_copy, verify_type_variant,
22459 verify_type): Member fns are on TYPE_FIELDS.
22460 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
22461 * tree-pretty-print.c (dump_generic_node): Likewise.
22463 2017-07-20 Jakub Jelinek <jakub@redhat.com>
22466 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
22467 V2TImode and V4TImode.
22468 (ix86_expand_vector_extract): Likewise.
22469 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
22470 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
22471 (ssescalarmode): Handle V4TImode and V2TImode.
22472 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
22473 (*vec_extractv2ti, *vec_extractv4ti): New insns.
22474 (VEXTRACTI128_MODE): New mode iterator.
22475 (splitter for *vec_extractv?ti first element): New.
22476 (VEC_INIT_MODE): New mode iterator.
22477 (vec_init<mode>): Consolidate 3 expanders into one using
22478 VEC_INIT_MODE mode iterator.
22480 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22482 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
22483 non_spilled_static_chain_regno_p.
22485 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22487 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
22489 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
22491 * bb-reorder.c (connect_traces): Allow copying of blocks within
22494 2017-07-20 Richard Biener <rguenther@suse.de>
22496 * gimple.h (gimple_phi_result): Add gphi * overload.
22497 (gimple_phi_result_ptr): Likewise.
22498 (gimple_phi_arg): Likewise. Adjust index assert to only
22499 allow actual argument accesses rather than all slots available
22501 (gimple_phi_arg_def): Add gphi * overload.
22502 * tree-phinodes.c (make_phi_node): Initialize only actual
22504 (resize_phi_node): Clear memory not covered by old node,
22505 do not initialize excess argument slots.
22506 (reserve_phi_args_for_new_edge): Initialize new argument slot
22509 2017-07-20 Bin Cheng <bin.cheng@arm.com>
22511 PR tree-optimization/81388
22513 2016-07-21 Bin Cheng <bin.cheng@arm.com>
22515 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
22516 by removing computation of may_be_zero.
22518 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22519 Tom de Vries <tom@codesourcery.com>
22521 PR middle-end/81030
22522 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
22523 when gimple level profile disagrees with what RTL expander did.
22525 2017-07-20 Richard Biener <rguenther@suse.de>
22527 PR tree-optimization/61171
22528 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
22529 (vect_analyze_stmt): Add slp instance parameter.
22530 (vectorizable_reduction): Likewise.
22531 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
22532 (vect_is_simple_reduction): Deal with chains not detected
22533 as SLP reduction chain, specifically not properly associated
22534 chains containing a mix of plus/minus.
22535 (get_reduction_op): Remove.
22536 (get_initial_defs_for_reduction): Simplify, pass in whether
22537 this is a reduction chain, pass in the SLP node for the PHIs.
22538 (vect_create_epilog_for_reduction): Get the SLP instance as
22540 (vectorizable_reduction): Get the SLP instance as arg.
22541 During analysis remember the SLP node with the PHIs in the
22542 instance. Simplify getting at the vectorized reduction PHIs.
22543 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
22544 through SLP instance.
22545 (vect_slp_analyze_operations): Likewise.
22546 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
22547 (vect_transform_stmt): Likewise.
22549 2017-07-20 Tom de Vries <tom@codesourcery.com>
22551 PR tree-optimization/81489
22552 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
22553 read of phi arg location to before loop that modifies phi.
22555 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22557 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
22560 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22562 PR middle-end/81331
22563 * except.c (execute): Fix ordering issue.
22565 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22567 PR rtl-optimization/81423
22568 * combine.c (make_compound_operation_int): Don't try to optimize
22569 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
22571 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22573 PR rtl-optimization/81423
22574 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
22575 with a constant that is -1 in the truncated to mode.
22577 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22579 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
22580 (determine_unlikely_bbs): ... here.
22581 * predict.h (propagate_unlikely_bbs_forward): Declare.
22582 * cfgexpand.c (pass_expand::execute): Use it.
22583 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
22585 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
22586 propagate_unlikely_bbs_forward.
22588 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22590 PR middle-end/81331
22591 * except.c (maybe_add_nop_after_section_switch): New function.
22594 2017-07-19 Tom de Vries <tom@codesourcery.com>
22596 * gimple.h (gimple_phi_set_arg): Make assert more strict.
22598 2017-07-19 Tom de Vries <tom@codesourcery.com>
22600 * gimple.h (gimple_phi_arg): Make assert more strict.
22602 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
22604 * config.gcc (powerpc*-*-*): Add mmintrin.h.
22605 * config/rs6000/mmintrin.h: New file.
22606 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
22608 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22610 PR tree-optimization/81346
22611 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
22613 2017-07-19 Tom de Vries <tom@codesourcery.com>
22615 * config/nvptx/nvptx.md (VECIM): Add V2DI.
22617 2017-07-19 Tom de Vries <tom@codesourcery.com>
22619 * config/nvptx/nvptx-modes.def: Add V2DImode.
22620 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
22621 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
22622 (nvptx_output_mov_insn): Handle lack of mov.b128.
22623 (nvptx_print_operand): Handle 'H' and 'L' codes.
22624 (nvptx_vector_mode_supported): Allow V2DImode.
22625 (nvptx_preferred_simd_mode): New function.
22626 (nvptx_data_alignment): New function.
22627 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
22628 nvptx_preferred_simd_mode.
22629 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
22631 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
22633 2017-07-19 Tom de Vries <tom@codesourcery.com>
22635 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
22636 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
22637 (nvptx_vector_mode_supported): New function. Allow V2SImode.
22638 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
22639 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
22640 (mov<VECIM>_insn): New define_insn.
22641 (define_expand "mov<VECIM>): New define_expand.
22643 2017-07-19 Tom de Vries <tom@codesourcery.com>
22645 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
22647 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22649 PR tree-optimization/81346
22650 * fold-const.h (fold_div_compare, range_check_type): Declare.
22651 * fold-const.c (range_check_type): New function.
22652 (build_range_check): Use range_check_type.
22653 (fold_div_compare): No longer static, rewritten into
22654 a match.pd helper function.
22655 (fold_comparison): Don't call fold_div_compare here.
22656 * match.pd (X / C1 op C2): New optimization using fold_div_compare
22657 as helper function.
22659 2017-07-19 Nathan Sidwell <nathan@acm.org>
22661 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
22662 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
22663 * tree.c (find_decls_types_r, verify_type): Use
22664 TYPE_{MIN,MAX}_VALUE_RAW.
22665 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
22666 (hash_tree): Likewise.
22667 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
22669 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
22672 2017-07-18 Tom de Vries <tom@codesourcery.com>
22674 PR middle-end/81464
22675 * omp-expand.c (expand_omp_for_static_chunk): Handle
22676 equal-argument loop exit phi.
22678 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
22681 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
22682 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
22683 operand 2 predicate.
22684 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
22685 operand 2 predicate.
22686 (ror,rol -> rorx splitters): Use const_int_operand as
22687 operand 2 predicate.
22689 2017-06-18 Richard Biener <rguenther@suse.de>
22691 PR tree-optimization/81410
22692 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
22693 the gap in the ! slp_perm SLP case after each group.
22695 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22697 PR middle-end/81463
22698 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
22701 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22703 PR middle-end/81462
22704 * predict.c (set_even_probabilities): Cleanup; do not affect
22705 probabilities that are already known.
22706 (combine_predictions_for_bb): Call even when count is set.
22708 2017-07-18 Nathan Sidwell <nathan@acm.org>
22710 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
22713 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22716 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
22717 optimization for loop niter analysis.
22719 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
22722 * config/avr/avr.c (avr_optimize_casesi): Don't use
22723 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
22725 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
22727 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
22728 body_cost_vec from _vect_peel_extended_info.
22729 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
22730 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
22733 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22735 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
22737 2017-07-18 Richard Biener <rguenther@suse.de>
22739 PR tree-optimization/80620
22740 PR tree-optimization/81403
22741 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
22742 info when re-using a VN table entry.
22744 2017-07-18 Richard Biener <rguenther@suse.de>
22746 PR tree-optimization/81418
22747 * tree-vect-loop.c (vectorizable_reduction): Properly compute
22748 vectype_in. Verify that with lane-reducing reduction operations
22749 we have a single def-use cycle.
22751 2017-07-17 Carl Love <cel@us.ibm.com>
22753 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
22755 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22756 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22757 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22758 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22759 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22760 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22761 VMULOSW): New enum "unspec" values.
22762 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22763 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
22764 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22765 altivec_vmulosw): New patterns.
22766 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22767 VMULOSW): Add definitions.
22769 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
22771 * config/alpha/alpha.c: Include predict.h.
22773 2017-07-17 Yury Gribov <tetra2005@gmail.com>
22775 * tree-vrp.c (compare_assert_loc): Fix comparison function
22776 to return predictable results.
22778 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22780 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
22782 (subdi3): Likewise.
22783 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
22784 * doc/invoke.texi (mexpand-adddi): Update text.
22786 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22788 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
22789 that also clobbers the CC register. The old expand code is moved
22791 (*arc_clzsi2): ... here.
22792 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
22793 the CC register. The old expand code is moved to ...
22794 (arc_ctzsi2): ... here.
22796 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22798 * config/arc/arc.opt (mindexed-loads): Use initial value
22799 TARGET_INDEXED_LOADS_DEFAULT.
22800 (mauto-modify-reg): Use initial value
22801 TARGET_AUTO_MODIFY_REG_DEFAULT.
22802 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22803 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22804 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22805 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22807 2017-07-17 Martin Liska <mliska@suse.cz>
22810 * opts.c (finish_options): Do not allow -fgnu-tm
22811 w/ -fsanitize={kernel-,}address. Say sorry.
22813 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22816 * tree-loop-distribution.c (classify_partition): Only assert on
22817 numer of iterations.
22818 (merge_dep_scc_partitions): Delete prameter. Update function call.
22819 (distribute_loop): Remove code handling loop with unknown niters.
22820 (pass_loop_distribution::execute): Skip loop with unknown niters.
22822 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22825 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
22826 function sort_partitions_by_post_order.
22828 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22830 PR tree-optimization/81374
22831 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
22832 the max index of basic blocks, rather than number of basic blocks.
22834 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22836 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
22838 (arc_legitimate_pic_operand_p): Likewise.
22839 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
22841 (arc_needs_pcl_p): Likewise.
22842 (arc_legitimate_pc_offset_p): Likewise.
22843 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
22844 function is also used in constrains.md.
22845 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
22846 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
22847 PLUS. Only return true/false in known cases, otherwise assert.
22848 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
22849 is already called in arc_legitimate_constant_p.
22850 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
22852 (LEGITIMATE_PIC_OPERAND_P): Use
22853 arc_raw_symbolic_reference_mentioned_p function.
22854 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
22859 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22860 Andrew Burgess <andrew.burgess@embecosm.com>
22862 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
22863 (arc_return_address_register): New function.
22864 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
22865 (arc_handle_fndecl_attribute): Add naked attribute.
22866 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22867 (TARGET_WARN_FUNC_RETURN): Likewise.
22868 (arc_allocate_stack_slots_for_args): New function.
22869 (arc_warn_func_return): Likewise.
22870 (machine_function): Change type fn_type.
22871 (arc_compute_function_type): Consider new naked function type,
22872 change function return type.
22873 (arc_must_save_register): Adapt to handle new
22874 arc_compute_function_type's return type.
22875 (arc_expand_prologue): Likewise.
22876 (arc_expand_epilogue): Likewise.
22877 (arc_return_address_regs): Delete.
22878 (arc_return_address_register): New function.
22879 (arc_epilogue_uses): Use above function.
22880 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
22881 (arc_function_type): Change encoding, add naked type.
22882 (ARC_INTERRUPT_P): Change to handle the new encoding.
22883 (ARC_FAST_INTERRUPT_P): Likewise.
22884 (ARC_NORMAL_P): Define.
22885 (ARC_NAKED_P): Likewise.
22886 (arc_compute_function_type): Delete prototype.
22887 * config/arc/arc.md (in_ret_delay_slot): Use
22888 arc_return_address_register function.
22889 (simple_return): Likewise.
22890 (p_return_i): Likewise.
22892 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22894 PR tree-optimization/81428
22895 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
22896 can't be built for those types.
22898 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
22900 Remove stuff dead since r239246.
22902 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
22903 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
22904 (avr_inform_devices): Remove dead stuff.
22906 2017-07-17 Tamar Christina <tamar.christina@arm.com>
22908 * config/arm/arm_neon.h: Fix softp typo.
22910 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22912 PR tree-optimization/81365
22913 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
22914 aggregate moves onto bb predecessor edges, make sure there are no
22915 loads that could alias the lhs in between the start of bb and the
22918 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
22921 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
22922 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
22923 [LSHIFTRT, outer_code = TRUNCATE]: Same.
22925 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22927 PR tree-optimization/81396
22928 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
22929 (init_symbolic_number): Initialize it to 1.
22930 (perform_symbolic_merge): Add n_ops from both operands into the new
22932 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
22933 without base_addr as useless if they need more than one operation.
22934 (bswap_replace): Handle !bswap case for NULL base_addr.
22936 2017-07-17 Tom de Vries <tom@codesourcery.com>
22939 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
22942 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
22944 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
22945 conditional builtin define __FIX_LEON3FT_B2BST.
22947 2017-07-17 Daniel Cederman <cederman@gaisler.com>
22949 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
22950 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
22953 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
22955 PR rtl-optimization/81424
22956 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
22957 to remove potential trapping from operands if -fnon-call-exceptions.
22959 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22961 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
22962 profile_proability for scalling.
22963 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
22965 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22967 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
22969 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22971 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
22972 fixpoint arithmetics.
22974 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22976 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
22977 fixpoint arithmetics.
22979 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22981 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
22982 fixpoint arithmetics.
22984 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22986 * profile-count.h (profile_probability::from_reg_br_prob_note,
22987 profile_probability::to_reg_br_prob_note): New functions.
22988 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
22989 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
22990 * predict.c (probability_reliable_p): Update.
22991 (edge_probability_reliable_p): Update.
22992 (br_prob_note_reliable_p): Update.
22993 (invert_br_probabilities): Update.
22994 (add_reg_br_prob_note): New function.
22995 (combine_predictions_for_insn): Update.
22996 * asan.c (asan_clear_shadow): Update.
22997 * cfgbuild.c (compute_outgoing_frequencies): Update.
22998 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22999 (update_br_prob_note): Update.
23000 (rtl_verify_edges): Update.
23001 (purge_dead_edges): Update.
23002 (fixup_reorder_chain): Update.
23003 * emit-rtl.c (try_split): Update.
23004 * ifcvt.c (cond_exec_process_insns): Update.
23005 (cond_exec_process_if_block): Update.
23006 (dead_or_predicable): Update.
23007 * internal-fn.c (expand_addsub_overflow): Update.
23008 (expand_neg_overflow): Update.
23009 (expand_mul_overflow): Update.
23010 * loop-doloop.c (doloop_modify): Update.
23011 * loop-unroll.c (compare_and_jump_seq): Update.
23012 * optabs.c (emit_cmp_and_jump_insn_1): Update.
23013 * predict.h: Update.
23014 * reorg.c (mostly_true_jump): Update.
23016 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
23017 * config/alpha/alpha.c (emit_unlikely_jump): Update.
23018 * config/arc/arc.c: (emit_unlikely_jump): Update.
23019 * config/arm/arm.c: (emit_unlikely_jump): Update.
23020 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
23021 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
23022 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
23023 (ix86_print_operand): Update.
23024 (ix86_split_fp_branch): Update.
23025 (predict_jump): Update.
23026 * config/ia64/ia64.c (ia64_print_operand): Update.
23027 * config/mmix/mmix.c (mmix_print_operand): Update.
23028 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
23029 (rs6000_expand_split_stack_prologue): Update.
23030 * config/rs6000/rs6000.c: Update.
23031 * config/s390/s390.c (s390_expand_vec_strlen): Update.
23032 (s390_expand_vec_movstr): Update.
23033 (s390_expand_cs_tdsi): Update.
23034 (s390_expand_split_stack_prologue): Update.
23035 * config/sh/sh.c (sh_print_operand): Update.
23036 (expand_cbranchsi4): Update.
23037 (expand_cbranchdi4): Update.
23038 * config/sparc/sparc.c (output_v9branch): Update.
23039 * config/spu/spu.c (get_branch_target): Update.
23040 (ea_load_store_inline): Update.
23041 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
23042 * config/tilepro/tilepro.c: Update.
23044 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23046 * gimplify.c (mostly_copy_tree_r): Revert latest change.
23047 (gimplify_save_expr): Likewise.
23049 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23051 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
23053 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23055 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
23057 * timevar.def (TV_IPA_FNSUMMARY): Define.
23059 2017-07-16 Daniel Cederman <cederman@gaisler.com>
23061 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
23062 to back store errata sensitive sequence from being generated.
23063 (sqrtdf2_fix): Likewise.
23065 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23067 * tree-ssa-threadupdate.c (compute_path_counts,
23068 update_joiner_offpath_counts): Use profile_probability.
23070 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
23073 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23075 * config/arm/arm-c.c (arm_cpu_builtins): Define
23076 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23078 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23080 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23081 array entries to represent __ieee128 versions of the
23082 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
23083 scalar_extract_sig, and scalar_insert_exp built-in functions.
23084 (altivec_resolve_overloaded_builtin): Add special case handling
23085 for the __builtin_scalar_insert_exp function, as represented by
23086 the P9V_BUILTIN_VEC_VSIEDP constant.
23087 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
23088 exponent support for __ieee128 argument.
23089 (VSESQP): Add scalar extract signature support for __ieee128
23091 (VSTDCNQP): Add scalar test negative support for __ieee128
23093 (VSIEQP): Add scalar insert exponent support for __int128 argument
23094 with __ieee128 result.
23095 (VSIEQPF): Add scalar insert exponent support for __ieee128
23096 argument with __ieee128 result.
23097 (VSTDCQP): Add scalar test data class support for __ieee128
23099 (VSTDCNQP): Add overload support for scalar test negative with
23100 __ieee128 argument.
23101 (VSTDCQP): Add overload support for scalar test data class
23102 __ieee128 argument.
23103 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
23104 UNSPEC_VSX_SXSIGDP.
23105 (UNSPEC_VSX_SIEXPQP): New constant.
23106 (xsxexpqp): New insn for VSX scalar extract exponent quad
23108 (xsxsigqp): New insn for VSX scalar extract significand quad
23110 (xsiexpqpf): New insn for VSX scalar insert exponent quad
23111 precision with floating point argument.
23112 (xststdcqp): New expand for VSX scalar test data class quad
23114 (xststdcnegqp): New expand for VSX scalar test negative quad
23116 (xststdcqp): New insn to match expansions for VSX scalar test data
23117 class quad precision and VSX scalar test negative quad precision.
23118 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
23119 special case operand checking to enforce that second operand of
23120 VSX scalar test data class with quad precision argument is a 7-bit
23122 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23123 prototypes and descriptions of __ieee128 versions of
23124 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
23125 scalar_test_data_class, and scalar_test_neg built-in functions.
23127 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23129 PR tree-optimization/81162
23130 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
23131 replace a negate with an add.
23133 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
23135 * doc/invoke.texi (arm/-mcpu): Document +crypto.
23137 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23139 * config/arm/arm-c.c (arm_cpu_builtins): Define
23140 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23142 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23144 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
23145 (armv8-r): Set ARM Cortex-R52 as default CPU.
23146 * config/arm/arm-tables.opt: Regenerate.
23147 * config/arm/arm-tune.md: Regenerate.
23148 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23150 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
23151 extension for -mcpu=cortex-r52.
23153 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23155 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
23156 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
23157 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
23158 (fp-armv8): Define it as FP_ARMv8 only.
23159 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
23160 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
23162 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
23163 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
23164 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
23165 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
23166 than TARGET_FPU_ARMV8.
23167 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
23168 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
23169 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
23171 * config/arm/neon.md (neon_vrint): Likewise.
23172 (neon_vcvt): Likewise.
23173 (neon_<fmaxmin_op><mode>): Likewise.
23174 (<fmaxmin><mode>3): Likewise.
23175 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
23176 * config/arm/predicates.md (arm_cond_move_operator): Check against
23177 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
23179 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
23181 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
23182 to top of function.
23184 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23186 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
23187 loop in comment with memset.
23189 2017-07-14 Martin Liska <mliska@suse.cz>
23191 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
23192 * dwarf2out.c (is_java): Remove the function.
23193 (output_pubname): Remove usage of the function.
23194 (lower_bound_default): Remove usage of DW_LANG_Java.
23195 (gen_compile_unit_die): Likewise.
23196 * gcc.c: Remove compiler defaults for .java and .zip files.
23197 * gimple-expr.c (remove_suffix): Change as there's no longer
23198 extension than 4-letter one.
23199 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
23200 (gimplify_save_expr): Likewise.
23201 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
23202 as it's possible even for other languages than Java.
23203 * langhooks.h (struct lang_hooks): Remove Java from a comment.
23204 * lto-opts.c (lto_write_options): Remove reference to Java.
23205 * opts.c (strip_off_ending): Update file extension handling.
23206 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
23207 * tree-eh.c (lower_resx): Likewise.
23208 * tree.c (free_lang_data_in_type): Remove dead code.
23209 (find_decls_types_r): Likewise.
23210 (build_common_builtin_nodes): Remove Java from a comment.
23211 (verify_type): Remove dead code.
23212 * varasm.c (assemble_external): Remove Java from a comment.
23214 2017-07-14 Martin Liska <mliska@suse.cz>
23216 * opts.c (finish_options): Add quotes.
23217 (common_handle_option): Likewise.
23219 2017-07-14 Martin Liska <mliska@suse.cz>
23221 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
23222 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
23223 Remove N_SO_PASCAL.
23224 * dwarf2out.c (lower_bound_default): Do not handle
23226 (gen_compile_unit_die): Likewise.
23227 * gcc.c: Remove default extension binding for GNU Pascal.
23228 * stmt.c: Remove Pascal language from a comment.
23229 * xcoffout.c: Likewise.
23231 2017-07-13 David Malcolm <dmalcolm@redhat.com>
23234 * diagnostic-show-locus.c (fixit_cmp): New function.
23235 (layout::layout): Sort m_fixit_hints.
23236 (column_range::column_range): Assert that the values are valid.
23237 (struct char_span): New struct.
23238 (correction::overwrite): New method.
23239 (struct source_line): New struct.
23240 (line_corrections::add_hint): Add assertions. Reimplement memcpy
23241 calls in terms of classes source_line and char_span, and
23242 correction::overwrite.
23243 (selftest::test_overlapped_fixit_printing_2): New function.
23244 (selftest::diagnostic_show_locus_c_tests): Call it.
23246 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
23248 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
23249 early if there is no lhs.
23251 2017-07-13 Martin Liska <mliska@suse.cz>
23253 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
23254 (gen_reference_type_die): Likewise.
23255 * stor-layout.c: Remove Pascal-related comment.
23257 2017-07-13 Martin Liska <mliska@suse.cz>
23259 * opts.c (finish_options): Add quotes to error messages.
23260 (parse_sanitizer_options): Likewise.
23262 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
23264 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
23266 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
23268 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
23270 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23272 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
23274 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
23276 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
23279 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
23280 provides the hardware capability bits, define the macro
23281 __BUILTIN_CPU_SUPPORTS__.
23282 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
23283 if GLIBC does not provide the hardware capability bits. Add a
23284 gcc_unreachable call if the built-in cpu function is neither
23285 __builtin_cpu_is nor __builtin_cpu_supports.
23286 (rs6000_get_function_versions_dispatcher): Change the warning
23287 that an old GLIBC is used which does not export the capability
23288 bits to be an error.
23289 * doc/extend.texi (target_clones attribute): Document the
23290 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
23291 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
23292 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
23293 the macros defined by GCC if the newer GLIBC is available.
23295 2017-07-12 Jeff Law <law@redhat.com>
23297 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
23298 remaining includes slightly.
23299 * config/riscv/riscv-builtins.c: Include profile-count.h.
23301 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23304 * config/avr/avr.c (avr_set_current_function): In diagnostic
23305 messages: Quote keywords and (parts of) identifiers.
23306 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
23309 2017-07-12 Carl Love <cel@us.ibm.com>
23311 * config/rs6000/rs6000-c.c: Add support for built-in functions
23312 vector bool char vec_revb (vector bool char);
23313 vector bool short vec_revb (vector short char);
23314 vector bool int vec_revb (vector bool int);
23315 vector bool long long vec_revb (vector bool long long);
23316 * doc/extend.texi: Update the built-in documentation file for the
23317 new built-in functions.
23319 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23321 * config/s390/s390.md: Remove movcc splitter.
23323 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23325 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
23326 load/store on condition.
23328 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23331 * config/avr/avr.c (avr_encode_section_info)
23332 [progmem && !TREE_READONLY]: Error if progmem object needs
23335 2017-07-11 Michael Collison <michael.collison@arm.com>
23337 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
23340 2017-07-11 Carl Love <cel@us.ibm.com>
23342 * config/rs6000/rs6000-c.c: Add support for builtins
23343 vector unsigned int vec_parity_lsbb (vector signed int);
23344 vector unsigned int vec_parity_lsbb (vector unsigned int);
23345 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
23346 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
23347 vector unsigned long long vec_parity_lsbb (vector signed long long);
23348 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
23349 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
23350 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
23351 * doc/extend.texi: Update the built-in documentation file for the
23352 new built-in functions.
23354 2017-07-11 David Malcolm <dmalcolm@redhat.com>
23356 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
23357 (layout::m_primary_loc): New field.
23358 (layout::layout): Initialize new field. Move location filtering
23359 logic from here to...
23360 (layout::maybe_add_location_range): ...this new method. Add
23361 support for filtering to just the lines already specified by other
23363 (layout::will_show_line_p): New method.
23364 (gcc_rich_location::add_location_if_nearby): New method.
23365 (selftest::test_add_location_if_nearby): New test function.
23366 (selftest::diagnostic_show_locus_c_tests): Call it.
23367 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
23370 2017-07-11 Tom de Vries <tom@codesourcery.com>
23372 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
23373 (bb_first_real_insn): New function.
23374 (nvptx_single): Add extra initialization of broadcasted condition
23377 2017-07-11 Nathan Sidwell <nathan@acm.org>
23379 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
23381 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
23383 * doc/extend.texi (AVR Function Attributes): Remove weblink to
23384 Binutils doc as TEXI will mess them up.
23385 * doc/invoke.texi (AVR Options): Same here.
23387 2017-07-11 Daniel Cederman <cederman@gaisler.com>
23389 * config/sparc/sparc.opt (mfix-ut700): New option.
23390 (mfix-gr712rc): Likewise.
23391 (sparc_fix_b2bst): New variable.
23392 * doc/invoke.texi (SPARC options): Document them.
23393 (ARM options): Fix warnings.
23394 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
23395 instructions to prevent sequences that can trigger the store-store
23396 errata for certain LEON3FT processors.
23397 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
23398 (sparc_option_override): Set sparc_fix_b2bst appropriately.
23399 * config/sparc/sparc.md (fix_b2bst): New attribute.
23400 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
23402 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
23405 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
23407 (*rsqrtsf2_sse): Ditto.
23409 (div<mode>3): Macroize insn from divdf3 and divsf3
23410 using MODEF mode iterator.
23412 2017-07-10 Martin Sebor <msebor@redhat.com>
23414 PR tree-optimization/80397
23415 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
23416 instead of testing for equality to INTEGER_TYPE.
23418 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
23420 * config.gcc: Remove uclibc from arc target spec.
23422 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
23424 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
23426 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23429 * lto-wrapper.c (remove_option): New function.
23430 (merge_and_complain): Merge PIC/PIE options more realistically.
23432 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23434 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
23438 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
23439 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
23440 * config.in: Regenerate.
23441 * configure: Regenerate.
23442 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
23443 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
23444 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
23445 (TARGET_GASISR_PROLOGUES): ...target mask.
23446 * common/config/avr/avr-common.c
23447 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
23448 Set -mgas-isr-prologues.
23449 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
23450 INSERT_PASS_BEFORE for it.
23451 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
23452 * config/avr/avr.c (avr_option_override)
23453 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
23454 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
23455 (avr_attribute_table) <no_gccisr>: Add new function attribute.
23456 (avr_set_current_function) <is_no_gccisr>: Init machine field.
23457 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
23459 (make_avr_pass_pre_proep): New function.
23460 (emit_push_sfr) <treg>: Add argument to function and use it
23461 instead of TMP_REG.
23462 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
23463 and set machine->gasisr.yes.
23464 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
23465 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
23466 __gcc_isr.n_pushed to .L__stack_usage.
23467 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
23468 (avr_asm_final_postscan_insn): ...this new static function.
23469 * config/avr/avr.h (machine_function)
23470 <is_no_gccisr, use_L__stack_usage>: New fields.
23471 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
23472 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
23473 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
23474 (gasisr, *gasisr): New expander and insn.
23475 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23476 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
23477 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
23479 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
23481 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
23484 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23486 Move jump-tables out of .text again.
23489 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
23490 (ASM_OUTPUT_ADDR_VEC): New function.
23491 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
23492 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
23493 INSN_ADDRESSes as asm comment.
23494 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
23495 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
23496 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
23497 * config/avr/avr.md (*tablejump): Adjust comment.
23498 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
23499 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
23501 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
23502 (avr_output_addr_vec): New proto.
23503 (avr_log_t) <insn_addresses>: New field.
23505 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
23508 * config/i386/i386.c (ix86_function_arg_advance): Set
23509 outgoing_args_on_stack to true if there are outgoing arguments
23511 (ix86_function_arg): Likewise.
23512 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
23513 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
23514 * config/i386/i386.h (machine_function): Add
23515 outgoing_args_on_stack.
23517 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
23519 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23520 supporting pthreds.
23521 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
23523 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
23525 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
23526 (REAL_H): Remove $(MACHMODE_H).
23527 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
23529 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
23530 $(MACHMODE_H) and double-int.h.
23531 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
23533 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
23536 2017-07-07 Andrew Pinski <apinski@cavium.com>
23538 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
23539 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
23541 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23543 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
23544 Add warning if GCC was not configured to link against a GLIBC that
23545 exports the hardware capability bits.
23546 (make_resolver_func): Make resolver function private and not a
23547 COMDAT function. Create the name with clone_function_name instead
23548 of make_unique_name.
23551 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
23552 correct operand in doing the split.
23554 2017-07-07 Carl Love <cel@us.ibm.com>
23556 * config/rs6000/rs6000-c: Add support for built-in function
23557 vector unsigned short vec_pack_to_short_fp32 (vector float,
23559 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
23560 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
23561 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
23562 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
23563 (convert_4f32_8i16): Add define_expand.
23564 * doc/extend.texi: Update the built-in documentation file for the
23565 new built-in function.
23567 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23569 * config/sparc/m8.md: New file.
23570 * config/sparc/sparc.md: Include m8.md.
23572 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23574 * config/sparc/sparc.opt: New option -mvis4b.
23575 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
23576 (sparc_option_override): Handle VIS4B.
23577 (enum sparc_builtins): Define
23578 SPARC_BUILTIN_DICTUNPACK{8,16,32},
23579 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
23580 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
23581 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
23582 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
23583 (check_constant_argument): New function.
23584 (sparc_vis_init_builtins): Define builtins
23585 __builtin_vis_dictunpack{8,16,32},
23586 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
23587 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
23588 __builtin_vis_fpcmpde{8,16,32}shl and
23589 __builtin_vis_fpcmpur{8,16,32}shl.
23590 (sparc_expand_builtin): Check that the constant operands to
23591 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
23592 constant and in range.
23593 * config/sparc/sparc-c.c (sparc_target_macros): Handle
23595 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
23596 (SPARC_IMM5_P): Likewise.
23597 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
23598 (enabled): Handle vis4b.
23599 (UNSPEC_DICTUNPACK): New unspec.
23600 (UNSPEC_FPCMPSHL): Likewise.
23601 (UNSPEC_FPUCMPSHL): Likewise.
23602 (UNSPEC_FPCMPDESHL): Likewise.
23603 (UNSPEC_FPCMPURSHL): Likewise.
23604 (cpu_feature): New CPU feature `vis4b'.
23605 (dictunpack{8,16,32}): New insns.
23606 (FPCSMODE): New mode iterator.
23607 (fpcscond): New code iterator.
23608 (fpcsucond): Likewise.
23609 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
23610 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
23611 (fpcmpde{8,16,32}{si,di}shl): Likewise.
23612 (fpcmpur{8,16,32}{si,di}shl): Likewise.
23613 * config/sparc/constraints.md: Define constraints `q' for unsigned
23614 2-bit integer constants and `t' for unsigned 5-bit integer
23616 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
23618 (imm5_operand_dictunpack16): Likewise.
23619 (imm5_operand_dictunpack32): Likewise.
23620 (imm2_operand): Likewise.
23621 * doc/invoke.texi (SPARC Options): Document -mvis4b.
23622 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
23623 ditunpack* and fpcmp*shl builtins.
23625 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23627 * config.gcc: Handle m8 in --with-{cpu,tune} options.
23628 * config.in: Add HAVE_AS_SPARC6 define.
23629 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
23631 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
23633 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
23634 (CPP_CPU_SPEC): Handle m8.
23635 (ASM_CPU_SPEC): Likewise.
23636 * config/sparc/sparc-opts.h (enum processor_type): Add
23638 * config/sparc/sparc.c (m8_costs): New struct.
23639 (sparc_option_override): Handle TARGET_CPU_m8.
23640 (sparc32_initialize_trampoline): Likewise.
23641 (sparc64_initialize_trampoline): Likewise.
23642 (sparc_issue_rate): Likewise.
23643 (sparc_register_move_cost): Likewise.
23644 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
23645 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
23646 (ASM_CPU64_DEFAULT_SPEC): Likewise.
23647 (CPP_CPU_SPEC): Handle M8.
23648 (ASM_CPU_SPEC): Likewise.
23649 (AS_M8_FLAG): Define.
23650 * config/sparc/sparc.md: Add m8 to the cpu attribute.
23651 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
23652 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
23654 * configure: Regenerate.
23655 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
23658 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23660 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
23662 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
23663 ("*movdi_insn_sp32"): Do not set v3pipe.
23664 ("*movsi_insn"): Likewise.
23665 ("*movdi_insn_sp64"): Likewise.
23666 ("*movsf_insn"): Likewise.
23667 ("*movdf_insn_sp32"): Likewise.
23668 ("*movdf_insn_sp64"): Likewise.
23669 ("*zero_extendsidi2_insn_sp64"): Likewise.
23670 ("*sign_extendsidi2_insn"): Likewise.
23671 ("*mov<VM32:mode>_insn"): Likewise.
23672 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23673 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23674 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23675 ("<vlop:code><VL:mode>3"): Likewise.
23676 ("*not_<vlop:code><VL:mode>3"): Likewise.
23677 ("*nand<VL:mode>_vis"): Likewise.
23678 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
23679 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
23680 ("one_cmpl<VL:mode>2"): Likewise.
23681 ("faligndata<VM64:mode>_vis"): Likewise.
23682 ("alignaddrsi_vis"): Likewise.
23683 ("alignaddrdi_vis"): Likweise.
23684 ("alignaddrlsi_vis"): Likewise.
23685 ("alignaddrldi_vis"): Likewise.
23686 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23687 ("bmaskdi_vis"): Likewise.
23688 ("bmasksi_vis"): Likewise.
23689 ("bshuffle<VM64:mode>_vis"): Likewise.
23690 ("cmask8<P:mode>_vis"): Likewise.
23691 ("cmask16<P:mode>_vis"): Likewise.
23692 ("cmask32<P:mode>_vis"): Likewise.
23693 ("pdistn<P:mode>_vis"): Likewise.
23694 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23696 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23698 * config/sparc/sparc.md ("subtype"): New insn attribute.
23699 ("*wrgsr_sp64"): Set insn subtype.
23700 ("*rdgsr_sp64"): Likewise.
23701 ("alignaddrsi_vis"): Likewise.
23702 ("alignaddrdi_vis"): Likewise.
23703 ("alignaddrlsi_vis"): Likewise.
23704 ("alignaddrldi_vis"): Likewise.
23705 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23706 ("fexpand_vis"): Likewise.
23707 ("fpmerge_vis"): Likewise.
23708 ("faligndata<VM64:mode>_vis"): Likewise.
23709 ("bshuffle<VM64:mode>_vis"): Likewise.
23710 ("cmask8<P:mode>_vis"): Likewise.
23711 ("cmask16<P:mode>_vis"): Likewise.
23712 ("cmask32<P:mode>_vis"): Likewise.
23713 ("fchksm16_vis"): Likewise.
23714 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
23715 ("fmean16_vis"): Likewise.
23716 ("fp<plusminus_insn>64_vis"): Likewise.
23717 ("<plusminus_insn>v8qi3"): Likewise.
23718 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23719 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
23720 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
23721 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
23722 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
23723 ("*movqi_insn"): Likewise.
23724 ("*movhi_insn"): Likewise.
23725 ("*movsi_insn"): Likewise.
23726 ("movsi_pic_gotdata_op"): Likewise.
23727 ("*movdi_insn_sp32"): Likewise.
23728 ("*movdi_insn_sp64"): Likewise.
23729 ("movdi_pic_gotdata_op"): Likewise.
23730 ("*movsf_insn"): Likewise.
23731 ("*movdf_insn_sp32"): Likewise.
23732 ("*movdf_insn_sp64"): Likewise.
23733 ("*zero_extendhisi2_insn"): Likewise.
23734 ("*zero_extendqihi2_insn"): Likewise.
23735 ("*zero_extendqisi2_insn"): Likewise.
23736 ("*zero_extendqidi2_insn"): Likewise.
23737 ("*zero_extendhidi2_insn"): Likewise.
23738 ("*zero_extendsidi2_insn_sp64"): Likewise.
23739 ("ldfsr"): Likewise.
23740 ("prefetch_64"): Likewise.
23741 ("prefetch_32"): Likewise.
23742 ("tie_ld32"): Likewise.
23743 ("tie_ld64"): Likewise.
23744 ("*tldo_ldub_sp32"): Likewise.
23745 ("*tldo_ldub1_sp32"): Likewise.
23746 ("*tldo_ldub2_sp32"): Likewise.
23747 ("*tldo_ldub_sp64"): Likewise.
23748 ("*tldo_ldub1_sp64"): Likewise.
23749 ("*tldo_ldub2_sp64"): Likewise.
23750 ("*tldo_ldub3_sp64"): Likewise.
23751 ("*tldo_lduh_sp32"): Likewise.
23752 ("*tldo_lduh1_sp32"): Likewise.
23753 ("*tldo_lduh_sp64"): Likewise.
23754 ("*tldo_lduh1_sp64"): Likewise.
23755 ("*tldo_lduh2_sp64"): Likewise.
23756 ("*tldo_lduw_sp32"): Likewise.
23757 ("*tldo_lduw_sp64"): Likewise.
23758 ("*tldo_lduw1_sp64"): Likewise.
23759 ("*tldo_ldx_sp64"): Likewise.
23760 ("*mov<VM32:mode>_insn"): Likewise.
23761 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23762 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23764 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23766 * config/sparc/sparc.md ("type"): New insn type viscmp.
23767 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
23769 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
23770 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
23771 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23772 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
23774 ("n7_vis_logical_11cycle"): Likewise.
23775 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
23776 * config/sparc/niagara2.md ("niag3_vis": Likewise.
23777 * config/sparc/niagara.md ("niag_vis"): Likewise.
23778 * config/sparc/ultra3.md ("us3_fga"): Likewise.
23779 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
23781 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23783 * config/sparc/sparc.md: New instruction type `bmask'.
23784 (bmaskdi_vis): Use the `bmask' type.
23785 (bmasksi_vis): Likewise.
23786 * config/sparc/ultra3.md (us3_array): Likewise.
23787 * config/sparc/niagara7.md (n7_array): Likewise.
23788 * config/sparc/niagara4.md (n4_array): Likewise.
23789 * config/sparc/niagara2.md (niag2_vis): Likewise.
23790 (niag3_vis): Likewise.
23791 * config/sparc/niagara.md (niag_vis): Likewise.
23793 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23795 * ipa-comdats.c: Remove optimize check from gate.
23796 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
23797 for functions not optimized.
23798 (ipa_fn_summary_read): Skip optimize check.
23799 (ipa_fn_summary_write): Likewise.
23800 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
23802 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
23804 (can_inline_edge_p): Check flag_pcc_struct_return for match.
23805 (check_callers): Give up on caller which is not optimized.
23806 (inline_small_functions): Likewise.
23807 (ipa_inline): Do not give up when not optimizing.
23808 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
23809 away unoptimizes cdtors.
23810 (whole_program_function_and_variable_visibility): Do
23811 ipa_discover_readonly_nonaddressable_vars in LTO mode.
23812 * ipa.c (process_references): Do not check optimize.
23813 (symbol_table::remove_unreachable_nodes): Update optimize check.
23814 (set_writeonly_bit): Update optimize check.
23815 (pass_ipa_cdtor_merge::gate): Do not check optimize.
23816 (pass_ipa_single_use::gate): Remove.
23818 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23820 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
23821 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
23822 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
23823 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
23824 permute_load, permute_store, adjust_extract, adjust_splat,
23825 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
23826 replace_swap_with_copy, dump_swap_insn_table,
23827 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
23828 recombine_lvx_pattern, recombine_stvx_pattern,
23829 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
23830 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
23831 to file rs6000-p8swap.c.
23832 * config/rs6000/rs6000-p8swap.c: New file.
23833 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
23834 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
23835 and rs6000*-*-* targets.
23837 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23839 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
23841 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23843 * lto-wrapper.c (merge_and_complain): Do not merge
23844 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
23845 fsigned_zeros, ftrapping_math, fwrapv.
23846 (append_compiler_options): Do not track these options.
23847 (append_linker_options): Likewie
23849 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23851 * cgraphunit.c (cgraph_node::finalize_function): When
23852 !flag_toplevel_reorde set no_reorder flag.
23853 (varpool_node::finalize_decl): Likewise.
23854 (symbol_table::compile): Drop no toplevel reorder path.
23856 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23858 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
23859 edges; zero probability is not better than uninitialized.
23861 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23863 * asan.h (asan_sanitize_allocas_p): Declare.
23864 * asan.c (asan_sanitize_allocas_p): New function.
23865 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
23866 (handle_builtin_alloca): Likewise.
23867 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
23868 if !asan_sanitize_allocas_p.
23869 * params.def (asan-instrument-allocas): Add new option.
23870 * params.h (ASAN_PROTECT_ALLOCAS): Define.
23871 * opts.c (common_handle_option): Disable allocas sanitization for
23874 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23876 * asan.c: Include gimple-fold.h.
23877 (get_last_alloca_addr): New function.
23878 (handle_builtin_stackrestore): Likewise.
23879 (handle_builtin_alloca): Likewise.
23880 (asan_emit_allocas_unpoison): Likewise.
23881 (get_mem_refs_of_builtin_call): Add new parameter, remove const
23882 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
23883 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
23884 (instrument_builtin_call): Pass gimple iterator to
23885 get_mem_refs_of_builtin_call.
23886 (last_alloca_addr): New global.
23887 * asan.h (asan_emit_allocas_unpoison): Declare.
23888 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
23889 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
23890 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
23891 if function calls alloca.
23892 * gimple-fold.c (replace_call_with_value): Remove static keyword.
23893 * gimple-fold.h (replace_call_with_value): Declare.
23894 * internal-fn.c: Include asan.h.
23895 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
23896 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
23898 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23900 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
23901 (C_SELFTEST_FLAGS): New.
23902 (CPP_SELFTEST_FLAGS): New.
23903 (SELFTEST_DEPS): New, from deps of s-selftest.
23904 (C_SELFTEST_DEPS): New, from deps of s-selftest.
23905 (CPP_SELFTEST_DEPS): New.
23906 (selftest): Add dependency on s-selftest-c++.
23907 (s-selftest): Rename to...
23908 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
23909 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
23910 than SELFTEST_FLAGS.
23911 (selftest-gdb): Rename to...
23912 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
23914 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
23915 (selftest-valgrind): Rename to...
23916 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
23918 (selftest-valgrind): Reintroduce as an alias for
23919 selftest-c-valgrind.
23920 (s-selftest-c++): New.
23921 (selftest-c++-gdb): New.
23922 (selftest-c++-valgrind): New.
23924 2017-07-06 Olivier Hainque <hainque@adacore.com>
23926 * gcc.c (process_command): When deciding if undefined variables
23927 should be ignored when processing specs, accept "gcc -v" as well.
23929 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23931 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
23932 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
23934 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
23936 * config/arm/arm-cpus.in (armv8-r): Add new entry.
23937 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
23938 * config/arm/arm-tables.opt: Regenerate.
23939 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
23941 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
23943 2017-07-06 Carl Love <cel@us.ibm.com>
23945 * ChangeLog: Clean up from mid air collision
23947 2017-07-06 Carl Love <cel@us.ibm.com>
23949 * config/rs6000/rs6000-c.c: Add support for built-in functions
23950 vector signed int vec_subc (vector signed int, vector signed int);
23951 vector signed __int128 vec_subc (vector signed __int128,
23952 vector signed __int128);
23953 vector unsigned __int128 vec_subc (vector unsigned __int128,
23954 vector unsigned __int128);
23955 vector signed int vec_sube (vector signed int, vector signed int,
23956 vector signed int);
23957 vector unsigned int vec_sube (vector unsigned int,
23958 vector unsigned int,
23959 vector unsigned int);
23960 vector signed __int128 vec_sube (vector signed __int128,
23961 vector signed __int128,
23962 vector signed__int128);
23963 vector unsigned __int128 vec_sube (vector unsigned __int128,
23964 vector unsigned __int128,
23965 vector unsigned __int128);
23966 vector signed int vec_subec (vector signed int, vector signed int,
23967 vector signed int);
23968 vector unsigned int vec_subec (vector unsigned int,
23969 vector unsigned int,
23970 vector unsigned int);
23971 vector signed __int128 vec_subec (vector signed __int128,
23972 vector signed __int128,
23973 vector signed__int128);
23974 vector unsigned __int128 vec_subec (vector unsigned __int128,
23975 vector unsigned __int128,
23976 vector unsigned __int128);
23977 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
23978 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
23979 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
23980 BU_ALTIVEC_OVERLOAD_X definitions.
23981 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
23982 * doc/extend.texi: Update the built-in documentation file for the new
23983 built-in functions.
23985 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23988 * diagnostic-show-locus.c (layout::layout): Use start and finish
23989 spelling location for the start and finish of each range.
23990 * genmatch.c (linemap_client_expand_location_to_spelling_point):
23991 Add unused aspect param.
23992 * input.c (expand_location_1): Add "aspect" param, and use it
23993 to access the correct part of the location.
23994 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
23996 (expand_location_to_spelling_point): Likewise.
23997 (linemap_client_expand_location_to_spelling_point): Add "aspect"
23998 param, and pass it to expand_location_1.
24000 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
24002 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
24003 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
24004 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
24005 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
24006 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
24007 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
24008 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
24009 _mm_maskz_getmant_ss): New intrinsics.
24010 (__builtin_ia32_getexpss128_mask): Changed to ...
24011 __builtin_ia32_getexpss128_round ... this.
24012 (__builtin_ia32_getexpsd128_mask): Changed to ...
24013 __builtin_ia32_getexpsd128_round ... this.
24014 * config/i386/i386-builtin-types.def
24015 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
24016 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
24017 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
24018 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
24019 __builtin_ia32_getmantss_mask_round): New builtins.
24020 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
24021 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
24022 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
24023 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
24024 * config/i386/sse.md
24025 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
24026 avx512f_sgetexp<mode><mask_scalar_name>
24027 <round_saeonly_scalar_name> ... this.
24028 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
24029 %0, %1, %2<round_saeonly_op3>}): Changed to ...
24030 vgetexp<ssescalarmodesuffix>
24031 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24032 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
24033 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
24034 avx512f_vgetmant<mode><mask_scalar_name>
24035 <round_saeonly_scalar_name> ... this.
24036 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24037 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
24038 vgetmant<ssescalarmodesuffix>
24039 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
24040 %0<mask_scalar_operand4>, %1, %2
24041 <round_saeonly_scalar_mask_op4>, %3} ... this.
24042 * config/i386/subst.md (mask_scalar_operand4,
24043 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
24044 round_saeonly_scalar_nimm_predicate): New subst attributes.
24046 2017-07-06 Julia Koval <julia.koval@intel.com>
24048 * config/i386/i386.c (ix86_erase_embedded_rounding):
24049 Remove code for old rounding pattern.
24051 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
24053 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
24055 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
24057 * doc/sourcebuild.texi (Test Directives, Variants of
24058 dg-require-support): Add documentation for dg-require-stack-check.
24060 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
24062 * config/i386/subst.md (mask_scalar, round_scalar,
24063 round_saeonly_scalar): New meta-templates.
24064 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
24065 round_scalar_mask_operand3, round_scalar_mask_op3,
24066 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
24067 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
24068 round_saeonly_scalar_constraint,
24069 round_saeonly_scalar_prefix): New subst attribute.
24070 * config/i386/sse.md
24071 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
24072 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
24073 <round_scalar_name> ... this.
24074 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
24075 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
24076 <round_scalar_name> ... this.
24077 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
24078 <sse>_vm<code><mode>3<mask_scalar_name>
24079 <round_saeonly_scalar_name> ... this.
24080 (v<plusminus_mnemonic><ssescalarmodesuffix>
24081 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24082 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24083 v<plusminus_mnemonic><ssescalarmodesuffix>
24084 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24085 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24086 (v<multdiv_mnemonic><ssescalarmodesuffix>
24087 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24088 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24089 v<multdiv_mnemonic><ssescalarmodesuffix>
24090 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24091 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24092 (v<maxmin_float><ssescalarmodesuffix>
24093 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
24094 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
24095 v<maxmin_float><ssescalarmodesuffix>
24096 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24097 %0<mask_scalar_operand3>, %1, %<iptr>2
24098 <round_saeonly_scalar_mask_op3>} ... this.
24100 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
24102 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
24103 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
24105 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
24106 Alan Hayward <alan.hayward@arm.com>
24107 David Sherwood <david.sherwood@arm.com>
24109 * combine.c (simplify_if_then_else): Remove "enum" before
24111 * compare-elim.c (can_eliminate_compare): Likewise.
24112 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
24114 (aarch64_lookup_simd_builtin_type): Likewise.
24115 (aarch64_simd_builtin_type): Likewise.
24116 (aarch64_init_simd_builtin_types): Likewise.
24117 (aarch64_simd_expand_args): Likewise.
24118 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
24120 (aarch64_reverse_mask): Likewise.
24121 (aarch64_simd_emit_reg_reg_move): Likewise.
24122 (aarch64_gen_adjusted_ldpstp): Likewise.
24123 (aarch64_ccmp_mode_to_code): Likewise.
24124 (aarch64_operands_ok_for_ldpstp): Likewise.
24125 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24126 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
24128 (aarch64_min_divisions_for_recip_mul): Likewise.
24129 (aarch64_reassociation_width): Likewise.
24130 (aarch64_get_condition_code_1): Likewise.
24131 (aarch64_simd_emit_reg_reg_move): Likewise.
24132 (aarch64_simd_attr_length_rglist): Likewise.
24133 (aarch64_reverse_mask): Likewise.
24134 (aarch64_operands_ok_for_ldpstp): Likewise.
24135 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24136 (aarch64_gen_adjusted_ldpstp): Likewise.
24137 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
24139 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
24140 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
24141 (arm_lookup_simd_builtin_type): Likewise.
24142 (arm_simd_builtin_type): Likewise.
24143 (arm_init_simd_builtin_types): Likewise.
24144 (arm_expand_builtin_args): Likewise.
24145 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
24146 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
24147 (ft32_setup_incoming_varargs): Likewise.
24148 (ft32_function_arg): Likewise.
24149 (ft32_function_arg_advance): Likewise.
24150 (ft32_pass_by_reference): Likewise.
24151 (ft32_arg_partial_bytes): Likewise.
24152 (ft32_valid_pointer_mode): Likewise.
24153 (ft32_addr_space_pointer_mode): Likewise.
24154 (ft32_addr_space_legitimate_address_p): Likewise.
24155 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
24157 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
24158 (ix86_emit_outlined_ms2sysv_restore): Likewise.
24159 (iamcu_alignment): Likewise.
24160 (canonicalize_vector_int_perm): Likewise.
24161 (ix86_noce_conversion_profitable_p): Likewise.
24162 (ix86_mpx_bound_mode): Likewise.
24163 (ix86_operands_ok_for_move_multiple): Likewise.
24164 * config/microblaze/microblaze-protos.h
24165 (microblaze_expand_conditional_branch_reg): Likewise.
24166 * config/microblaze/microblaze.c
24167 (microblaze_expand_conditional_branch_reg): Likewise.
24168 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
24170 (rs6000_reassociation_width): Likewise.
24171 (rs6000_invalid_binary_op): Likewise.
24172 (fusion_p9_p): Likewise.
24173 (emit_fusion_p9_load): Likewise.
24174 (emit_fusion_p9_store): Likewise.
24175 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
24177 (riscv_hard_regno_mode_ok_p): Likewise.
24178 (riscv_address_insns): Likewise.
24179 (riscv_split_symbol): Likewise.
24180 (riscv_legitimize_move): Likewise.
24181 (riscv_function_value): Likewise.
24182 (riscv_hard_regno_nregs): Likewise.
24183 (riscv_expand_builtin): Likewise.
24184 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
24185 (riscv_build_integer): Likewise.
24186 (riscv_split_integer): Likewise.
24187 (riscv_legitimate_constant_p): Likewise.
24188 (riscv_cannot_force_const_mem): Likewise.
24189 (riscv_regno_mode_ok_for_base_p): Likewise.
24190 (riscv_valid_base_register_p): Likewise.
24191 (riscv_valid_offset_p): Likewise.
24192 (riscv_valid_lo_sum_p): Likewise.
24193 (riscv_classify_address): Likewise.
24194 (riscv_legitimate_address_p): Likewise.
24195 (riscv_address_insns): Likewise.
24196 (riscv_load_store_insns): Likewise.
24197 (riscv_force_binary): Likewise.
24198 (riscv_split_symbol): Likewise.
24199 (riscv_force_address): Likewise.
24200 (riscv_legitimize_address): Likewise.
24201 (riscv_move_integer): Likewise.
24202 (riscv_legitimize_const_move): Likewise.
24203 (riscv_legitimize_move): Likewise.
24204 (riscv_address_cost): Likewise.
24205 (riscv_subword): Likewise.
24206 (riscv_output_move): Likewise.
24207 (riscv_canonicalize_int_order_test): Likewise.
24208 (riscv_emit_int_order_test): Likewise.
24209 (riscv_function_arg_boundary): Likewise.
24210 (riscv_pass_mode_in_fpr_p): Likewise.
24211 (riscv_pass_fpr_single): Likewise.
24212 (riscv_pass_fpr_pair): Likewise.
24213 (riscv_get_arg_info): Likewise.
24214 (riscv_function_arg): Likewise.
24215 (riscv_function_arg_advance): Likewise.
24216 (riscv_arg_partial_bytes): Likewise.
24217 (riscv_function_value): Likewise.
24218 (riscv_pass_by_reference): Likewise.
24219 (riscv_setup_incoming_varargs): Likewise.
24220 (riscv_print_operand): Likewise.
24221 (riscv_elf_select_rtx_section): Likewise.
24222 (riscv_save_restore_reg): Likewise.
24223 (riscv_for_each_saved_reg): Likewise.
24224 (riscv_register_move_cost): Likewise.
24225 (riscv_hard_regno_mode_ok_p): Likewise.
24226 (riscv_hard_regno_nregs): Likewise.
24227 (riscv_class_max_nregs): Likewise.
24228 (riscv_memory_move_cost): Likewise.
24229 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
24230 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
24231 (rl78_addr_space_address_mode): Likewise.
24232 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
24234 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
24235 (rs6000_reassociation_width): Likewise.
24236 (rs6000_invalid_binary_op): Likewise.
24237 (fusion_p9_p): Likewise.
24238 (emit_fusion_p9_load): Likewise.
24239 (emit_fusion_p9_store): Likewise.
24240 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
24241 (ok_for_simple_move_operands): Likewise.
24242 (ok_for_simple_move_strict_operands): Likewise.
24243 (ok_for_simple_arith_logic_operands): Likewise.
24244 (visium_legitimize_reload_address): Likewise.
24245 (visium_select_cc_mode): Likewise.
24246 (output_cbranch): Likewise.
24247 (visium_split_double_move): Likewise.
24248 (visium_expand_copysign): Likewise.
24249 (visium_expand_int_cstore): Likewise.
24250 (visium_expand_fp_cstore): Likewise.
24251 * config/visium/visium.c (visium_pass_by_reference): Likewise.
24252 (visium_function_arg): Likewise.
24253 (visium_function_arg_advance): Likewise.
24254 (visium_libcall_value): Likewise.
24255 (visium_setup_incoming_varargs): Likewise.
24256 (visium_legitimate_constant_p): Likewise.
24257 (visium_legitimate_address_p): Likewise.
24258 (visium_legitimize_address): Likewise.
24259 (visium_secondary_reload): Likewise.
24260 (visium_register_move_cost): Likewise.
24261 (visium_memory_move_cost): Likewise.
24262 (prepare_move_operands): Likewise.
24263 (ok_for_simple_move_operands): Likewise.
24264 (ok_for_simple_move_strict_operands): Likewise.
24265 (ok_for_simple_arith_logic_operands): Likewise.
24266 (visium_function_value_1): Likewise.
24267 (rtx_ok_for_offset_p): Likewise.
24268 (visium_legitimize_reload_address): Likewise.
24269 (visium_split_double_move): Likewise.
24270 (visium_expand_copysign): Likewise.
24271 (visium_expand_int_cstore): Likewise.
24272 (visium_expand_fp_cstore): Likewise.
24273 (visium_split_cstore): Likewise.
24274 (visium_select_cc_mode): Likewise.
24275 (visium_split_cbranch): Likewise.
24276 (output_cbranch): Likewise.
24277 (visium_print_operand_address): Likewise.
24278 * expmed.c (flip_storage_order): Likewise.
24279 * expmed.h (emit_cstore): Likewise.
24280 (flip_storage_order): Likewise.
24281 * genrecog.c (validate_pattern): Likewise.
24282 * hsa-gen.c (gen_hsa_addr): Likewise.
24283 * internal-fn.c (expand_arith_overflow): Likewise.
24284 * ira-color.c (allocno_copy_cost_saving): Likewise.
24285 * lra-assigns.c (find_hard_regno_for_1): Likewise.
24286 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
24287 (process_invariant_for_inheritance): Likewise.
24288 * lra-eliminations.c (move_plus_up): Likewise.
24289 * omp-low.c (lower_oacc_reductions): Likewise.
24290 * simplify-rtx.c (simplify_subreg): Likewise.
24291 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
24292 (TARGET_CHKP_BOUND_MODE): Likewise..
24293 * targhooks.c (default_chkp_bound_mode): Likewise.
24294 (default_setup_incoming_vararg_bounds): Likewise.
24295 * targhooks.h (default_chkp_bound_mode): Likewise.
24296 (default_setup_incoming_vararg_bounds): Likewise.
24297 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
24298 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
24299 (have_whole_vector_shift): Likewise.
24300 * tree-vect-stmts.c (vectorizable_load): Likewise.
24301 * doc/tm.texi: Regenerate.
24303 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24305 Graceful degrade if Binutils PR21472 is not available.
24308 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
24309 .rodata in flash test fails.
24310 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
24311 * confgure: Regenerate.
24312 * config.in: Regenerate.
24313 * config/avr/avr.c (avr_asm_named_section)
24314 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
24315 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
24316 (avr_asm_init_sections): Same.
24318 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24320 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
24321 (fma<VH:mode>4_intrinsic): Likewise.
24322 (*fmsub<VCVTF:mode>4): Likewise.
24323 (*fmsub<VH:mode>4_intrinsic): Likewise.
24325 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24328 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
24329 Don't depend on "optimize > 0".
24330 (out_movhi_r_mr, out_movqi_mr_r): Same.
24331 (out_movhi_mr_r, out_movqi_r_mr): Same.
24332 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
24333 io_address_operand on "optimize > 0".
24335 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24337 * tree-loop-distribution.c: Add general explanantion on the pass.
24338 (generate_loops_for_partition): Mark distributed loop.
24339 (pg_add_dependence_edges): New parameter. Handle alias data
24340 dependence specially and record it in the parameter if asked.
24341 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
24342 (init_partition_graph_vertices, add_partition_graph_edge): New.
24343 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
24344 (free_partition_graph_vdata, build_partition_graph): New.
24345 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
24346 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
24347 (data_ref_segment_size, latch_dominated_by_data_ref): New.
24348 (compute_alias_check_pairs, version_loop_by_alias_check): New.
24349 (version_for_distribution_p, finalize_partitions): New.
24350 (distribute_loop): Handle alias data dependence specially. Factor
24351 out loop fusion code as functions and call these functions.
24353 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24355 * tree-loop-distribution.c (classify_partition): New parameter and
24356 better handle reduction statement.
24357 (rdg_build_partitions): Revise comment.
24358 (distribute_loop): Compute statements in all partitions and pass it
24359 to classify_partition.
24361 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24363 * tree-loop-distribution.c (enum partition_type): New.
24364 (struct partition): New field type.
24365 (partition_merge_into): Add parameter. Update partition type.
24366 (data_dep_in_cycle_p, update_type_for_merge): New functions.
24367 (build_rdg_partition_for_vertex): Compute partition type.
24368 (rdg_build_partitions): Dump partition type.
24369 (distribute_loop): Update calls to partition_merge_into.
24371 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24373 * tree-loop-distribution.c (struct ddr_hasher): New.
24374 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
24376 (classify_partition): Call get_data_dependence.
24377 (pg_add_dependence_edges): Ditto.
24378 (distribute_loop): Release data dependence hash table.
24380 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24382 * tree-loop-distribution.c (ref_base_address): Delete.
24383 (similar_memory_accesses): Rename ...
24384 (share_memory_accesses): ... to this. Check if partitions access
24385 the same memory reference.
24386 (distribute_loop): Call share_memory_accesses.
24388 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24390 * tree-loop-distribution.c (struct partition): New field recording
24391 its data reference.
24392 (partition_alloc, partition_free): Init and release data refs.
24393 (partition_merge_into): Merge data refs.
24394 (build_rdg_partition_for_vertex): Collect data refs for partition.
24395 (pg_add_dependence_edges): Change parameters from vector to bitmap.
24397 (distribute_loop): Remve data refs from vertice data of partition
24400 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24402 * tree-loop-distribution.c (params.h): Include header file.
24403 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
24404 (datarefs_vec): New global var.
24405 (create_rdg_vertices): Use datarefs_vec directly.
24406 (free_rdg): Don't free data references.
24407 (build_rdg): Update use. Don't free data references.
24408 (distribute_loop): Compute global variable for data references.
24409 Bail out if there are too many data references.
24411 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24413 * tree-loop-distribution.c (loop_nest): New global var.
24414 (build_rdg): Use loop directly, rather than loop nest.
24415 (pg_add_dependence_edges): Remove loop nest parameter. Use global
24417 (distribute_loop): Compute global variable loop nest. Update use.
24419 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24421 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
24422 (partition_merge_into): New parameter. Dump reason for fusion.
24423 (distribute_loop): Update use of partition_merge_into.
24425 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24427 * tree-loop-distribution.c (bb_top_order_index): New.
24428 (bb_top_order_index_size, bb_top_order_cmp): New.
24429 (stmts_from_loop): Use topological order.
24430 (pass_loop_distribution::execute): Compute and release topological
24431 order for basic blocks.
24433 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24435 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
24438 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24440 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
24441 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
24442 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
24443 * internal-fn.def (LOOP_DIST_ALIAS): New.
24444 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
24445 (fold_loop_internal_call): ... this.
24446 (vect_loop_dist_alias_call): New function.
24447 (set_uid_loop_bbs): Call fold_loop_internal_call.
24448 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
24451 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24454 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
24455 Require dead FLAGS_REG at the beginning of a peephole.
24457 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24460 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
24461 arguments in the call to __builtin_ia32_sbb_u32.
24462 (_subborrow_u64): Swap _X and _Y arguments in the call to
24463 __builtin_ia32_sbb_u64.
24465 2017-07-04 Jakub Jelinek <jakub@redhat.com>
24468 * tree-vrp.c (compare_assert_loc): Turn into a function template
24469 with stable template parameter. Only test if a->e is NULL,
24470 !a->e == !b->e has been verified already. Use e == NULL or
24471 e != NULL instead of e or ! e tests. If stable is true, don't use
24472 iterative_hash_expr, on the other side allow a or b or both NULL
24473 and sort the NULLs last.
24474 (process_assert_insertions): Sort using compare_assert_loc<false>
24475 instead of compare_assert_loc, later sort using
24476 compare_assert_loc<true> before calling process_assert_insertions_for
24477 in a loop. Use break instead of continue once seen NULL pointer.
24479 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24481 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24482 Cortex-R7 and Cortex-R8 processors.
24484 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24486 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
24487 uninitialized while src is not.
24489 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
24491 * common/config/arm/arm-common.c: Adjust include path for
24493 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
24494 (arm-cpu.h): Create in build directory. Adjust dependency rules.
24495 (arm-cpu-data.h): Likewise.
24496 (arm-cpu-cdata.h): Likewise.
24497 * config/arm/arm-cpu.h: Delete.
24498 * config/arm/arm-cpu-cdata.h: Delete.
24499 * config/arm/arm-cpu-data.h: Delete.
24501 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
24503 * config/arm/arm-cpus.in (cortex-a55): New.
24504 (cortex-a75): Likewise.
24505 (cortex-a75.cortex-a55): Likewise.
24506 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
24508 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
24509 * config/arm/arm-cpu-cdata.h: Regenerate.
24510 * config/arm/arm-cpu-data.h: Regenerate.
24511 * config/arm/arm-cpu.h: Regenerate.
24512 * config/arm/arm-tables.opt: Regenerate.
24513 * config/arm/arm-tune.md: Regenerate.
24515 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24517 * haifa-sched.c (sched_create_recovery_edges): Update profile.
24519 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24521 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
24524 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
24526 PR tree-optimization/81292
24527 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
24528 full_string_p, also call adjust_related_strinfos if the adjustment
24529 is simple, otherwise invalidate related strinfos.
24531 2017-07-04 Martin Liska <mliska@suse.cz>
24534 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
24535 newly created variable as DECL_IGNORED_P.
24537 2017-07-04 Martin Liska <mliska@suse.cz>
24540 * ipa-inline.c (inline_small_functions):
24541 Use xstrdup_for_dump.
24543 2017-07-04 Tom de Vries <tom@codesourcery.com>
24545 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
24547 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24550 * config/darwin.c (darwin_function_switched_text_sections):
24553 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
24555 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
24557 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24559 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
24561 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24563 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
24564 min_profitable_iters, and th as inclusive lower bounds.
24565 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
24566 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
24567 for min_profitable_iters and min_profitable_estimate.
24568 (vect_transform_loop): Treat th as an inclusive lower bound.
24569 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24571 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24574 * config/darwin.c (darwin_function_switched_text_sections):
24575 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
24576 in two pieces, and suppress the use of buf.
24578 2017-07-03 Nathan Sidwell <nathan@acm.org>
24580 * hash-table.h (hash_table_mod1): Fix indentation.
24582 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24584 PR middle-end/81290
24585 * predict.c (force_edge_cold): Be more careful about propagation
24587 * profile-count.h (profile_probability::guessed,
24588 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
24590 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
24592 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
24594 * doc/invoke.texi (rcpc architecture extension): Document it.
24596 2017-07-03 Richard Biener <rguenther@suse.de>
24598 PR tree-optimization/60510
24599 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
24600 the scalar reduction PHI and use it.
24601 (vectorizable_reduction): Properly guard the single_defuse_cycle
24602 path for non-SLP reduction chains where we cannot use it.
24603 Rework reduc_def/index and vector type deduction. Rework
24604 vector operand gathering during reduction op code-gen.
24605 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
24606 chains dissolve the chain and leave it to non-SLP reduction
24609 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24611 * tree-data-ref.h (dr_alignment): Declare.
24612 * tree-data-ref.c (dr_alignment): New function.
24613 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
24614 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
24616 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
24618 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24620 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
24621 and base_misalignment fields.
24622 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
24623 * tree-data-ref.c: Include builtins.h.
24624 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
24625 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
24626 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
24627 * tree-vect-data-refs.c: Include tree-cfg.h.
24628 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
24629 fields instead of calculating an alignment here.
24630 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
24631 innermost_loop_behavior fields.
24633 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24635 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
24637 (DR_STEP_ALIGNMENT): New macro.
24638 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
24639 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
24640 (create_data_ref): Print it.
24641 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
24642 to tell whether the step preserves vector (mis)alignment.
24643 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24644 Move the check for an integer step and generalise to all INTEGER_CST.
24645 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
24646 Print the outer step alignment.
24648 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24650 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
24651 with offset_alignment.
24652 (DR_ALIGNED_TO): Delete.
24653 (DR_OFFSET_ALIGNMENT): New macro.
24654 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
24655 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
24656 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
24657 (create_data_ref): Likewise.
24658 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24659 (vect_analyze_data_refs): Likewise.
24660 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
24661 creating dummy innermost behavior.
24663 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24665 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
24666 with a "innermost_loop_behavior *" and refeence tree.
24667 * tree-data-ref.c (dr_analyze_innermost): Likewise.
24668 (create_data_ref): Update call accordingly.
24669 * tree-predcom.c (find_looparound_phi): Likewise.
24671 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24673 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
24674 fields with dr_wrt_vec_loop.
24675 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
24676 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
24677 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
24678 (vect_dr_behavior): New function.
24679 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24680 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
24681 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
24682 track whether the step preserves the misalignment.
24683 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24684 Use vect_dr_behavior.
24685 (vect_setup_realignment): Update call accordingly.
24686 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
24687 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
24688 call to vect_create_addr_base_for_vector_ref.
24689 (vect_create_cond_for_align_checks): Likewise.
24690 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
24691 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
24692 (vect_recog_mask_conversion_pattern): Likewise.
24693 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
24694 (new_stmt_vec_info): Remove redundant zeroing.
24696 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24698 * common/config/arm/arm-common.c (arm_be8_option): New function.
24699 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
24700 (ISA_ARMv6): Add isa_bit_be8.
24701 * config/arm/arm.h (arm_be8_option): Add prototype.
24702 (BE8_SPEC_FUNCTION): New define.
24703 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
24704 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
24705 (mlittle-endian): Similarly.
24706 (mbe8, mbe32): New options.
24707 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
24708 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
24710 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24712 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
24714 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24716 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
24717 (cleanup_tree_cfg_bb): Use it.
24718 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
24720 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
24722 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24725 * loop-doloop.c (add_test): Update profile.
24727 2017-07-03 Martin Liska <mliska@suse.cz>
24730 * sanopt.c (rewrite_usage_of_param): New function.
24731 (sanitize_rewrite_addressable_params): Likewise.
24732 (pass_sanopt::execute): Call rewrite_usage_of_param.
24734 2017-07-03 Richard Biener <rguenther@suse.de>
24736 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
24737 back to using VIEW_CONVERT_EXPR.
24739 2017-07-03 Martin Liska <mliska@suse.cz>
24742 * doc/extend.texi: Document when a resolver function is
24743 generated for target_clones.
24745 2017-07-03 Martin Liska <mliska@suse.cz>
24747 * asan.c (asan_emit_stack_protection): Unpoison just red zones
24748 and shadow memory of auto variables which are subject of
24749 use-after-scope sanitization.
24750 (asan_expand_mark_ifn): Add do set only when is_poison.
24752 2016-07-03 Richard Biener <rguenther@suse.de>
24754 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
24756 (vect_force_simple_reduction): Record reduction def -> phi mapping.
24757 (vectorizable_reduction): Perform reduction PHI creation when
24758 visiting a reduction PHI and adjust and simplify code generation
24759 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
24760 (vect_transform_loop): Visit reduction PHIs.
24761 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
24762 defs into the SLP tree.
24763 (vect_build_slp_tree): Reduction defs terminate the recursion.
24764 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
24766 (vect_get_vec_defs_for_stmt_copy): Export.
24767 (vect_get_vec_defs): Likewise.
24768 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
24770 (vect_get_vec_defs_for_stmt_copy): Declare.
24771 (vect_get_vec_defs): Likewise.
24773 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24775 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
24776 parameter with a "loop" parameter and use it instead of the
24777 loop containing DR_STMT. Don't check simple_iv when doing
24778 BB analysis. Describe the two analysis modes in the comment.
24780 2017-07-03 Tom de Vries <tom@codesourcery.com>
24782 PR tree-optimization/69468
24783 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
24784 (find_same_succ_bb): Handle ignore_edge_flags.
24786 2017-07-03 Tom de Vries <tom@codesourcery.com>
24788 PR tree-optimization/81192
24789 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
24791 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
24793 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
24795 2017-07-03 Tom de Vries <tom@codesourcery.com>
24797 PR tree-optimization/81192
24798 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
24799 BB_SAME_SUCC (bb) == NULL.
24801 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24803 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
24806 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24808 * dumpfile.c: Include profile-count.h
24809 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
24811 (insert_cond_bb): Update profile.
24812 * tree-cfg.h (insert_cond_bb): Update prototype.
24813 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
24814 * tree-dump.c: Do not include tree-cfg.
24816 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24818 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
24820 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24822 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
24825 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24827 * tree-complex.c (expand_complex_div_wide): update profile.
24829 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24830 Alan Hayward <alan.hayward@arm.com>
24831 David Sherwood <david.sherwood@arm.com>
24833 * Makefile.in (MACHMODE_H): Remove insn-modes.h
24834 (CORETYPES_H): New define.
24835 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
24836 (insn-modes-inline.h, s-modes-inline-h): New rules.
24837 (generated_files): Add insn-modes-inline.h.
24838 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
24839 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
24840 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
24841 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
24842 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
24843 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
24844 (build/gencodes.o, build/genconditions.o): Likewise.
24845 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
24846 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
24847 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
24848 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
24849 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
24850 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
24851 * coretypes.h: Include everything up to real.h for generators.
24852 Include insn-modes.h first. Include wide-int-print.h after
24853 wide-int.h. Include insn-modes-inline.h and then machmode.h.
24854 * machmode.h: Don't include insn-modes.h here.
24855 * function-tests.c: Remove includes of signop.h, machmode.h,
24856 double-int.h and wide-int.h.
24858 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
24860 * optc-save-gen.awk: Likewise.
24861 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
24862 * godump.c: Remove include of wide-int-print.h.
24863 * pretty-print.h: Likewise.
24864 * wide-int-print.cc: Likewise.
24865 * wide-int.cc: Likewise.
24866 * hash-map-tests.c: Remove include of signop.h.
24867 * hash-set-tests.c: Likewise.
24868 * rtl-tests.c: Likewise.
24869 * mkconfig.sh: Remove include of machmode.h.
24870 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
24872 (emit_insn_modes_inline_h): ...this new function. Emit the code
24873 into an insn-modes-inline.h header file, adding appropriate
24874 include guards and end comments.
24875 (emit_insn_modes_c_header): Remove include of machmode.h.
24876 (emit_min_insn_modes_c_header): Include coretypes.h rather than
24878 (main): Handle -i flag and call emit_insn_modes_inline_h when
24881 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24883 * tree-ssa-strlen.c (strinfo): Rename the length field to
24884 nonzero_chars. Add a full_string_p field.
24885 (compare_nonzero_chars, zero_length_string_p): New functions.
24886 (get_addr_stridx): Add an offset_out parameter.
24887 Use compare_nonzero_chars.
24888 (get_stridx): Update accordingly. Use compare_nonzero_chars.
24889 (new_strinfo): Update after above changes to strinfo.
24890 (set_endptr_and_length): Set full_string_p.
24891 (get_string_length): Update after above changes to strinfo.
24892 (unshare_strinfo): Update call to new_strinfo.
24893 (maybe_invalidate): Likewise.
24894 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
24895 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
24896 as a uhwi instead of an shwi. Update after above changes to
24897 strinfo and new_strinfo.
24898 (zero_length_string): Assert that chainsi contains full strings.
24899 Use zero_length_string_p. Update call to new_strinfo.
24900 (adjust_related_strinfos): Update after above changes to strinfo.
24901 Copy full_string_p from origsi.
24902 (adjust_last_stmt): Use zero_length_string_p.
24903 (handle_builtin_strlen): Update after above changes to strinfo and
24904 new_strinfo. Install the lhs as the string length if the previous
24905 entry didn't describe a full string.
24906 (handle_builtin_strchr): Update after above changes to strinfo
24908 (handle_builtin_strcpy): Likewise.
24909 (handle_builtin_strcat): Likewise.
24910 (handle_builtin_malloc): Likewise.
24911 (handle_pointer_plus): Likewise.
24912 (handle_builtin_memcpy): Likewise. Track nonzero characters
24913 that aren't necessarily followed by a nul terminator.
24914 (handle_char_store): Likewise.
24916 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24918 PR tree-optimization/80769
24919 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
24920 for malloc and calloc. Document the new invariant that all related
24921 strinfos have delayed lengths or none do.
24922 (verify_related_strinfos): Move earlier in file.
24923 (set_endptr_and_length): New function, split out from...
24924 (get_string_length): ...here. Also set the lengths of related
24926 (zero_length_string): Assert that chainsi has known (rather than
24928 (adjust_related_strinfos): Likewise.
24930 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24932 PR tree-optimization/81136
24933 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
24934 assert that two references with the same misalignment have the same
24935 compile-time misalignment if those compile-time misalignments
24938 2017-07-01 Andi Kleen <ak@linux.intel.com>
24940 * print-tree.c (print_node): Print all attributes.
24942 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24944 * cfg.c (scale_bbs_frequencies): New function.
24945 * cfg.h (scale_bbs_frequencies): Declare it.
24946 * cfgloopanal.c (single_likely_exit): Cleanup.
24947 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
24949 (scale_loop_profile): Likewise.
24950 (loop_version): Likewise.
24951 (create_empty_loop_on_edge): Update.
24952 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
24953 scale_loop_frequencies, scale_loop_profile, loopify,
24954 loop_version): Update prototypes.
24955 * modulo-sched.c (sms_schedule): Update.
24956 * predict.c (unlikely_executed_edge_p): Also check probability.
24957 (probably_never_executed_edge_p): Fix typo.
24958 * tree-if-conv.c (version_loop_for_if_conversion): Update.
24959 * tree-parloops.c (gen_parallel_loop): Update.
24960 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
24961 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24962 * tree-ssa-loop-split.c (split_loop): Update.
24963 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
24964 * tree-vect-loop-manip.c (vect_do_peeling): Update.
24965 (vect_loop_versioning): Update.
24966 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24968 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24970 * trans-mem.c (split_bb_make_tm_edge): Update profile.
24972 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24974 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
24975 to keep profile consistent.
24977 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24979 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
24980 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
24981 * profile-count.h (max_safe_multiplier): Make unsigned.
24982 (profile_count::guessed_zero): New.
24984 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24986 * bb-reorder.c (fix_up_crossing_landing_pad,
24987 fix_crossing_conditional_branches): Use make_single_succ_edge
24988 to keep profile consistent.
24990 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24992 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
24995 2017-07-01 Jakub Jelinek <jakub@redhat.com>
24998 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
24999 the right scopes, make sure cond_jump isn't preserved between multiple
25000 iterations. Search for fallthru edge whenever there are 3+ edges and
25001 use find_fallthru_edge for it.
25003 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25005 Patch by Alexander Monakov <amonakov@ispras.ru>
25006 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
25007 probabilities consistently.
25009 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25011 * pa.c (pa_expand_compare_and_swap_loop): Update call of
25012 emit_cmp_and_jump_insns.
25014 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25017 * tree-inline.c (expand_call_inline): Combine profile statuses.
25019 2017-06-30 Andrew Pinski <apinski@cavium.com>
25021 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
25022 fold_stmt returned true.
25024 2017-06-30 Nathan Sidwell <nathan@acm.org>
25026 * ggc.h (empty_string): Delete.
25027 * cfgexpand.c (expand_asm_stmt): Use plain "".
25028 * optabs.c (expand_asm_memory_barrier): Likewise.
25029 * stringpool.c (empty_string): Delete.
25030 (digit_vector, digit_string): Delete.
25031 (ggc_alloc_string): Use plain "", don't optimize single digit
25032 strings. Use ggc_alloc_atomic.
25034 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
25036 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
25037 comparison set and one other set, use the cost of the non-comparison
25040 2017-06-30 Nathan Sidwell <nathan@acm.org>
25042 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
25045 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
25047 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
25048 loops. Remove now unneeded calls to gimple_switch_set_label() that
25049 just set removed labels to NULL_TREE.
25051 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
25053 * tree-ssanames.c (set_range_info_raw): Abstract from ...
25054 (set_range_info): ...here. Only call set_range_info_raw if domain
25056 (set_nonzero_bits): Call set_range_info_raw.
25057 * tree-ssanames.h (set_range_info_raw): New.
25059 2017-06-30 Jakub Jelinek <jakub@redhat.com>
25062 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
25063 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
25064 of nonimmediate_operand and <store_mask_constraint> instead of m
25065 for the input operand. For V8FI iterator, always split if input
25066 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
25067 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
25068 <store_mask_predicate> instead of register_operand and
25069 <store_mask_constraint> instead of v for the input operand. Make
25070 sure both operands aren't MEMs for if not <mask_applied>.
25072 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
25074 * lto-wrapper.c (copy_file) Close both file descriptors before
25077 2017-06-30 Martin Liska <mliska@suse.cz>
25080 * multiple_target.c (create_dispatcher_calls): Make ifunc
25081 also for function that don't have calls or are not referenced.
25083 2017-06-30 Richard Biener <rguenther@suse.de>
25085 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
25086 analyze the first scalar stmt. Move vector type computation
25087 for the BB case here from ...
25088 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
25089 live operation processing in the SLP case properly.
25091 2017-06-30 Richard Biener <rguenther@suse.de>
25093 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
25095 2017-06-30 Martin Liska <mliska@suse.cz>
25098 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
25099 before BUILT_IN_UNWIND_RESUME when ASAN is used.
25101 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
25103 * doc/invoke.texi (AArch64): Add missing options and remove redundant
25106 2017-06-30 Richard Biener <rguenther@suse.de>
25108 PR tree-optimization/81249
25109 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
25110 condition reduction result to original scalar type.
25112 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25114 * profile-count.h (enum profile_quality): Fix typos and whitespace
25117 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25119 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
25120 type for branch probabilities.
25122 2017-06-29 Julian Brown <julian@codesourcery.com>
25123 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25125 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
25126 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
25127 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
25128 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
25130 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25132 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
25133 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
25134 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
25135 CC usage from generic code to here.
25136 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
25137 CC usage into the target macros.
25139 2017-06-29 Maya Rashish <coypu@sdf.org>
25141 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
25144 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25146 * arm/arm-builtins.c: Include profile-count.h
25147 * except.c (sjlj_emit_function_enter): Use
25148 profile_probability::unlikely.
25150 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25152 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
25153 and tocrel_offset be pointer args rather than implicitly using
25155 (legitimate_constant_pool_address_p, rs6000_emit_move,
25156 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
25157 tocrel_offset and use in toc_relative_expr_p call.
25158 (print_operand, print_operand_address): Use static tocrel_base_oac
25159 and tocrel_offset_oac.
25160 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
25163 2017-06-29 Maya Rashish <coypu@sdf.org>
25165 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
25167 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
25169 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
25170 objects, take into account only the alignment of 'op0' and 'mode1' if
25173 2017-06-29 Steve Ellcey <sellcey@cavium.com>
25175 * ccmp.c (ccmp_tree_comparison_p): New function.
25176 (ccmp_candidate_p): Update to use above function.
25177 (get_compare_parts): New function.
25178 (expand_ccmp_next): Update to use new functions.
25179 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
25181 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
25182 take mode as argument.
25183 * ccmp.h (expand_ccmp_expr): Add mode as argument.
25184 * expr.c (expand_expr_real_1): Pass mode as argument.
25186 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
25188 * combine.c (combine_instructions): Print insns to dump_file, together
25191 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25193 * asan.c (asan_emit_stack_protection): Update.
25194 (create_cond_insert_point): Update.
25195 * auto-profile.c (afdo_propagate_circuit): Update.
25196 * basic-block.h (struct edge_def): Turn probability to
25197 profile_probability.
25198 (EDGE_FREQUENCY): Update.
25199 * bb-reorder.c (find_traces_1_round): Update.
25200 (better_edge_p): Update.
25201 (sanitize_hot_paths): Update.
25202 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
25203 (make_single_succ_edge): Update.
25204 (check_bb_profile): Update.
25205 (dump_edge_info): Update.
25206 (update_bb_profile_for_threading): Update.
25207 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
25208 probabilitycount to 0.
25209 * cfgbuild.c (compute_outgoing_frequencies): Update.
25210 * cfgcleanup.c (try_forward_edges): Update.
25211 (outgoing_edges_match): Update.
25212 (try_crossjump_to_edge): Update.
25213 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
25214 (expand_gimple_tailcall): Update.
25215 (construct_init_block): Use make_single_succ_edge.
25216 (construct_exit_block): Use make_single_succ_edge.
25217 * cfghooks.c (verify_flow_info): Update.
25218 (redirect_edge_succ_nodup): Update.
25219 (split_edge): Update.
25220 (account_profile_record): Update.
25221 * cfgloopanal.c (single_likely_exit): Update.
25222 * cfgloopmanip.c (scale_loop_profile): Update.
25223 (set_zero_probability): Remove.
25224 (duplicate_loop_to_header_edge): Update.
25225 * cfgloopmanip.h (loop_version): Update prototype.
25226 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
25227 (force_nonfallthru_and_redirect): Update.
25228 (update_br_prob_note): Update.
25229 (rtl_verify_edges): Update.
25230 (purge_dead_edges): Update.
25231 (rtl_lv_add_condition_to_bb): Update.
25232 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
25233 * cgraphunit.c (init_lowered_empty_function): Update.
25234 (cgraph_node::expand_thunk): Update.
25235 * cilk-common.c: Include profile-count.h
25236 * dojump.c (inv): Remove.
25237 (jumpifnot): Update.
25238 (jumpifnot_1): Update.
25239 (do_jump_1): Update.
25241 (do_jump_by_parts_greater_rtx): Update.
25242 (do_compare_rtx_and_jump): Update.
25243 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
25244 do_jump_1. do_compare_rtx_and_jump): Update prototype.
25245 * dwarf2cfi.c: Include profile-count.h
25246 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
25247 (sjlj_emit_dispatch_table): Likewise.
25248 * explow.c: Include profile-count.h
25249 * expmed.c (emit_store_flag_force): Update.
25250 (do_cmp_and_jump): Update.
25251 * expr.c (compare_by_pieces_d::generate): Update.
25252 (compare_by_pieces_d::finish_mode): Update.
25253 (emit_block_move_via_loop): Update.
25254 (store_expr_with_bounds): Update.
25255 (store_constructor): Update.
25256 (expand_expr_real_2): Update.
25257 (expand_expr_real_1): Update.
25258 * expr.h (try_casesi, try_tablejump): Update prototypes.
25259 * gimple-pretty-print.c (dump_probability): Update.
25260 (dump_profile): New.
25261 (dump_gimple_label): Update.
25262 (dump_gimple_bb_header): Update.
25263 * graph.c (draw_cfg_node_succ_edges): Update.
25264 * hsa-gen.c (convert_switch_statements): Update.
25265 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
25266 (find_if_case_1): Update.
25267 (find_if_case_2): Update.
25268 * internal-fn.c (expand_arith_overflow_result_store): Update.
25269 (expand_addsub_overflow): Update.
25270 (expand_neg_overflow): Update.
25271 (expand_mul_overflow): Update.
25272 (expand_vector_ubsan_overflow): Update.
25273 * ipa-cp.c (good_cloning_opportunity_p): Update.
25274 * ipa-split.c (split_function): Use make_single_succ_edge.
25275 * ipa-utils.c (ipa_merge_profiles): Update.
25276 * loop-doloop.c (add_test): Update.
25277 (doloop_modify): Update.
25278 * loop-unroll.c (compare_and_jump_seq): Update.
25279 (unroll_loop_runtime_iterations): Update.
25280 * lra-constraints.c (lra_inheritance): Update.
25281 * lto-streamer-in.c (input_cfg): Update.
25282 * lto-streamer-out.c (output_cfg): Update.
25283 * mcf.c (adjust_cfg_counts): Update.
25284 * modulo-sched.c (sms_schedule): Update.
25285 * omp-expand.c (expand_omp_for_init_counts): Update.
25286 (extract_omp_for_update_vars): Update.
25287 (expand_omp_ordered_sink): Update.
25288 (expand_omp_for_ordered_loops): Update.
25289 (expand_omp_for_generic): Update.
25290 (expand_omp_for_static_nochunk): Update.
25291 (expand_omp_for_static_chunk): Update.
25292 (expand_cilk_for): Update.
25293 (expand_omp_simd): Update.
25294 (expand_omp_taskloop_for_outer): Update.
25295 (expand_omp_taskloop_for_inner): Update.
25296 * omp-simd-clone.c (simd_clone_adjust): Update.
25297 * optabs.c (expand_doubleword_shift): Update.
25298 (expand_abs): Update.
25299 (emit_cmp_and_jump_insn_1): Update.
25300 (expand_compare_and_swap_loop): Update.
25301 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
25302 * predict.c (predictable_edge_p): Update.
25303 (edge_probability_reliable_p): Update.
25304 (set_even_probabilities): Update.
25305 (combine_predictions_for_insn): Update.
25306 (combine_predictions_for_bb): Update.
25307 (propagate_freq): Update.
25308 (estimate_bb_frequencies): Update.
25309 (force_edge_cold): Update.
25310 * profile-count.c (profile_count::dump): Add missing space into dump.
25311 (profile_count::debug): Add newline.
25312 (profile_count::differs_from_p): Explicitly convert to unsigned.
25313 (profile_count::stream_in): Update.
25314 (profile_probability::dump): New member function.
25315 (profile_probability::debug): New member function.
25316 (profile_probability::differs_from_p): New member function.
25317 (profile_probability::differs_lot_from_p): New member function.
25318 (profile_probability::stream_in): New member function.
25319 (profile_probability::stream_out): New member function.
25320 * profile-count.h (profile_count_quality): Rename to ...
25321 (profile_quality): ... this one.
25322 (profile_probability): New.
25323 (profile_count): Update.
25324 * profile.c (compute_branch_probabilities): Update.
25325 * recog.c (peep2_attempt): Update.
25326 * sched-ebb.c (schedule_ebbs): Update.
25327 * sched-rgn.c (find_single_block_region): Update.
25328 (compute_dom_prob_ps): Update.
25329 (schedule_region): Update.
25330 * sel-sched-ir.c (compute_succs_info): Update.
25331 * stmt.c (struct case_node): Update.
25332 (do_jump_if_equal): Update.
25333 (get_outgoing_edge_probs): Update.
25334 (conditional_probability): Update.
25335 (emit_case_dispatch_table): Update.
25336 (expand_case): Update.
25337 (expand_sjlj_dispatch_table): Update.
25338 (emit_case_nodes): Update.
25339 * targhooks.c: Update.
25340 * tracer.c (better_p): Update.
25341 (find_best_successor): Update.
25342 * trans-mem.c (expand_transaction): Update.
25343 * tree-call-cdce.c: Update.
25344 * tree-cfg.c (gimple_split_edge): Upate.
25345 (move_sese_region_to_fn): Upate.
25346 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
25347 * tree-eh.c (lower_resx): Upate.
25348 (cleanup_empty_eh_move_lp): Upate.
25349 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25350 * tree-inline.c (copy_edges_for_bb): Update.
25351 (copy_cfg_body): Update.
25352 * tree-parloops.c (gen_parallel_loop): Update.
25353 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
25354 (gimple_gen_time_profiler): Update.
25355 * tree-ssa-dce.c (remove_dead_stmt): Update.
25356 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25357 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
25358 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
25359 (unloop_loops): Update.
25360 (try_peel_loop): Update.
25361 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25362 * tree-ssa-loop-split.c (connect_loops): Update.
25363 (split_loop): Update.
25364 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25365 (hoist_guard): Update.
25366 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
25367 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
25368 (value_replacement): Update.
25369 * tree-ssa-reassoc.c (branch_fixup): Update.
25370 * tree-ssa-tail-merge.c (replace_block_by): Update.
25371 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
25372 (create_edge_and_update_destination_phis): Update.
25373 (compute_path_counts): Update.
25374 (recompute_probabilities): Update.
25375 (update_joiner_offpath_counts): Update.
25376 (freqs_to_counts_path): Update.
25377 (duplicate_thread_path): Update.
25378 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
25379 (struct switch_conv_info): Update.
25380 (gen_inbound_check): Update.
25381 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25382 (vect_do_peeling): Update.
25383 (vect_loop_versioning): Update.
25384 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25385 (optimize_mask_stores): Update.
25386 * ubsan.c (ubsan_expand_null_ifn): Update.
25387 * value-prof.c (gimple_divmod_fixed_value): Update.
25388 (gimple_divmod_fixed_value_transform): Update.
25389 (gimple_mod_pow2): Update.
25390 (gimple_mod_pow2_value_transform): Update.
25391 (gimple_mod_subtract): Update.
25392 (gimple_mod_subtract_transform): Update.
25393 (gimple_ic): Update.
25394 (gimple_stringop_fixed_value): Update.
25395 (gimple_stringops_transform): Update.
25396 * value-prof.h: Update.
25398 2017-06-29 Carl Love <cel@us.ibm.com>
25400 * config/rs6000/rs6000-c.c: Add support for built-in functions
25401 vector signed int vec_signed (vector float);
25402 vector signed long long vec_signed (vector double);
25403 vector signed int vec_signed2 (vector double, vector double);
25404 vector signed int vec_signede (vector double);
25405 vector signed int vec_signedo (vector double);
25406 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
25407 instruction generator.
25408 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25409 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
25410 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
25412 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
25413 vunsignede_v2df): Add define_expands.
25414 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
25415 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
25416 VEC_UNSIGNEDO): Add definitions.
25417 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25418 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
25419 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
25420 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
25421 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
25422 * config/rs6000/altivec.h (vec_signed, vec_signed2,
25423 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
25424 vec_unsignede, vec_unsignedo): Add builtin defines.
25425 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
25427 * doc/extend.texi: Update the built-in documentation file for the
25428 new built-in functions.
25430 2017-06-29 Richard Biener <rguenther@suse.de>
25432 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
25433 reduction chains to LOOP_VINFO_REDUCTIONS.
25434 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
25435 SLP reductions after processing reduction chains.
25437 2017-06-29 Nathan Sidwell <nathan@acm.org>
25439 * builtins.c (fold_builtin_FUNCTION): Use
25440 lang_hooks.decl_printable_name.
25442 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
25444 PR middle-end/81194
25445 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
25446 with only one label.
25447 * stmt.c (expand_case): Assert NCASES is greater than one.
25449 2017-06-29 Richard Biener <rguenther@suse.de>
25451 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
25453 (group_case_labels): Likewise.
25454 (find_taken_edge): Push sanity checking on val to workers...
25455 (find_taken_edge_cond_expr): ... here
25456 (find_taken_edge_switch_expr): ... and here, handle cases
25457 with just a default label.
25458 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
25459 (group_case_labels): Likewise.
25460 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
25461 group_case_labels does anything cleanup the CFG again.
25463 2017-06-29 Bin Cheng <bin.cheng@arm.com>
25465 PR tree-optimization/81196
25466 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
25467 exit condition comparing two IVs.
25469 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
25471 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
25472 profile to the dummy entry at the end of the list of architectures.
25473 * config/arm/arm-cpu-cdata.h: Regenerated.
25475 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25476 Michael Collison <michael.collison@arm.com>
25479 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
25481 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
25482 (*aarch64_reg_<mode>3_minus_mask): New pattern.
25483 (*aarch64_<optab>_reg_di3_mask2): New pattern.
25484 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
25485 of shift when the shift amount is masked with constant equal to
25486 the size of the mode.
25487 * config/aarch64/predicates.md (subreg_lowpart_operator): New
25490 2017-06-29 Martin Liska <mliska@suse.cz>
25492 * config/i386/i386.opt: Change range from [1,5] to [0,5].
25494 2017-06-29 Yury Gribov <tetra2005@gmail.com>
25497 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
25500 (edge_growth_cache_entry::edge_growth_cache_entry): New
25502 (reset_edge_growth_cache): Update to use constructor.
25504 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25506 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
25507 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25508 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25510 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
25512 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
25513 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
25515 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
25517 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
25518 (*-linux-uclibc*): Add t-uclibc tmake_file.
25519 * config/t-musl: New.
25520 * config/t-uclibc: New.
25522 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
25524 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
25526 (gen_comm_data): Emit architectural setting of arch_prof.
25527 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
25529 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
25530 (armv8-m.base, armv8-m.main): Likewise.
25531 * arm-protos.h (arm_build_target): Add profile field.
25532 (arch_option): Likewise.
25533 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
25535 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
25536 arm_active_target.profile.
25538 2017-06-28 Richard Biener <rguenther@suse.de>
25540 PR middle-end/81227
25541 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
25542 TYPE_OVERFLOW_WRAPS.
25543 * match.pd (negate_expr_p): Likewise.
25544 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
25545 fold_build2, not fold_binary.
25547 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25549 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
25550 Convert memory address to Pmode.
25551 (aarch64_print_operand): Assert MEM operands are always Pmode.
25553 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25556 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
25557 Remove redundant if.
25558 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
25559 * config/arm/aarch-common-protos.h
25560 (aarch_forward_to_shift_is_not_shifted_re): Remove.
25561 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
25563 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
25566 * multiple_target.c (create_dispatcher_calls): Set the default
25567 clone to be static, not public.
25569 2017-06-28 Richard Biener <rguenther@suse.de>
25571 * tree-vect-loop.c (vectorizable_reduction): Move special
25572 cond reduction IV var creation ...
25573 (vect_create_epilog_for_reduction): ... here. Remove induction_index
25574 parameter. Use STMT_VINFO_VECTYPE.
25575 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
25578 2017-06-28 Martin Liska <mliska@suse.cz>
25581 * ipa-visibility.c (non_local_p): Handle visibility.
25583 2017-06-28 Martin Liska <mliska@suse.cz>
25586 * common.opt: Add IntegerRange to various options.
25587 * opt-functions.awk (integer_range_info): New function.
25588 * optc-gen.awk: Add integer_range_info to cl_options struct.
25589 * opts-common.c (decode_cmdline_option): Handle
25590 CL_ERR_INT_RANGE_ARG.
25591 (cmdline_handle_error): Likewise.
25592 * opts.c (print_filtered_help): Show valid interval in
25593 when --help is provided.
25594 * opts.h (struct cl_option): Add range_min and range_max fields.
25595 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
25597 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
25599 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
25600 (x * C EQ/NE y * C): New transformation.
25602 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
25604 * genmultilib (combination_space): Accept '+' in option names.
25606 2017-06-28 Martin Liska <mliska@suse.cz>
25609 * asan.c (instrument_derefs): Bail out inner references
25610 that are hard register variables.
25612 2017-06-28 Jakub Jelinek <jakub@redhat.com>
25615 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
25616 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
25618 2017-06-28 Richard Biener <rguenther@suse.de>
25620 * tree-vectorizer.h (vect_get_vec_defs): Remove.
25621 (vect_get_slp_defs): Adjust.
25622 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
25624 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
25626 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
25627 get_initial_defs_for_reduction instead of vect_get_vec_defs.
25628 (vectorizable_reduction): Adjust.
25629 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
25631 (vect_get_slp_defs): Likewise.
25632 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
25633 (vectorizable_bswap): Adjust.
25634 (vectorizable_call): Likewise.
25635 (vectorizable_conversion): Likewise.
25636 (vectorizable_assignment): Likewise.
25637 (vectorizable_shift): Likewise.
25638 (vectorizable_operation): Likewise.
25639 (vectorizable_store): Likewise.
25640 (vectorizable_condition): Likewise.
25641 (vectorizable_comparison): Likewise.
25643 2017-06-28 Michael Collison <michael.collison@arm.com>
25646 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
25648 (arm_gen_movmemqi): Removed unused variable 'i'.
25649 Convert 'for' loop into 'while' loop.
25650 (arm_expand_prologue): Remove last unnecessary set of insn.
25651 (thumb_pop): Remove unused variable 'pushed_words'.
25652 (thumb_exit): Remove last unnecessary set of regs_to_pop.
25654 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25656 * config/s390/predicates.md: Use s390_rel_address_ok_p.
25657 * config/s390/s390-protos.h: Add prototype of
25658 s390_rel_address_ok_p.
25659 * config/s390/s390.c (s390_got_symbol): New function.
25660 (s390_rel_address_ok_p): New function.
25661 (legitimize_pic_address): Use s390_rel_address_ok_p.
25662 (s390_load_got): Use s390_got_symbol.
25663 (s390_option_override): Issue error if
25664 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
25665 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
25667 * config/s390/s390.opt: New option mpic-data-is-text-relative.
25669 2017-06-27 Andrew Pinski <apinski@cavium.com>
25671 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
25672 (X * copysign (1.0, X)): New pattern.
25673 (X * copysign (1.0, -X)): New pattern.
25674 (copysign (-1.0, CST)): New pattern.
25676 2017-06-27 Joseph Myers <joseph@codesourcery.com>
25678 * genmultilib (combination_space): Remove variable.
25679 Validate reuse rules against regular expression for any sequence
25680 of multilib options in any order.
25682 2017-06-27 Michael Collison <michael.collison@arm.com>
25684 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
25685 call aarch64_split_simd_combine.
25686 * (aarch64_combine_internal<mode>): Delete pattern.
25687 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
25688 Allow register and subreg operands.
25690 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25692 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
25693 specific need, just fallback on defaults.
25694 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
25696 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25697 Olivier Hainque <hainque@adacore.com>
25699 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
25701 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
25702 targets. Pick a default if no particular attempt applied.
25703 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
25706 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25708 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
25709 (x86_64-wrs-vxworks7): Likewise.
25711 2017-06-27 Marek Polacek <polacek@redhat.com>
25714 * ubsan.c (instrument_null): Check get_base_address's result for null.
25716 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25718 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
25720 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25722 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
25723 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
25724 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
25725 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
25726 New function types.
25727 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
25728 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
25729 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
25730 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
25731 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
25732 BUILT_IN_FEUPDATEENV): New builtins.
25733 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
25734 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
25735 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
25736 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
25738 (builtin_structptr_types): Adjust size.
25739 * tree.c (builtin_structptr_types): Add four entries.
25741 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25742 Olivier Hainque <hainque@adacore.com>
25744 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
25745 (TLS_SYM): New local macro, forcing reference to __tls__ on
25746 link command lines for VxWorks 7 RTPs, triggering initialization
25748 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
25749 OS features TLS support, true for RTPs on VxWorks 7.
25750 * config/vxworks.c (vxworks_override_options): Setup emutls
25753 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25755 * predict.c (test_prediction_value_range): Use -1U instead of -1
25756 to avoid narrowing conversion warning.
25757 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
25758 to avoid narrowing conversion warning.
25759 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
25761 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
25763 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25765 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
25766 64bit configurations.
25767 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
25768 (SIZE_TYPE): Likewise.
25769 * config/vxworks.c (vxworks_emutls_var_fields): Use
25770 long_unsigned_type_node instead of unsigned_type_node as the offset
25771 field type, which is "pointer" mode in emutls.c.
25773 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25776 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
25778 PR middle-end/81207
25779 * gimple-fold.c (replace_call_with_call_and_fold): Handle
25780 gimple_vuse copying separately from gimple_vdef copying.
25782 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25784 * value-prof.c (free_hist): Remove call to memset and the enclosing if
25787 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25788 Olivier Hainque <hainque@adacore.com>
25790 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
25791 for all vxworks7 targets.
25792 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
25793 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
25794 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
25795 variations for VX6/VX7 and 32/64bits later on in ...
25796 (VXWORKS_LIB_SPEC): Leverage new macros.
25797 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
25798 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
25800 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25802 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
25803 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
25805 2017-06-26 Carl Love <cel@us.ibm.com>
25807 * config/rs6000/rs6000-c.c: Add support for built-in functions
25808 vector bool char vec_reve (vector bool char);
25809 vector signed char vec_reve (vector signed char);
25810 vector unsigned char vec_reve (vector unsigned char);
25811 vector bool int vec_reve (vector bool int);
25812 vector signed int vec_reve (vector signed int);
25813 vector unsigned int vec_reve (vector unsigned int);
25814 vector bool long long vec_reve (vector bool long long);
25815 vector signed long long vec_reve (vector signed long long);
25816 vector unsigned long long vec_reve (vector unsigned long long);
25817 vector bool short vec_reve (vector bool short);
25818 vector signed short vec_reve (vector signed short);
25819 vector double vec_reve (vector double);
25820 vector float vec_reve (vector float);
25821 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
25822 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
25823 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
25824 (altivec_vreve): New pattern.
25825 * config/rs6000/altivec.h (vec_reve): New define.
25826 * doc/extend.texi (vec_rev): Update the built-in documentation file
25827 for the new built-in functions.
25829 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25831 PR tree-optimization/71815
25832 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
25834 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
25836 (slsr_process_phi): Likewise.
25837 (replace_uncond_cands_and_profitable_phis): Don't replace a
25838 multiply candidate with a stride of 1 (copy or cast).
25839 (phi_incr_cost): Call uses_consumed_by_stmt rather than
25841 (lowest_cost_path): Likewise.
25842 (total_savings): Likewise.
25844 2017-06-26 Richard Biener <rguenther@suse.de>
25847 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
25848 Use def_builtin_pure for all gather builtins.
25850 2017-06-26 Richard Biener <rguenther@suse.de>
25852 PR tree-optimization/81203
25853 * tree-tailcall.c (find_tail_calls): Do not move stmts into
25854 non-dominating BBs.
25856 2017-06-26 Marek Polacek <polacek@redhat.com>
25859 * doc/invoke.texi: Document -Wmultistatement-macros.
25861 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
25863 * doc/sourcebuild.texi (ARM-specific attributes): Document new
25864 arm_neon_ok_no_float_abi effective target.
25866 2017-06-26 Richard Biener <rguenther@suse.de>
25868 PR tree-optimization/80928
25869 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
25870 (copy_bbs): Set BB_DUPLICATED flag early.
25871 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
25873 (execute_on_shrinking_pred): Likewise.
25874 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
25875 BB_DUPLICATED blocks.
25876 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
25877 iterate over all PHIs considering removal of *gsi.
25879 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
25881 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
25884 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25886 * config/rs6000/rs6000-string.c: (expand_block_clear,
25887 do_load_for_compare, select_block_compare_mode,
25888 compute_current_alignment, expand_block_compare,
25889 expand_strncmp_align_check, expand_strn_compare,
25890 expand_block_move, rs6000_output_load_multiple)
25891 Move functions related to string/block move/compare
25892 to a separate file.
25893 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
25894 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
25895 for this function which is now used in two files.
25896 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
25897 * config.gcc: Add rs6000-string.o to extra_objs for
25898 targets powerpc*-*-* and rs6000*-*-*.
25900 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
25903 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
25904 32-bit, since indexed is not valid for DImode.
25905 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
25906 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
25907 (define_peephole2 for Altivec d-form load): Add 32-bit support.
25908 (define_peephole2 for Altivec d-form store): Likewise.
25911 * multiple_target.c (create_dispatcher_calls): Only create the
25912 dispatcher call if the function is the default clone of a
25913 versioned function.
25915 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
25917 PR middle-end/80902
25918 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
25919 a call, force the call to not be a tail call.
25921 2017-06-23 Jeff Law <law@redhat.com>
25923 * doc/contrib.texi: Add entry for Steven Pemberton's work on
25926 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
25928 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
25929 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
25930 handling for early expansion of vector shifts (sl,sr,sra,rl).
25931 (builtin_function_type): Add vector shift right instructions
25932 to the unsigned argument list.
25934 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
25936 rtl-optimizatoin/79286
25937 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
25938 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
25939 trap. PIC register plus a const unspec without offset can never trap.
25941 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
25943 * tree.h (builtin_structptr_type): New type.
25944 (builtin_structptr_types): Declare new array.
25945 * tree.c (builtin_structptr_types): New array.
25946 (free_lang_data, build_common_tree_nodes): Use it.
25948 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
25951 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
25954 2017-06-22 Matt Turner <mattst88@gmail.com>
25956 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
25957 Lake models to skylake case. Assume skylake for unknown
25958 models with clflushopt.
25960 2017-06-22 Jeff Law <law@redhat.com>
25962 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
25963 frame sizes that do not satisfy aarch64_uimm12_shift.
25965 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
25967 * profile-count.h (apply_probability,
25968 apply_scale, probability_in): Fix checks for zero.
25970 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25972 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
25973 * doc/cppdiropts.texi (-I @var{dir}): Document it.
25975 2016-06-22 Richard Biener <rguenther@suse.de>
25977 * tree-vect-loop.c (vect_model_reduction_cost): Handle
25978 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
25979 REDUC_MAX_EXPR support.
25980 (vectorizable_reduction): Likewise.
25981 (vect_create_epilog_for_reduction): Likewise.
25983 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
25985 * match.pd (A / (1 << B) -> A >> B): New.
25986 * generic-match-head.c: Include optabs-tree.h.
25987 * gimple-match-head.c: Likewise.
25988 * optabs-tree.h (target_supports_op_p): New.
25989 * optabs-tree.c (target_supports_op_p): New.
25991 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25993 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
25994 $gcc_cv_ld --help output.
25995 (gcc_cv_ld_demangle): Likewise.
25996 (gcc_cv_ld_eh_frame_hdr): Likewise.
25997 (gcc_cv_ld_pie): Likewise.
25998 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
25999 (gcc_cv_ld_buildid): Likewise.
26000 (gcc_cv_ld_sysroot): Likewise.
26001 (ld_bndplt_support): Likewise.
26002 (ld_pushpopstate_support): Likewise.
26003 * configure: Regenerate.
26004 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
26006 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26009 * config/i386/sse.md (round<mode>2): Renumber match_dup and
26010 operands indexes to avoid gap between operands and match_dups.
26012 2017-06-21 Andrew Pinski <apinski@cavium.com>
26014 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
26015 Increment Arith_shift and Arith_shift_reg by 1.
26016 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
26018 * config/aarch64/aarch64.c (thunderx_tunings): Enable
26019 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
26020 (aarch64_strip_extend): Add new argument and test for it.
26021 (aarch64_cheap_mult_shift_p): New function.
26022 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
26023 add a cost if it is true.
26024 Update calls to aarch64_strip_extend.
26025 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
26027 2017-06-21 Andrew Pinski <apinski@cavium.com>
26029 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
26031 (thunderxt88): Likewise.
26032 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
26033 (thunderx_prefetch_tune): New variable.
26034 (thunderx2t99_prefetch_tune): Update for the correct values.
26035 (thunderxt88_tunings): New variable.
26036 (thunderx_tunings): Use thunderx_prefetch_tune instead of
26037 generic_prefetch_tune.
26038 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
26040 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26042 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
26043 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
26044 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
26045 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
26046 (aarch64_atomic_cas<mode>, GPI): Likewise.
26048 2017-06-21 Martin Liska <mliska@suse.cz>
26050 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
26051 statements on cold and hot labels.
26052 * predict.c (tree_estimate_probability_bb): Remove the
26053 prediction from this place.
26055 2017-06-21 Martin Liska <mliska@suse.cz>
26057 PR tree-optimization/79489
26058 * gimplify.c (maybe_add_early_return_predict_stmt): New
26060 (gimplify_return_expr): Call the function.
26061 * predict.c (tree_estimate_probability_bb): Remove handling
26063 * predict.def: Update comment about early return predictor.
26064 * gimple-predict.h (is_gimple_predict): New function.
26065 * predict.def: Change default value of early return to 66.
26066 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
26068 * passes.def: Put pass_strip_predict_hints to the beginning of
26071 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
26073 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
26074 FUNCTION_DECL declarations.
26075 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
26077 (dwaf2out_decl): Likewise.
26078 * godump.c (go_early_global_decl): Skip call to the real debug hook
26079 for FUNCTION_DECL declarations.
26080 * passes.c (rest_of_decl_compilation): Skip call to the
26081 early_global_decl debug hook for FUNCTION_DECL declarations, unless
26082 -fdump-go-spec is passed.
26084 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26086 * config/i386/i386.c (struct builtin_isa): New field pure_p.
26087 Reorder for compactness.
26088 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
26089 (def_builtin_pure, def_builtin_pure2): New functions.
26090 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
26092 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26094 * match.pd (nop_convert): New predicate.
26095 ((A +- CST1) +- CST2): Allow some NOP conversions.
26097 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26100 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
26101 with ctors/dtors if GOVD_SHARED is set.
26103 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26105 * config/aarch64/aarch64.md (movti_aarch64):
26106 Emit mov rather than orr.
26107 (movtf_aarch64): Likewise.
26108 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
26109 Emit mov rather than orr.
26111 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26113 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
26114 Swap alternatives, make integer dup more expensive.
26116 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26118 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
26119 Return true for non-tls symbols.
26121 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
26123 * config/aarch64/aarch64-cores.def (cortex-a55): New.
26124 (cortex-a75): Likewise.
26125 (cortex-a75.cortex-a55): Likewise.
26126 * config/aarch64/aarch64-tune.md: Regenerate.
26127 * doc/invoke.texi (-mtune): Document new values for -mtune.
26129 2017-06-21 Tom de Vries <tom@codesourcery.com>
26131 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
26132 stack_size feature.
26133 (Effective-Target Keywords, Other attributes): Suggest using
26134 dg-add-options stack_size feature to get stack limit in stack_size
26135 effective target documentation.
26137 2017-06-21 Julian Brown <julian@codesourcery.com>
26138 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26140 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
26141 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
26142 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
26144 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
26145 attribute type list for neon_multiply.
26146 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
26147 attribute type list for neon_multiply.
26148 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
26149 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
26150 attribute type list for neon_multiply.
26151 * config/arm/types.md (crypto_pmull): Add.
26152 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
26153 attribute type list.
26155 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
26157 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
26160 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26162 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
26163 to make sure not to dereference a NULL cost_classes_ptr pointer.
26165 2017-06-20 Carl Love <cel@us.ibm.com>
26167 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26168 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
26169 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
26170 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
26171 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
26172 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
26173 VMULOSW): New enum "unspec" values.
26174 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
26175 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
26176 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
26177 altivec_vmulosw): New patterns.
26178 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
26179 VMULOSW): Add definitions.
26181 2017-06-20 Julia Koval <julia.koval@intel.com>
26183 * config/i386/i386.c: Fix rounding expand for new pattern.
26184 * config/i386/subst.md: Fix pattern (parallel -> unspec).
26186 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26188 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
26189 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
26191 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26193 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
26196 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26198 * config/aarch64/aarch64-cores.def: Rearrange to sort by
26199 architecture, then by implementer ID.
26200 * config/aarch64/aarch64-tune.md: Regenerate.
26202 2017-06-20 Richard Biener <rguenther@suse.de>
26204 PR middle-end/81097
26205 * fold-const.c (split_tree): Fold to type before negating.
26207 2017-06-20 David Malcolm <dmalcolm@redhat.com>
26209 * diagnostic-show-locus.c
26210 (selftest::test_fixit_deletion_affecting_newline): New function.
26211 (selftest::diagnostic_show_locus_c_tests): Call it.
26213 2017-06-20 Andreas Schwab <schwab@suse.de>
26216 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
26219 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26221 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
26222 __ARM_FEATURE_COPROC according to support.
26224 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26226 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26227 Rewritten to avoid overflow for > 32-bit pointers.
26230 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
26231 by removing enum keyword.
26232 (ubsan_type_descriptor): Likewise. Formatting fix.
26235 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
26236 splitter): Require TARGET_SSE2 in the condition.
26238 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
26241 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
26242 for doing vector set of SFmode on ISA 3.0.
26243 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
26244 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
26246 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
26247 SFmode value into a V4SF variable that was extracted from another
26248 V4SF variable without converting the element to double precision
26249 and back to single precision vector format.
26250 (vsx_insert_extract_v4sf_p9_2): Likewise.
26252 2017-06-19 Jakub Jelinek <jakub@redhat.com>
26254 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
26255 in UWHI to avoid undefined overflow.
26258 * ubsan.h (enum ubsan_encode_value_phase): New.
26259 (ubsan_encode_value): Change second argument to
26260 enum ubsan_encode_value_phase with default value of
26261 UBSAN_ENCODE_VALUE_GENERIC.
26262 * ubsan.c (ubsan_encode_value): Change second argument to
26263 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
26264 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
26265 create_tmp_var_raw instead of create_tmp_var and use a
26267 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
26268 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
26269 ubsan_encode_value callers.
26272 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
26273 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
26274 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
26276 2017-06-19 Richard Biener <rguenther@suse.de>
26278 PR middle-end/81118
26279 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
26280 estimates if we changed anything.
26282 2017-06-19 Richard Biener <rguenther@suse.de>
26284 PR tree-optimization/80887
26285 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
26286 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
26287 simplified lookups, then reset mprts_hook.
26288 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
26290 (try_to_simplify): Likewise.
26292 2017-06-19 Martin Liska <mliska@suse.cz>
26295 * gimplify.c (gimplify_switch_expr):
26296 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
26298 2017-06-19 Martin Liska <mliska@suse.cz>
26300 * doc/install.texi: Document that PGO runs in 4 stages.
26302 2017-06-19 Martin Liska <mliska@suse.cz>
26305 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
26306 to dispatcher function name.
26307 * multiple_target.c (replace_function_decl): New function.
26308 (create_dispatcher_calls): Redirect both edges and references.
26310 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
26312 * profile-count.c (profile_count::dump): Dump quality.
26313 (profile_count::differs_from_p): Update for unsigned val.
26314 * profile-count.h (profile_count_quality): New enum.
26315 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
26317 2017-06-19 Richard Biener <rguenther@suse.de>
26319 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
26320 struct function as arg.
26321 (estimate_numbers_of_iterations): Export overload with loop arg.
26322 (free_numbers_of_iterations_estimates_loop): Use an overload of
26323 free_numbers_of_iterations_estimates instead.
26324 * tree-cfg.c (remove_bb): Adjust.
26325 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
26326 * tree-parloops.c (gen_parallel_loop): Likewise.
26327 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
26329 (tree_unroll_loops_completely): Likewise.
26330 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
26331 Use an overload instead and export.
26332 (estimated_loop_iterations): Adjust.
26333 (max_loop_iterations): Likewise.
26334 (likely_max_loop_iterations): Likewise.
26335 (estimate_numbers_of_iterations): Take struct function as arg
26337 (loop_exits_before_overflow): Adjust.
26338 (free_numbers_of_iterations_estimates_loop): Use an overload.
26339 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
26340 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
26342 2017-06-19 Richard Biener <rguenther@suse.de>
26345 * ipa-prop.c (find_constructor_constant_at_offset): Handle
26346 RANGE_EXPR conservatively.
26348 2017-06-16 Carl Love <cel@us.ibm.com>
26350 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26351 definitions for vec_float, vec_float2, vec_floato,
26352 vec_floate built-ins.
26353 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
26354 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
26356 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
26357 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
26358 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
26359 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
26360 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
26361 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
26362 vec_floato): Add builtin defines.
26363 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
26364 Update the built-in documentation file for the new built-in
26367 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26369 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
26370 (mthumb): Mark as the negative of -marm.
26372 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26374 * doc/invoke.texi (ARM Options, -mcpu): Document supported
26376 (ARM Options, -mtune): Document that this accepts the same
26377 extension options as -mcpu.
26378 (ARM Options, -mfpu): Document addition of -mfpu=auto.
26380 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26382 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
26383 permitted extensions.
26385 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26387 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
26388 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
26389 (armv8-m.main): Add option +nodsp.
26390 * config/arm/arm-cpu-cdata.h: Regenerated.
26392 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26394 * config/arm/t-fuchsia: New file.
26395 * config.gcc (arm*-*-fuchsia*): Use it.
26397 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26399 * config/arm/t-symbian: Rewrite for new option infrastructure.
26401 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26403 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
26404 (MULTILIB_REQUIRED): Likewise.
26406 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26408 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
26409 (MULTILIB_RESUE): Likewise.
26410 (MULTILIB_MATCHES): Likewise.
26411 (MULTLIB_REQUIRED): Likewise.
26413 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26415 * config/arm/t-rtems: Rewrite for new option framework.
26417 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26419 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
26420 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
26421 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
26422 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
26423 * config/arm/t-multilib: ... here.
26424 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
26425 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
26426 armv7-a and armv8*-a when A-profile libraries have not been built.
26427 * config/arm/t-rmprofile: Rewrite.
26429 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26431 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
26432 with a backslash. Remove the backslash after substituting unescaped
26434 * doc/fragments.texi (MULTILIB_REUSE): Document it.
26436 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26438 * config.gcc: (arm*-*-*): When building a-profile libraries, force
26439 the driver to pass through the default setting of -mfloat-abi.
26440 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
26442 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
26443 (all_feat_combs): New rule.
26444 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
26446 * config/arm/t-aprofile: Rewrite.
26448 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26450 * config/arm/arm.h (FPUTYPE_AUTO): Define.
26451 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
26452 fpu is not specified by the user/command-line.
26453 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
26454 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
26455 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
26456 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
26457 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
26458 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
26460 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26462 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
26463 * config/arm/t-arm-elf: Rewritten.
26465 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26467 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
26468 have some floating-point instructions.
26469 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
26470 (TARGET_MAYBE_HARD_FLOAT): New macro.
26471 * config/arm/arm-builtins.c (arm_init_builtins): Use
26472 TARGET_MAYBE_HARD_FLOAT.
26473 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
26475 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26477 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
26478 (configargs.h): Include it.
26479 (arm_print_hint_for_fpu_option): New function.
26480 (arm_parse_fpu_option): New function.
26481 (candidate_extension): New class.
26482 (arm_canon_for_multilib): New function.
26483 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
26484 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
26485 (ARCH_CANONICAL_SPECS): New macro.
26486 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
26488 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26490 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
26491 are set after handling multilib fragments. Set target_cpu_default2
26494 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26496 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
26498 (arm*-*-*): Set target_cpu_default2 to a quoted string.
26499 * config/arm/parsecpu.awk (check_cpu): Validate any extension
26501 (check_arch): Likewise.
26502 * config/arm/arm.c (arm_configure_build_target): Handle
26503 TARGET_CPU_DEFAULT being a string constant. Scan any feature
26504 options in the default.
26506 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26508 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
26509 when an option is an alias of another.
26510 * config/arm/parsecpu.awk (optalias): New parser token.
26511 (gen_comm_data): Mark non-alias options as such. Emit entries
26512 for extension aliases.
26513 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
26514 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
26515 (armv6kz, armv6zk, armv6t2): Likewise.
26516 (armv7): Make vfpv3-d16 an alias.
26517 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
26519 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
26520 Sort in canonical order.
26521 (armv8-a): Sort in canonical order.
26522 (armv8.1-a, armv8.2-a): Likewise.
26523 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
26525 (cortex-a9): Sort in canonical order.
26526 * config/arm/arm.c (selftests.h): Include it.
26527 (arm_test_cpu_arch_data): New function.
26528 (arm_run_self_tests): New function.
26529 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
26530 (targetm): Move declaration to the end of the file.
26531 * arm-cpu-cdata.h: Regenerated.
26533 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26535 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
26536 call to target_mode_check describing the type of option passed.
26537 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
26538 (arm_target_thumb_only): Use arm_parse_arch_option_name or
26539 arm_parse_cpu_option_name to match parameters against list of
26541 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
26542 arm_arch_core_flags data structure.
26543 * config/arm/arm-cpu_cdata.h: Regenerated.
26545 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26547 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
26549 (arm_print_hint_for_cpu_option): Likewise.
26550 (arm_print_hint_for_arch_option): Likewise.
26551 (arm_parse_cpu_option_name): Likewise.
26552 (arm_parse_arch_option_name): Likewise.
26553 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
26554 of entries in the all_fpus list.
26555 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
26556 (arm_parse_cpu_option_name): Declare.
26557 (arm_parse_arch_option_name): Declare.
26558 (arm_parse_option_features): Declare.
26559 (arm_intialize_isa): Declare.
26560 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
26562 (gen_comm_data): ... here. Make definitions non-static.
26563 * config/arm/arm-cpu-data.h: Regenerated.
26564 * config/arm/arm-cpu-cdata.h: Regenerated.
26566 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26568 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
26569 (cpu_arch_extension): New structure.
26570 (cpu_arch_option, arch_option, cpu_option): New structures.
26571 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
26572 architecture types.
26573 (gen_data): Generate new format data tables.
26574 * config/arm/arm.c (cpu_tune): New structure.
26575 (cpu_option, processors): Delete.
26576 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
26577 (arm_print_hint_for_cpu_option): ... this and ...
26578 (arm_print_hint_for_arch_option): ... this.
26579 (arm_parse_arch_cpu_name): Delete. Replace with ...
26580 (arm_parse_cpu_option_name): ... this and ...
26581 (arm_parse_arch_option_name): ... this.
26582 (arm_unrecognized_feature): Change type of target parameter to
26584 (arm_parse_arch_cpu_features): Delete. Replace with ...
26585 (arm_parse_option_features): ... this.
26586 (arm_configure_build_target): Rework to use new configuration data
26588 (arm_print_tune_info): Rework for new configuration data tables.
26589 * config/arm/arm-cpu-data.h: Regenerated.
26590 * config/arm/arm-cpu.h: Regenerated.
26592 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26594 * Makefile.in (OBJS): Move sbitmap.o from here ...
26595 (OBJS-libcommon): ... to here.
26597 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26599 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
26600 (ISA_ALL_CRYPTO): New macro.
26601 (ISA_ALL_SIMD): New macro
26602 (ISA_ALL_FP): New macro.
26603 * config/arm/arm.c (fpu_bitlist): Update initializer.
26604 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
26606 (arm9e): Add fpu. Add option for nofp
26607 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
26608 (arm926ej-s, arm1026ej-s): Likewise.
26609 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
26610 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
26611 neon-fp16, neon-vfpv4, nofp and nosimd.
26612 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
26613 (cortex-a8): Add fpu. Add option for nofp.
26614 (cortex-a9): Add fpu. Add options for nosimd and nofp.
26615 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
26616 (cortex-r4f): Add fpu.
26617 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
26618 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
26620 (cortex-r8): Likewise.
26621 (cortex-m4): Add fpu. Add option for nofp.
26622 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
26623 (cortex-a17.cortex-a7): Likewise.
26624 (cortex-a32): Add fpu. Add options for crypto and nofp.
26625 (cortex-a35, cortex-a53): Likewise.
26626 (cortex-a57): Add fpu. Add option for crypto.
26627 (cortex-a72, cortex-a73): Likewise.
26628 (exynos-m1): Likewise.
26629 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
26630 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
26631 (cortex-m33): Add fpu. Add option for nofp.
26632 * config/arm/arm-cpu-cdata.h: Regenerated
26633 * config/arm/arm-cpu-data.h: Regenerated.
26635 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26637 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
26638 (armv5te, armv5tej): Likewise.
26639 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
26640 (armv7): Add options fp and vfpv3-d16.
26641 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
26642 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
26644 (armv7ve): Likewise.
26645 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
26646 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
26647 (armv8-a): Add nocrypto option.
26648 (armv8.1-a, armv8.2-a): Likewise.
26649 (armv8-m.main): add options fp, fp.dp and nofp.
26651 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26653 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
26655 (armv8-a+crc): Delete.
26656 (armv8.1-a): Add options simd, crypto and nofp.
26657 (armv8.2-a): Add options fp16, simd, crypto and nofp.
26658 (armv8.2-a+fp16): Delete.
26659 (armv8-m.main): Add option dsp.
26660 (armv8-m.main+dsp): Delete.
26661 (cortex-a8): Add fpu. Add nofp option.
26662 (cortex-a9): Add fpu. Add nofp and nosimd options.
26663 * config/arm/parsecpu.awk (gen_data): Generate option tables and
26664 link to main cpu and architecture data structures.
26665 (gen_comm_data): Only put isa attributes from the main architecture
26667 (option): New statement for architecture and CPU entries.
26668 * arm.c (struct cpu_option): New structure.
26669 (struct processors): Add entry for options.
26670 (arm_unrecognized_feature): New function.
26671 (arm_parse_arch_cpu_name): Ignore any characters after the first
26673 (arm_parse_arch_cpu_feature): New function.
26674 (arm_configure_build_target): Separate out any CPU and architecture
26675 features and parse separately. Don't error out if -mfpu=auto is
26676 used with only an architecture string.
26677 (arm_print_asm_arch_directives): New function.
26678 (arm_file_start): Call it.
26679 * config/arm/arm-cpu-cdata.h: Regenerated.
26680 * config/arm/arm-cpu-data.h: Likewise.
26681 * config/arm/arm-tables.opt: Likewise.
26683 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26685 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
26686 assembler when it is not -mfpu=auto.
26688 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26690 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
26691 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
26692 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
26693 (ASM_CPU_SPEC): Rewrite.
26694 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
26695 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
26696 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
26697 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
26698 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
26699 copied string is NUL-terminated. Also strip any characters prefixed
26701 (arm_rewrite_selected_arch): New function.
26702 (arm_rewrite_march): New function.
26704 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26706 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
26707 (x_arm_cpu_string, x_arm_tune_string): Likewise.
26708 (march, mcpu, mtune): Convert to string-based options.
26709 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
26710 (arm_parse_arch_cpu_name): New function.
26711 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
26712 identify selected architecture or CPU.
26713 (arm_option_save): New function.
26714 (TARGET_OPTION_SAVE): Redefine.
26715 (arm_option_restore): Restore string options.
26716 (arm_option_print): Print string options.
26718 2017-06-16 Martin Sebor <msebor@redhat.com>
26720 PR tree-optimization/80933
26721 PR tree-optimization/80934
26722 * builtins.c (fold_builtin_3): Do not handle bcmp here.
26723 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
26724 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
26725 (gimple_fold_builtin): Call them.
26727 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26729 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
26730 as unlikely; update profile.
26732 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26734 * predict.c (force_edge_cold): Handle declaring edges impossible
26737 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26739 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
26741 (try_unroll_loop_completely): Fix reporting.
26743 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26745 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
26747 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
26750 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
26751 if given a non-constant argument for an intrinsic which requires a
26754 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26756 * profile.c (compare_freqs): New function.
26757 (branch_prob): Sort edge list.
26758 (find_spanning_tree): Assume that the list is priority sorted.
26760 2017-06-16 Richard Biener <rguenther@suse.de>
26762 PR tree-optimization/81090
26763 * passes.def (pass_record_bounds): Remove.
26764 * tree-pass.h (make_pass_record_bounds): Likewise.
26765 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
26766 make_pass_record_bounds): Likewise.
26767 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
26768 not free niter estimates at the beginning but at the end.
26769 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
26771 2017-06-16 Richard Biener <rguenther@suse.de>
26773 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
26774 initializer to workaround ICE in host GCC 4.8.
26776 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26778 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
26780 (clone_inlined_nodes): Update.
26782 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26784 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
26785 prefetch settings, and enable prefetching by default at -O3.
26787 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26789 * config/aarch64/aarch64.c (aarch64_override_options_internal):
26790 Set flag_prefetch_loop_arrays according to tuning data.
26792 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26794 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
26795 New tune structure.
26796 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
26797 [Unrelated to main purpose of the patch] Place the pointer field last
26798 to enable type checking errors when tune structure are wrongly merged.
26799 * config/aarch64/aarch64.c (generic_prefetch_tune,)
26800 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
26801 (thunderx2t99_prefetch_tune): New tune constants.
26802 (tune_params *_tunings): Update all tunings (no functional change).
26803 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
26804 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
26805 from tunings structures.
26807 2017-06-16 Jakub Jelinek <jakub@redhat.com>
26810 * ubsan.c (instrument_null): Add T argument, use it instead
26811 of computing it based on IS_LHS.
26812 (instrument_object_size): Likewise.
26813 (pass_ubsan::execute): Adjust instrument_null and
26814 instrument_object_size callers to pass gimple_get_lhs or
26815 gimple_assign_rhs1 result to it. Use instrument_null instead of
26816 calling get_base_address and instrument_mem_ref. Handle
26817 aggregate call arguments for object-size sanitization.
26819 2017-06-16 Yury Gribov <tetra2005@gmail.com>
26821 PR tree-optimization/81089
26822 * tree-vrp.c (is_masked_range_test): Validate operands of
26825 2017-06-15 Martin Sebor <msebor@redhat.com>
26828 * dumpfile.c (dump_register): Avoid calling memset to initialize
26829 a class with a default ctor.
26830 * gcc.c (struct compiler): Remove const qualification.
26831 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
26832 * hash-table.h: Ditto.
26833 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
26835 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
26836 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
26838 * params.h (struct param_info): Make struct members non-const.
26839 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
26840 with default initialization.
26841 * vec.h (vec_copy_construct, vec_default_construct): New helper
26843 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
26844 with vec_copy_construct.
26845 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
26846 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
26847 * doc/invoke.texi (-Wclass-memaccess): Document.
26849 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26851 * emit-rtl.h (is_leaf): Update comment about local
26852 register allocator.
26854 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26857 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
26858 for a variable to have a section before checking if the section has a
26860 Set section to.persistent if persistent attribute is set.
26861 Warn if .persistent attribute is used on an automatic variable.
26863 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
26865 PR rtl-optimization/80474
26866 * reorg.c (update_block): Do not ignore instructions in a delay slot.
26868 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
26870 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
26873 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
26875 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
26876 (casesi): Emit bounds checking as RTL.
26877 (casesi_internal_mips16_<mode>): Remove bounds checking.
26879 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
26881 * config/xtensa/xtensa.c (xtensa_option_override): Append
26882 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
26883 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
26884 xtensa_doloop_hooks): Define unconditionally.
26885 (xtensa_reorg_loops): Only call reorg_loops in the presence of
26887 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
26888 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
26889 for it in xtensa_option_override.
26890 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
26891 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
26893 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
26895 * doc/cppopts.texi: Document '-' special value to -MF.
26897 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
26899 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
26900 (cortex_a53_fconst): Likewise.
26901 (cortex_a53_fpmul): Likewise.
26902 (cortex_a53_f_load_64): Likewise.
26903 (cortex_a53_f_load_many): Likewise.
26904 (cortex_a53_advsimd_alu): Likewise.
26905 (cortex_a53_advsimd_alu_q): Likewise.
26906 (cortex_a53_advsimd_mul): Likewise.
26907 (cortex_a53_advsimd_mul_q): Likewise.
26908 (fpmac bypass): Add new bypass for fpmac-fpmac case.
26909 Add missing fmul, r2f_cvt and fconst cases.
26911 2017-06-14 Richard Biener <rguenther@suse.de>
26913 PR middle-end/81088
26914 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
26916 (fold_binary_loc): When associating do not treat pre-existing
26917 TREE_OVERFLOW on literal constants as a reason to allow
26918 TREE_OVERFLOW on associated literal constants.
26920 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26922 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
26923 (MASK_FEATURES): New macro.
26924 * config/sparc/sparc.c (sparc_option_override): Remove the special
26925 handling of -mfpu and generalize it to all MASK_FEATURES switches.
26927 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26929 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
26930 a division of 0 if non-call exceptions are enabled.
26932 2017-06-14 Andrew Pinski <apinski@cavium.com>
26933 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26936 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
26937 Improve vector initialization code gen for only variable case.
26939 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26941 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
26943 2017-06-14 Richard Biener <rguenther@suse.de>
26945 PR tree-optimization/81083
26946 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
26949 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26951 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
26952 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
26953 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
26954 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
26955 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
26956 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
26958 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26960 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
26961 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
26963 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26965 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
26967 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26969 * config/rs6000/t-rtems: Don't handle SPE.
26971 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26973 * config/rs6000/t-linux: Don't handle SPE.
26975 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26977 * config/rs6000/eabispe.h: Delete file.
26979 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26981 * config/rs6000/t-spe: Delete file.
26983 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26985 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
26986 (rs6000_legitimate_offset_address_p): Return false for anything in
26987 V2SImode or V2SFmode.
26989 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26991 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
26992 except V2SF and V2SI. Rearrange the vector modes, and add comments.
26993 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
26995 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
26996 (rs6000_legitimate_offset_address_p): Ditto.
26997 (rs6000_emit_move): Ditto.
26998 (rs6000_init_builtins): Remove V4HI_type_node.
27000 2017-06-13 Martin Liska <mliska@suse.cz>
27003 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
27004 (gate_asan): Likewise.
27005 * asan.h (asan_no_sanitize_address_p): Remove the function.
27006 (sanitize_flags_p): New function.
27007 * builtins.def: Fix coding style.
27008 * common.opt: Use renamed enum value.
27009 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
27010 * doc/extend.texi: Document no_sanitize attribute.
27011 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
27012 to SANITIZE_UNDEFINED_NONDEFAULT.
27013 * gcc.c (sanitize_spec_function): Use the renamed enum value.
27014 * gimple-fold.c (optimize_atomic_compare_exchange_p):
27015 Use sanitize_flags_p.
27016 * gimplify.c (gimplify_function_tree): Likewise.
27017 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
27018 * opts.c (parse_no_sanitize_attribute): New function.
27019 (common_handle_option): Use renamed enum value.
27020 * opts.h (parse_no_sanitize_attribute): Declare.
27021 * tree.c (sanitize_flags_p): New function.
27022 * tree.h: Declared here.
27023 * tsan.c: Use sanitize_flags_p.
27024 * ubsan.c (ubsan_expand_null_ifn): Likewise.
27025 (instrument_mem_ref): Likewise.
27026 (instrument_bool_enum_load): Likewise.
27027 (do_ubsan_in_current_function): Remove the function.
27028 (pass_ubsan::execute): Use sanitize_flags_p.
27029 * ubsan.h: Remove do_ubsan_in_current_function
27030 * tree-cfg.c (print_no_sanitize_attr_value): New function.
27031 (dump_function_to_file): Use it here.
27033 2017-06-13 Martin Jambor <mjambor@suse.cz>
27035 PR tree-optimization/80803
27036 PR tree-optimization/81063
27037 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
27038 (propagate_subaccesses_across_link): Enqueue subtree whenever
27039 necessary instead of relying on the caller.
27041 2017-06-13 Martin Jambor <mjambor@suse.cz>
27043 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
27044 that have a first_link.
27045 (sort_and_splice_var_accesses): Do not check first_link before
27047 (subtree_mark_written_and_enqueue): Likewise.
27048 (propagate_all_subaccesses): Likewise and do not stop at first
27049 parent with a first_link.
27051 2017-06-13 Martin Jambor <mjambor@suse.cz>
27053 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
27056 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27058 * match.pd: New pattern.
27060 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27062 * tree-vrp.c (is_masked_range_test): New function.
27063 (register_edge_assert_for): Determine ranges for
27066 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27068 PR tree-optimization/67328
27069 * fold-const.c (maskable_range_p): New function.
27070 (build_range_check): Generate bittests if possible.
27072 2017-06-13 Martin Liska <mliska@suse.cz>
27074 * gimple-pretty-print.c (dump_probability): Add new argument.
27075 (dump_edge_probability): Dump both probability and count.
27076 (dump_gimple_label): Likewise.
27077 (dump_gimple_bb_header): Likewise.
27079 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
27082 * config/avr/avr-devices.c: Fix indentation.
27083 * config/avr/gen-avr-mmcu-specs.c: Dito.
27085 2017-06-13 Richard Biener <rguenther@suse.de>
27087 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
27088 instead get vector type from stmt_info.
27089 (vectorizable_reduction): Adjust. Remove dead code.
27091 2017-06-13 Richard Biener <rguenther@suse.de>
27093 PR middle-end/81065
27094 * fold-const.c (extract_muldiv_1): Remove bogus distribution
27095 case of C * (x * C2 + C3).
27096 (fold_addr_of_array_ref_difference): Properly fold index difference.
27098 2017-06-12 David S. Miller <davem@davemloft.net>
27101 * config/sparc/sparc.md (return expander): Emit frame blockage if
27102 function uses alloca.
27104 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
27106 * combine.c (make_field_assignment): Check len rather than the mode
27107 precision when calling force_to_mode.
27109 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
27111 Support multilibs and devices that see flash in RAM address range.
27114 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
27115 (avr_mcu_t) <flash_pm_offset>: New field.
27116 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
27117 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
27118 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
27119 (AVR_TINY_PM_OFFSET): Remove macro.
27120 * config/avr/avr.opt (-mshort-calls): New option.
27121 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
27122 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
27123 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
27124 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
27125 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
27126 instead of avr_arch->have_jmp_call.
27127 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
27128 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
27129 avr_arch->flash_pm_offset to define.
27130 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
27131 new field flash_pm_offset. Add entry for avrxmega3.
27132 (avr_texinfo): Add entry for avrxmega3.
27133 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
27134 attiny212, attiny214,
27135 attiny412, attiny414, attiny416, attiny417,
27136 attiny814, attiny816, attiny817,
27137 attiny1614, attiny1616, attiny1617,
27138 attiny3214, attiny3216, attiny3217.
27139 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
27140 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
27141 (avr_print_operand_address) [AVR_TINY]: Same.
27142 (avr_asm_init_sections) <readonly_data_section>: Only patch
27143 callback if avr_arch->flash_pm_offset = 0.
27144 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
27145 for rodata if avr_arch->flash_pm_offset != 0.
27146 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
27147 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
27148 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
27149 (m_options): Append opt_rcall.
27150 (m_dirnames): Append dir_rcall.
27151 * config/avr/t-multilib: Regenerate.
27153 * configure.ac [target=avr]: Check whether avrxmega3 default
27154 linker description file works as needed.
27155 * configure: Regenerate.
27156 * doc/avr-mmcu.texi: Regenerate.
27157 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
27158 <__AVR_ARCH__>: Document avrxmega3 and 103.
27159 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
27160 <__AVR_SHORT_CALLS__>: Document it.
27161 <__AVR_PM_BASE_ADDRESS__>: Document it.
27162 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
27163 (AVR Variable Attributes) <progmem>: Document this is
27164 not needed for avrxmega3.
27165 (AVR Named Address Spaces) <__flash>: Dito.
27167 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
27169 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
27171 2017-06-12 Doug Rupp <rupp@adacore.com>
27173 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
27174 Append vxworks-stdint.h to the tm_file list.
27175 * config/vxworks-stdint.h: New file.
27177 2017-06-12 Martin Liska <mliska@suse.cz>
27179 PR tree-optimization/81041
27180 * tree-profile.c (gimple_gen_ic_func_profiler):
27181 Create an extra BB in profile-generate
27182 (gimple_gen_time_profiler): Likewise.
27184 2017-06-12 Jakub Jelinek <jakub@redhat.com>
27186 PR tree-optimization/81003
27187 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
27188 (update_range_test): Use it instead of force_gimple_operand_gsi.
27190 2017-06-12 Richard Biener <rguenther@suse.de>
27192 PR tree-optimization/81053
27193 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
27194 with backedge value not defined in loop. Simplify def stmt
27197 2017-06-11 Tom de Vries <tom@codesourcery.com>
27200 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
27202 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
27203 nvptx_cannot_force_const_mem.
27205 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27207 * opts.c (finish_options): Move test for flag_split_stack after
27208 it has been initialized.
27210 2017-06-11 Jason Merrill <jason@redhat.com>
27212 * tree.h (id_equal): New.
27213 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
27214 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
27215 instead of strcmp of IDENTIFIER_POINTER.
27217 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27219 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
27220 (mark_all_inlined_calls_cdtor): Fix formating.
27221 (inline_transform): Rescale profile before inlining.
27223 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27225 * cgraph.h (cgraph_edge::clone): Update prototype.
27226 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
27227 (cgraph_node::create_clone): Update.
27228 (cgraph_node::create_version_clone): Update.
27229 * tree-inline.c (copy_bb): Update.
27230 (expand_call_inline): Update.
27232 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27234 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
27235 factored out from ...
27236 (rs6000_emit_prologue): ... here.
27238 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27240 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
27241 factored out from ...
27242 (rs6000_emit_prologue): ... here.
27244 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27246 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
27248 (handle_missing_profiles): Fix computation of tp_first_run.
27249 (counts_to_freqs): Do not touch freqs when count is 0.
27251 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27253 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
27256 2017-06-10 Tom de Vries <tom@codesourcery.com>
27258 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27259 attributes): Document signal effective target.
27261 2017-06-10 Tom de Vries <tom@codesourcery.com>
27263 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27264 Document effective target stack_size.
27266 2017-06-09 David Malcolm <dmalcolm@redhat.com>
27268 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
27269 to the edit_context if they can be auto-applied.
27271 2017-06-9 Ian Lance Taylor <iant@golang.org>
27273 * opts.c (finish_options): If -fsplit-stack, disable implicit
27274 -forder-blocks-and-partition.
27275 * doc/invoke.texi (Optimize Options): Document that when using
27276 -fsplit-stack -forder-blocks-and-partition is not implicitly
27279 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27281 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
27282 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
27283 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
27284 * builtins.def (abort, trap, unreachable): Declare cold.
27285 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
27286 * tree-core.h (ECF_COLD): New.
27287 * tree.c (set_call_expr_flags): Handle ECF_COLD.
27288 (build_common_builtin_nodes): Mark unreachable and abort as cold.
27290 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27292 * predict.c (unlikely_executed_stmt_p): Cleanup.
27294 2017-06-09 Richard Biener <rguenther@suse.de>
27296 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
27297 model if the ref is always written to.
27299 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27301 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
27303 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27305 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
27308 2017-06-09 Tom de Vries <tom@codesourcery.com>
27311 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
27312 "target cannot support label values" when encountering LABEL_REF.
27314 2017-06-09 Martin Liska <mliska@suse.cz>
27316 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
27317 (gimple_gen_ic_func_profiler): Emit direct comparison
27318 of __gcov_indirect_call_callee with NULL.
27319 (gimple_gen_time_profiler): Change probability from
27320 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
27322 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27324 * profile.c (edge_gcov_counts): Turn to pointer.
27325 (compute_branch_probabilities, compute_branch_probabilities): Update.
27326 (branch_prob): Do not clear edge_gcov_count.
27327 * profile.h (edge_gcov_counts): Turn to pointer.
27328 (edge_gcov_count): Update.
27330 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27332 * gimple.h (gimple_check_failed): Mark cold.
27334 2017-06-09 Richard Biener <rguenther@suse.de>
27336 PR tree-optimization/66623
27337 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
27338 refactor check_reduction into two parts, properly computing
27339 whether we have to check reduction validity for outer loop
27342 2017-06-09 Richard Biener <rguenther@suse.de>
27344 PR tree-optimization/79483
27345 * graphite-scop-detection.c (order): New global.
27346 (get_order): Compute bb to order mapping that satisfies code
27347 generation constraints.
27348 (cmp_pbbs): New helper.
27349 (build_scops): Start domwalk at entry block, sort generated
27352 2017-06-09 Richard Biener <rguenther@suse.de>
27354 PR middle-end/81007
27355 * ipa-polymorphic-call.c
27356 (ipa_polymorphic_call_context::restrict_to_inner_class):
27357 Skip FIELD_DECLs with error_mark_node type.
27358 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
27361 2017-06-09 Martin Liska <mliska@suse.cz>
27363 * predict.c (struct branch_predictor): New struct.
27364 (test_prediction_value_range): New test.
27365 (predict_c_tests): New function.
27366 * selftest-run-tests.c (selftest::run_tests): Run the function.
27367 * selftest.h: Declare new tests.
27369 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
27372 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
27373 gen_add3_insn did not fail.
27374 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
27375 r0, construct that number in a temporary reg and add that reg to r0.
27376 If asked to put the result in r0 as well, fail.
27378 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
27380 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27381 for early expansion of vec_eqv.
27383 2017-06-08 Jakub Jelinek <jakub@redhat.com>
27385 PR middle-end/81005
27386 * ubsan.c (instrument_null): Avoid pointless code temporary.
27387 (pass_ubsan::execute): Instrument aggregate arguments of calls.
27389 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
27393 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
27396 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
27399 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27401 * predict.c (unlikely_executed_edge_p): Move ahead.
27402 (probably_never_executed_edge_p): Use it.
27404 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27406 PR middle-end/79988
27407 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
27408 gimple_call_builtin_p call.
27410 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27412 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
27413 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
27414 rtl_check_failed_type2, rtl_check_failed_code1,
27415 rtl_check_failed_code2, rtl_check_failed_code_mode,
27416 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
27417 rtvec_check_failed_bounds, rtl_check_failed_flag,
27418 _fatal_insn_not_found, _fatal_insn): Likewise.
27419 * tree.h (tree_contains_struct_check_failed,
27420 tree_check_failed, tree_not_check_failed,
27421 tree_class_check_failed, tree_range_check_failed,
27422 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
27423 tree_vec_elt_check_failed, phi_node_elt_check_failed,
27424 tree_operand_check_failed, omp_clause_check_failed,
27425 omp_clause_operand_check_failed, omp_clause_range_check_failed):
27428 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27430 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
27431 flag_branch_probabilities.
27432 * ipa-inline.c (edge_badness): Likewise.
27433 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
27434 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
27435 * predict.c (maybe_hot_frequency_p): Likewise.
27436 (probably_never_executed): Likewise.
27437 * sched-ebb.c (schedule_ebbs): Likewise.
27438 * sched-rgn.c (find_single_block_region): Likewise.
27439 * tracer.c (tail_duplicate): Likewise.
27441 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27443 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
27444 longer requires x_flag_profile_use.
27446 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27448 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
27449 instead of flag_reorder_blocks_and_partition.
27450 * dbxout.c (dbxout_function_end): Likewise.
27451 * dwarf2out.c (gen_subprogram_die): Likewise.
27452 * haifa-sched.c (sched_create_recovery_edges): Likewise.
27453 * hw-doloop.c (reorg_loops): Likewise.
27454 * varasm.c (assemble_start_function,
27455 assemble_end_function): Likewise.
27456 (decide_function_section): Do not check for
27457 flag_reorder_blocks_and_partition.
27459 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27461 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27463 (chkp_get_hard_register_fake_addr_expr): Ditto.
27464 (chkp_build_addr_expr): Add check for hard reg case.
27465 (chkp_parse_array_and_component_ref): Ditto.
27466 (chkp_find_bounds_1): Ditto.
27467 (chkp_process_stmt): Don't generate bounds store for
27470 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27472 * predict.c (maybe_hot_bb_p): Do not check profile status.
27473 (maybe_hot_edge_p): Likewise.
27474 (probably_never_executed): Check for zero counts even if profile
27476 (unlikely_executed_edge_p): New function.
27477 (unlikely_executed_stmt_p): New function.
27478 (unlikely_executed_bb_p): New function.
27479 (set_even_probabilities): Use unlikely predicates.
27480 (combine_predictions_for_bb): Likewise.
27481 (predict_paths_for_bb): Likewise.
27482 (predict_paths_leading_to_edge): Likewise.
27483 (determine_unlikely_bbs): New function.
27484 (estimate_bb_frequencies): Use it.
27485 (compute_function_frequency): Use zero counts even if profile is
27487 * profile-count.h: Fix typo.
27489 2017-08-08 Julia Koval <julia.koval@intel.com>
27491 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
27492 _mm512_mask_cvtsepi16_storeu_epi8,
27493 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
27494 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
27495 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
27496 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
27497 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
27498 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
27499 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
27500 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
27501 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
27502 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
27503 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
27504 __builtin_ia32_pmovuswb256mem_mask,
27505 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
27506 __builtin_ia32_pmovwb512mem_mask): New builtins.
27508 2017-08-08 Julia Koval <julia.koval@intel.com>
27510 PR target/73350,80862
27511 * config/i386/subst.md (round): Fix round pattern.
27512 * config/i386/i386.c (ix86_erase_embedded_rounding):
27513 Fix erasing rounding for the fixed pattern.
27515 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27517 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
27519 2017-06-08 Martin Liska <mliska@suse.cz>
27521 PR gcov-profile/80911
27522 * gcov.c (block_info::block_info): New constructor.
27524 2017-06-07 Carl Love <cel@us.ibm.com>
27526 * config/rs6000/rs6000-c: The return type of the following
27527 built-in functions was implemented as int not long long. Fix sign
27528 of return value for the unsigned version of vec_mulo and vec_mule.
27529 vector unsigned long long vec_bperm (vector unsigned long long,
27530 vector unsigned char)
27531 vector signed long long vec_mule (vector signed int,
27533 vector unsigned long long vec_mule (vector unsigned int,
27534 vector unsigned int)
27535 vector signed long long vec_mulo (vector signed int,
27537 vector unsigned long long vec_mulo (vector unsigned int,
27538 vector unsigned int)
27539 * doc/extend.texi: Fix the documentation for the built-in
27542 2017-06-07 Carl Love <cel@us.ibm.com>
27545 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
27548 2017-06-07 Carl Love <cel@us.ibm.com>
27550 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
27551 support, Generate doublehv for signed int/float for BE case only.
27553 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
27555 * doc/invoke.texi (mcx16): Rewrite.
27557 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27559 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
27560 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
27561 *mov<mode>_softfloat, and an anonymous splitter): Use
27562 nonimmediate_operand instead of rs6000_nonimmediate_operand.
27564 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27566 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
27568 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27569 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
27570 (rs6000_debug_reg_global): Adjust.
27571 (rs6000_init_hard_regno_mode_ok): Adjust.
27572 (rs6000_dbx_register_number): Adjust.
27573 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
27574 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27575 Remove SPE_ACC and SPEFSCR.
27576 (REG_ALLOC_ORDER): Ditto.
27577 (FRAME_POINTER_REGNUM): Change to 111.
27578 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
27579 (REG_CLASS_NAMES): Ditto.
27580 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
27581 (REGISTER_NAMES): Ditto.
27582 (ADDITIONAL_REG_NAMES): Ditto.
27583 (rs6000_reg_names): Ditto.
27584 * config/rs6000/rs6000.md: Renumber some register number
27587 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27589 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
27591 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27592 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
27594 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
27595 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27596 Delete the SPE high registers.
27597 (REG_ALLOC_ORDER): Ditto.
27598 (enum reg_class): Remove SPE_HIGH_REGS.
27599 (REG_CLASS_NAMES): Ditto.
27600 (REG_CLASS_CONTENTS): Delete the SPE high registers.
27601 (REGISTER_NAMES): Ditto.
27602 (rs6000_reg_names): Ditto.
27603 * doc/tm.texi.in: Remove SPE as example.
27604 * doc/tm.texi: Regenerate.
27606 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27608 * config/rs6000/8540.md (ppc8540_brinc): Delete.
27609 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
27610 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
27611 * config/rs6000/rs6000.md (type): Remove "brinc".
27613 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27615 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
27616 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
27617 * config/rs6000/linuxspe.h: Delete file.
27618 * config/rs6000/rs6000.md: Don't include spe.md.
27619 * config/rs6000/spe.h: Delete file.
27620 * config/rs6000/spe.md: Delete file.
27621 * config/rs6000/t-rs6000: Remove spe.md.
27623 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27625 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
27626 (reg_or_none500mem_operand): Delete.
27627 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
27628 instead of reg_or_none500mem_operand.
27630 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27632 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27633 handling of SPE flags.
27634 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
27636 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27638 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
27640 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
27641 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
27642 paired_divv2sf3): Similar.
27643 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
27644 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
27645 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
27647 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
27648 Rename the paired_* instruction patterns.
27649 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
27651 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
27652 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
27653 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
27654 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
27655 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
27656 PAIRED_VECTOR_MODE.
27657 (struct machine_function): Delete field spe_insn_chain_scanned_p.
27658 (spe_func_has_64bit_regs_p): Delete.
27659 (spe_expand_predicate_builtin): Delete.
27660 (spe_expand_evsel_builtin): Delete.
27661 (TARGET_DWARF_REGISTER_SPAN): Do not define.
27662 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
27663 (invalid_e500_subreg): Delete.
27664 (rs6000_legitimize_address): Always force_reg op2 as well, for
27665 paired single memory accesses.
27666 (rs6000_member_type_forces_blk): Delete.
27667 (rs6000_spe_function_arg): Delete.
27668 (rs6000_expand_unop_builtin): Delete SPE handling.
27669 (rs6000_expand_binop_builtin): Ditto.
27670 (spe_expand_stv_builtin): Delete.
27671 (bdesc_2arg_spe): Delete.
27672 (spe_expand_builtin): Delete.
27673 (spe_expand_predicate_builtin): Delete.
27674 (spe_expand_evsel_builtin): Delete.
27675 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
27676 (spe_init_builtins): Delete.
27677 (spe_func_has_64bit_regs_p): Delete.
27678 (savres_routine_name): Delete "info" parameter. Adjust callers.
27679 (rs6000_emit_stack_reset): Ditto.
27680 (rs6000_dwarf_register_span): Delete.
27681 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
27682 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
27683 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
27685 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
27687 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
27688 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
27689 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
27690 mulv2sf3, divv2sf3): Delete expanders.
27692 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27694 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
27696 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27698 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
27699 * config/rs6000/rs6000.c: Ditto.
27701 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27703 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
27704 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
27705 comparison_operator.
27707 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27709 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
27710 * config/rs6000/rs6000.opt: Ditto.
27711 * config/rs6000/t-rtems: Ditto.
27713 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27715 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
27716 TARGET_E500_SINGLE by 0, simplify.
27717 * config/rs6000/rs6000.c: Ditto.
27718 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
27719 (spe_build_register_parallel): Delete.
27720 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
27721 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
27722 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
27723 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
27724 (E500_CONVERT): Delete.
27725 * config/rs6000/spe.md: Remove many patterns and all define_constants.
27727 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27729 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
27730 * config/rs6000/dfp.md: Ditto.
27731 (negdd2, *negdd2_fpr): Merge.
27732 (absdd2, *absdd2_fpr): Merge.
27733 (negtd2, *negtd2_fpr): Merge.
27734 (abstd2, *abstd2_fpr): Merge.
27735 * config/rs6000/e500.h: Delete file.
27736 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
27737 TARGET_FPRS by 1 and simplify.
27738 * config/rs6000/rs6000-c.c: Ditto.
27739 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
27740 TARGET_DF_SPE by 0.
27741 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
27743 * config/rs6000/rs6000.md: Ditto.
27744 (floatdidf2, *floatdidf2_fpr): Merge.
27745 (move_from_CR_gt_bit): Delete.
27746 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
27747 (E500_CR_IOR_COMPARE): Delete.
27748 (All patterns that require !TARGET_FPRS): Delete.
27749 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
27751 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27753 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
27755 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27757 * graphds.c (add_edge): Intitialize edge's attached data.
27758 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
27759 pointer parameter. Call pointed function on each edge during
27760 graph traversing. Skip traversing the edge when the function
27762 (graphds_dfs, graphds_scc): Ditto.
27763 (for_each_edge): New parameter. Pass the new parameter to callback
27765 * graphds.h (skip_edge_callback): New function pointer type.
27766 (graphds_dfs, graphds_scc): New function pointer parameter.
27767 (graphds_edge_callback, for_each_edge): New parameter.
27769 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27771 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
27772 out code checking if runtime alias check is possible to below ...
27773 Call the new function.
27774 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
27775 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
27777 2017-06-07 Marek Polacek <polacek@redhat.com>
27780 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
27781 TYPE_OVERFLOW_WRAPS check.
27783 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27785 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
27786 if versioning is required.
27787 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
27788 peeling with the check for versioning.
27790 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27792 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
27793 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
27794 Set true to new parameter if new ssa variable is defined.
27795 (vect_gen_vector_loop_niters): Refactor. Set range information
27796 for the new vector loop bound variable.
27797 (vect_do_peeling): Ditto.
27799 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27801 * tree-affine.c (ssa.h): Include header file.
27802 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
27803 has wrapping overflow behavior.
27805 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27807 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
27809 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27811 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
27812 (tree_to_aff_combination): ... here.
27814 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27816 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
27817 reg_pressure model function.
27818 (ivopts_global_cost_for_size): Delete.
27819 (determine_set_costs, iv_ca_recount_cost): Call new model function
27820 ivopts_estimate_reg_pressure.
27822 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27824 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
27825 expensive than udiv. Remove floating point cases from mod.
27827 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27829 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
27830 Increase idiv cost.
27832 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27834 * config/aarch64/aarch64.md
27835 (copysignsf3): Fix mask generation.
27837 2017-06-07 Jakub Jelinek <jakub@redhat.com>
27839 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
27841 (class dump_manager): Add register_dumps method.
27842 * dumpfile.c: Include langhooks.h.
27843 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
27844 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
27845 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
27846 (dump_manager::dump_register): Start with 512 entries instead of 32.
27847 (dump_manager::register_dumps): New method.
27848 * toplev.c (general_init): Instead of invoking register_dumps
27849 langhook, invoke register_dumps method on the dump manager.
27850 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
27853 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
27855 * doc/md.texi: Clarify the restrictions on a define_insn condition.
27856 Say that # requires an associated define_split to exist, and that
27857 the define_split must be suitable for use after register allocation.
27859 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27861 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
27862 (compute_outgoing_frequencies): Also initialize zero counts.
27863 (find_many_sub_basic_blocks): Do not produce uninitialized profile
27864 around loops; preserve more of profile when nothing changes.
27866 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
27868 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
27870 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
27871 * config/arm/arm-cpu-cdata.h: Regenerate.
27872 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
27873 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
27874 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
27875 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
27876 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
27878 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
27880 * config/arm/t-rmprofile: Likewise.
27881 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
27883 2017-06-06 David S. Miller <davem@davemloft.net>
27886 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
27887 blockage if function uses alloca.
27889 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27891 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
27892 New "uid" fields to hold pretty-print IDs of group and ref.
27893 Memory references are now identified as <group_id>:<ref_id>
27894 instead of using [random] addresses.
27895 (dump_mem_details): Simplify, no functional change.
27896 (dump_mem_ref): Simplify and make output more concise.
27897 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
27898 (find_or_create_group): Initialize group uid.
27899 (record_ref): Initialize ref uid. Improve debug output.
27900 (prune_group_by_reuse, should_issue_prefetch_p,)
27901 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
27902 (mark_nontemporal_store, determine_loop_nest_reuse):
27903 Improve debug output.
27905 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27907 * dbgcnt.def (prefetch): New debug counter.
27908 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
27909 (schedule_prefetches): Stop issueing prefetches if debug counter
27912 2017-06-06 Tom de Vries <tom@codesourcery.com>
27914 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
27915 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
27917 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27919 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
27920 Use aarch64_reg_or_zero predicate for operand 4.
27921 (aarch64_compare_and_swap<mode> define_insn_and_split):
27922 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
27923 (aarch64_store_exclusive<mode>): Likewise for operand 2.
27925 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
27927 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
27928 (arm_compute_save_core_reg_mask): This.
27929 (thumb1_compute_save_reg_mask): Rename into ...
27930 (thumb1_compute_save_core_reg_mask): This.
27931 (arm_compute_save_reg0_reg12_mask): Adapt comment.
27932 (arm_compute_frame_layout): Likewise.
27934 2017-06-06 Richard Biener <rguenther@suse.de>
27936 PR tree-optimization/80974
27937 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
27938 keep or clear leaders SSA info.
27940 2017-06-06 Tom de Vries <tom@codesourcery.com>
27942 * config/nvptx/nvptx.c (split_mode_p): New function.
27943 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
27945 2017-06-06 Tom de Vries <tom@codesourcery.com>
27947 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
27949 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27952 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
27955 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27957 * shrink-wrap.c (handle_simple_exit): Update profile.
27958 (try_shrink_wrapping): Upate profile.
27960 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27962 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
27963 (tree_guess_outgoing_edge_probabilities): New.
27964 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
27965 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
27967 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27969 * ipa-split.c (split_function): Initialize return bb profile.
27971 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27973 * profile.c (compute_branch_probabilities): Also initialize
27974 EXIT_BLOCK profile.
27976 2017-06-06 Richard Biener <rguenther@suse.de>
27978 PR tree-optimization/80928
27979 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
27980 (vect_analyze_loop_operations): Properly guard analysis for
27982 (vect_transform_loop): Likewise.
27983 (vect_analyze_loop_2): Also reset SLP type on PHIs.
27984 (vect_model_induction_cost): Do not cost for pure SLP.
27985 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
27986 of induction in inner loop vectorization.
27987 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
27988 (vect_get_and_check_slp_defs): Handle vect_induction_def.
27989 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
27991 (vect_analyze_slp_cost_1): Cost induction.
27992 (vect_detect_hybrid_slp_stmts): Handle PHIs.
27993 (vect_get_slp_vect_defs): Likewise.
27994 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
27995 (vect_transform_stmt): Handle SLP reductions.
27996 * tree-vectorizer.h (vectorizable_induction): Adjust.
27998 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28000 * config/rs6000/rs6000.c (make_resolver_func): Update
28001 init_lowered_empty_function call.
28003 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
28005 * doc/invoke.texi: Document the -fprofile-abs-path option.
28006 * common.opt (fprofile-abs-path): New option.
28007 * gcov-io.h (gcov_write_filename): Declare.
28008 * gcov-io.c (gcov_write_filename): New function.
28009 * coverage.c (coverage_begin_function): Use gcov_write_filename.
28010 * profile.c (output_location): Likewise.
28012 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28014 * shring-wrap.c: Revert accidental commit.
28016 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
28018 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
28020 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28022 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
28024 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
28025 profile in callgraph edge.
28026 * profile-count.h (apply_probability): If THIS is 0, then result is 0
28027 (apply_scale): Likewise.
28028 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
28029 Also scale profile when inlining function with zero profile.
28030 (initialize_cfun): Update exit block profile even when it is zero.
28031 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
28032 when profile is read.
28034 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28036 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
28037 (CLONE_*): New constants to define the processors we can generate
28038 code for with the target_clone attribute.
28039 (rs6000_clone_map): New array to identify which clone processors
28040 the current program is running on.
28041 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
28042 target_clone attribute.
28043 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
28044 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
28045 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
28046 (cpu_expand_builtin): Add support for target_clone attribute.
28047 (rs6000_valid_attribute_p): Allow "default" attribute.
28048 (get_decl_name): New debug function to simplify printing the
28049 current function name in debugging statements.
28050 (rs6000_clone_priority): New functions to support the target_clone
28051 attribute, and be able to generate code to switch between ISA 2.05
28052 through ISA 3.0 (power6 through power9).
28053 (rs6000_compare_version_priority): Likewise.
28054 (rs6000_get_function_versions_dispatcher): Likewise.
28055 (make_resolver_func): Likewise.
28056 (add_condition_to_bb): Likewise.
28057 (dispatch_function_versions): Likewise.
28058 (rs6000_generate_version_dispatcher_body): Likewise.
28059 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
28060 (fusion_gpr_load_p): Fix a spacing issue.
28061 * doc/extend.texi (Common Function Attributes): Document that the
28062 PowerPC supports the target_clone attribute.
28064 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28066 * config/arm/arm.h: explain F symbol found in description of ARM
28067 register allocation in its legend.
28069 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28071 * config/mips/frame-header-opt.c: Include profile-count.h.
28072 * config/riscv/riscv.c: Include profile-count.h
28074 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28076 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
28078 (sm_set_flag_if_changed): Add bbs field.
28079 (execute_sm_if_changed_flag_set): Pass BBS.
28080 (execute_sm): Update.
28082 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28084 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
28087 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28089 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
28090 (peephole2): New peephole2 to emit the above.
28091 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
28093 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28095 * config/aarch64/aarch64.c (define_peephole2 above
28096 *sub_<shift>_<mode>): New peephole.
28098 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28100 * config/i386/i386.c (make_resolver_func): Update.
28101 * Makefile.in: Add profile-count.h and profile-count.o
28102 * auto-profile.c (afdo_indirect_call): Update to new API.
28103 (afdo_set_bb_count): Update.
28104 (afdo_propagate_edge): Update.
28105 (afdo_propagate_circuit): Update.
28106 (afdo_calculate_branch_prob): Update.
28107 (afdo_annotate_cfg): Update.
28108 * basic-block.h: Include profile-count.h
28109 (struct edge_def): Turn count to profile_count.
28110 (struct basic_block_def): Likewie.
28111 (REG_BR_PROB_BASE): Move to profile-count.h
28112 (RDIV): Move to profile-count.h
28113 * bb-reorder.c (max_entry_count): Turn to profile_count.
28114 (find_traces): Update.
28115 (rotate_loop):Update.
28116 (connect_traces):Update.
28117 (sanitize_hot_paths):Update.
28118 * bt-load.c (migrate_btr_defs): Update.
28119 * cfg.c (RDIV): Remove.
28120 (init_flow): Use alloc_block.
28121 (alloc_block): Uninitialize count.
28122 (unchecked_make_edge): Uninitialize count.
28123 (check_bb_profile): Update.
28124 (dump_edge_info): Update.
28125 (dump_bb_info): Update.
28126 (update_bb_profile_for_threading): Update.
28127 (scale_bbs_frequencies_int): Update.
28128 (scale_bbs_frequencies_gcov_type): Update.
28129 (scale_bbs_frequencies_profile_count): New.
28130 * cfg.h (update_bb_profile_for_threading): Update.
28131 (scale_bbs_frequencies_profile_count): Declare.
28132 * cfgbuild.c (compute_outgoing_frequencies): Update.
28133 (find_many_sub_basic_blocks): Update.
28134 * cfgcleanup.c (try_forward_edges): Update.
28135 (try_crossjump_to_edge): Update.
28136 * cfgexpand.c (expand_gimple_tailcall): Update.
28137 (construct_exit_block): Update.
28138 * cfghooks.c (verify_flow_info): Update.
28139 (dump_bb_for_graph): Update.
28140 (split_edge): Update.
28141 (make_forwarder_block): Update.
28142 (duplicate_block): Update.
28143 (account_profile_record): Update.
28144 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
28145 (get_estimated_loop_iterations): Update.
28146 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
28147 (single_likely_exit): Update.
28148 * cfgloopmanip.c (scale_loop_profile): Update.
28150 (set_zero_probability): Update.
28151 (lv_adjust_loop_entry_edge): Update.
28152 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
28153 (purge_dead_edges): Update.
28154 (rtl_account_profile_record): Update.
28155 * cgraph.c (cgraph_node::create): Uninitialize count.
28156 (symbol_table::create_edge): Uninitialize count.
28157 (cgraph_update_edges_for_call_stmt_node): Update.
28158 (cgraph_edge::dump_edge_flags): Update.
28159 (cgraph_node::dump): Update.
28160 (cgraph_edge::maybe_hot_p): Update.
28161 * cgraph.h: Include profile-count.h
28162 (create_clone), create_edge, create_indirect_edge): Update.
28163 (cgraph_node): Turn count to profile_count.
28164 (cgraph_edge0: Likewise.
28165 (make_speculative, clone): Update.
28166 (create_edge): Update.
28167 (init_lowered_empty_function): Update.
28168 * cgraphclones.c (cgraph_edge::clone): Update.
28169 (duplicate_thunk_for_node): Update.
28170 (cgraph_node::create_clone): Update.
28171 * cgraphunit.c (cgraph_node::analyze): Update.
28172 (cgraph_node::expand_thunk): Update.
28173 * final.c (dump_basic_block_info): Update.
28174 * gimple-streamer-in.c (input_bb): Update.
28175 * gimple-streamer-out.c (output_bb): Update.
28176 * graphite.c (print_global_statistics): Update.
28177 (print_graphite_scop_statistics): Update.
28178 * hsa-brig.c: Include basic-block.h.
28179 * hsa-dump.c: Include basic-block.h.
28180 * hsa-gen.c (T sum_slice): Update.
28181 (convert_switch_statements):Update.
28182 * hsa-regalloc.c: Include basic-block.h.
28183 * ipa-chkp.c (chkp_produce_thunks): Update.
28184 * ipa-cp.c (struct caller_statistics): Update.
28185 (init_caller_stats): Update.
28186 (gather_caller_stats): Update.
28187 (ipcp_cloning_candidate_p): Update.
28188 (good_cloning_opportunity_p): Update.
28189 (get_info_about_necessary_edges): Update.
28190 (dump_profile_updates): Update.
28191 (update_profiling_info): Update.
28192 (update_specialized_profile): Update.
28193 (perhaps_add_new_callers): Update.
28194 (decide_about_value): Update.
28195 (ipa_cp_c_finalize): Update.
28196 * ipa-devirt.c (struct odr_type_warn_count): Update.
28197 (struct decl_warn_count): Update.
28198 (struct final_warning_record): Update.
28199 (possible_polymorphic_call_targets): Update.
28200 (ipa_devirt): Update.
28201 * ipa-fnsummary.c (redirect_to_unreachable): Update.
28202 * ipa-icf.c (sem_function::merge): Update.
28203 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
28204 * ipa-inline.c (compute_uninlined_call_time): Update.
28205 (compute_inlined_call_time): Update.
28206 (want_inline_small_function_p): Update.
28207 (want_inline_self_recursive_call_p): Update.
28208 (edge_badness): Update.
28209 (lookup_recursive_calls): Update.
28210 (recursive_inlining): Update.
28211 (inline_small_functions): Update.
28212 (dump_overall_stats): Update.
28213 (dump_inline_stats): Update.
28214 * ipa-profile.c (ipa_profile_generate_summary): Update.
28215 (ipa_propagate_frequency): Update.
28216 (ipa_profile): Update.
28217 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28218 * ipa-utils.c (ipa_merge_profiles): Update.
28219 * loop-doloop.c (doloop_modify): Update.
28220 * loop-unroll.c (report_unroll): Update.
28221 (unroll_loop_runtime_iterations): Update.
28222 * lto-cgraph.c (lto_output_edge): Update.
28223 (lto_output_node): Update.
28224 (input_node): Update.
28225 (input_edge): Update.
28226 (merge_profile_summaries): Update.
28227 * lto-streamer-in.c (input_cfg): Update.
28228 * lto-streamer-out.c (output_cfg): Update.
28229 * mcf.c (create_fixup_graph): Update.
28230 (adjust_cfg_counts): Update.
28231 (sum_edge_counts): Update.
28232 * modulo-sched.c (sms_schedule): Update.
28233 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
28234 * predict.c (maybe_hot_count_p): Update.
28235 (probably_never_executed): Update.
28236 (dump_prediction): Update.
28237 (combine_predictions_for_bb): Update.
28238 (propagate_freq): Update.
28239 (handle_missing_profiles): Update.
28240 (counts_to_freqs): Update.
28241 (rebuild_frequencies): Update.
28242 (force_edge_cold): Update.
28243 * predict.h: Include profile-count.h
28244 (maybe_hot_count_p, counts_to_freqs): UPdate.
28245 * print-rtl-function.c: Do not include cfg.h
28246 * print-rtl.c: Include basic-block.h
28247 * profile-count.c: New file.
28248 * profile-count.h: New file.
28249 * profile.c (is_edge_inconsistent): Update.
28250 (correct_negative_edge_counts): Update.
28251 (is_inconsistent): Update.
28252 (set_bb_counts): Update.
28253 (read_profile_edge_counts): Update.
28254 (compute_frequency_overlap): Update.
28255 (compute_branch_probabilities): Update; Initialize and deinitialize
28257 (branch_prob): Update.
28258 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
28259 (edge_gcov_count): New.
28260 (bb_gcov_count): New.
28261 * shrink-wrap.c (try_shrink_wrapping): Update.
28262 * tracer.c (better_p): Update.
28263 * trans-mem.c (expand_transaction): Update.
28264 (ipa_tm_insert_irr_call): Update.
28265 (ipa_tm_insert_gettmclone_call): Update.
28266 * tree-call-cdce.c: Update.
28267 * tree-cfg.c (gimple_duplicate_sese_region): Update.
28268 (gimple_duplicate_sese_tail): Update.
28269 (gimple_account_profile_record): Update.
28270 (execute_fixup_cfg): Update.
28271 * tree-inline.c (copy_bb): Update.
28272 (copy_edges_for_bb): Update.
28273 (initialize_cfun): Update.
28274 (freqs_to_counts): Update.
28275 (copy_cfg_body): Update.
28276 (expand_call_inline): Update.
28277 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
28278 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
28279 (try_unroll_loop_completely): Update.
28280 (try_peel_loop): Update.
28281 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
28282 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
28283 * tree-ssa-loop-split.c (connect_loops): Update.
28284 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
28285 * tree-ssa-reassoc.c (branch_fixup): Update.
28286 * tree-ssa-tail-merge.c (replace_block_by): Update.
28287 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
28288 (compute_path_counts): Update.
28289 (update_profile): Update.
28290 (recompute_probabilities): Update.
28291 (update_joiner_offpath_counts): Update.
28292 (estimated_freqs_path): Update.
28293 (freqs_to_counts_path): Update.
28294 (clear_counts_path): Update.
28295 (ssa_fix_duplicate_block_edges): Update.
28296 (duplicate_thread_path): Update.
28297 * tree-switch-conversion.c (case_bit_test_cmp): Update.
28298 (struct switch_conv_info): Update.
28299 * tree-tailcall.c (decrease_profile): Update.
28300 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
28301 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
28302 * value-prof.c (check_counter): Update.
28303 (gimple_divmod_fixed_value): Update.
28304 (gimple_mod_pow2): Update.
28305 (gimple_mod_subtract): Update.
28306 (gimple_ic_transform): Update.
28307 (gimple_stringop_fixed_value): Update.
28308 * value-prof.h (gimple_ic): Update.
28310 2017-06-02 Carl Love <cel@us.ibm.com>
28312 * config/rs6000/rs6000-c: Add support for built-in functions
28313 vector double vec_doublee (vector signed int);
28314 vector double vec_doublee (vector unsigned int);
28315 vector double vec_doublee (vector float);
28316 vector double vec_doubleh (vector signed int);
28317 vector double vec_doubleh (vector unsigned int);
28318 vector double vec_doubleh (vector float);
28319 vector double vec_doublel (vector signed int);
28320 vector double vec_doublel (vector unsigned int);
28321 vector double vec_doublel (vector float);
28322 vector double vec_doubleo (vector signed int);
28323 vector double vec_doubleo (vector unsigned int);
28324 vector double vec_doubleo (vector float);.
28325 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
28326 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
28328 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
28329 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
28330 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
28332 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
28333 vec_doublel, vec_doubleh.
28334 * doc/extend.texi: Update the built-in documentation file for the
28335 new built-in functions.
28337 2017-06-02 David Malcolm <dmalcolm@redhat.com>
28340 * ipa-inline-analysis.c (free_growth_caches): Set
28341 edge_removal_hook_holder to NULL after removing it.
28343 2017-06-02 Sudakshina Das <sudi.das@arm.com>
28345 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
28346 comparision with zero.
28348 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28349 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28350 for early expansion of vec_min and vec_max builtins.
28351 (builtin_function_type): Add min/max unsigned variants to those
28352 identified as having unsigned arguments.
28354 2017-06-02 Olivier Hainque <hainque@adacore.com>
28356 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
28358 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28360 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
28361 Use VALL_F16 iterator rather than VALL.
28363 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28365 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
28366 Emit CBNZ inside loop when doing a strong exchange and comparing
28367 against zero. Generate the CC flags after the loop.
28369 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
28371 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
28372 (dl_section_ref): New.
28373 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
28374 On AIX, append an expression to subtract the size of the
28375 section length to dl_section_ref.
28377 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28379 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28380 for early expansion of vector absolute builtins.
28382 2017-06-02 Richard Biener <rguenther@suse.de>
28384 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
28385 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
28387 2017-06-02 Richard Biener <rguenther@suse.de>
28389 PR tree-optimization/80948
28390 * tree-tailcall.c (find_tail_calls): Track stmts to move in
28391 stmt order as well.
28393 2017-06-02 Richard Biener <rguenther@suse.de>
28395 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
28397 * tree-vect-stmts.c (process_use): Do not mark backedge defs
28398 for inductions as relevant.
28400 2017-06-02 Richard Biener <rguenther@suse.de>
28402 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
28403 (vectorizable_induction): ... this. Remove dead code.
28405 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
28407 * builtins. (expand_builtin_alloca): Remove second parameter and
28408 infer its value from the first parameter instead.
28409 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
28411 2017-06-02 Jakub Jelinek <jakub@redhat.com>
28413 PR rtl-optimization/80903
28414 * loop-doloop.c (add_test): Unshare sequence.
28416 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
28418 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
28420 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
28422 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
28424 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
28425 xlogue_layout::get_instance, logue_layout::xlogue_layout,
28426 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
28427 (xlogue_layout::get_stub_rtx): Make static.
28428 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
28429 (xlogue_layout::compute_stub_managed_regs): Rename to...
28430 (xlogue_layout::count_stub_managed_regs): ...this.
28431 (xlogue_layout::is_stub_managed_reg): New function.
28432 (xlogue_layout::m_stub_names): Rename to...
28433 (xlogue_layout::s_stub_names): ...this, make static.
28434 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
28435 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
28436 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
28437 xlogue_layout::s_stub_names): Instantiate statics.
28438 (stub_managed_regs): Remove.
28439 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
28440 (disable_call_ms2sysv_xlogues): Rename to...
28441 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
28442 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
28444 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
28445 change after reload_completed.
28446 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
28448 (ix86_expand_prologue): Likewise.
28449 (ix86_expand_epilogue): Likewise.
28450 (ix86_expand_split_stack_prologue): Likewise.
28451 (ix86_compute_frame_layout): Remove frame parameter ...
28452 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
28453 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
28455 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
28456 (ix86_frame): Move from here ...
28457 * config/i386/i386.h (ix86_frame): ... to here.
28458 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
28459 complete ix86_frame data structure instead. Remove some_ld_name.
28461 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28463 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
28464 symbols that hold a DECL_VALUE_EXPR.
28466 2017-06-01 Martin Jambor <mjambor@suse.cz>
28468 PR tree-optimization/80898
28469 * tree-sra.c (process_subtree_disqualification): Removed.
28470 (disqualify_candidate): Do not acll
28471 process_subtree_disqualification.
28472 (subtree_mark_written_and_enqueue): New function.
28473 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
28474 RHS has been disqualified and re-queue LHS if necessary. Apart
28475 from that, ignore disqualified RHS.
28477 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28479 * config/s390/s390.c (s390_emit_epilogue): Disable early return
28480 address fetch for z10 or later.
28482 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28484 * config/arc/arc.md (tst_movb): Add guard when splitting.
28486 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28488 * config/arc/arc.c (arc_can_eliminate): Test against
28489 arc_frame_pointer_needed.
28491 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28493 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
28494 to prevent store reordering.
28495 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
28496 (type): Add block type.
28497 (stack_tie): Define special instruction to be used in
28500 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28502 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
28503 constraint. It is not valid for the pattern.
28504 (noncommutative_binary_comparison): Likewise.
28506 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28508 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
28511 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28513 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
28514 be used by the reg-alloc.
28516 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28518 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
28519 reg-alloc when having mul64 or mul32x16 instructions.
28520 (mulsidi3): Likewise.
28521 (umulsidi3): Likewise.
28522 (mulsi32x16): New pattern.
28523 (mulsi64): Likewise.
28524 (mulsidi64): Likewise.
28525 (umulsidi64): Likewise.
28526 (MUL32x16_REG): Define.
28527 (mul64_600): Use MUL32x16_REG.
28528 (mac64_600): Likewise.
28529 (umul64_600): Likewise.
28530 (umac64_600): Likewise.
28532 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28534 * config/arc/arc.md (mulsi3_700): Make it commutative.
28536 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
28538 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
28540 (*sign_extendsidi2_insn): Likewise for movstosw.
28542 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28544 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
28545 the type of the input discriminant value. Convert the
28546 discriminant value of signedness vary.
28548 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
28550 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
28551 Add to -Wall section.
28553 2017-06-01 Richard Biener <rguenther@suse.de>
28555 PR middle-end/66313
28556 * fold-const.c (fold_plusminus_mult_expr): If the factored
28557 factor may be zero use a wrapping type for the inner operation.
28558 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
28559 and handle moved defs.
28560 (process_assignment): Properly guard the unary op case. Return a
28561 tri-state indicating that moving the stmt before the call may allow
28562 to continue. Pass through to_move.
28563 (find_tail_calls): Handle moving unrelated defs before
28566 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
28569 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
28570 splitter result in the canonical way.
28572 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
28574 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
28575 also for 32bit target. Update insn attributes.
28576 (zero-extendsidi2 splitter): Allow all registers for operand 1.
28578 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
28580 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
28581 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
28582 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
28583 (_mm_maskz_min_ss): New intrinsics.
28585 2017-05-31 Martin Liska <mliska@suse.cz>
28587 * tree-vect-loop.c (vect_create_epilog_for_reduction):
28588 Change comment style to one we normally use.
28589 (vectorizable_reduction): Likewise.
28590 (vectorizable_induction): Likewise.
28591 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
28592 (vectorizable_call): Likewise.
28593 (vectorizable_simd_clone_call): Likewise.
28594 (vectorizable_conversion): Likewise.
28595 (vectorizable_assignment): Likewise.
28596 (vectorizable_shift): Likewise.
28597 (vectorizable_operation): Likewise.
28598 (vectorizable_store): Likewise.
28599 (vectorizable_load): Likewise.
28600 * tree-vectorizer.h: Likewise.
28602 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
28604 * passes.c (emergency_dump_function): New.
28605 * tree-pass.h (emergency_dump_function): Declare.
28606 * plugin.c (plugins_internal_error_function): Remove.
28607 * plugin.h (plugins_internal_error_function): Remove declaration.
28608 * toplev.c (internal_error_function): New static function. Use it...
28609 (general_init): ...here.
28611 2017-05-31 Graham Markall <graham.markall@embecosm.com>
28613 * config/arc/arc.c (arc_print_operand): Handle constant operands.
28614 (arc_rtx_costs): Add costs for new patterns.
28615 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
28616 * config/arc/predicates.md: Add _1_2_3_operand predicate.
28618 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28620 * tree-ssa-strlen.c (get_next_strinfo): New function.
28621 (get_stridx_plus_constant): Use it.
28622 (zero_length_string): Likewise.
28623 (adjust_related_strinfos): Likewise.
28624 (adjust_last_stmt): Likewise.
28626 2017-05-31 Richard Biener <rguenther@suse.de>
28629 * config/i386/i386.c (ix86_expand_builtin): Remove assert
28630 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
28632 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28634 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
28635 loop_vinfo argument and use of dependence distance vectors.
28636 Check instead whether the two references differ only in their
28637 initial value and assume that they have the same alignment if the
28638 difference is a multiple of the vector alignment.
28639 (vect_analyze_data_refs_alignment): Update call accordingly.
28641 2017-05-31 Martin Liska <mliska@suse.cz>
28644 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
28646 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28648 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
28649 (create_intersect_range_checks): Move from ...
28650 * tree-data-ref.c (create_intersect_range_checks_index)
28651 (create_intersect_range_checks): ... to here.
28652 (create_runtime_alias_checks): New function factored from ...
28653 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
28654 here. Call above function.
28655 * tree-data-ref.h (create_runtime_alias_checks): New function.
28657 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28659 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
28660 segment length for dr_b and compute it in wide_int.
28662 2017-05-31 Richard Biener <rguenther@suse.de>
28664 PR tree-optimization/80906
28665 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
28666 and pass through iv_map.
28667 (copy_bb_and_scalar_dependences): Adjust.
28668 (translate_pending_phi_nodes): Likewise.
28669 (copy_loop_close_phi_args): Handle code-generating IVs instead
28672 2017-05-30 David Malcolm <dmalcolm@redhat.com>
28674 * diagnostic-color.c (color_dict): Add "type-diff".
28675 (parse_gcc_colors): Update comment.
28676 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
28677 -fdiagnostics-show-template-tree and -fno-elide-type.
28678 (GCC_COLORS): Add type-diff to example.
28680 (-fdiagnostics-show-template-tree): New.
28681 (-fno-elide-type): New.
28682 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
28683 the pp_format_decoder callback. Call any m_format_postprocessor's
28685 (pretty_printer::pretty_printer): Initialize
28686 m_format_postprocessor.
28687 (pretty_printer::~pretty_printer): Delete any
28688 m_format_postprocessor.
28689 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
28690 (class format_postprocessor): New class.
28691 (struct pretty_printer::format_decoder): Document the new parameters.
28692 (struct pretty_printer::m_format_postprocessor): New field.
28693 * tree-diagnostic.c (default_tree_printer): Update for new
28694 bool and const char ** params.
28695 * tree-diagnostic.h (default_tree_printer): Likewise.
28697 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
28699 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
28700 (lwa_operand): Delete rs6000_gen_cell_microcode test.
28701 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28702 rs6000_gen_cell_microcode code.
28703 (rs6000_final_prescan_insn): Delete.
28704 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
28705 "warn-cell-microcode" entries.
28706 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
28707 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
28708 throughout. Change cc_reg_not_micro_cr0_operand to
28709 cc_reg_not_cr0_operand throughout.
28710 (*extendhi<mode>2_noload): Delete.
28711 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
28712 (mwarn-cell-microcode): Delete.
28713 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
28714 -mgen-cell-microcode and -mwarn-cell-microcode.
28716 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
28719 * config/i386/constraints.md (Yd): New constraint.
28721 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
28722 and (?Yd, r) alternatives. Update insn attributes.
28723 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
28724 and (?*Yd, r) alternatives. Update insn attributes.
28725 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
28727 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
28729 * gimplify.c (gimplify_modify_expr): Don't create a
28730 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
28733 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
28735 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
28737 2017-05-30 Richard Biener <rguenther@suse.de>
28739 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
28740 and reduc_def fields.
28741 (STMT_VINFO_REDUC_TYPE): New define.
28742 (STMT_VINFO_REDUC_DEF): Likewise.
28743 (vect_force_simple_reduction): Adjust prototype.
28744 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
28745 (vect_is_simple_reduction): Remove check_reduction argument.
28746 (vect_force_simple_reduction): Adjust and set
28747 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28748 (vectorizable_reduction): Do not re-do reduction analysis
28749 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28750 * tree-parloops.c (gather_scalar_reductions): Adjust.
28752 2017-05-30 Richard Biener <rguenther@suse.de>
28754 PR middle-end/80901
28755 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
28758 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
28760 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
28761 Introduce unknown_misalignment parameter and remove vf.
28762 (vect_peeling_hash_get_lowest_cost):
28763 Pass unknown_misalignment parameter.
28764 (vect_enhance_data_refs_alignment):
28765 Fix unsupportable data ref treatment.
28767 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28769 * tree-vect-data-refs.c (vect_get_data_access_cost):
28770 Workaround for SLP handling.
28771 (vect_enhance_data_refs_alignment):
28772 Compute costs for doing no peeling at all, compare to the best
28773 peeling costs so far and avoid peeling if cheaper.
28775 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28777 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
28778 Return peeling info and set costs to zero for unlimited cost
28780 (vect_enhance_data_refs_alignment): Also inspect all datarefs
28781 with unknown misalignment. Compute and costs for unknown
28782 misalignment, compare them to the costs for known misalignment
28783 and choose the cheapest for peeling.
28785 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28787 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
28788 (vect_get_peeling_costs_all_drs): Create function.
28789 (vect_peeling_hash_get_lowest_cost):
28790 Use vect_get_peeling_costs_all_drs.
28791 (vect_peeling_supportable): Create function.
28792 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
28794 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28796 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
28797 DR_HAS_NEGATIVE_STEP.
28798 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
28799 (vect_enhance_data_refs_alignment): Use.
28800 (vect_duplicate_ssa_name_ptr_info): Use.
28801 * tree-vectorizer.h (dr_misalignment): Use.
28802 (known_alignment_for_access_p): Use.
28804 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28807 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
28809 (has_section_name): New function.
28811 2017-05-30 Martin Liska <mliska@suse.cz>
28814 * auto-profile.c (get_function_decl_from_block): Fix
28817 2017-05-30 Richard Biener <rguenther@suse.de>
28819 PR middle-end/80876
28820 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
28822 2017-05-30 Martin Liska <mliska@suse.cz>
28824 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
28825 * dumpfile.h (struct dump_file_info): Remove ctors.
28827 2017-05-30 Martin Liska <mliska@suse.cz>
28829 * predict.def: Fix GNU coding style.
28831 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28833 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28834 Mark 'to' argument with ATTRIBUTE_UNUSED.
28836 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28838 * config/xtensa/xtensa.c (xtensa_emit_call): Use
28839 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
28840 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
28843 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
28845 * doc/install.texi (Options specification): Restore entry of
28846 --enable-sjlj-exceptions.
28848 2017-05-27 Michael Eager <eager@eagercon.com>
28851 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
28853 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
28855 * config/microblaze/microblaze.h
28856 (FIXED_REGISTERS): Update in macro.
28857 (CALL_USED_REGISTERS): Update in macro.
28859 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
28861 * doc/install.texi: Add links to macOS binary distributions.
28863 2017-05-27 Jakub Jelinek <jakub@redhat.com>
28867 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28869 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
28871 2017-05-26 Martin Liska <mliska@suse.cz>
28873 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
28875 2017-05-26 Martin Liska <mliska@suse.cz>
28877 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
28878 always leading ';; '.
28879 (dump_bb_info): Likewise.
28880 (brief_dump_cfg): Likewise.
28881 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
28882 * dumpfile.c: Remove usage of TDF_VERBOSE.
28883 * dumpfile.h (enum dump_kind): Likewise.
28884 (dump_gimple_bb_header): Do not use TDF_COMMENT.
28885 * print-tree.c (debug_verbose): Remove.
28886 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
28887 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
28888 * tree-diagnostic.c (default_tree_printer): Replace
28889 TDF_DIAGNOSTIC with TDF_SLIM.
28891 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28893 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
28894 in parameter loop, rather than loop_vinfo.
28895 (create_intersect_range_checks): Ditto.
28896 (vect_create_cond_for_alias_checks): Update call to above functions.
28898 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28900 PR tree-optimization/80815
28901 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
28902 for merging runtime alias checks. Handle negative DR_STEPs.
28904 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28906 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
28908 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
28909 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
28910 out code pruning runtime alias checks.
28911 * tree-data-ref.c (prune_runtime_alias_test_list): New function
28912 factored out from above.
28913 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28915 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28917 (prune_runtime_alias_test_list): New decalaration.
28919 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28921 * tree-vect-data-refs.c (compare_tree): Rename and move ...
28922 * tree-data-ref.c (data_ref_compare_tree): ... to here.
28923 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
28924 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
28925 (operator==, comp_dr_with_seg_len_pair): Ditto.
28926 (vect_prune_runtime_alias_test_list): Ditto.
28928 2017-05-26 Martin Liska <mliska@suse.cz>
28931 * params.def: Bound partial-inlining-entry-probability param.
28933 2017-05-26 Marek Polacek <polacek@redhat.com>
28936 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
28939 2017-05-26 Richard Biener <rguenther@suse.de>
28941 PR tree-optimization/80842
28942 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
28945 2017-05-26 Richard Biener <rguenther@suse.de>
28947 PR tree-optimization/80844
28948 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
28950 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
28952 * doc/md.texi (Machine Constraints): Update x86 family
28953 machine constraints section to match 'config/i386/constraints.md'.
28955 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
28957 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
28959 2017-05-25 Nathan Sidwell <nathan@acm.org>
28961 * doc/invoke.texi (--enable-languages): Update documentation.
28963 2017-05-25 Martin Liska <mliska@suse.cz>
28965 * dumpfile.c: Add TDF_FOLDING.
28966 * dumpfile.h (enum dump_kind): Likewise.
28967 * genmatch.c (dt_simplify::gen_1): Use it.
28969 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28971 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
28973 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28975 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
28976 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
28978 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28980 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
28981 * match.pd (X == C): Rewrite it here.
28982 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
28983 with_certain_nonzero_bits2): New predicates.
28984 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
28986 2017-05-24 Nathan Sidwell <nathan@acm.org>
28988 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
28991 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
28994 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
28996 * config/powerpcspe: New port. Files are copied from the rs6000
28997 port, with "rs6000" in filenames replaced by "powerpcspe".
28999 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
29001 PR rtl-optimization/80754
29002 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
29004 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
29006 * config/sparc/sparc.md (length): Return the correct value for -mflat
29007 sibcalls to match output_sibcall.
29009 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29013 * config/rs6000/rs6000.c (struct machine_function): Add new field
29015 (rs6000_get_separate_components): Init that field, use it.
29016 (rs6000_components_for_bb): Use the field.
29018 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
29020 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
29022 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
29024 PR middle-end/80823
29025 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
29027 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29030 * config/s390/s390.c (s390_check_qrst_address): Check incoming
29031 address against address_operand predicate.
29032 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
29034 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
29036 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
29037 parameters passed indirectly.
29039 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
29041 * config/i386/i386.md (*movdi_internal): Remove SSE4
29042 alternative 18 (?r, *v). Update insn attributes.
29043 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
29044 Update insn attributes.
29045 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
29046 Update insn attributes.
29047 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
29048 alternative 1 (r, v). Remove isa attribute.
29049 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
29050 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
29051 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
29053 2017-05-23 Tom de Vries <tom@codesourcery.com>
29055 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
29058 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29060 * cgraphunit.c (symbol_table::process_new_functions): Update.
29061 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
29062 (inline_generate_summary): Rename to ...
29063 (ipa_fn_summary_generate): ... this one.
29064 (inline_read_summary): Rename to ...
29065 (ipa_fn_summary_read): ... this one.
29066 (inline_write_summary): Rename to ...
29067 (ipa_fn_summary_write): ... this one.
29068 (inline_free_summary): Rename to ...
29069 (ipa_free_fn_summary): ... this one.
29070 (pass_data_local_fn_summary, pass_local_fn_summary,
29071 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
29072 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
29073 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
29074 make_pass_ipa_fn_summary): New.
29075 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
29076 inline_write_summary, inline_free_summary): Remove.
29077 (ipa_free_fn_summary) : New.
29078 * ipa-inline.c (ipa_inline): Update.
29079 (pass_ipa_inline): Do not generate summaries.
29080 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
29082 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
29083 and add pass_ipa_fn_summary.
29084 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
29086 (make_pass_inline_parameters): Remove.
29088 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
29090 * omp-low.c (struct omp_context): Remove "default_kind" member.
29093 * omp-offload.c (execute_oacc_device_lower): Remove the
29094 parallelism dimensions function attributes for unparallelized
29095 OpenACC kernels constructs.
29097 2017-05-23 Martin Liska <mliska@suse.cz>
29099 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
29101 (cgraph_edge::make_speculative): Likewise.
29102 (cgraph_edge::resolve_speculation): Likewise.
29103 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29104 (cgraph_node::dump): Likewise.
29105 * cgraph.h: Likewise.
29106 * cgraphunit.c (analyze_functions): Likewise.
29107 (symbol_table::compile): Likewise.
29108 * ipa-cp.c (print_all_lattices): Likewise.
29109 (determine_versionability): Likewise.
29110 (initialize_node_lattices): Likewise.
29111 (ipcp_verify_propagated_values): Likewise.
29112 (estimate_local_effects): Likewise.
29113 (update_profiling_info): Likewise.
29114 (create_specialized_node): Likewise.
29115 (perhaps_add_new_callers): Likewise.
29116 (decide_about_value): Likewise.
29117 (decide_whether_version_node): Likewise.
29118 (identify_dead_nodes): Likewise.
29119 (ipcp_store_bits_results): Likewise.
29120 * ipa-devirt.c (dump_targets): Likewise.
29121 (ipa_devirt): Likewise.
29122 * ipa-icf.c (sem_item::dump): Likewise.
29123 (sem_function::equals): Likewise.
29124 (sem_variable::equals): Likewise.
29125 (sem_item_optimizer::read_section): Likewise.
29126 (sem_item_optimizer::execute): Likewise.
29127 (congruence_class::dump): Likewise.
29128 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
29129 (dump_inline_summary): Likewise.
29130 (estimate_node_size_and_time): Likewise.
29131 (inline_analyze_function): Likewise.
29132 * ipa-inline-transform.c (inline_call): Likewise.
29133 * ipa-inline.c (report_inline_failed_reason): Likewise.
29134 (want_early_inline_function_p): Likewise.
29135 (edge_badness): Likewise.
29136 (update_edge_key): Likewise.
29137 (inline_small_functions): Likewise.
29138 * ipa-profile.c (ipa_profile): Likewise.
29139 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
29140 (ipa_make_edge_direct_to_target): Likewise.
29141 (remove_described_reference): Likewise.
29142 (ipa_impossible_devirt_target): Likewise.
29143 (propagate_controlled_uses): Likewise.
29144 (ipa_print_node_params): Likewise.
29145 (ipcp_transform_function): Likewise.
29146 * ipa-pure-const.c (pure_const_read_summary): Likewise.
29147 (propagate_pure_const): Likewise.
29148 * ipa-reference.c (generate_summary): Likewise.
29149 (read_write_all_from_decl): Likewise.
29150 (propagate): Likewise.
29151 (ipa_reference_read_optimization_summary): Likewise.
29152 * ipa-utils.c (ipa_merge_profiles): Likewise.
29153 * ipa.c (walk_polymorphic_call_targets): Likewise.
29154 (symbol_table::remove_unreachable_nodes): Likewise.
29155 (ipa_single_use): Likewise.
29156 * passes.c (execute_todo): Likewise.
29157 * predict.c (drop_profile): Likewise.
29158 * symtab.c (symtab_node::get_dump_name): New function.
29159 (symtab_node::dump_name): Likewise.
29160 (symtab_node::dump_asm_name): Likewise.
29161 (symtab_node::dump_references): Likewise.
29162 (symtab_node::dump_referring): Likewise.
29163 (symtab_node::dump_base): Likewise.
29164 (symtab_node::debug_symtab): Likewise.
29165 * tree-sra.c (convert_callers_for_node): Likewise.
29166 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
29167 * value-prof.c (init_node_map): Likewise.
29169 2017-05-23 Martin Liska <mliska@suse.cz>
29171 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
29172 and symtab_node::debug_symtab to symbol_table::debug.
29173 * cgraphunit.c (analyze_functions): Use the renamed function.
29174 (symbol_table::compile): Likewise.
29175 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
29176 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
29177 * passes.c (execute_todo): Likewise.
29178 * symtab.c (symbol_table::dump): New function.
29179 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
29181 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29183 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
29184 that nonconst implies exec.
29186 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29188 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29189 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29190 (inline_edge_summary_vec): Turn into ...
29191 (ipa_call_summaries): ... this one.
29192 (redirect_to_unreachable, edge_set_predicate,
29193 evaluate_properties_for_edge, inline_summary_alloc,
29194 reset_ipa_call_summary, reset_inline_summary,
29195 inline_summary_t::duplicate): Update.
29196 (inline_edge_duplication_hook): Turn to ...
29197 (ipa_call_summary_t::duplicate): ... this one.
29198 (inline_edge_removal_hook): Turn to ...
29199 (ipa_call_summary_t::remove): ... this one.
29200 (dump_inline_edge_summary): Turn to ...
29201 (dump_ipa_call_summary): ... this one.
29202 (estimate_function_body_sizes): Update.
29203 (inline_update_callee_summaries): Update.
29204 (remap_edge_change_prob): Update.
29205 (remap_edge_summaries): Update.
29206 (inline_merge_summary): Update.
29207 (do_estimate_edge_time): Update.
29208 (inline_generate_summary): Update.
29209 (inline_read_section): Update.
29210 (inline_read_summary): Update.
29211 (inline_free_summary): Update.
29212 * ipa-inline.c (can_inline_edge_p): Update.
29213 (compute_inlined_call_time): Update.
29214 (want_inline_small_function_p): Update.
29215 (edge_badness): Update.
29216 (early_inliner): Update.
29217 * ipa-inline.h (inline_edge_summary): Turn to ...
29218 (ipa_call_summary): ... this one.
29219 (ipa_call_summary_t): New class.
29220 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29221 (ipa_call_summaries): New.
29222 (inline_edge_summary): Remove.
29223 (estimate_edge_growth): Update.
29224 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29225 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29226 * ipa-split.c (execute_split_functions): Update.
29227 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29229 2017-05-23 Tom de Vries <tom@codesourcery.com>
29231 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29232 attributes): Document rdrand effective target.
29234 2017-05-23 Tom de Vries <tom@codesourcery.com>
29236 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29237 attributes): Sort alphabetically.
29239 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
29241 * config/avr/genmultilib.awk: Use gsub instead of gensub.
29243 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29246 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
29247 V2DF/V2DI splat into two separate patterns, one that handles
29248 registers, and the other that only handles memory. Drop support
29249 for splatting from a GPR on ISA 2.07 and then splitting the
29250 splat into direct move and splat.
29251 (vsx_splat_<mode>_reg): Likewise.
29252 (vsx_splat_<mode>_mem): Likewise.
29254 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
29256 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
29258 2017-05-22 Jakub Jelinek <jakub@redhat.com>
29260 PR middle-end/80809
29261 * omp-low.c (finish_taskreg_remap): New function.
29262 (finish_taskreg_scan): If unit size of ctx->record_type
29263 is non-constant, unshare the size expression and replace
29264 decls in it with possible outer var refs.
29266 PR middle-end/80809
29267 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
29268 GOVD_SHARED rather than GOVD_PRIVATE with it.
29269 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
29270 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
29272 PR middle-end/80853
29273 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
29274 as last argument to build_outer_var_ref for pointer bases of array
29275 section reductions.
29277 2017-05-19 Martin Sebor <msebor@redhat.com>
29279 * print-tree.c (print_node): Print DECL_READ_P flag.
29281 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29283 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
29284 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
29285 * cgraph.c: Likewise.
29286 * cgraphunit.c: Likewise.
29287 * gengtype.c: Likewise.
29288 * ipa-cp.c: Likewise.
29289 * ipa-devirt.c: Likewise.
29290 * ipa-icf.c: Likewise.
29291 * ipa-predicate.c: Likewise.
29292 * ipa-profile.c: Likewise.
29293 * ipa-prop.c: Likewise.
29294 * ipa-split.c: Likewise.
29296 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
29297 edge_predicate_pool, dump_inline_hints,
29298 inline_summary::account_size_time, redirect_to_unreachable,
29299 edge_set_predicate, set_hint_predicate,
29300 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
29301 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
29302 inline_summary_t::remove, remap_hint_predicate_after_duplication,
29303 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
29304 ipa_call_summary_t::remove, initialize_growth_caches,
29305 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
29306 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
29307 mark_modified, unmodified_parm_1, unmodified_parm,
29308 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
29309 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
29310 compute_bb_predicates, will_be_nonconstant_expr_predicate,
29311 will_be_nonconstant_predicate, record_modified_bb_info,
29312 get_minimal_bb, record_modified, param_change_prob,
29313 phi_result_unknown_predicate, predicate_for_phi_result,
29314 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
29315 estimate_function_body_sizes, compute_inline_parameters,
29316 compute_inline_parameters_for_curren, pass_data_inline_parameters,
29317 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
29318 inline_update_callee_summaries, remap_edge_change_prob,
29319 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
29320 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
29321 inline_analyze_function, inline_summary_t::insert,
29322 inline_generate_summary, read_ipa_call_summary, inline_read_section,
29323 inline_read_summary, write_ipa_call_summary, inline_write_summary,
29324 inline_free_summary): Move to ipa-fnsummary.h
29325 (predicate_t): Remove.
29326 * ipa-fnsummary.c: New file.
29327 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
29328 (enum inline_hints_vals, inline_hints, agg_position_info,
29329 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
29330 inline_summaries, ipa_call_summary, ipa_call_summary_t,
29331 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
29332 dump_inline_summary, dump_inline_hints, inline_generate_summary,
29333 inline_read_summary, inline_write_summary, inline_free_summary,
29334 inline_analyze_function, initialize_inline_failed,
29335 inline_merge_summary, inline_update_overall_summary,
29336 compute_inline_parameters): Move to ipa-fnsummary.h
29337 * ipa-fnsummary.h: New file.
29338 * ipa-inline-transform.h: Include ipa-inline.h.
29339 * ipa-inline.c: LIkewise.
29341 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29343 * ipa-inline.c (edge_badness): Use inlined_time instead of
29344 inline_summaries->get.
29346 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29348 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
29350 2017-05-22 Nathan Sidwell <nathan@acm.org>
29352 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
29353 (fdump-lang): Document 'raw' option.
29354 * dumpfile.h (TDI_tu): Delete.
29355 * dumpfile.c (dump_files): Remove translation-unit.
29356 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
29358 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
29360 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
29361 command option from $(AWK) call.
29362 * config/avr/genmultilib.awk: Simplify and rewrite so that it
29363 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
29364 [FORMAT]: Remove handling of variable.
29365 * config/avr/t-multilib: Regenerate.
29367 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29369 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
29371 (dump_inline_summary): Do not print self_time.
29372 (estimate_function_body_sizes): Do not set self_time.
29373 (compute_inline_parameters): Likewise.
29374 (inline_read_section, inline_write_summary): Do not stream self_time.
29375 * ipa-inline.h (inline_summary): Drop self_time.
29377 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29379 * ipa-inline-analysis.c (account_size_time): Rename to ...
29380 (inline_summary::account_size_time): ... this one.
29381 (reset_ipa_call_summary): Turn to ...
29382 (ipa_call_summary::reset): ... this one.
29383 (reset_inline_summary): Turn to ...
29384 (inline_summary::reset): ... this one.
29385 (inline_summary_t::remove): Update.
29386 (inline_summary_t::duplicate): Update.
29387 (ipa_call_summary_t::remove): Update.
29388 (dump_inline_summary): Update.
29389 (estimate_function_body_sizes): Update.
29390 (compute_inline_parameters): Update.
29391 (estimate_node_size_and_time): Update.
29392 (inline_merge_summary): Update.
29393 (inline_update_overall_summary): Update.
29394 (inline_read_section): Update.
29395 (inline_write_summary): Update.
29396 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
29397 add account_size_time and reset member functions.
29398 (ipa_call_summary): Add reset function.
29399 * ipa-predicate.h (predicate::operator &): Constify.
29401 2017-05-22 Richard Biener <rguenther@suse.de>
29403 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
29405 2017-05-19 Jason Merrill <jason@redhat.com>
29407 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
29409 2017-05-19 Marek Polacek <polacek@redhat.com>
29412 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
29413 TYPE_OVERFLOW_WRAPS checks.
29415 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
29417 * tree-core.h (enum omp_clause_default_kind): Add
29418 "OMP_CLAUSE_DEFAULT_PRESENT".
29419 * tree-pretty-print.c (dump_omp_clause): Handle it.
29420 * gimplify.c (enum gimplify_omp_var_data): Add
29421 "GOVD_MAP_FORCE_PRESENT".
29422 (gimplify_adjust_omp_clauses_1): Map it to
29423 "GOMP_MAP_FORCE_PRESENT".
29424 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
29426 * gimplify.c (oacc_default_clause): Clarify.
29428 2017-05-19 Nathan Sidwell <nathan@acm.org>
29430 LANG_HOOK_REGISTER_DUMPS
29431 * toplev.c (general_init): Call register dump lang hook.
29432 * doc/invoke.texi: Document -fdump-lang option family.
29433 * dumpfile.c (dump_files): Remove class dump here.
29434 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
29435 * dumpfile.h (tree_dump_index): Remove TDI_class.
29436 * langhooks-def.h (lhd_register_dumps): Declare.
29437 (LANG_HOOKS_REGISTER_DUMPS): Define.
29438 (LANG_HOOKS_INITIALIZER): Add it.
29439 * langhooks.c (lhd_register_dumps): Define.
29440 * langhooks.h (struct lang_hooks): Add register_dumps.
29442 2017-05-19 Nathan Sidwell <nathan@acm.org>
29444 * context.h (context::set_passes): New.
29445 * context.c (context::context): Do not create pass manager.
29446 * toplev.c (general_init): Create pass manager here.
29448 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
29450 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
29451 use this splitter if two add or or instructions would also work for
29452 the constant we want to generate.
29454 2017-05-19 Richard Biener <rguenther@suse.de>
29457 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
29458 predicate evaluation.
29460 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29462 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
29464 * ipa-inline.c (want_inline_small_function_p): Do not cast to
29467 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29469 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29470 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29471 (inline_edge_summary_vec): Turn into ...
29472 (ipa_call_summaries): ... this one.
29473 (redirect_to_unreachable, edge_set_predicate,
29474 evaluate_properties_for_edge, inline_summary_alloc,
29475 reset_ipa_call_summary, reset_inline_summary,
29476 inline_summary_t::duplicate): Update.
29477 (inline_edge_duplication_hook): Turn to ...
29478 (ipa_call_summary_t::duplicate): ... this one.
29479 (inline_edge_removal_hook): Turn to ...
29480 (ipa_call_summary_t::remove): ... this one.
29481 (dump_inline_edge_summary): Turn to ...
29482 (dump_ipa_call_summary): ... this one.
29483 (estimate_function_body_sizes): Update.
29484 (inline_update_callee_summaries): Update.
29485 (remap_edge_change_prob): Update.
29486 (remap_edge_summaries): Update.
29487 (inline_merge_summary): Update.
29488 (do_estimate_edge_time): Update.
29489 (inline_generate_summary): Update.
29490 (inline_read_section): Update.
29491 (inline_read_summary): Update.
29492 (inline_free_summary): Update.
29493 * ipa-inline.c (can_inline_edge_p): Update.
29494 (compute_inlined_call_time): Update.
29495 (want_inline_small_function_p): Update.
29496 (edge_badness): Update.
29497 (early_inliner): Update.
29498 * ipa-inline.h (inline_edge_summary): Turn to ...
29499 (ipa_call_summary): ... this one.
29500 (ipa_call_summary_t): New class.
29501 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29502 (ipa_call_summaries): New.
29503 (inline_edge_summary): Remove.
29504 (estimate_edge_growth): Update.
29505 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29506 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29507 * ipa-split.c (execute_split_functions): Update.
29508 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29510 2017-05-19 Richard Biener <rguenther@suse.de>
29512 PR middle-end/80764
29513 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
29515 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
29517 * config/rs6000/rs6000.c (struct machine_function): Add field
29518 fpr_is_wrapped_separately.
29519 (rs6000_get_separate_components): Use 64 components. Handle the
29520 new FPR components.
29521 (rs6000_components_for_bb): Handle the FPR components.
29522 (rs6000_emit_prologue_components): Handle the FPR components.
29523 (rs6000_emit_epilogue_components): Handle the FPR components.
29524 (rs6000_set_handled_components): Handle the FPR components.
29525 (rs6000_emit_prologue): Don't output prologue code for those FPRs
29526 that are already separately shrink-wrapped.
29527 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
29528 that are already separately shrink-wrapped.
29530 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29533 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
29536 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
29537 (define_peephole2 for Altivec d-form load): Add peepholes to catch
29538 cases where the register allocator uses a move and an offsettable
29539 memory operation to/from a FPR register on ISA 2.06/2.07.
29540 (define_peephole2 for Altivec d-form store): Likewise.
29542 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
29545 * config/i386/mmx.md (*mov<mode>_internal): Enable
29546 alternatives 11, 12, 13 and 14 also for 32bit targets.
29547 Remove alternatives 15, 16, 17 and 18.
29548 * config/i386/sse.md (vec_concatv2di): Change
29549 alternative (!x, *y) to (x, ?!*Yn).
29551 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
29553 * dumpfile.h (enum dump_kind): Remove stray comma.
29555 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29557 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
29558 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
29559 predicate::num_conditions
29560 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
29561 (CHANGED): turn into predicate::changed.
29562 (agg_position_info): Move to ipa-predicate.h
29563 (add_condition, predicate::add_clause, predicate::operator &=,
29564 predicate::or_with, predicate::evaluate, predicate::probability,
29565 dump_condition, dump_clause, predicate::dump,
29566 predicate::remap_after_duplication, predicate::remap_after_inlining,
29567 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
29568 (evaluate_conditions_for_known_args): Update.
29569 (set_cond_stmt_execution_predicate): Update.
29570 * ipa-inline.h: Include ipa-predicate.h
29571 (condition, inline_param_summary, conditions, agg_position_info,
29572 predicate): Move to ipa-predicate.h
29573 * ipa-predicate.c: New file.
29574 * ipa-predicate.h: New file.
29576 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
29578 * final.c (leaf_function_p): Check we are not in a sequence.
29580 2017-05-18 Martin Liska <mliska@suse.cz>
29582 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
29583 * dumpfile.c (dump_register): Use new enum dump_kind.
29584 (get_dump_file_name): Likewise.
29585 (dump_enable_all): Likewise.
29586 (dump_switch_p_1): Likewise.
29587 (enable_rtl_dump_file): Remove usage of TDF_RTL.
29588 * dumpfile.h (enum dump_kind): New enum type.
29589 (struct dump_file_info): Create constructor and
29590 format fields and comments.
29591 * passes.c (pass_manager::register_one_dump_file):
29593 * statistics.c (statistics_early_init): Likewise.
29594 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
29595 TDF_TREE with TDF_SLIM.
29596 (gather_memory_references_ref): Likewise.
29598 2017-05-18 Martin Liska <mliska@suse.cz>
29600 * vec.h (struct vnull): Use it.
29602 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29604 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
29605 (true_predicate, false_predicate, true_predicate_p,
29606 false_predicate_p): Remove.
29607 (single_cond_predicate, not_inlined_predicate): Turn to member function
29609 (add_condition): Update.
29610 (add_clause): Turn to...
29611 (predicate::add_clause): ... this one; update; allow passing NULL
29613 (and_predicates): Turn to ...
29614 (predicate::operator &=): ... this one.
29615 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
29616 (or_predicates): Turn to ...
29617 (predicate::or_with): ... this one.
29618 (evaluate_predicate): Turn to ...
29619 (predicate::evaluate): ... this one.
29620 (predicate_probability): Turn to ...
29621 (predicate::probability): ... this one.
29622 (dump_condition): Update.
29623 (dump_predicate): Turn to ...
29624 (predicate::dump): ... this one.
29625 (account_size_time): Update.
29626 (edge_set_predicate): Update.
29627 (set_hint_predicate): UPdate.
29628 (evaluate_conditions_for_known_args): Update.
29629 (evaluate_properties_for_edge): Update.
29630 (remap_predicate_after_duplication): Turn to...
29631 (predicate::remap_after_duplication): ... this one.
29632 (remap_hint_predicate_after_duplication): Update.
29633 (inline_summary_t::duplicate): UPdate.
29634 (dump_inline_edge_summary): Update.
29635 (dump_inline_summary): Update.
29636 (set_cond_stmt_execution_predicate): Update.
29637 (set_switch_stmt_execution_predicate): Update.
29638 (compute_bb_predicates): Update.
29639 (will_be_nonconstant_expr_predicate): Update.
29640 (will_be_nonconstant_predicate): Update.
29641 (phi_result_unknown_predicate): Update.
29642 (predicate_for_phi_result): Update.
29643 (array_index_predicate): Update.
29644 (estimate_function_body_sizes): Update.
29645 (estimate_node_size_and_time): Update.
29646 (estimate_ipcp_clone_size_and_time): Update.
29647 (remap_predicate): Rename to ...
29648 (predicate::remap_after_inlining): ... this one.
29649 (remap_hint_predicate): Update.
29650 (inline_merge_summary): Update.
29651 (inline_update_overall_summary): Update.
29652 (estimate_size_after_inlining): Update.
29653 (read_predicate): Rename to ...
29654 (predicate::stream_in): ... this one.
29655 (read_inline_edge_summary): Update.
29656 (write_predicate): Rename to ...
29657 (predicate::stream_out): ... this one.
29658 (write_inline_edge_summary): Update.
29659 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
29660 (clause_t): Turn to uint32_t
29661 (predicate): Turn to class; implement constructor and operators
29663 (size_time_entry): Update.
29664 (inline_summary): Update.
29665 (inline_edge_summary): Update.
29667 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
29669 * fold-const.c (fold_binary_loc): Move transformation...
29670 * match.pd (C - X CMP X): ... here.
29672 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
29674 * config/sparc/sparc.c (sparc_option_override): Set function
29675 alignment for -mcpu=niagara7 to 64 to match the I$ line.
29676 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
29678 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
29680 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
29682 2017-05-18 Marek Polacek <polacek@redhat.com>
29685 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
29686 (pass_ubsan::execute): Call gimple_assign_single_p instead of
29687 gimple_assign_load_p.
29689 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
29691 PR middle-end/80692
29692 * real.c (do_compare): Give decimal_do_compare preference over
29693 comparing just the signs.
29695 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
29697 * doc/md.texi (Canonicalization of Instructions): Describe the
29698 canonical form of instructions that inherently set a condition
29701 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
29703 PR middle-end/80775
29704 * tree-cfg.c: Move deletion of unreachable case statements to after
29705 the merging of consecutive case labels.
29707 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
29709 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
29710 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
29711 restoring of callee-saved registers.
29713 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
29715 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
29716 * config/visium/visium.c (single_set_and_flags): Likewise.
29717 * config/visium/visium.md (Substitutions): Likewise.
29719 2017-05-17 Martin Liska <mliska@suse.cz>
29721 * cfg.c: Introduce dump_flags_t type and
29722 use it instead of int type.
29724 * cfghooks.c: Likewise.
29725 * cfghooks.h (struct cfg_hooks): Likewise.
29726 * cfgrtl.c: Likewise.
29727 * cfgrtl.h: Likewise.
29728 * cgraph.c (cgraph_node::get_body): Likewise.
29729 * coretypes.h: Likewise.
29730 * domwalk.c: Likewise.
29731 * domwalk.h: Likewise.
29732 * dumpfile.c (struct dump_option_value_info): Likewise.
29733 (dump_enable_all): Likewise.
29734 (dump_switch_p_1): Likewise.
29735 (opt_info_switch_p): Likewise.
29736 * dumpfile.h (enum tree_dump_index): Likewise.
29737 (struct dump_file_info): Likewise.
29738 * genemit.c: Likewise.
29739 * generic-match-head.c: Likewise.
29740 * gengtype.c (open_base_files): Likewise.
29741 * gimple-pretty-print.c: Likewise.
29742 * gimple-pretty-print.h: Likewise.
29743 * graph.c (print_graph_cfg): Likewise.
29744 * graphite-scop-detection.c (dot_all_sese): Likewise.
29745 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
29746 * loop-unroll.c (report_unroll): Likewise.
29747 * passes.c (pass_manager::register_one_dump_file): Likewise.
29748 * print-tree.c: Likewise.
29749 * statistics.c: Likewise.
29750 * tree-cfg.c: Likewise.
29751 * tree-cfg.h: Likewise.
29752 * tree-dfa.c: Likewise.
29753 * tree-dfa.h: Likewise.
29754 * tree-dump.c (dump_function): Likewise.
29755 * tree-dump.h (struct dump_info): Likewise.
29756 * tree-pretty-print.c: Likewise.
29757 * tree-pretty-print.h: Likewise.
29758 * tree-ssa-live.c: Likewise.
29759 * tree-ssa-live.h: Likewise.
29760 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
29761 * tree-vect-loop.c: Likewise.
29762 * tree-vect-slp.c: Likewise.
29764 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
29765 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29767 PR tree-optimization/80457
29768 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
29769 of all arguments to a statement as scalar_to_vec operations.
29770 (vectorizable_call): Adjust call to vect_model_simple_cost for
29772 (vectorizable_conversion): Likewise.
29773 (vectorizable_assignment): Likewise.
29774 (vectorizable_shift): Likewise.
29775 (vectorizable_operation): Likewise.
29776 (vectorizable_comparison): Likewise.
29777 (vect_is_simple_cond): Record the def types for operands.
29778 (vectorizable_condition): Likewise, call vect_model_simple_cost.
29779 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
29780 for statement argument count.
29782 2017-05-16 Carl Love <cel@us.ibm.com>
29784 * config/rs6000/rs6000-c: Add support for built-in functions
29785 vector unsigned long long vec_bperm (vector unsigned long long,
29786 vector unsigned char)
29787 vector signed long long vec_mule (vector signed int,
29789 vector unsigned long long vec_mule (vector unsigned int,
29790 vector unsigned int)
29791 vector signed long long vec_mulo (vector signed int,
29793 vector unsigned long long vec_mulo (vector unsigned int,
29794 vector unsigned int)
29795 vector signed char vec_sldw (vector signed char,
29796 vector signed char,
29798 vector unsigned char vec_sldw (vector unsigned char,
29799 vector unsigned char,
29801 vector signed short vec_sldw (vector signed short,
29802 vector signed short,
29804 vector unsigned short vec_sldw (vector unsigned short,
29805 vector unsigned short,
29807 vector signed int vec_sldw (vector signed int,
29810 vector unsigned int vec_sldw (vector unsigned int,
29811 vector unsigned int,
29813 vector signed long long vec_sldw (vector signed long long,
29814 vector signed long long,
29816 vector unsigned long long vec_sldw (vector unsigned long long,
29817 vector unsigned long long,
29819 * config/rs6000/rs6000-c: Add support for built-in functions
29820 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
29821 * config/rs6000/altivec.h: Add defintion for vec_sldw.
29822 * doc/extend.texi: Update the built-in documentation for the
29823 new built-in functions.
29825 2017-05-16 Marek Polacek <polacek@redhat.com>
29829 * tree.c (save_expr): Don't fold the expression.
29831 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
29833 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
29834 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
29835 and (?*y,m). Update insn attributes.
29837 2017-05-16 Martin Liska <mliska@suse.cz>
29839 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
29840 flags argument of print_gimple_stmt, print_gimple_expr,
29841 print_generic_stmt and print_generic_expr.
29842 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
29843 * coretypes.h: Likewise.
29844 * except.c (dump_eh_tree): Likewise.
29845 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
29846 * gimple-pretty-print.h: Likewise.
29847 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
29848 (backprop::push_to_worklist): Likewise.
29849 (backprop::pop_from_worklist): Likewise.
29850 (backprop::process_use): Likewise.
29851 (backprop::intersect_uses): Likewise.
29852 (note_replacement): Likewise.
29853 * gimple-ssa-store-merging.c
29854 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
29855 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
29856 (pass_store_merging::execute): Likewise.
29857 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
29858 (ssa_base_cand_dump_callback): Likewise.
29859 (dump_incr_vec): Likewise.
29860 (replace_refs): Likewise.
29861 (replace_mult_candidate): Likewise.
29862 (create_add_on_incoming_edge): Likewise.
29863 (create_phi_basis): Likewise.
29864 (insert_initializers): Likewise.
29865 (all_phi_incrs_profitable): Likewise.
29866 (introduce_cast_before_cand): Likewise.
29867 (replace_one_candidate): Likewise.
29868 * gimplify.c (gimplify_expr): Likewise.
29869 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
29870 (set_rename): Likewise.
29871 (rename_uses): Likewise.
29872 (copy_loop_phi_nodes): Likewise.
29873 (add_close_phis_to_merge_points): Likewise.
29874 (copy_loop_close_phi_args): Likewise.
29875 (copy_cond_phi_args): Likewise.
29876 (graphite_copy_stmts_from_block): Likewise.
29877 (translate_pending_phi_nodes): Likewise.
29878 * graphite-poly.c (print_pdr): Likewise.
29879 (dump_gbb_cases): Likewise.
29880 (dump_gbb_conditions): Likewise.
29881 (print_scop_params): Likewise.
29882 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
29883 (build_cross_bb_scalars_use): Likewise.
29884 (gather_bbs::before_dom_children): Likewise.
29885 * hsa-dump.c (dump_hsa_immed): Likewise.
29886 * ipa-cp.c (print_ipcp_constant_value): Likewise.
29887 (get_replacement_map): Likewise.
29888 * ipa-inline-analysis.c (dump_condition): Likewise.
29889 (estimate_function_body_sizes): Likewise.
29890 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
29891 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
29892 * ipa-prop.c (ipa_dump_param): Likewise.
29893 (ipa_print_node_jump_functions_for_edge): Likewise.
29894 (ipa_modify_call_arguments): Likewise.
29895 (ipa_modify_expr): Likewise.
29896 (ipa_dump_param_adjustments): Likewise.
29897 (ipa_dump_agg_replacement_values): Likewise.
29898 (ipcp_modif_dom_walker::before_dom_children): Likewise.
29899 * ipa-pure-const.c (check_stmt): Likewise.
29900 (pass_nothrow::execute): Likewise.
29901 * ipa-split.c (execute_split_functions): Likewise.
29902 * omp-offload.c (dump_oacc_loop_part): Likewise.
29903 (dump_oacc_loop): Likewise.
29904 * trans-mem.c (tm_log_emit): Likewise.
29905 (tm_memopt_accumulate_memops): Likewise.
29906 (dump_tm_memopt_set): Likewise.
29907 (dump_tm_memopt_transform): Likewise.
29908 * tree-cfg.c (gimple_verify_flow_info): Likewise.
29909 (print_loop): Likewise.
29910 * tree-chkp-opt.c (chkp_print_addr): Likewise.
29911 (chkp_gather_checks_info): Likewise.
29912 (chkp_get_check_result): Likewise.
29913 (chkp_remove_check_if_pass): Likewise.
29914 (chkp_use_outer_bounds_if_possible): Likewise.
29915 (chkp_reduce_bounds_lifetime): Likewise.
29916 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
29917 (chkp_mark_completed_bounds): Likewise.
29918 (chkp_register_incomplete_bounds): Likewise.
29919 (chkp_mark_invalid_bounds): Likewise.
29920 (chkp_maybe_copy_and_register_bounds): Likewise.
29921 (chkp_build_returned_bound): Likewise.
29922 (chkp_get_bound_for_parm): Likewise.
29923 (chkp_build_bndldx): Likewise.
29924 (chkp_get_bounds_by_definition): Likewise.
29925 (chkp_generate_extern_var_bounds): Likewise.
29926 (chkp_get_bounds_for_decl_addr): Likewise.
29927 * tree-chrec.c (chrec_apply): Likewise.
29928 * tree-data-ref.c (dump_data_reference): Likewise.
29929 (dump_subscript): Likewise.
29930 (dump_data_dependence_relation): Likewise.
29931 (analyze_overlapping_iterations): Likewise.
29932 * tree-inline.c (expand_call_inline): Likewise.
29933 (tree_function_versioning): Likewise.
29934 * tree-into-ssa.c (dump_defs_stack): Likewise.
29935 (dump_currdefs): Likewise.
29936 (dump_names_replaced_by): Likewise.
29937 (dump_update_ssa): Likewise.
29938 (update_ssa): Likewise.
29939 * tree-object-size.c (pass_object_sizes::execute): Likewise.
29940 * tree-parloops.c (build_new_reduction): Likewise.
29941 (try_create_reduction_list): Likewise.
29942 (ref_conflicts_with_region): Likewise.
29943 (oacc_entry_exit_ok_1): Likewise.
29944 (oacc_entry_exit_single_gang): Likewise.
29945 * tree-pretty-print.h: Likewise.
29946 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
29947 (get_scalar_evolution): Likewise.
29948 (add_to_evolution): Likewise.
29949 (get_loop_exit_condition): Likewise.
29950 (analyze_evolution_in_loop): Likewise.
29951 (analyze_initial_condition): Likewise.
29952 (analyze_scalar_evolution): Likewise.
29953 (instantiate_scev): Likewise.
29954 (number_of_latch_executions): Likewise.
29955 (gather_chrec_stats): Likewise.
29956 (final_value_replacement_loop): Likewise.
29957 (scev_const_prop): Likewise.
29958 * tree-sra.c (dump_access): Likewise.
29959 (disqualify_candidate): Likewise.
29960 (create_access): Likewise.
29961 (reject): Likewise.
29962 (maybe_add_sra_candidate): Likewise.
29963 (create_access_replacement): Likewise.
29964 (analyze_access_subtree): Likewise.
29965 (analyze_all_variable_accesses): Likewise.
29966 (sra_modify_assign): Likewise.
29967 (initialize_constant_pool_replacements): Likewise.
29968 (find_param_candidates): Likewise.
29969 (decide_one_param_reduction): Likewise.
29970 (replace_removed_params_ssa_names): Likewise.
29971 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
29972 * tree-ssa-copy.c (dump_copy_of): Likewise.
29973 (copy_prop_visit_cond_stmt): Likewise.
29974 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
29975 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
29976 (record_equivalences_from_stmt): Likewise.
29977 * tree-ssa-dse.c (compute_trims): Likewise.
29978 (delete_dead_call): Likewise.
29979 (delete_dead_assignment): Likewise.
29980 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
29981 (forward_propagate_into_cond): Likewise.
29982 (pass_forwprop::execute): Likewise.
29983 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
29984 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
29986 (move_computations_worker): Likewise.
29987 (execute_sm): Likewise.
29988 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
29989 (remove_exits_and_undefined_stmts): Likewise.
29990 (remove_redundant_iv_tests): Likewise.
29991 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
29992 (adjust_iv_update_pos): Likewise.
29993 * tree-ssa-math-opts.c (bswap_replace): Likewise.
29994 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
29995 (value_replacement): Likewise.
29996 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
29997 * tree-ssa-pre.c (print_pre_expr): Likewise.
29998 (get_representative_for): Likewise.
29999 (create_expression_by_pieces): Likewise.
30000 (insert_into_preds_of_block): Likewise.
30001 (eliminate_insert): Likewise.
30002 (eliminate_dom_walker::before_dom_children): Likewise.
30003 (eliminate): Likewise.
30004 (remove_dead_inserted_code): Likewise.
30005 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
30006 * tree-ssa-reassoc.c (get_rank): Likewise.
30007 (eliminate_duplicate_pair): Likewise.
30008 (eliminate_plus_minus_pair): Likewise.
30009 (eliminate_not_pairs): Likewise.
30010 (undistribute_ops_list): Likewise.
30011 (eliminate_redundant_comparison): Likewise.
30012 (update_range_test): Likewise.
30013 (optimize_range_tests_var_bound): Likewise.
30014 (optimize_vec_cond_expr): Likewise.
30015 (rewrite_expr_tree): Likewise.
30016 (rewrite_expr_tree_parallel): Likewise.
30017 (linearize_expr): Likewise.
30018 (break_up_subtract): Likewise.
30019 (linearize_expr_tree): Likewise.
30020 (attempt_builtin_powi): Likewise.
30021 (attempt_builtin_copysign): Likewise.
30022 (transform_stmt_to_copy): Likewise.
30023 (transform_stmt_to_multiply): Likewise.
30024 (dump_ops_vector): Likewise.
30025 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
30026 (print_scc): Likewise.
30027 (set_ssa_val_to): Likewise.
30028 (visit_reference_op_store): Likewise.
30029 (visit_use): Likewise.
30030 (sccvn_dom_walker::before_dom_children): Likewise.
30031 (run_scc_vn): Likewise.
30032 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
30034 (expr_hash_elt::print): Likewise.
30035 (const_and_copies::pop_to_marker): Likewise.
30036 (const_and_copies::record_const_or_copy_raw): Likewise.
30037 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
30038 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
30039 (dump_predicates): Likewise.
30040 (find_uninit_use): Likewise.
30041 (warn_uninitialized_phi): Likewise.
30042 (pass_late_warn_uninitialized::execute): Likewise.
30043 * tree-ssa.c (verify_vssa): Likewise.
30044 (verify_ssa): Likewise.
30045 (maybe_optimize_var): Likewise.
30046 * tree-vrp.c (dump_value_range): Likewise.
30047 (dump_all_value_ranges): Likewise.
30048 (dump_asserts_for): Likewise.
30049 (register_edge_assert_for_2): Likewise.
30050 (vrp_visit_cond_stmt): Likewise.
30051 (vrp_visit_switch_stmt): Likewise.
30052 (vrp_visit_stmt): Likewise.
30053 (vrp_visit_phi_node): Likewise.
30054 (simplify_cond_using_ranges_1): Likewise.
30055 (fold_predicate_in): Likewise.
30056 (evrp_dom_walker::before_dom_children): Likewise.
30057 (evrp_dom_walker::push_value_range): Likewise.
30058 (evrp_dom_walker::pop_value_range): Likewise.
30059 (execute_early_vrp): Likewise.
30061 2017-05-16 Richard Biener <rguenther@suse.de>
30063 * dwarf2out.c (loc_list_from_tree_1): Do not create
30064 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
30066 2017-05-16 Richard Biener <rguenther@suse.de>
30068 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
30070 (note_variable_value_in_expr): If we resolved the decl ref
30071 do not push to the stack.
30073 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
30075 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
30076 operations in fast-math mode.
30077 (vaddq_f16): Likewise.
30078 (vmul_f16): Likewise.
30079 (vmulq_f16): Likewise.
30080 (vsub_f16): Likewise.
30081 (vsubq_f16): Likewise.
30082 * config/arm/neon.md (add<mode>3): New.
30084 (fma:<VH:mode>3): New. Also remove outdated comment.
30087 2017-05-16 Martin Liska <mliska@suse.cz>
30091 * ipa-devirt.c (warn_types_mismatch): Fix typo.
30092 (odr_types_equivalent_p): Likewise.
30094 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
30096 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
30098 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
30101 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
30102 non-interunit SSE move alternatives with '?'.
30103 (zero-extendsidi peephole2): New peephole to skip intermediate
30104 general register in SSE zero-extend sequence.
30106 2017-05-15 Jeff Law <law@redhat.com>
30108 * reorg.c (relax_delay_slots): Create a new variable to hold
30109 the temporary target rather than clobbering TARGET_LABEL.
30111 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
30112 missing argument to extract_bit_field call.
30113 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
30115 2017-05-15 Martin Liska <mliska@suse.cz>
30118 * gcc.c (process_command): Do not allow empty argument of -o option.
30120 2017-05-15 Renlin Li <renlin.li@arm.com>
30122 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
30123 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
30124 * config/aarch64/constraints.md (Usf): Add long call check.
30125 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
30126 (call_value): Likewise.
30127 (sibcall): Likewise.
30128 (sibcall_value): Likewise.
30130 (call_value_insn): New.
30131 (sibcall_insn): Update rtx pattern.
30132 (sibcall_value_insn): Likewise.
30133 (call_internal): Remove.
30134 (call_value_internal): Likewise.
30135 (sibcall_internal): Likewise.
30136 (sibcall_value_internal): Likewise.
30137 (call_reg): Likewise.
30138 (call_symbol): Likewise.
30139 (call_value_reg): Likewise.
30140 (call_value_symbol): Likewise.
30142 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
30145 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
30147 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
30149 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
30150 compatible with CCGOCmode and with CCZmode.
30152 2017-05-14 Martin Sebor <msebor@redhat.com>
30154 PR middle-end/77671
30155 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
30156 (gimple_fold_builtin_snprintf): Same.
30157 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
30158 (gimple_fold_builtin_snprintf): Same.
30159 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
30160 of character types.
30161 (is_call_safe): New function.
30162 (try_substitute_return_value): Call it.
30163 (try_simplify_call): New function.
30164 (pass_sprintf_length::handle_gimple_call): Call it.
30166 2017-05-14 Martin Sebor <msebor@redhat.com>
30168 PR middle-end/80669
30169 * builtins.c (expand_builtin_stpncpy): Simplify.
30171 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
30173 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
30174 * config/i386/i386.h
30175 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
30176 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
30177 (struct machine_function): Add new members call_ms2sysv,
30178 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
30179 (struct machine_frame_state): New fields sp_realigned and
30180 sp_realigned_offset.
30181 * config/i386/i386.c
30182 (enum xlogue_stub): New enum.
30183 (enum xlogue_stub_sets): New enum.
30184 (class xlogue_layout): New class.
30185 (struct ix86_frame): New fields stack_realign_allocate_offset,
30186 stack_realign_offset and outlined_save_offset. Modify comments to
30187 detail stack layout when using out-of-line stubs.
30188 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
30189 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
30190 -mcall-ms2sysv-xlogues.
30191 (stub_managed_regs): New static variable.
30192 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
30193 registers managed by out-of-line stub.
30194 (disable_call_ms2sysv_xlogues): New function.
30195 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
30196 m->call_ms2sysv when appropriate and compute frame layout for
30198 (sp_valid_at, fp_valid_at): New inline functions.
30199 (choose_basereg): New function.
30200 (choose_baseaddr): Add align parameter, use choose_basereg and modify
30202 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
30203 Use align parameter of choose_baseaddr to generated aligned SSE movs
30205 (pro_epilogue_adjust_stack): Modify to track
30206 machine_frame_state::sp_realigned.
30207 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
30208 (ix86_nsaved_sseregs): Likewise.
30209 (ix86_emit_save_regs): Likewise.
30210 (ix86_emit_save_regs_using_mov): Likewise.
30211 (ix86_emit_save_sse_regs_using_mov): Likewise.
30212 (get_scratch_register_on_entry): Likewise.
30213 (gen_frame_set): New function.
30214 (gen_frame_load): Likewise.
30215 (gen_frame_store): Likewise.
30216 (emit_outlined_ms2sysv_save): Likewise.
30217 (emit_outlined_ms2sysv_restore): Likewise.
30218 (ix86_expand_prologue): Modify stack re-alignment code and call
30219 emit_outlined_ms2sysv_save when appropriate.
30220 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
30221 parameter rtx_insn *insn, which allows the function to be used to only
30222 generate the notes.
30223 (ix86_expand_epilogue): Modify validity checks of frame and stack
30224 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
30225 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
30226 * config/i386/predicates.md
30227 (save_multiple): New predicate.
30228 (restore_multiple): Likewise.
30229 * config/i386/sse.md
30230 (save_multiple<mode>): New pattern.
30231 (save_multiple_realign<mode>): Likewise.
30232 (restore_multiple<mode>): Likewise.
30233 (restore_multiple_and_return<mode>): Likewise.
30234 (restore_multiple_leave_return<mode>): Likewise.
30235 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
30237 2017-05-14 Julia Koval <julia.koval@intel.com>
30239 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
30240 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
30241 (__builtin_ia32_xsetbv): New builtins.
30242 * config/i386/i386.c (ix86_expand_special_args_builtin):
30244 (ix86_expand_builtin): Special expand for new intrinsics.
30245 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
30246 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
30247 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
30249 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30251 * cfganal.c (inverted_post_order_compute): Change argument type
30253 * cfganal.h (inverted_post_order_compute): Adjust prototype.
30254 * df-core.c (rest_of_handle_df_initialize): Adjust.
30255 (rest_of_handle_df_finish): Likewise.
30256 (df_analyze_1): Likewise.
30257 (df_analyze): Likewise.
30258 (loop_inverted_post_order_compute): Change argument to be a vec *.
30259 (df_analyze_loop): Adjust.
30260 (df_get_n_blocks): Likewise.
30261 (df_get_postorder): Likewise.
30262 * df.h (struct df_d): Change field to be a vec.
30263 * lcm.c (compute_laterin): Adjust.
30264 (compute_available): Likewise.
30265 * lra-lives.c (lra_create_live_ranges_1): Likewise.
30266 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
30267 * tree-ssa-pre.c (compute_antic): Likewise.
30269 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30271 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
30272 (depth_first_search::depth_first_search): Change structure init
30273 function to this constructor.
30274 (depth_first_search::add_bb): Rename function to this member.
30275 (depth_first_search::execute): Likewise.
30276 (flow_dfs_compute_reverse_finish): Adjust.
30278 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30280 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
30281 (longest_simple_path): Likewise.
30282 * shrink-wrap.c (spread_components): Likewise.
30283 (disqualify_problematic_components): Likewise.
30284 (emit_common_heads_for_components): Likewise.
30285 (emit_common_tails_for_components): Likewise.
30286 (insert_prologue_epilogue_for_components): Likewise.
30288 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30290 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
30293 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30295 * df-core.c (df_set_blocks): Start using auto_bitmap.
30296 (df_compact_blocks): Likewise.
30297 * df-problems.c (df_rd_confluence_n): Likewise.
30298 * df-scan.c (df_insn_rescan_all): Likewise.
30299 (df_process_deferred_rescans): Likewise.
30300 (df_update_entry_block_defs): Likewise.
30301 (df_update_exit_block_uses): Likewise.
30302 (df_entry_block_bitmap_verify): Likewise.
30303 (df_exit_block_bitmap_verify): Likewise.
30304 (df_scan_verify): Likewise.
30305 * lra-constraints.c (lra_constraints): Likewise.
30306 (undo_optional_reloads): Likewise.
30307 (lra_undo_inheritance): Likewise.
30308 * lra-remat.c (calculate_gen_cands): Likewise.
30309 (do_remat): Likewise.
30310 * lra-spills.c (assign_spill_hard_regs): Likewise.
30311 (spill_pseudos): Likewise.
30312 * tree-ssa-pre.c (bitmap_set_and): Likewise.
30313 (bitmap_set_subtract_values): Likewise.
30315 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30317 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
30318 management with auto_bitmap.
30319 (fix_inter_tick): Likewise.
30320 (fix_recovery_deps): Likewise.
30321 * ira.c (add_store_equivs): Likewise.
30322 (find_moveable_pseudos): Likewise.
30323 (split_live_ranges_for_shrink_wrap): Likewise.
30324 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
30325 (rtx_reuse_manager::seen_def_p): Likewise.
30326 (rtx_reuse_manager::set_seen_def): Likewise.
30327 * print-rtl.h (class rtx_reuse_manager): Likewise.
30329 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30331 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
30333 (migrate_btr_def): Likewise.
30334 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
30335 * df-core.c (loop_post_order_compute): Likewise.
30336 (loop_inverted_post_order_compute): Likewise.
30337 * hsa-common.h: Likewise.
30338 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
30339 * init-regs.c (initialize_uninitialized_regs): Likewise.
30340 * ipa-inline.c (resolve_noninline_speculation): Likewise.
30341 (inline_small_functions): Likewise.
30342 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
30343 * ira.c (combine_and_move_insns): Likewise.
30344 (build_insn_chain): Likewise.
30345 * loop-invariant.c (find_invariants): Likewise.
30346 * lower-subreg.c (propagate_pseudo_copies): Likewise.
30347 * predict.c (tree_predict_by_opcode): Likewise.
30348 (predict_paths_leading_to): Likewise.
30349 (predict_paths_leading_to_edge): Likewise.
30350 (estimate_loops_at_level): Likewise.
30351 (estimate_loops): Likewise.
30352 * shrink-wrap.c (try_shrink_wrapping): Likewise.
30353 (spread_components): Likewise.
30354 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
30355 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
30356 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
30357 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
30358 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
30359 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
30360 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
30361 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
30362 (mark_threaded_blocks): Likewise.
30363 (thread_through_all_blocks): Likewise.
30364 * tree-ssa.c (verify_ssa): Likewise.
30365 (execute_update_addresses_taken): Likewise.
30366 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
30368 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30370 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
30372 (post_order_compute): Likewise.
30373 (inverted_post_order_compute): Likewise.
30374 (pre_and_rev_post_order_compute_fn): Likewise.
30376 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30378 * genrecog.c (int_set::int_set): Explicitly construct our
30379 auto_vec base class.
30380 * vec.h (auto_vec::auto_vec): New constructor.
30382 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30384 * bitmap.h (class auto_bitmap): New constructor taking
30385 bitmap_obstack * argument.
30387 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30389 * bitmap.h (class auto_bitmap): Change type of m_bits to
30390 bitmap_head, and adjust ctor / dtor and member operators.
30392 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
30394 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
30395 when returned register mode doesn't match original mode.
30397 2017-05-12 Jeff Law <law@redhat.com>
30398 Jakub Jelinek <jakub@redhat.com>
30400 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
30401 we look for cc setter after the compare-elim changes.
30402 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
30403 within the vector to match what compare-elim now expects.
30404 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
30405 (xorsi3_flags, one_cmplsi2_flags): Likewise.
30407 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
30408 after the compare-elim changes.
30409 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
30410 the vector to match what compare-elim now expects.
30411 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
30412 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
30413 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
30414 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
30415 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
30417 * config/visium/visium.c (single_set_and_flags): Fix where
30418 we look for cc setter after the compare-elim changes.
30419 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
30420 with the vector to match what compare-elim now expects.
30421 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
30422 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
30423 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
30424 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
30425 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
30426 (neg<mode>2_insn_set_overflow): Likewise.
30428 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
30430 PR middle-end/79794
30431 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
30432 maybe_expand_insn call, set ops[0].target. If still set after call,
30433 set alt_rtl. Add extra arg to recursive calls.
30434 (extract_bit_field): Add alt_rtl argument. Pass to
30436 * expmed.h (extract_bit_field): Fix prototype.
30437 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
30438 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
30439 to extract_bit_field_calls.
30440 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
30441 Pass alt_rtl to extract_bit_field calls.
30442 * calls.c (store_unaligned_arguments_into_psuedos)
30443 load_register_parameters): Pass extra NULL to extract_bit_field calls.
30444 * optabs.c (maybe_legitimize_operand): Clear op->target when call
30446 * optabs.h (struct expand_operand): Add target bitfield.
30448 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30450 * compare-elim.c (try_eliminate_compare): Canonicalize
30451 operation with embedded compare to
30452 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
30453 (set (reg) (operation)].
30455 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
30457 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30460 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
30461 cost of adding a carry flag for ADC instruction.
30462 [case MINUS]: Ignore the cost of subtracting a carry flag
30463 for SBB instruction.
30465 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
30467 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
30469 * config/rs6000/bmiintrin.h: New file.
30470 * config/rs6000/bmi2intrin.h: New file.
30471 * config/rs6000/x86intrin.h: New file.
30473 2017-05-12 Jeff Law <law@redhat.com>
30475 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
30478 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
30480 PR middle-end/80707
30481 * tree-cfg.c: Remove cfg edges of unreachable case statements.
30483 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30485 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30486 early expansion of vector divide builtins.
30487 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
30488 builtins identified as having unsigned arguments.
30490 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30492 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
30493 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
30494 expansion of vector logical operations (and, andc, or, xor,
30497 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30499 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
30500 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
30502 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30504 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30505 early GIMPLE expansion of vector multiplies.
30507 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30509 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
30510 TARGET_HAVE_MOVT conditional.
30511 (movt splitter): Likewise.
30513 2017-05-12 Richard Biener <rguenther@suse.de>
30515 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
30516 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30517 Fold all stmts not inplace.
30519 2017-05-12 Richard Biener <rguenther@suse.de>
30521 PR tree-optimization/80713
30522 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
30523 inserted_exprs bit for not removed stmts.
30525 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
30527 PR middle-end/69921
30528 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
30529 parallelized" attribute for parallelized OpenACC kernels.
30530 * omp-offload.c (execute_oacc_device_lower): Use it.
30532 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
30533 Set "oacc kernels" attribute.
30534 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
30535 parameter. Adjust all users.
30536 (oacc_fn_attrib_kernels_p): Remove function.
30537 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
30538 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
30539 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
30540 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
30541 assert "oacc kernels" attribute is set.
30543 2017-05-11 Carl Love <cel@us.ibm.com>
30545 * config/rs6000/rs6000-c: Add support for built-in functions
30546 vector unsigned char vec_popcnt (vector signed char)
30547 vector unsigned char vec_popcnt (vector unsigned char)
30548 vector unsigned short vec_popcnt (vector signed short)
30549 vector unsigned short vec_popcnt (vector unsigned short)
30550 vector unsigned int vec_popcnt (vector signed int)
30551 vector unsigned int vec_popcnt (vector unsigned int)
30552 vector unsigned long long vec_popcnt (vector signed long long)
30553 vector unsigned long long vec_popcnt (vector unsigned long long)
30554 vector signed long long vec_slo (vector signed long long,
30555 vector signed char)
30556 vector signed long long vec_slo (vector signed long long,
30557 vector unsigned char)
30558 vector unsigned long long vec_slo (vector unsigned long long,
30559 vector signed char)
30560 vector unsigned long long vec_slo (vector unsigned long long,
30561 vector unsigned char)
30562 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
30563 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
30564 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
30565 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
30566 * doc/extend.texi: Update the built-in documentation file for the
30567 new built-in functions.
30569 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
30571 * attribs.h (sorted_attr_string): Move machine independent
30572 functions for target clone support from the i386 port to common
30573 code. Rename ix86_function_versions to common_function_versions.
30574 Rename make_name to make_unique_name.
30575 (common_function_versions): Likewise.
30576 (make_unique_name): Likewise.
30577 (make_dispatcher_decl): Likewise.
30578 (is_function_default_version): Likewise.
30579 * attribs.c (attr_strcmp): Likewise.
30580 (sorted_attr_string): Likewise.
30581 (common_function_versions): Likewise.
30582 (make_unique_name): Likewise.
30583 (make_dispatcher_decl): Likewise.
30584 (is_function_default_version): Likewise.
30585 * config/i386/i386.c (attr_strcmp): Likewise.
30586 (sorted_attr_string): Likewise.
30587 (ix86_function_versions): Likewise.
30588 (make_name): Likewise.
30589 (make_dispatcher_decl): Likewise.
30590 (is_function_default_version): Likewise.
30591 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
30593 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30596 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
30597 Account for direct move costs for vec_construct of integer
30600 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
30603 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
30604 (UNSPEC_STX_ATOMIC): Ditto.
30605 (loaddi_via_sse): New insn.
30606 (storedi_via_sse): Ditto.
30607 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
30608 Update corresponding peephole2 patterns.
30609 (atomic_storedi_fpu): Ditto.
30611 2017-05-11 Julia Koval <julia.koval@intel.com>
30613 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
30614 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
30616 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
30617 (__builtin_ia32_rsqrt14ss_mask): New builtins.
30618 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
30620 2017-05-11 Nathan Sidwell <nathan@acm.org>
30622 * graphite-poly.c: Include dumpfile.h.
30624 * dumpfle.h (dump_function): Declare here ...
30625 * tree-dump.h (dump_function): ... not here.
30626 * dumpfile.c: #include tree-cfg.h.
30627 (dump_function): Move here from ...
30628 * tree-dump.c (dump_function): ... here.
30629 * gimplify.c: #include splay-tree.h, not tree-dump.h.
30630 * graphite-poly.c: Don't include tree-dump.h.
30631 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
30632 * print-tree.c: Likewise.
30633 * stor-layout.c: Likewise.
30634 * tree-nested.c: Likewise.
30636 * dumpfile.c (dump_start): Use TDF_FLAGS.
30637 (dump_enable_all): Fix TDF_KIND check thinko.
30639 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
30641 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30642 array entries to represent two legal parameterizations of the
30643 overloaded __builtin_cmpb function, as represented by the
30644 P6_OV_BUILTIN_CMPB constant.
30645 (altivec_resolve_overloaded_builtin): Add special case handling
30646 for the __builtin_cmpb function, as represented by the
30647 P6_OV_BUILTIN_CMPB constant.
30648 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
30649 (BU_P6_64BIT_2): New macro.
30650 (BU_P6_OVERLOAD_2): New macro
30651 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
30652 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
30653 (CMPB): Add overload support to represent both 32-bit and 64-bit
30654 compare-bytes function.
30655 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
30656 support for TARGET_CMPB.
30657 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
30658 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30659 documentation of the __builtin_cmpb overloaded built-in function.
30661 2017-05-11 Richard Biener <rguenther@suse.de>
30663 PR tree-optimization/80705
30664 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
30665 bases are not vectorizable.
30667 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30669 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
30670 when counting register pressure.
30672 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30674 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
30675 (iv_ca_more_deps): Renamed to ...
30676 (iv_ca_compare_deps): ... this.
30677 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
30679 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30681 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
30683 (determine_group_iv_costs): ... here.
30684 (find_inv_vars_cb): Record inv var if it's not recorded before.
30686 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30688 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
30689 (get_shiftadd_cost): Ditto.
30691 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30693 * tree-ssa-address.c: Include header file.
30694 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
30696 (add_to_parts): Refactor.
30697 (addr_to_parts): New parameter. Update use of move_hint_to_base.
30698 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
30701 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30703 PR tree-optimization/53090
30704 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
30706 (extract_cond_operands): Detect condition with IV on both sides
30707 and return COMP_IV_EXPR_2.
30708 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
30709 (rewrite_use_compare): Simplify by removing call to function
30710 extract_cond_operands.
30712 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30714 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
30715 (extract_cond_operands): Detect condition comparing against non-
30716 invariant bound and return appropriate enum value.
30717 (find_interesting_uses_cond): Update use of extract_cond_operands.
30718 Handle its return value accordingly.
30719 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
30721 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30723 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
30724 nonlinear iv_use computation in loop invariant sensitive way.
30726 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30728 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
30729 (find_iv_candidates): Call relate_compare_use_with_all_cands.
30731 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30733 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
30734 (dump_cand): Support iv_cand.inv_exprs.
30735 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
30737 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
30740 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30742 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
30744 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
30745 as local function. Include necessary header files.
30746 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
30748 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30750 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
30752 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30754 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
30755 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
30756 RSHIFT_EXPR and BIT_NOT_EXPR.
30758 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30760 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
30761 (adjust_setup_cost): New parameter supporting round up adjustment.
30762 (struct address_cost_data): Delete.
30763 (force_expr_to_var_cost): Don't bound cost with spill_cost.
30764 (split_address_cost, ptr_difference_cost): Delete.
30765 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
30766 (struct ainc_cost_data): New struct.
30767 (get_address_cost_ainc): New function.
30768 (get_address_cost, get_computation_cost): Reimplement.
30769 (determine_group_iv_cost_address): Record inv_expr for all uses of
30771 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
30772 (iv_ca_has_deps): Reimplemented to ...
30773 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
30775 (iv_ca_extend): Call iv_ca_more_deps.
30777 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30779 * tree-ssa-address.c (struct mem_address): Move to header file.
30780 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
30781 * tree-ssa-address.h (struct mem_address): Move from C file.
30782 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
30784 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30786 * tree-affine.h (aff_combination_type): New interface.
30787 (aff_combination_zero_p): Remove static.
30788 (aff_combination_const_p): New interface.
30789 (aff_combination_singleton_var_p): New interfaces.
30791 2017-05-11 Richard Biener <rguenther@suse.de>
30793 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30794 Skip unreachable blocks and destinations.
30795 (eliminate): Move stmt removal and fixup ...
30796 (fini_eliminate): ... here. Skip inserted exprs.
30797 (pass_pre::execute): Move fini_pre after fini_eliminate.
30798 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
30799 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
30800 PRE to get rid of dead code that has invalid SSA form and
30801 split critical edges again.
30803 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30805 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30807 2017-05-11 Richard Biener <rguenther@suse.de>
30809 * passes.c (execute_function_todo): Verify loops if they are
30810 said to be up-to-date.
30811 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
30812 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
30814 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
30817 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
30818 handle calling assemble_external ourself.
30821 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
30822 modes with zero size. Enhance comment.
30824 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30826 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
30827 built-ins for vec_xl and vec_xst with short and char pointer
30830 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30832 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
30833 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
30834 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
30835 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
30836 (_mm_maskz_min_round_ss): New intrinsics.
30837 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
30838 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
30839 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
30840 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
30841 (__builtin_ia32_minss_mask_round): New builtins.
30842 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30843 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30844 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
30846 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
30847 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
30849 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
30852 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30854 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
30855 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
30856 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
30857 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
30858 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
30859 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
30860 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
30861 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30862 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
30863 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
30864 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
30865 (__builtin_ia32_mulss_mask_round): New builtins.
30866 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30867 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30868 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
30870 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
30871 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
30873 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
30876 2017-05-10 Julia Koval <julia.koval@intel.com>
30878 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
30879 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
30880 (_mm256_setr_m128i): New intrinsics.
30882 2017-05-10 Julia Koval <julia.koval@intel.com>
30884 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
30885 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
30886 (_mm_maskz_rcp14_ss): New intrinsics.
30887 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
30888 (__builtin_ia32_rcp14ss_mask): New builtins.
30889 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
30891 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
30893 PR tree-optimization/51513
30894 * tree-cfg.c (gimple_seq_unreachable_p): New function.
30895 (assert_unreachable_fallthru_edge_p): Use it.
30896 (group_case_labels_stmt): Likewise.
30897 * tree-cfg.h: Prototype it.
30898 * stmt.c: Include cfghooks.h and tree-cfg.h.
30899 (emit_case_dispatch_table) <gap_label>: New local variable.
30900 Use it to fill dispatch table gaps.
30901 Test for default_label before updating probabilities.
30902 (expand_case) <default_label>: Remove unneeded initialization.
30903 Test for unreachable default case statement and remove its edge.
30904 Set default_label accordingly.
30905 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
30907 2017-05-10 Carl Love <cel@us.ibm.com>
30909 * config/rs6000/rs6000-c: Add support for built-in functions
30910 vector signed char vec_neg (vector signed char)
30911 vector signed short int vec_neg (vector short int)
30912 vector signed int vec_neg (vector signed int)
30913 vector signed long long vec_neg (vector signed long long)
30914 vector float vec_neg (vector float)
30915 vector double vec_neg (vector double)
30916 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
30918 * config/rs6000/altivec.h: Add define for vec_neg
30919 * doc/extend.texi: Update the built-in documentation for the
30920 new built-in functions.
30922 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30924 PR tree-optimization/77644
30925 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
30927 2017-05-10 Nathan Sidwell <nathan@acm.org>
30929 * dumpfile.h (TDI_lang_all): New.
30930 (TDF_KIND): New. Renumber others
30931 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
30933 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
30935 (get_dump_file_name): Adjust suffix generation.
30936 (dump_enable_all): Use TDF_KIND.
30937 * doc/invoke.texi (-fdump-lang-all): Document.
30939 * dumpfile.h: Tabify.
30941 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
30944 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
30945 Move member access before delete.
30947 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
30949 * tree-inline.c (expand_call_inline): Split block at stmt
30952 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
30955 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
30956 are now unused after splitting mov{sf,sd}_hardfloat.
30957 (f32_lr2): Likewise.
30958 (f32_lm): Likewise.
30959 (f32_lm2): Likewise.
30960 (f32_li): Likewise.
30961 (f32_li2): Likewise.
30962 (f32_lv): Likewise.
30963 (f32_sr): Likewise.
30964 (f32_sr2): Likewise.
30965 (f32_sm): Likewise.
30966 (f32_sm2): Likewise.
30967 (f32_si): Likewise.
30968 (f32_si2): Likewise.
30969 (f32_sv): Likewise.
30970 (f32_dm): Likewise.
30971 (f32_vsx): Likewise.
30972 (f32_av): Likewise.
30973 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
30974 For movsf, order stores so the VSX stores occur before the GPR
30975 store which encourages the register allocator to use a traditional
30976 FPR instead of a GPR. For movsd, order the stores so that the GPR
30977 store comes before the VSX stores to allow the power6 to work.
30978 This is due to the power6 not having a 32-bit integer store
30979 instruction from a FPR.
30980 (movsf_hardfloat): Likewise.
30981 (movsd_hardfloat): Likewise.
30983 2017-05-09 Martin Sebor <msebor@redhat.com>
30985 PR translation/80280
30986 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
30989 PR translation/80280
30990 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
30991 data member added in r247778.
30992 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
30994 2017-05-09 Nathan Sidwell <nathan@acm.org>
30996 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
30998 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
31001 2017-05-09 Marek Polacek <polacek@redhat.com>
31003 * doc/invoke.texi: Fix typo.
31005 2017-05-09 Richard Biener <rguenther@suse.de>
31007 * tree-vrp.c (vrp_val_is_max): Adjust comment.
31008 (vrp_val_is_min): Likewise.
31009 (set_value_range_to_value): Likewise.
31010 (set_value_range_to_nonnegative): Likewise.
31011 (gimple_assign_nonzero_p): Likewise.
31012 (gimple_stmt_nonzero_p): Likewise.
31013 (vrp_int_const_binop): Likewise. Remove unreachable case.
31014 (adjust_range_with_scev): Adjust comments.
31015 (compare_range_with_value): Likewise.
31016 (extract_range_from_phi_node): Likewise.
31017 (test_for_singularity): Likewise.
31019 2017-05-09 Richard Biener <rguenther@suse.de>
31021 * tree-vrp.c (get_single_symbol): Add assert that we don't
31022 get overflowed constants as invariant part.
31023 (compare_values_warnv): Add comment before the TREE_NO_WARNING
31024 checks. Use wi::cmp instead of recursing for integer constants.
31025 (compare_values): Just ignore whether we assumed undefined
31026 overflow instead of failing the compare.
31027 (extract_range_for_var_from_comparison_expr): Add comment before the
31028 TREE_NO_WARNING sets.
31029 (test_for_singularity): Likewise.
31030 (extract_range_from_comparison): Do not disable optimization
31031 when we assumed undefined overflow.
31032 (extract_range_basic): Remove init of unused var.
31034 2017-05-09 Richard Biener <rguenther@suse.de>
31036 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
31037 (extract_range_from_multiplicative_op_1): Adjust.
31038 (extract_range_from_binary_expr_1): Use int_const_binop.
31040 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
31043 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
31044 rs6000_store_data_bypass_p in seven define_bypass directives and
31045 in several comments.
31046 * config/rs6000/rs6000-protos.h: Add prototype for
31047 rs6000_store_data_bypass_p function.
31048 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
31049 function implements slightly different (rs6000-specific) semantics
31050 than store_data_bypass_p, returning false rather than aborting
31051 with assertion error when arguments do not satisfy the
31052 requirements of store data bypass.
31053 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
31054 rs6000_store_data_bypass_p.
31056 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
31058 * config/xtensa/xtensa-protos.h
31059 (xtensa_initial_elimination_offset): New declaration.
31060 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
31061 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
31062 macro definition, add case for FRAME_POINTER_REGNUM when
31063 FRAME_GROWS_DOWNWARD.
31064 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
31065 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
31066 xtensa_initial_elimination_offset.
31068 2017-05-08 Nathan Sidwell <nathan@acm.org>
31070 * doc/invoke.texi: Alphabetize -fdump options.
31072 2017-05-08 Martin Sebor <msebor@redhat.com>
31074 PR translation/80280
31075 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
31077 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
31079 * target.def (compute_frame_layout): New optional target hook.
31080 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
31081 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
31082 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
31084 * reload1.c (verify_initial_elim_offsets): Likewise.
31085 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
31086 (use_simple_return_p): Call arm_compute_frame_layout if needed.
31087 (arm_get_frame_offsets): Split up into this ...
31088 (arm_compute_frame_layout): ... and this function.
31090 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
31092 * config/aarch64/constraints.md (Usa): New constraint.
31093 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
31095 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
31097 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
31098 with_multilib_list after it has been checked.
31100 2017-05-08 Richard Biener <rguenther@suse.de>
31102 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
31103 (bitmap_set_subtract_values): Likewise.
31105 2017-05-08 Richard Biener <rguenther@suse.de>
31107 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
31108 (gimple_assign_nonzero): ... this and remove strict_overflow_p
31110 (gimple_stmt_nonzero_warnv_p): Rename to ...
31111 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
31113 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
31114 (extract_range_basic): Adjust, do not disable propagation on
31115 strict overflow sensitive simplification.
31116 (vrp_visit_cond_stmt): Likewise.
31118 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
31120 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
31121 body size unconditionally.
31123 2017-05-07 Jeff Law <law@redhat.com>
31126 2017-05-06 Jeff Law <law@redhat.com>
31127 PR tree-optimization/78496
31128 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31131 PR tree-optimization/78496
31132 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31133 (simplify_stmt_using_ranges): Call it.
31134 (vrp_dom_walker::before_dom_children): Extract equivalences
31135 from an ASSERT_EXPR with an equality comparison against a
31138 2017-05-06 Jeff Law <law@redhat.com>
31140 PR tree-optimization/78496
31141 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31144 PR tree-optimization/78496
31145 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31146 (simplify_stmt_using_ranges): Call it.
31147 (vrp_dom_walker::before_dom_children): Extract equivalences
31148 from an ASSERT_EXPR with an equality comparison against a
31151 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31153 * lra-constraints.c (lra_copy_reg_equiv): New function.
31154 (split_reg): Use it to copy equivalence information from the
31155 original register to the spill register.
31157 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31159 PR rtl-optimization/75964
31160 * simplify-rtx.c (simplify_const_relational_operation): Remove
31161 invalid handling of comparisons of integer ABS.
31163 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
31165 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
31166 initialize to zero.
31167 (init_regs): Remove declaration.
31168 (function_arg_advance_32): Initialize error_p as boolean variable.
31170 2017-05-05 Nathan Sidwell <nathan@acm.org>
31172 * store-motion.c (remove_reachable_equiv_notes): Reformat long
31173 lines. Use for (;;).
31175 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31177 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
31178 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
31179 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
31180 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
31181 VF=2 that require versioning.
31183 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31185 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
31188 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31190 * diagnostic.h (diagnostic_override_option_index): Convert from
31191 macro to inline function.
31193 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31195 * diagnostic.c (last_module_changed_p): New function.
31196 (set_last_module): New function.
31197 (diagnostic_report_current_module): Convert macro usage to
31198 the above functions.
31199 * diagnostic.h (diagnostic_context::last_module): Strengthen
31200 from const line_map * to const line_map_ordinary *.
31201 (diagnostic_last_module_changed): Delete macro.
31202 (diagnostic_set_last_module): Delete macro.
31204 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31206 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
31207 with diagnostic_report_diagnostic.
31208 (diagnostic_n_impl_richloc): Likewise.
31209 * diagnostic.h (report_diagnostic): Delete macro.
31210 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
31211 with diagnostic_report_diagnostic.
31212 * substring-locations.c (format_warning_va): Likewise.
31214 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31216 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
31217 save/restor of format_spec. Move option-printing code to...
31218 (print_option_information): ...this new function, and
31219 reimplement by simply printing to the pretty_printer,
31220 rather than appending to the format string.
31222 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31224 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
31225 handling logic into...
31226 (update_effective_level_from_pragmas): ...this new function.
31228 2017-05-04 Andrew Waterman <andrew@sifive.com>
31230 * config/riscv/riscv.opt (mstrict-align): New option.
31231 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
31232 (SLOW_UNALIGNED_ACCESS): Define.
31233 (riscv_slow_unaligned_access): Declare.
31234 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
31236 (riscv_slow_unaligned_access): New variable.
31237 (rocket_tune_info): Set slow_unaligned_access to true.
31238 (optimize_size_tune_info): Set slow_unaligned_access to false.
31239 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
31240 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
31241 (riscv_option_override): Set riscv_slow_unaligned_access.
31242 * doc/invoke.texi: Add -mstrict-align to RISC-V.
31244 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
31246 * config/riscv/riscv.md: Unify indentation.
31248 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
31253 * config/rs6000/rs6000.md (u code attribute): Add FIX and
31255 (extendsi<mode>2): Add support for doing sign extension via
31256 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
31257 don't have ISA 3.0 instructions.
31258 (extendsi<mode>2 splitter): Likewise.
31259 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
31260 generate the normal insns since SImode can now go in vector
31261 registers. Disallow the special UNSPECs needed for previous
31262 machines to hide SImode being used. Add new insns
31263 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
31264 (fix_trunc<mode>si2_stfiwx): Likewise.
31265 (fix_trunc<mode>si2_internal): Likewise.
31266 (fixuns_trunc<mode>si2): Likewise.
31267 (fixuns_trunc<mode>si2_stfiwx): Likewise.
31268 (fctiw<u>z_<mode>_smallint): Likewise.
31269 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
31270 of floating point to 32-bit integer from doing a direct move to
31271 the GPR registers to do a store.
31272 (fctiwz_<mode>): Break long line.
31274 2017-05-05 Bin Cheng <bin.cheng@arm.com>
31276 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
31277 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
31278 (addr_list, addr_offset_valid_p): New.
31279 (split_address_groups): Check offset validity with above function.
31280 (gt-tree-ssa-loop-ivopts.h): Include header file.
31282 2017-05-05 Nathan Sidwell <nathan@acm.org>
31284 * config.gcc (arm*-*-*): Add missing 'fi'.
31286 2017-05-05 Steve Ellcey <sellcey@cavium.com>
31288 * doc/invoke.texi (-fopt-info): Explicitly say order of options
31289 included in -fopt-info does not matter.
31290 * doc/optinfo.texi (-fopt-info): Fix description of default
31291 behavour. Explicitly say order of options included in -fopt-info
31294 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
31296 * config.gcc: Allow combinations of aprofile and rmprofile values for
31297 --with-multilib-list.
31298 * config/arm/t-multilib: New file.
31299 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
31300 variables. Remove setting of ISA and floating-point ABI in
31301 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
31302 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
31303 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
31304 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
31306 * config/arm/t-rmprofile: Likewise except for the matches changes.
31307 * doc/install.texi (--with-multilib-list): Document the combination of
31308 aprofile and rmprofile values and warn about pitfalls in doing that.
31310 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31312 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
31313 (movdi_aarch64): Likewise.
31315 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31317 PR tree-optimization/80632
31318 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
31320 (build_arrays): Initialize it for virtual phis.
31321 (fix_phi_nodes): Use it for virtual phis.
31323 PR tree-optimization/80558
31324 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
31325 [x, y] op z into [x op, y op z] for op & or | if conditions
31328 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
31329 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31332 * config/arm/arm.md (use_literal_pool): Remove.
31333 (64-bit immediate split): No longer takes cost into consideration
31334 if arm_disable_literal_pool is enabled.
31335 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
31336 used when arm_disable_literal_pool is enabled.
31337 (arm_max_const_double_inline_cost): Remove use of
31338 arm_disable_literal_pool.
31339 (push_minipool_fix): Add assert.
31340 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
31341 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
31342 (no_literal_pool_sf_immediate): New.
31344 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31346 PR tree-optimization/80613
31347 * tree-ssa-dce.c (propagate_necessity): Remove cases for
31348 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
31350 2017-05-05 Richard Biener <rguenther@suse.de>
31352 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
31354 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
31356 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
31357 of this flag from insn conditions due to removal from r247495.
31359 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31361 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
31363 (arm_early_store_addr_dep_ptr): Likewise.
31364 * config/arm/aarch-common-protos.h
31365 (arm_early_load_addr_dep_ptr): Add prototype.
31366 (arm_early_store_addr_dep_ptr): Likewise.
31367 * config/arm/cortex-a53.md: Add new bypasses.
31369 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31371 * tree.c (next_type_uid): Change type to unsigned.
31372 (type_hash_canon): Decrement back next_type_uid if
31373 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
31374 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
31377 2017-05-04 Martin Sebor <msebor@redhat.com>
31379 * builtins.c: Fix a trivial typo in a comment.
31381 PR middle-end/79234
31382 * builtins.c (check_sizes): Adjust to handle reading past the end.
31383 Avoid printing excessive upper bound of ranges. Use %E to print
31384 tree nodes instead of converting them to %wu.
31385 (expand_builtin_memchr): New function.
31386 (compute_dest_size): Rename...
31387 (compute_objsize): ...to this.
31388 (expand_builtin_memcpy): Adjust.
31389 (expand_builtin_mempcpy): Adjust.
31390 (expand_builtin_strcat): Adjust.
31391 (expand_builtin_strcpy): Adjust.
31392 (check_strncat_sizes): Adjust.
31393 (expand_builtin_strncat): Adjust.
31394 (expand_builtin_strncpy): Adjust and simplify.
31395 (expand_builtin_memset): Adjust.
31396 (expand_builtin_bzero): Adjust.
31397 (expand_builtin_memcmp): Adjust.
31398 (expand_builtin): Handle memcmp.
31399 (maybe_emit_chk_warning): Check strncat just once.
31401 2017-05-04 Martin Sebor <msebor@redhat.com>
31403 PR preprocessor/79214
31404 PR middle-end/79222
31405 PR middle-end/79223
31406 * builtins.c (check_sizes): Add inlining context and issue
31407 warnings even when -Wno-system-headers is set.
31408 (check_strncat_sizes): Same.
31409 (expand_builtin_strncat): Same.
31410 (expand_builtin_memmove): New function.
31411 (expand_builtin_stpncpy): Same.
31412 (expand_builtin): Handle memmove and stpncpy.
31414 2017-05-04 Bin Cheng <bin.cheng@arm.com>
31416 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
31417 which is not used any more.
31419 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31421 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
31423 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31425 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
31426 (cortexa53_tunings): Likewise.
31427 (cortexa57_tunings): Likewise.
31428 (cortexa72_tunings): Likewise.
31429 (cortexa73_tunings): Likewise.
31431 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31433 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
31434 Set loop alignment to 8.
31436 2017-05-04 Martin Sebor <msebor@redhat.com>
31438 PR translation/80280
31439 * builtins.c (expand_builtin_object_size): Add missing quoting to
31440 %D and like directives.
31441 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
31442 (hsa_type_for_tree_type): Same.
31443 (verify_function_arguments): Same.
31444 * symtab.c (symbol_table::change_decl_assembler_name): Same.
31445 * varasm.c (get_section): Same.
31448 2017-05-04 Martin Sebor <msebor@redhat.com>
31450 PR translation/80280
31451 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
31453 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31455 * config/aarch64/aarch64.c (generic_addrcost_table):
31456 Change HI/TI mode setting.
31458 2017-05-04 Martin Jambor <mjambor@suse.cz>
31460 PR tree-optimization/80622
31461 * tree-sra.c (comes_initialized_p): New function.
31462 (build_accesses_from_assign): Only set write lazily when
31463 comes_initialized_p is false.
31464 (analyze_access_subtree): Use comes_initialized_p.
31465 (propagate_subaccesses_across_link): Assert !comes_initialized_p
31466 instead of testing for PARM_DECL.
31468 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31470 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
31471 constraint on operand 0 to allow more general addressing modes.
31472 Adjust output template.
31473 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
31475 * config/aarch64/aarch64-protos.h
31476 (aarch64_address_valid_for_prefetch_p): Declare prototype.
31477 * config/aarch64/constraints.md (Dp): New address constraint.
31478 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
31481 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
31483 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
31484 update use of estimate_ipcp_clone_size_and_time.
31485 (estimate_local_effects): Update use of
31486 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
31487 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
31488 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
31489 Return nonspecialized time.
31491 2017-05-04 Richard Biener <rguenther@suse.de>
31493 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
31494 for the last VUSE which def dominates the PHI. Directly call
31496 (get_continuation_for_phi_1): Remove.
31498 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
31500 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
31501 to explain the use of truncating division. Cap the number of
31502 iterations to the maximum given by nb_iterations_upper_bound,
31505 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31507 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
31508 * configure: Regenerate.
31509 * config.in: Regenerate.
31510 * config/i386/driver-mingw32.c: new file.
31511 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
31512 * config.host: Link driver-mingw32.o on MinGW host.
31513 * doc/install.texi: Document new --enable-mingw-wildcard configure
31516 2017-05-04 Marek Polacek <polacek@redhat.com>
31518 PR tree-optimization/80612
31519 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
31521 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31522 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
31524 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
31525 (movt splitter): Likewise.
31526 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
31527 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
31528 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
31529 block for Thumb-1 with MOVT.
31530 (thumb2_legitimate_address_p): Move code block ...
31531 (can_avoid_literal_pool_for_label_p): ... into this new function.
31532 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
31534 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
31535 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
31536 "M-profile targets with the MOVT instruction".
31538 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31540 * config/arm/arm-builtins.c (arm_init_builtins): Rename
31541 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
31542 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
31544 2017-05-04 Martin Liska <mliska@suse.cz>
31546 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
31547 variable cond_code.
31549 2017-05-04 Richard Biener <rguenther@suse.de>
31551 * tree.c (array_at_struct_end_p): Handle arrays at struct
31552 end with flexarrays more conservatively. Refactor and treat
31553 arrays of arrays or aggregates more strict. Fix
31554 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
31555 * tree.c (array_at_struct_end_p): Adjust prototype.
31556 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
31557 * gimple-fold.c (get_range_strlen): Likewise.
31558 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
31560 2017-05-04 Richard Biener <rguenther@suse.de>
31562 PR tree-optimization/31130
31563 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
31565 (supports_overflow_infinity): Likewise.
31566 (is_negative_overflow_infinity): Likewise.
31567 (is_positive_overflow_infinity): Likewise.
31568 (is_overflow_infinity): Likewise.
31569 (stmt_overflow_infinity): Likewise.
31570 (overflow_infinity_range_p): Likewise.
31571 (usable_range_p): Remove as always returning true.
31572 (make_overflow_infinity): Remove.
31573 (negative_overflow_infinity): Likewise.
31574 (positive_overflow_infinity): Likewise.
31575 (avoid_overflow_infinity): Likewise.
31576 (set_value_range): Adjust accordingly.
31577 (set_value_range_to_nonnegative): Likewise, remove now unused
31578 overflow_infinity arg.
31579 (vrp_operand_equal_p): Adjust.
31580 (update_value_range): Likewise.
31581 (range_int_cst_singleton_p): Likewise.
31582 (operand_less_p): Likewise.
31583 (compare_values_warnv): Likewise.
31584 (extract_range_for_var_from_comparison_expr): Likewise.
31585 (vrp_int_const_binop): Likewise.
31586 (zero_nonzero_bits_from_vr): Likewise.
31587 (extract_range_from_multiplicative_op_1): Likewise.
31588 (extract_range_from_binary_expr_1): Likewise.
31589 (extract_range_from_unary_expr): Likewise.
31590 (extract_range_from_comparison): Likewise.
31591 (extract_range_basic): Likewise.
31592 (adjust_range_with_scev): Likewise.
31593 (compare_ranges): Likewise.
31594 (compare_range_with_value): Likewise.
31595 (dump_value_range): Likewise.
31596 (test_for_singularity): Likewise, remove strict_overflow_p parameter
31598 (simplify_cond_using_ranges): Adjust.
31600 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31602 * brig-builtins.def: Added a builtin for class_f64.
31603 * builtin-types.def: Added a builtin type needed by class_f64.
31605 2017-05-03 Jason Merrill <jason@redhat.com>
31607 * timevar.def: Add TV_CONSTEXPR.
31609 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31611 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
31613 2017-05-03 Martin Jambor <mjambor@suse.cz>
31615 * ipa-prop.c (ipa_update_after_lto_read): Removed.
31616 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
31617 * ipa-cp.c (ipcp_propagate_stage): Do not call
31618 ipa_update_after_lto_read.
31619 * ipa-inline.c (ipa_inline): Likewise.
31621 2017-05-03 Martin Jambor <mjambor@suse.cz>
31623 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
31624 tag. Added a default constructor and a destructor.
31625 (ipa_edge_args_sum_t): New class;
31626 (ipa_edge_args_sum): Declare.
31627 (ipa_edge_args_vector): Remove declaration.
31628 (IPA_EDGE_REF): Use ipa_edge_args_sum.
31629 (ipa_free_edge_args_substructures): Remove declaration.
31630 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
31631 (ipa_edge_args_info_available_for_edge_p): Likewise.
31632 * ipa-prop.c (ipa_edge_args_vector): Removed.
31633 (edge_removal_hook_holder): Likewise.
31634 (edge_duplication_hook_holder): Likewise.
31635 (ipa_edge_args_sum): New variable.
31636 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
31637 ipa_edge_args_vector.
31638 (ipa_free_edge_args_substructures): Likewise.
31639 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
31640 ipa_edge_args_vector.
31641 (ipa_edge_removal_hook): Turned into method
31642 ipa_edge_args_sum_t::remove.
31643 (ipa_edge_duplication_hook): Turned into method
31644 ipa_edge_args_sum_t::duplicate.
31645 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
31646 registering edge hooks.
31647 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
31648 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
31649 ipa_edge_args_sum instead of ipa_edge_args_vector.
31650 * ipa-profile.c (ipa_profile): Likewise.
31652 2017-05-03 Martin Jambor <mjambor@suse.cz>
31654 * symbol-summary.h (function_summary): New method exists.
31655 (function_summary::symtab_removal): Deallocate through release.
31656 (call_summary): New class.
31657 (gt_ggc_mx): New overload.
31658 (gt_pch_nx): Likewise.
31659 (gt_pch_nx): Likewise.
31661 2017-05-03 Jeff Law <law@redhat.com>
31663 PR tree-optimization/78496
31664 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
31665 from simplify_cond_using_ranges. Split off code to walk
31666 backwards through casts into ...
31667 (simplify_cond_using_ranges_2): New function.
31668 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
31669 (execute_vrp): After identifying jump threads, call
31670 simplify_cond_using_ranges_2.
31672 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
31675 * ipa-inline.h (inline_summary): Add ctor.
31676 (create_ggc): Do not use ggc_cleared_alloc.
31678 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
31679 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31681 * gcc.c (handle_braces): Support escaping in switch matching
31683 * doc/invoke.texi (Spec Files): Document it.
31684 Remove superfluous @code markup in items.
31686 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31688 * diagnostic-show-locus.c (struct column_range): New struct.
31689 (get_affected_columns): New function.
31690 (get_printed_columns): New function.
31691 (struct correction): New struct.
31692 (correction::ensure_capacity): New function.
31693 (correction::ensure_terminated): New function.
31694 (struct line_corrections): New struct.
31695 (line_corrections::~line_corrections): New dtor.
31696 (line_corrections::add_hint): New function.
31697 (layout::print_trailing_fixits): Reimplement in terms of the new
31699 (selftest::test_overlapped_fixit_printing): New function.
31700 (selftest::diagnostic_show_locus_c_tests): Call it.
31702 2017-05-03 Nathan Sidwell <nathan@acm.org>
31704 Canonicalize canonical type hashing
31705 * tree.h (type_hash_canon_hash): Declare.
31706 * tree.c (type_hash_list, attribute_hash_list): Move into
31707 type_hash_canon_hash.
31708 (build_type_attribute_qual_variant): Break out hash code calc into
31709 type_hash_canon_hash.
31710 (type_hash_canon_hash): New. Generic type hash computation.
31711 (build_range_type_1, build_array_type_1, build_function_type,
31712 build_method_type_directly, build_offset_type, build_complex_type,
31713 make_vector_type): Call it.
31715 2017-05-03 Richard Biener <rguenther@suse.de>
31717 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
31718 When all DRs have unknown misaligned do not always peel
31719 when there is a store but apply the same costing model as if
31720 there were only loads.
31722 2017-05-03 Richard Biener <rguenther@suse.de>
31725 PR tree-optimization/80492
31726 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
31727 compare_base_decls returning dont-know properly.
31729 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
31731 * config/arm/iterators.md (CCSI): New mode iterator.
31732 (arch): New mode attribute.
31733 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
31734 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
31735 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
31736 code iterator for success result mode.
31737 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
31738 the corresponding new insn generators.
31740 2017-05-03 Bin Cheng <bin.cheng@arm.com>
31743 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31744 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31746 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
31748 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
31749 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
31750 (DDR_A): Wrap DDR argument in brackets.
31751 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
31752 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
31753 (DDR_REVERSED_P): Likewise.
31755 2017-05-03 Jakub Jelinek <jakub@redhat.com>
31757 PR tree-optimization/79472
31758 * tree-switch-conversion.c (struct switch_conv_info): Add
31759 contiguous_range and default_case_nonstandard fields.
31760 (collect_switch_conv_info): Compute contiguous_range and
31761 default_case_nonstandard fields, don't clear final_bb if
31762 contiguous_range and only the default case doesn't have the required
31764 (check_all_empty_except_final): Set default_case_nonstandard instead
31765 of failing if contiguous_range and the default case doesn't have empty
31767 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
31768 and only the default case doesn't have the required constants. Skip
31770 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
31771 if default_case_nonstandard.
31772 (build_constructors): Build constant 1 just once. Assert that default
31773 values aren't inserted in between cases if contiguous_range. Skip
31775 (build_arrays): Skip virtual phis.
31776 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
31777 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
31778 Handle virtual phis.
31779 (gen_inbound_check): Handle default_case_nonstandard case.
31780 (process_switch): Adjust check_final_bb caller. Call
31781 gather_default_values with the first non-default case instead of
31782 default case if default_case_nonstandard.
31784 2017-05-02 Nathan Sidwell <nathan@acm.org>
31786 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
31787 check. Fix formatting.
31789 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
31791 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
31792 errors when comparing specialized and unspecialized times.
31794 2017-05-02 David Malcolm <dmalcolm@redhat.com>
31796 * diagnostic-show-locus.c
31797 (layout::should_print_annotation_line_p): Make private.
31798 (layout::print_annotation_line): Make private.
31799 (layout::annotation_line_showed_range_p): Make private.
31800 (layout::show_ruler): Make private.
31801 (layout::print_source_line): Make private. Pass in line and
31802 line_width, rather than calling location_get_source_line. Drop
31804 (layout::print_leading_fixits): New method.
31805 (layout::print_any_fixits): Rename to...
31806 (layout::print_trailing_fixits): ...this, and make private.
31807 Don't print newline fixits.
31808 (diagnostic_show_locus): Move logic for printing one row into...
31809 (layout::print_line): ...this new function. Move the
31810 location_get_source_line call and error-handling from
31811 print_source_line to here. Call print_leading_fixits, and rename
31812 print_any_fixits to print_trailing_fixits.
31813 (selftest::test_fixit_insert_containing_newline): Update now that
31814 newlines are partially supported.
31815 (selftest::test_fixit_insert_containing_newline_2): New test.
31816 (selftest::test_fixit_replace_containing_newline): Update comments.
31817 (selftest::diagnostic_show_locus_c_tests): Call the new test.
31818 * edit-context.c (class added_line): New class.
31819 (class edited_line): Describe newline handling in comment.
31820 (edited_line::actually_edited_p): New method.
31821 (edited_line::print_content): Delete redundant decl.
31822 (edited_line::m_predecessors): New field.
31823 (edited_file::print_content): Call edited_line::print_content.
31824 (edited_file::print_diff): Update to support newlines.
31825 (edited_file::print_diff_hunk): Likewise.
31826 (edited_file::print_run_of_changed_lines): New function.
31827 (edited_file::print_diff_line): Convert to...
31828 (print_diff_line): ...this.
31829 (edited_file::get_effective_line_count): New function.
31830 (edited_line::edited_line): Initialize new field m_predecessors.
31831 (edited_line::~edited_line): Clean up m_predecessors.
31832 (edited_line::apply_fixit): Handle newlines.
31833 (edited_line::get_effective_line_count): New function.
31834 (edited_line::print_content): New function.
31835 (edited_line::print_diff_lines): New function.
31836 (selftest::test_applying_fixits_insert_containing_newline): New
31838 (selftest::test_applying_fixits_replace_containing_newline): New
31840 (selftest::insert_line): New function.
31841 (selftest::test_applying_fixits_multiple_lines): Add example of
31843 (selftest::edit_context_c_tests): Call the new tests.
31845 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31847 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
31848 parameter cand. Update dump information.
31849 (get_computation_cost): Update uses.
31851 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31853 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
31854 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
31855 (get_computation_at, rewrite_use_address): Update use of
31856 get_computation_aff.
31858 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31860 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
31861 (get_computation): Delete.
31862 (get_computation_cost): Implement like get_computation_cost_at.
31863 Use get_computation_at.
31864 (get_computation_cost_at): Delete.
31865 (rewrite_use_nonlinear_expr): Use get_computation_at.
31866 (rewrite_use_compare, remove_unused_ivs): Ditto.
31868 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31870 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
31872 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31874 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
31875 (ivopts_global_cost_for_size): Rename parameter and update uses.
31876 (iv_ca_recount_cost): Update uses.
31877 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
31878 candidates seperately in n_invs and n_cands.
31879 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
31881 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31883 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
31884 (find_inv_vars_cb): New.
31885 (find_depends): Renamed to ...
31886 (find_inv_vars): ... this.
31887 (add_candidate_1, force_var_cost): Call find_inv_vars.
31888 (split_address_cost, determine_group_iv_cost_cond): Ditto.
31890 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31892 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
31893 inv_vars. Add inv_exprs.
31894 (struct iv_cand): Rename depends_on to inv_vars.
31895 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
31896 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
31897 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
31898 (dump_cand): Dump inv_vars.
31899 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
31900 (record_invariant, find_depends, add_candidate_1): Ditto.
31901 (set_group_iv_cost, force_var_cost): Ditto.
31902 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
31903 (get_computation_cost_at, get_computation_cost): Ditto.
31904 (determine_group_iv_cost_generic): Ditto.
31905 (determine_group_iv_cost_address): Ditto.
31906 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
31907 (determine_group_iv_costs): Ditto.
31908 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
31909 (iv_ca_set_remove_invariants): Renamed to ...
31910 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
31911 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
31912 (iv_ca_set_add_invariants): Renamed to ...
31913 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
31914 (iv_ca_set_cp): Use iv_ca_set_add_invs.
31915 (iv_ca_has_deps): Support inv_vars and inv_exprs.
31916 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
31917 (create_new_ivs): Remove useless dump.
31919 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31921 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
31923 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
31924 (iv_ca_set_no_cp, create_new_iv): Ditto.
31926 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31928 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31930 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
31932 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
31933 function tree_check2.
31935 2017-05-02 Martin Liska <mliska@suse.cz>
31937 * doc/gcov.texi: Add missing preposition.
31938 * gcov.c (function_info::function_info): Properly fill up
31939 all member variables.
31941 2017-05-02 Tamar Christina <tamar.christina@arm.com>
31943 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
31945 2017-05-02 Tamar Christina <tamar.christina@arm.com>
31947 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
31949 2017-05-02 Martin Liska <mliska@suse.cz>
31952 * lto-streamer-in.c (lto_read_tree_1): Remove
31953 LTO_STREAMER_DEBUG.
31954 * lto-streamer.c (struct tree_hash_entry): Likewise.
31955 (struct tree_entry_hasher): Likewise.
31956 (tree_entry_hasher::hash): Likewise.
31957 (tree_entry_hasher::equal): Likewise.
31958 (lto_streamer_init): Likewise.
31959 (lto_orig_address_map): Likewise.
31960 (lto_orig_address_get): Likewise.
31961 (lto_orig_address_remove): Likewise.
31962 * lto-streamer.h: Likewise.
31963 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
31964 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
31966 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
31968 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
31969 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
31970 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
31971 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
31972 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
31973 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
31974 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
31975 (mm_maskz_sub_ss): New intrinsics.
31976 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31977 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31978 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
31979 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
31980 (__builtin_ia32_subss_mask_round): New builtins.
31981 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31982 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31983 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
31985 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
31986 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31988 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31991 2017-05-02 Martin Jambor <mjambor@suse.cz>
31993 PR tree-optimization/78687
31994 * tree-sra.c (access): New field parent.
31995 (process_subtree_disqualification): New function.
31996 (disqualify_candidate): Call it.
31997 (build_accesses_from_assign): Reset write flag if creating an
31999 (build_access_subtree): Fill in parent field and also prpagate
32000 down grp_write flag.
32001 (create_artificial_child_access): New parameter set_grp_write, set
32002 grp_write to its value.
32003 (propagate_subaccesses_across_link): Also propagate grp_write flag
32005 (propagate_all_subaccesses): Push the closest parent back to work
32006 queue if add_access_to_work_queue returned true.
32008 2017-05-02 Richard Biener <rguenther@suse.de>
32010 * common.opt (fstrict-overflow): Alias negative to fwrapv.
32011 * doc/invoke.texi (fstrict-overflow): Remove all traces of
32012 -fstrict-overflow documentation.
32013 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
32014 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
32015 flag_strict_overflow.
32016 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
32017 * lto-opts.c (lto_write_options): Do not stream it.
32018 * lto-wrapper.c (merge_and_complain): Do not handle it.
32019 * opts.c (default_options_table): Do not set -fstrict-overflow.
32020 (finish_options): Likewise do not clear it when sanitizing.
32021 * simplify-rtx.c (simplify_const_relational_operation): Do not
32022 test flag_strict_overflow.
32024 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32026 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
32027 using enabled attribute.
32028 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
32029 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
32030 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
32031 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
32032 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
32033 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
32034 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
32035 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
32036 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
32037 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
32039 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32041 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
32043 2017-05-02 Richard Biener <rguenther@suse.de>
32045 PR tree-optimization/80591
32047 2017-04-10 Richard Biener <rguenther@suse.de>
32049 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32052 2017-05-02 Richard Biener <rguenther@suse.de>
32054 PR tree-optimization/80549
32055 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
32056 (cleanup_tree_cfg_noloop): Create forwarders to known loop
32057 headers if they do not have a preheader.
32059 2017-05-02 Martin Liska <mliska@suse.cz>
32062 * common.opt: Fix typo.
32063 * doc/invoke.texi: Likewise.
32065 2017-05-01 Jan Beulich <jbeulich@suse.com>
32067 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
32068 swapping, add (x,x,m,x,n) alternative.
32070 2017-05-01 Nathan Sidwell <nathan@acm.org>
32072 * calls.c (combine_pending_stack_adjustment_and_call): Remove
32073 unnecessary unadjusted_alignment check.
32075 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32078 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
32080 * gimplify.c (gimplify_cilk_detach): New function.
32081 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
32082 * tree-core.h: Document EXPR_CILK_SPAWN.
32083 * tree.h (EXPR_CILK_SPAWN): Define.
32085 2017-05-01 David Malcolm <dmalcolm@redhat.com>
32087 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
32088 to use new fixit_hint representation, using the "replace" logic.
32089 (get_line_span_for_fixit_hint): Likewise.
32090 (layout::print_any_fixits): Likewise.
32091 (selftest::test_one_liner_many_fixits): Rename to...
32092 (selftest::test_one_liner_many_fixits_1): ...this, and update
32093 comment and expected output to reflect that the multiple fix-it
32094 hints are now consolidated into one insertion.
32095 (selftest::test_one_liner_many_fixits_2): New test.
32096 (selftest::test_diagnostic_show_locus_one_liner): Update for
32098 (selftest::test_fixit_consolidation): Update for fix-it API
32100 * diagnostic.c (print_parseable_fixits): Likewise.
32101 * edit-context.c (edited_line::m_line_events): Convert from
32102 auto_vec <line_event *> to auto_vec <line_event>.
32103 (class line_event): Convert from abstract base class to a concrete
32104 class, taking over the role of replace_event.
32105 (class insert_event): Delete.
32106 (class replace_event): Rename to class line_event. Convert to
32108 (edit_context::add_fixits): Reimplement.
32109 (edit_context::apply_insert): Delete.
32110 (edit_context::apply_replace): Rename to...
32111 (edit_context::apply_fixit): ...this. Convert to half-open range.
32112 (edited_file::apply_insert): Delete.
32113 (edited_file::apply_replace): Rename to...
32114 (edited_file::apply_fixit): ...this.
32115 (edited_line::~edited_line): Drop deletion of events.
32116 (edited_line::apply_insert): Delete.
32117 (edited_line::apply_replace): Rename to...
32118 (edited_line::apply_fixit): ...this. Convert to half-open range.
32119 Update for change to type of m_line_events.
32120 * edit-context.h (edit_context::apply_insert): Delete.
32121 (edit_context::apply_replace): Rename to...
32122 (edit_context::apply_fixit): ...this.
32124 2017-05-01 Martin Sebor <msebor@redhat.com>
32126 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
32129 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
32132 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
32133 __get_cpuid_max returns 0.
32134 (__get_cpuid_count): Ditto.
32136 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
32138 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
32139 replacement expression is another instance of one of its arguments.
32141 2017-05-01 Jakub Jelinek <jakub@redhat.com>
32144 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
32145 check for stack push/pop autoinc.
32146 * config/i386/i386.c (ix86_agi_dependent): Return false
32147 if the only reason why modified_in_p returned true is that
32148 addr is SP based and set_insn is a push or pop.
32150 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32152 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
32155 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32158 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
32159 (account_size_time): Use two predicates - exec_pred and
32161 (evaluate_conditions_for_known_args): Compute both clause and
32163 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
32164 (inline_summary_t::duplicate): Update.
32165 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
32167 (compute_inline_parameters): Likewise.
32168 (estimate_edge_size_and_time): Update caluclation of time.
32169 (estimate_node_size_and_time): Compute both time and nonspecialized
32171 (estimate_ipcp_clone_size_and_time): Update.
32172 (inline_merge_summary): Update.
32173 (do_estimate_edge_time): Update.
32174 (do_estimate_edge_size): Update.
32175 (do_estimate_edge_hints): Update.
32176 (inline_read_section, inline_write_summary): Stream both new predicates.
32177 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
32179 (compute_inlined_call_time): Cleanup.
32180 (big_speedup_p): Update.
32181 (edge_badness): Update.
32182 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
32183 (size_time_entry): Replace predicate by exec_predicate and
32184 nonconst_predicate.
32185 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
32186 (estimate_edge_time): Return also nonspec_time.
32187 (reset_edge_growth_cache): Update.
32189 2017-04-29 Jakub Jelinek <jakub@redhat.com>
32191 PR rtl-optimization/80491
32192 * ifcvt.c (noce_process_if_block): When looking for x setter
32193 with missing else_bb, don't check only the insn right before
32194 cond_earliest, but look for the last insn that x is modified in
32195 within the same bb.
32197 PR rtl-optimization/80491
32198 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
32200 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
32202 PR tree-optimization/80487
32203 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
32205 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32207 PR tree-optimization/79697
32208 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
32209 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
32210 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
32212 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
32213 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
32215 2017-04-28 Martin Sebor <msebor@redhat.com>
32217 PR tree-optimization/80523
32218 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
32219 (init_target_to_host_charmap, target_to_host, target_strtol10): New
32221 (maybe_warn, format_directive, parse_directive): Use new functions.
32222 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
32224 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
32226 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
32228 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
32230 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
32231 target_header_dir): Set correctly.
32232 * configure: Regenerated.
32233 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32234 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32235 instead of SYSTEM_HEADER_DIR.
32237 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
32239 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
32240 (estimate_local_effects): Likewise.
32241 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
32242 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
32243 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
32244 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
32245 do_estimate_edge_time, estimate_edge_time): Likewise.
32246 * ipa-inline-analysis.c (estimate_node_size_and_time,
32247 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
32248 (estimate_time_after_inlining): Remove.
32250 2017-04-28 Martin Liska <mliska@suse.cz>
32252 * doc/gcov.texi: Enhance documentation of gcov.
32254 2017-04-28 Martin Liska <mliska@suse.cz>
32256 * doc/gcov.texi: Sort options in alphabetic order.
32257 * doc/gcov-dump.texi: Likewise.
32258 * doc/gcov-tool.texi: Likewise.
32259 * gcov.c (print_usage): Likewise.
32260 * gcov-dump.c (print_usage): Likewise.
32261 * gcov-tool.c (print_merge_usage_message): Likewise.
32262 (print_rewrite_usage_message): Likewise.
32263 (print_overlap_usage_message): Likewise.
32265 2017-04-28 Martin Liska <mliska@suse.cz>
32267 PR gcov-profile/53915
32268 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
32270 2017-04-28 Martin Liska <mliska@suse.cz>
32272 PR gcov-profile/79891
32273 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
32274 is marked by compiler as living on a line.
32275 (get_cycles_count): Remove usage of the union.
32276 (output_intermediate_file): Likewise.
32277 (find_source): Fix GNU coding style.
32278 (accumulate_line_counts): Remove old non-all block mode.
32279 (output_lines): Remove usage of the union.
32280 * profile.c (output_location): Include all BBs, even if
32281 belonging to a same line (and file) as a previous BB.
32283 2017-04-28 Martin Liska <mliska@suse.cz>
32285 * gcov.c (process_args): Handle new argument 'w'.
32286 (read_graph_file): Assign ID to BBs.
32287 (output_branch_count): Display BB # if verbose flag is set.
32288 (output_lines): Likewise for arcs.
32289 (print_usage): Add '--verbose' option help.
32290 * doc/gcov.texi: Document --verbose (-w) option.
32292 2017-04-28 Martin Liska <mliska@suse.cz>
32294 * gcov.c (struct block_location_info): New struct.
32295 (process_file): Fill up the new structure.
32296 (read_graph_file): Replace usage of encoding by the newly added
32298 (add_line_counts): Likewise.
32299 (accumulate_line_counts): Remove usage of the union.
32300 (function_info::function_info): New function.
32301 (function_info::~function_info): Likewise.
32302 (process_file): Call delete instead of release_function.
32303 (release_function): Release the function.
32304 (release_structures): Call delete instead of release_function.
32305 (solve_flow_graph): Replace usage of num_blocks.
32306 (find_exception_blocks): Likewise.
32307 (output_lines): Fix GNU coding style.
32309 2017-04-28 Martin Liska <mliska@suse.cz>
32312 * coverage.c (coverage_remove_note_file): New function.
32313 * coverage.h: Declare the function.
32314 * toplev.c (finalize): Clean if an error has been seen.
32316 2017-04-28 Martin Liska <mliska@suse.cz>
32318 PR gcov-profile/80031
32319 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
32320 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
32321 * gcov.c (read_graph_file): Read just number of blocks.
32322 * profile.c (branch_prob): Do not stream 0 flags per a basic
32325 2017-04-28 Martin Liska <mliska@suse.cz>
32327 * gcov-dump.c (tag_*): Add new argument to declarations.
32328 (dump_gcov_file): Likewise.
32329 (tag_blocks): Add and use new argument depth.
32330 (tag_arcs): Likewise.
32331 (tag_lines): Likewise.
32332 (tag_counters): Likewise.
32333 (tag_summary): Likewise.
32334 (dump_working_sets): Use depth to do a proper indentation.
32336 2017-04-28 Jakub Jelinek <jakub@redhat.com>
32339 * cgraph.h (symtab_node::debug_symtab): No longer inline.
32340 * symtab.c (symtab_node::debug_symtab): Move definition here.
32342 2017-04-28 Richard Biener <rguenther@suse.de>
32344 * lto-streamer.h (LTO_major_version): Bump to 7.
32346 2017-04-28 Richard Biener <rguenther@suse.de>
32348 * tree-vrp.c (assert_info): New struct.
32349 (add_assert_info): New helper.
32350 (register_edge_assert_for_2): Refactor to add asserts to a vector
32352 (register_edge_assert_for_1): Likewise.
32353 (register_edge_assert_for): Likewise.
32354 (finish_register_edge_assert_for): New helper actually registering
32355 asserts where live on edge.
32356 (find_conditional_asserts): Adjust.
32357 (find_switch_asserts): Likewise.
32358 (evrp_dom_walker::try_find_new_range): Generalize.
32359 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
32361 2017-04-27 Marek Polacek <polacek@redhat.com>
32364 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
32365 arg10 and arg11 to itype.
32367 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
32369 * doc/extend.texi (Object Size Checking): Improve grammar.
32371 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
32374 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
32375 that the logic for permitting reciprocal estimates matches that
32378 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32381 * tree.c (type_cache_hasher::equal): Only compare
32382 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
32383 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
32384 non-aggregate element types.
32385 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
32386 about the flag on ARRAY_TYPEs in the comment, formatting fix.
32388 2017-04-27 Richard Biener <rguenther@suse.de>
32390 PR middle-end/80533
32391 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
32392 stripping ARRAY_REFs from MEM_EXPR make sure we're not
32393 keeping a reference to a trailing array.
32395 2017-04-27 Richard Biener <rguenther@suse.de>
32397 PR middle-end/80539
32398 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
32399 being in loop-closed SSA form conservatively.
32400 (chrec_fold_multiply_poly_poly): Likewise.
32402 2017-04-27 Tamar Christina <tamar.christina@arm.com>
32404 PR middle-end/79665
32405 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
32406 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
32408 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32411 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
32412 (aarch64_function_arg_alignment): Return unsigned int again, but still
32413 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
32414 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32415 Don't emit -Wpsabi note.
32416 (aarch64_function_arg_boundary): Likewise.
32417 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32420 2017-04-26 Nathan Sidwell <nathan@acm.org>
32422 * tree.h (crc32_unsigned_n): Declare.
32423 (crc32_unsigned, crc32_unsigned): Make inline.
32424 * tree.c (crc32_unsigned_bits): Replace with ...
32425 (crc32_unsigned_n): ... this.
32426 (crc32_unsigned, crc32_byte): Remove.
32427 (crc32_string): Remove unnecessary braces.
32429 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32431 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
32432 * ipa-inline-analysis.c (MAX_TIME): Remove.
32433 (account_size_time): Use sreal for time.
32434 (dump_inline_summary): Update.
32435 (estimate_function_body_sizes): Update.
32436 (estimate_edge_size_and_time): Update.
32437 (estimate_calls_size_and_time): Update.
32438 (estimate_node_size_and_time): Update.
32439 (inline_merge_summary): Update.
32440 (inline_update_overall_summary): Update.
32441 (estimate_time_after_inlining): Update.
32442 (inline_read_section): Update.
32443 (inline_write_summary): Update.
32444 * ipa-inline.c (compute_uninlined_call_time): Update.
32445 (compute_inlined_call_time): Update.
32446 (recursive_inlining): Update.
32447 (inline_small_functions): Update.
32448 (dump_overall_stats): Update.
32449 * ipa-inline.h: Include sreal.h.
32450 (size_time_entry): Turn time to sreal.
32451 (inline_summary): Turn self_time nad time to sreal.
32453 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32455 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
32457 (sreal::stream_out, sreal::stream_in): New.
32458 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
32460 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32462 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
32465 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
32468 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
32470 (dimode_scalar_chain::compute_convert_gain): Ditto.
32471 (dimode_scalar_chain::make_vector_copies): Ditto.
32472 (dimode_scalar_chain::convert_reg): Ditto.
32473 (dimode_scalar_chain::convert_insn): Ditto.
32474 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
32475 (VI248_AVX512BW_1): New mode iterator.
32476 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
32477 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
32480 2017-04-25 Martin Sebor <msebor@redhat.com>
32482 PR tree-optimization/80497
32483 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
32484 constants are representable in HOST_WIDE_INT.
32485 (parse_directive): Ditto.
32487 2017-04-25 Martin Sebor <msebor@redhat.com>
32490 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
32491 (new_zero_array): Adjust signature.
32492 (dom_info::dom_init): Used unsigned rather that size_t.
32493 (dom_info::dom_info): Same.
32495 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32496 Jakub Jelinek <jakub@redhat.com>
32499 * config/arm/arm.c: Include gimple.h.
32500 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32501 returns negative, increment ncrn only if it returned positive.
32502 (arm_needs_doubleword_align): Return int instead of bool,
32503 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
32504 members, but if there is any such non-FIELD_DECL
32505 > PARM_BOUNDARY aligned decl, return -1 instead of false.
32506 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32507 returns negative, increment nregs only if it returned positive.
32508 (arm_setup_incoming_varargs): Likewise.
32509 (arm_function_arg_boundary): Emit -Wpsabi note if
32510 arm_needs_doubleword_align returns negative, return
32511 DOUBLEWORD_ALIGNMENT only if it returned positive.
32513 2017-04-25 Marek Polacek <polacek@redhat.com>
32516 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32517 first argument to type.
32519 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
32522 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
32523 type checks to test for compatibility instead of equality.
32525 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32526 Jakub Jelinek <jakub@redhat.com>
32529 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
32531 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
32532 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
32533 the alignment computation, but return their maximum in warn_alignment.
32534 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32535 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
32537 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
32538 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32541 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32543 * config/arc/simdext.md (dmpyh): Fix typo.
32545 2017-04-25 Richard Biener <rguenther@suse.de>
32547 PR tree-optimization/80492
32548 * alias.c (compare_base_decls): Handle registers with asm
32549 specification conservatively.
32550 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32551 compare_base_decls returning dont-know properly.
32553 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32555 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
32556 (legitimate_offset_address_p): New function.
32557 (arc_legitimate_address_p): Use above function.
32559 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32561 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
32563 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32565 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
32566 ACCH registers whenever they are available.
32568 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32570 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
32571 double regs fix when not used.
32573 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32575 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
32577 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
32578 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
32580 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32582 * config/arc/arc.c (arc_output_addsi): Check for h-register class
32583 when emitting short ADD instructions.
32585 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32587 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
32589 (cmpsi_cc_c_insn): Likewise.
32590 (cbranchsi4_scratch): Compute proper instruction length using
32591 compact_hreg_operand.
32592 * config/arc/predicates.md (compact_hreg_operand): New predicate.
32594 2017-04-25 Richard Biener <rguenther@suse.de>
32596 PR middle-end/80509
32597 * passes.c (pass_manager::pass_manager): Initialize
32598 m_name_to_pass_map.
32600 2017-04-25 Richard Biener <rguenther@suse.de>
32602 PR tree-optimization/79201
32603 * tree-ssa-sink.c (statement_sink_location): Handle calls.
32605 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32608 * config/s390/vector.md: Split MEM->GPR vector moves for
32609 non-s_operand addresses.
32611 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32614 * config/s390/predicates.md (reload_const_wide_int_operand): New
32616 * config/s390/s390.md ("movti"): Remove d/P alternative.
32617 ("movti_bigconst"): New pattern definition.
32619 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32622 * s390-protos.h (s390_expand_cs_hqi): Removed.
32623 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
32624 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
32625 modes as well as CCZ1mode and CCZmode.
32626 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
32627 signature of s390_emit_compare_and_swap.
32628 (s390_expand_cs_hqi): Likewise, make static.
32629 (s390_expand_cs_tdsi): Generate an explicit compare before trying
32630 compare-and-swap, in some cases.
32631 (s390_expand_cs): Wrapper function.
32632 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
32634 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
32635 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
32636 patterns for small and large integers. Forbid symref memory operands.
32637 Move expander to s390.c. Require cc register.
32638 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
32639 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
32640 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
32641 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
32642 symref memory operands. Remove CC mode and call s390_match_ccmode
32644 ("atomic_exchange<mode>"): Allow and implement all integer modes.
32646 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32648 * config/s390/s390.md (define_peephole2): New peephole to help
32649 combining the load-and-test pattern with volatile memory.
32651 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32653 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
32654 with CCZmode for TARGET_Z196.
32656 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32658 PR rtl-optimization/80501
32659 * combine.c (make_compound_operation_int): Set subreg_code to SET
32660 even for AND with mask of the sign bit of mode.
32662 PR rtl-optimization/80500
32663 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
32664 sum's initial value.
32666 2017-04-25 Julian Brown <julian@codesourcery.com>
32667 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32669 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
32671 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
32673 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
32675 2017-04-25 Julian Brown <julian@codesourcery.com>
32676 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32678 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
32679 (thunderx2t99_sha): New Reservation.
32681 2017-04-25 Julian Brown <julian@codesourcery.com>
32682 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32684 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
32685 type for 1-element load.
32687 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
32689 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
32691 2017-04-24 Martin Jambor <mjambor@suse.cz>
32693 PR tree-optimization/80293
32694 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
32695 char arrays not totally scalarizable if it is false.
32696 (analyze_all_variable_accesses): Pass correct value in the new
32697 parameter. Add a statistics counter.
32699 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
32701 PR middle-end/79931
32702 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
32704 2017-04-24 Richard Biener <rguenther@suse.de>
32706 PR tree-optimization/80494
32707 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
32708 out for complex types.
32710 2017-04-24 Richard Biener <rguenther@suse.de>
32712 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
32713 * tree-ssa-sccvn.c (print_scc): Print SCC size.
32714 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
32715 (DFS): Adjust and never fail.
32716 (sccvn_dom_walker::fail): Remove.
32717 (sccvn_dom_walker::before_dom_children): Adjust.
32718 (run_scc_vn): Likewise and never fail.
32719 * tree-ssa-pre.c (pass_pre::execute): Adjust.
32720 (pass_fre::execute): Likewise.
32722 2017-04-24 Richard Biener <rguenther@suse.de>
32724 PR tree-optimization/79725
32725 * tree-ssa-sink.c (statement_sink_location): Return whether
32726 failure reason was zero uses. Move that check later.
32727 (sink_code_in_bb): Deal with zero uses by removing the stmt
32730 2017-04-24 Richard Biener <rguenther@suse.de>
32733 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
32734 pointer-based references.
32736 2017-04-24 Richard Biener <rguenther@suse.de>
32739 * pass_manager.h (pass_manager::operator new): Remove.
32740 (pass_manager::operator delete): Likewise.
32741 * passes.c (pass_manager::operator new): Remove.
32742 (pass_manager::operator delete): Likewise.
32743 (pass_manager::pass_manager): Zero individual pass members.
32745 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
32748 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
32749 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
32750 Check "XEXP (src, 1)" operand here.
32751 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
32752 Check "XEXP (src, 1)" operand here.
32753 (dimode_scalar_chain::make_vector_copies): Detect count register
32754 of a shift instruction. Zero extend count register from QImode
32755 to DImode to satisfy vector shift pattern count operand predicate.
32756 Substitute vector shift count operand with a DImode copy.
32757 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
32760 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
32762 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
32763 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32764 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
32765 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32766 (UNSPEC_NOREX_MEM): Remove definition.
32768 2017-04-21 Richard Biener <rguenther@suse.de>
32770 PR tree-optimization/79547
32771 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
32772 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
32773 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
32774 without any constraints.
32776 2017-04-21 Richard Biener <rguenther@suse.de>
32778 PR tree-optimization/78847
32779 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
32781 2017-04-21 Richard Biener <rguenther@suse.de>
32783 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
32784 (build_distinct_type_copy): Likewise.
32785 (build_variant_type_copy): Likewise.
32786 * tree.c (build_qualified_type): Pass down mem-stat info.
32787 (build_distinct_type_copy): Likewise.
32788 (build_variant_type_copy): Likewise.
32790 2017-04-21 Richard Biener <rguenther@suse.de>
32792 PR tree-optimization/80237
32793 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
32795 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
32796 for a simplified result.
32798 2016-04-21 Richard Biener <rguenther@suse.de>
32800 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
32801 sth as strict as a simple_iv but a chrec without symbols and an
32802 operand defined in the loop we are peeling (and not some subloop).
32803 (propagate_constants_for_unrolling): Propagate all constants.
32805 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32808 * config/i386/i386.c (print_reg): Remove assert for disalowed
32809 regno values, call output_operand_lossage instead.
32811 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32814 * config/i386/constraints.md (Yc): New register constraint.
32815 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
32816 Use Yc constraint for alternative 2 of operand 0. Remove
32817 preferred_for_speed attribute.
32819 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32821 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
32822 lastprivate clauses in SIMT case.
32824 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
32826 * doc/invoke.texi (-Wextra-semi): Document new warning option.
32828 2017-04-20 Richard Biener <rguenther@suse.de>
32830 PR tree-optimization/57796
32831 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
32832 as N scalar stores.
32833 (vect_model_load_cost): Cost gathers as N scalar loads.
32835 2017-04-20 Richard Biener <rguenther@suse.de>
32837 * ggc-page.c (ggc_allocated_p): Rename to ...
32838 (safe_lookup_page_table_entry): ... this and return the lookup
32840 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
32842 2017-04-20 Richard Biener <rguenther@suse.de>
32844 PR tree-optimization/80453
32845 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
32846 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
32847 from the conditions.
32848 (vn_phi_eq): Pass them down.
32849 (vn_phi_lookup): Record them.
32850 (vn_phi_insert): Likewise.
32852 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
32854 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
32855 uninitialized variable warning to avoid buffer overrun.
32857 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32860 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
32861 is suppressed for '{ 0 }' in C.
32863 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32865 * BASE-VER: Set to 8.0.0.
32867 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
32869 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
32870 priority .init_array and .fini_array section with SECTION_NOTYPE
32873 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32875 PR middle-end/80423
32876 * tree.h (build_array_type): Add typeless_storage default argument.
32877 * tree.c (type_cache_hasher::equal): Also compare
32878 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
32879 (build_array_type): Add typeless_storage argument, set
32880 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
32882 (build_nonshared_array_type): Adjust build_array_type_1 caller.
32883 (build_array_type): Likewise. Add typeless_storage argument.
32885 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
32886 Jakub Jelinek <jakub@redhat.com>
32888 PR tree-optimization/80426
32889 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
32890 operation on symbolic operands, also compute the overflow for the
32891 invariant part when the operation degenerates into a negation.
32893 2017-04-19 Jakub Jelinek <jakub@redhat.com>
32896 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
32897 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
32900 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
32902 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
32905 * config/avr/avr.c (tree.h): Include it.
32906 (cgraph.h): Include it.
32907 (avr_encode_section_info): Don't warn for uninitialized progmem
32908 variable if it's just an alias.
32910 2017-04-19 Richard Biener <rguenther@suse.de>
32913 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
32914 when needed by AutoPGO.
32916 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
32919 * doc/lto.texi: Remove an extra 'that'.
32921 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
32923 PR rtl-optimization/80429
32924 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
32925 are only used in debug insns.
32927 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
32928 Vladimir Makarov <vmakarov@redhat.com>
32930 * config/sparc/predicates.md (input_operand): Add comment. Return
32931 true for any memory operand when LRA is in progress.
32932 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
32934 2017-04-18 Jeff Law <law@redhat.com>
32937 * mips.md ({return,simple_return}_internal): Do not overwrite
32940 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32942 PR tree-optimization/80443
32943 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
32944 instead of adding 1, subtract -1 and similarly instead of subtracting
32947 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
32949 PR rtl-optimization/80357
32950 * haifa-sched.c (tmp_bitmap): New variable.
32951 (model_recompute): Handle duplicate use records.
32952 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
32953 (free_global_sched_pressure_data): Free it.
32955 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
32958 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
32959 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32960 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32961 instead of SYSTEM_HEADER_DIR.
32963 2017-04-18 Jeff Law <law@redhat.com>
32965 PR middle-end/80422
32966 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
32967 predecessors after walking up the insn chain.
32969 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32972 * dwarf2out.c (modified_type_die): Try harder not to emit internal
32973 sizetype type into debug info.
32975 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32978 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
32979 unneeded test for TARGET_UPPER_REGS_SF.
32980 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
32982 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32985 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
32986 instead of gsi_after_labels.
32988 2017-04-18 Jeff Law <law@redhat.com>
32990 * regcprop.c (maybe_mode_change): Avoid creating copies of the
32994 2017-04-13 Jeff Law <law@redhat.com>
32995 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
32996 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
32998 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
33001 * config/avr/avr.c (intl.h): Include it.
33002 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
33004 2017-04-18 Martin Liska <mliska@suse.cz>
33006 PR gcov-profile/78783
33007 * gcov-tool.c (gcov_output_files): Validate that destination
33008 file is either removed by the tool or by a user.
33010 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33011 Guy Benyei <guybe@mellanox.com>
33013 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
33014 block, and do not negate it, the stored id is already negative.
33016 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33018 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
33020 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33023 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
33024 masks of options that should be turned off if the VSX vector
33025 options are turned off.
33026 (OTHER_P8_VECTOR_MASKS): Likewise.
33027 (OTHER_VSX_VECTOR_MASKS): Likewise.
33028 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
33029 rs6000_disable_incompatible_switches to validate no type switches
33031 (rs6000_incompatible_switch): New function to disallow turning on
33032 other vector options if -mno-vsx, -mno-power8-vector, or
33033 -mno-power9-vector are specified.
33035 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33037 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
33039 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33041 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
33042 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
33043 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
33044 (ARG_POINTER_CFA_OFFSET): Likewise.
33046 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33048 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
33049 conditions to take advantage of various optimizations.
33051 2017-04-13 Jeff Law <law@redhat.com>
33053 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33054 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33055 (zero_extendsidi2_dext): Likewise.
33057 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33060 * fold-const.c (fold_ternary_loc): Revert
33061 use op0 instead of fold_convert_loc (loc, type, arg0) part of
33064 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
33066 PR rtl-optimization/80343
33067 * lra-remat.c (update_scratch_ops): Assign original hard reg to
33068 new scratch pseudo.
33070 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
33073 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
33074 to ubsan_encode_value.
33076 2017-04-13 Jeff Law <law@redhat.com>
33078 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
33079 appearing in DEBUG_INSNs.
33081 2017-04-13 Martin Liska <mliska@suse.cz>
33083 PR gcov-profile/80413
33084 * gcov-io.c (gcov_write_string): Copy to buffer just when
33085 allocated size is greater than zero.
33087 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33090 * dwarf2out.c (decls_for_scope): Ignore declarations of
33091 current_function_decl in BLOCK_NONLOCALIZED_VARS.
33093 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
33096 * ipa-visibility.c (non_local_p): Fix typos.
33097 (localize_node): When localizing symbol in same comdat group,
33098 dissolve the group only when we know external symbols are going
33100 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
33102 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33104 PR tree-optimization/79390
33105 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
33106 order does not result in usable sequence, retry with reversed operand
33112 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
33113 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
33114 op0 instead of fold_convert_loc (loc, type, arg0).
33116 2017-04-12 Jeff Law <law@redhat.com>
33118 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
33119 has a delay slot in the generated code.
33121 * config/cris/cris.md (cris_preferred_reload_class): Return
33122 GENNONACR_REGS rather than GENERAL_REGS.
33124 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33127 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
33128 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
33129 signedness of the result type.
33131 2017-04-12 Richard Biener <rguenther@suse.de>
33132 Jeff Law <law@redhat.com>
33134 PR tree-optimization/80359
33135 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
33136 trim stores to TARGET_MEM_REFs.
33138 2017-04-12 Richard Biener <rguenther@suse.de>
33140 PR tree-optimization/79390
33141 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
33142 threading case even more.
33144 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
33147 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
33148 for quad_address_p for TImode, instead of just not indexed_address.
33150 2017-04-12 Richard Biener <rguenther@suse.de>
33151 Bernd Edlinger <bernd.edlinger@hotmail.de>
33153 PR middle-end/79671
33154 * alias.c (component_uses_parent_alias_set_from): Handle
33155 TYPE_TYPELESS_STORAGE.
33156 (get_alias_set): Likewise.
33157 * tree-core.h (tree_type_common): Add typeless_storage flag.
33158 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
33159 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
33160 for types containing members with TYPE_TYPELESS_STORAGE.
33161 (place_field): Likewise.
33162 (layout_type): Likewise for ARRAY_TYPE.
33163 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
33164 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
33165 TYPE_TYPELESS_STORAGE.
33166 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
33168 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33171 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33172 first argument to type.
33174 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33178 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
33179 CONST0_RTX (mode) rather than const0_rtx where appropriate.
33180 (rs6000_expand_binop_builtin): Likewise.
33181 (rs6000_expand_ternop_builtin): Likewise; also add missing
33182 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
33183 vshasigma built-ins.
33184 * doc/extend.texi: Document that vec_xxpermdi's third argument
33185 must be a constant.
33187 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
33189 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
33190 Use shift_const cost parameter when calculating gain of STV shifts.
33192 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
33194 PR rtl-optimization/70478
33195 * lra-constraints.c (process_alt_operands): Check memory for
33196 disfavoring memory insn operand.
33198 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33200 PR middle-end/80100
33201 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
33202 left shift in unsigned HOST_WIDE_INT type.
33204 PR rtl-optimization/80385
33205 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
33206 (not (neg X)) into (plus X -1) for complex or non-integral modes.
33209 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
33210 if they have any depend clauses.
33212 2017-04-11 Martin Liska <mliska@suse.cz>
33215 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
33216 * ipa-split.c (split_function): Create a local comdat symbol
33217 if caller is in a comdat group.
33219 2017-04-11 Martin Liska <mliska@suse.cz>
33222 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
33225 2017-04-11 Martin Sebor <msebor@redhat.com>
33227 PR middle-end/80364
33228 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
33229 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
33231 (directive::set_width, directive::set_precision, format_character):
33233 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
33236 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
33239 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
33240 conflict, set target->arch_name instead of target->cpu_name.
33242 2017-04-11 Richard Biener <rguenther@suse.de>
33244 PR tree-optimization/80374
33245 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
33246 build_zero_cst, remove fold_convertible_p check again.
33248 2017-04-11 Martin Liska <mliska@suse.cz>
33251 * ubsan.c (instrument_object_size): Do not instrument register
33254 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33257 * config/i386/i386-builtin-types.def
33258 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
33259 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
33260 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
33261 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
33262 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
33263 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
33264 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
33265 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
33266 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
33267 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
33268 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
33269 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
33270 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
33271 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
33272 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
33273 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
33274 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
33275 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
33276 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
33277 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
33278 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
33279 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
33280 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
33281 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
33282 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
33283 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
33284 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
33285 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
33286 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
33287 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
33288 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
33289 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
33290 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
33291 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
33292 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
33293 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
33294 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
33295 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
33296 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
33297 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
33298 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
33299 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
33300 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
33301 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
33302 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
33303 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
33305 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
33306 flag to second_arg_count, handle 4 argument function type _COUNT
33307 aliases, handle second_arg_count on second argument rather than last.
33309 2017-04-10 Jeff Law <law@redhat.com>
33311 PR tree-optimization/80374
33312 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
33313 record anything if we can not convert integer_zero_node to the
33316 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33319 * config/rs6000/rs6000.c (rs6000_option_override_internal):
33320 Enhance special handling given to the TARGET_P9_MINMAX option in
33321 relation to certain other options.
33323 2017-04-10 Bin Cheng <bin.cheng@arm.com>
33325 PR tree-optimization/80153
33326 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
33327 remove POINTER_PLUS_EXPR's base part directly, rather than through
33330 2017-04-10 Richard Biener <rguenther@suse.de>
33331 Bin Cheng <bin.cheng@arm.com>
33333 PR tree-optimization/80153
33334 * tree-affine.c (aff_combination_to_tree): Get base pointer from
33335 the first element of pointer type aff_tree. Build result expr in
33337 (add_elt_to_tree): Convert to type unconditionally. Remove other
33338 fold_convert calls.
33339 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
33340 (rewrite_use_nonlinear_expr): Check invariant using iv information.
33342 2017-04-10 Richard Biener <rguenther@suse.de>
33344 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33347 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
33349 PR rtl-optimization/70478
33350 * lra-constraints.c (curr_small_class_check): New.
33351 (update_and_check_small_class_inputs): New.
33352 (process_alt_operands): Update curr_small_class_check. Disfavor
33353 alternative insn memory operands. Check available regs for small
33356 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33359 * config/mips/mips.opt (-mvirt): Update description.
33360 * doc/invoke.texi (-mvirt): Likewise.
33362 2017-04-10 Richard Biener <rguenther@suse.de>
33364 PR middle-end/80362
33365 * fold-const.c (fold_binary_loc): Look at unstripped ops when
33366 looking for NEGATE_EXPR in -A / -B to A / B folding.
33368 2017-04-10 Martin Liska <mliska@suse.cz>
33370 PR gcov-profile/80224
33371 * gcov.c (print_usage): Fix usage string.
33372 (get_gcov_intermediate_filename): Remove.
33373 (output_gcov_file): Use both for normal and intermediate format.
33374 (generate_results): Do not initialize special file for
33375 intermediate format.
33377 2017-04-10 Richard Biener <rguenther@suse.de>
33379 PR tree-optimization/80304
33380 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
33383 2017-04-10 Nathan Sidwell <nathan@acm.org>
33386 * config/rs6000/rs6000.c (rs6000_vector_type): New.
33387 (rs6000_init_builtins): Use it.
33389 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33391 * config/arm/arm.md (<mrc>): Add mode to SET source.
33392 (<mrrc>): Likewise.
33394 2017-04-10 Richard Biener <rguenther@suse.de>
33396 PR middle-end/80344
33397 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
33399 2017-04-10 Jakub Jelinek <jakub@redhat.com>
33402 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
33403 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
33404 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
33405 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
33406 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
33407 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
33408 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
33409 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
33410 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
33411 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
33412 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
33413 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
33414 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
33415 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
33416 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
33417 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
33418 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
33419 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
33420 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
33421 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
33422 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
33423 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
33424 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
33426 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
33428 PR rtl-optimization/70478
33429 * lra-constraints.c: Reverse the last patch.
33431 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
33433 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
33434 Add comment for WCHAR_T.
33436 2017-04-08 Martin Liska <mliska@suse.cz>
33439 2017-04-07 Martin Liska <mliska@suse.cz>
33442 * ipa-split.c (split_function): Add function part to a same comdat
33445 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33448 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
33450 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
33452 * rs6000/rs6000.c (vec_load_pendulum): Rename...
33453 (vec_pairing): ...to this.
33454 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
33455 (rs6000_sched_init): Adjust for name change.
33456 (struct rs6000_sched_context): Likewise.
33457 (rs6000_init_sched_context): Likewise.
33458 (rs6000_set_sched_context): Likewise.
33460 2017-04-07 Jakub Jelinek <jakub@redhat.com>
33466 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
33468 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
33469 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
33470 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
33472 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
33474 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
33476 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33478 PR rtl-optimization/70703
33479 * ira-color.c (update_conflict_hard_regno_costs): Use
33480 int64_t instead of HOST_WIDE_INT.
33482 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33484 PR rtl-optimization/70478
33485 * lra-constraints.c (process_alt_operands): Disfavor alternative
33486 insn memory operands.
33488 2017-04-07 Jeff Law <law@redhat.com>
33490 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
33491 CALL and NOTE_INSN_CALL_ARG_LOCATION.
33493 2017-04-07 Martin Liska <mliska@suse.cz>
33496 * config/aarch64/aarch64.c (aarch64_process_target_attr):
33497 Show error message instead of an ICE.
33499 2017-04-07 Martin Liska <mliska@suse.cz>
33502 * ipa-split.c (split_function): Add function part to a same comdat
33505 2017-04-07 Richard Biener <rguenther@suse.de>
33507 PR middle-end/80341
33508 * tree.c (get_unwidened): Also handle ! for_type case for
33510 * convert.c (do_narrow): Split out from ...
33511 (convert_to_integer_1): ... here. Do not pass final truncation
33512 type to get_unwidened for TRUNC_DIV_EXPR.
33514 2017-04-07 Richard Biener <rguenther@suse.de>
33516 * tree-affine.c (wide_int_ext_for_comb): Take type rather
33518 (aff_combination_const): Adjust.
33519 (aff_combination_scale): Likewise.
33520 (aff_combination_add_elt): Likewise.
33521 (aff_combination_add_cst): Likewise.
33522 (aff_combination_convert): Likewise.
33523 (add_elt_to_tree): Likewise. Remove unused argument.
33524 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
33526 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
33528 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
33530 * config/arm/arm.c (arm_default_short_enums): Use
33531 ARM_DEFAULT_SHORT_ENUMS.
33532 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
33534 2017-04-06 Jakub Jelinek <jakub@redhat.com>
33537 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
33538 members with redundant out-of-class redeclaration.
33540 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33543 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
33544 * config/i386/i386.md (*zero_extendsidi2):
33545 Add (?*x,*x) and (?*v,*v) alternatives.
33547 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33550 * config/i386/i386.c (ix86_expand_builtin)
33551 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
33552 mode from insn data. Convert operands to insn operand mode.
33553 Copy operands that don't satisfy insn predicate to a register.
33555 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
33557 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
33560 2017-04-06 Richard Biener <rguenther@suse.de>
33562 PR tree-optimization/80334
33563 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
33564 preserve alignment of accesses.
33566 2017-04-06 Richard Biener <rguenther@suse.de>
33568 PR tree-optimization/80262
33569 * tree-sra.c (build_ref_for_offset): Preserve address-space
33571 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
33572 Drop useless address-space information on MEM_REF offsets.
33574 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
33576 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
33578 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33580 PR rtl-optimization/70703
33581 * ira-color.c (update_conflict_hard_regno_costs): Use
33582 HOST_WIDE_INT instead of long.
33584 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
33587 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
33588 not defined for x86_64 target. Add -mmmx target option when __SSE2__
33590 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
33591 for x86_64 target. Handle -m3dnowa option.
33593 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33595 PR rtl-optimization/70703
33596 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
33597 (update_conflict_hard_regno_costs): Use long instead of unsigned
33598 arithmetic for cost calculation.
33600 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33601 Bernd Edlinger <bernd.edlinger@hotmail.de>
33604 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
33607 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
33610 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
33611 ptr_mode with Pmode throughout.
33612 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
33613 into probe_stack_range and use DImode.
33615 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
33618 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
33619 call_eh_return is true.
33621 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33623 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
33624 Initialize last_match_fntype_index.
33626 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33629 * tree-nvr.c: Include internal-fn.h.
33630 (pass_return_slot::execute): Ignore internal calls without
33633 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33634 Richard Biener <rguenther@suse.de>
33637 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
33638 captures used multiple times, except for the last use.
33639 * generic-match-head.c: Include gimplify.h.
33641 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33643 PR tree-optimization/79390
33644 * target.h (struct noce_if_info): Declare.
33645 * targhooks.h (default_noce_conversion_profitable_p): Declare.
33646 * target.def (noce_conversion_profitable_p): New target hook.
33647 * ifcvt.h (struct noce_if_info): New type, moved from ...
33648 * ifcvt.c (struct noce_if_info): ... here.
33649 (noce_conversion_profitable_p): Renamed to ...
33650 (default_noce_conversion_profitable_p): ... this. No longer
33652 (noce_try_store_flag_constants, noce_try_addcc,
33653 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
33654 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
33655 instead of noce_conversion_profitable_p.
33656 * config/i386/i386.c: Include ifcvt.h.
33657 (ix86_option_override_internal): Don't override
33658 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
33659 (ix86_noce_conversion_profitable_p): New function.
33660 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
33661 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
33662 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
33663 * doc/tm.texi: Regenerated.
33665 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33667 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
33670 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
33673 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
33674 instructions for small multiply cores.
33676 2017-04-04 Jeff Law <law@redhat.com>
33678 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
33680 (mips_expand_vec_perm_const): Initialize elements in orig_perm
33681 that are not set by the loop over the elements.
33683 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33686 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
33687 int mode, convert_modes it to mode as unsigned, otherwise use
33688 lowpart_subreg to mode rather than SImode.
33689 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
33690 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
33691 Use DImode instead of SImode for the shift count operand.
33692 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
33695 2017-04-04 Richard Biener <rguenther@suse.de>
33697 PR middle-end/80281
33698 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
33699 arithmetic done for the negate or the plus. Simplify.
33700 (A - (-B) -> A + B): Likewise.
33701 * fold-const.c (split_tree): Make sure to not negate pointers.
33703 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
33705 PR rtl-optimization/60818
33706 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
33707 a compare of comparisons with the thing compared if this results
33708 in a different machine mode.
33710 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
33712 * alias.c (base_alias_check): Fix typo in comment.
33713 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
33714 * cgraphunit.c (symbol_table::compile): Likewise.
33715 * collect2.c (maybe_run_lto_and_relink): Likewise.
33716 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
33717 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
33718 * config/avr/avr.c (avr_map_op_t): Likewise.
33719 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
33720 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
33721 * config/epiphany/epiphany.md (movcc): Likewise.
33722 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
33723 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
33725 * config/mips/mips.c (mips_save_restore_reg): Likewise.
33726 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
33727 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
33728 * config/sh/sh.c (sh_rtx_costs): Likewise.
33729 * fold-const.c (fold_truth_andor): Likewise.
33730 * genautomata.c (collapse_flag): Likewise.
33731 * gengtype.h (struct type::u::s): Likewise.
33732 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
33733 * input.c (FORMAT_AMOUNT): Likewise.
33734 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
33735 (known_aggs_to_agg_replacement_list): Likewise.
33736 * ipa-inline-analysis.c: Likewise.
33737 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
33738 * ipa-polymorphic-call.c
33739 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
33740 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
33741 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
33743 * modulo-sched.c (apply_reg_moves): Likewise.
33744 * omp-expand.c (build_omp_regions_1): Likewise.
33745 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
33746 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
33747 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
33748 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
33749 * value-prof.c: Likewise.
33750 * var-tracking.c (val_reset): Likewise.
33752 2017-04-03 Richard Biener <rguenther@suse.de>
33754 PR tree-optimization/80275
33755 * fold-const.c (split_address_to_core_and_offset): Handle
33758 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
33760 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
33761 descriptors is at least equal to that of functions.
33763 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33765 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
33767 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33770 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
33771 (mov<IMOD4:mode>): New expander.
33772 (*mov<IMOD4:mode>_internal): New insn and split pattern.
33774 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
33776 PR rtl-optimization/79405
33777 * fwprop.c (propagations_left): New variable.
33778 (forward_propagate_into): Decrement it.
33779 (fwprop_init): Initialize it.
33780 (fw_prop): If the variable has reached zero, stop propagating.
33781 (fwprop_addr): Ditto.
33783 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33786 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
33787 a FUNCTION_DECL, pass it as decl instead of origin to
33790 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
33792 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
33795 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
33798 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
33799 TARGET_VSX_SMALL_INTEGER.
33801 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33803 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
33804 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
33806 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33808 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
33809 extraction from odd-numbered MSA register.
33811 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33813 PR middle-end/80173
33814 * expmed.c (store_bit_field_1): Don't attempt to create
33815 a word subreg out of hard registers wider than word if they
33816 have HARD_REGNO_NREGS of 1 for their mode.
33818 PR middle-end/80163
33819 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
33820 conversions to integer types wider than word and pointer.
33823 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
33824 (rtx_equal_for_cselib_p): Pass 0 to it.
33825 * cselib.c (cselib_hasher::equal): Likewise.
33826 (rtx_equal_for_cselib_1): Add depth argument. If depth
33827 is 128, don't look up VALUE locs and punt. Increment
33828 depth in recursive calls when walking VALUE locs.
33830 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
33832 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
33833 (make_gcov_file_name): Use the canonical path name for generating
33835 (read_line): Fix handling of files with ascii null bytes.
33837 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
33839 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
33840 to initialise a vector register instead
33841 of using a const_int.
33843 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33845 PR translation/80189
33846 * gimplify.c (omp_default_clause): Use %qs instead of %s in
33847 diagnostic messages.
33849 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
33852 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
33853 (dfp_diex_<mode>): Update mode of operand 1.
33854 * doc/extend.texi (dxex, dxexq): Document change to return type.
33855 (diex, diexq): Document change to argument type.
33857 2017-03-30 Martin Jambor <mjambor@suse.cz>
33860 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
33861 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
33862 it reflects the signature changes performed at the callee side.
33863 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
33864 to cgraph_build_function_type_skip_args.
33865 (build_function_decl_skip_args): Adjust call to the above function.
33867 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33870 * config/i386/sse.md
33871 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
33872 register as dest whenever it is a MEM not rtx_equal_p to the
33873 corresponding dup operand, and when forcing into reg move the
33874 reg into the memory afterwards.
33875 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
33876 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
33877 for the force_reg mode.
33878 (avx512vl_vextractf128<mode>): Use register as dest either
33879 always when a MEM, or when it is a MEM not rtx_equal_p to the
33880 corresponding dup operand, or even not when it is a CONST_VECTOR
33881 depending on the mode and lo vs. hi.
33882 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
33884 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
33885 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
33886 Likewise. Require that operands[2] is even.
33887 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
33888 Remove extraneous parens. Require that operands[2] is a multiple
33890 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
33891 operands[0] is a MEM if <mask_applied>, the predicates/constraints
33892 disallow memory then.
33894 2017-03-30 Richard Biener <rguenther@suse.de>
33896 PR tree-optimization/77498
33897 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
33898 to non-constants over backedges.
33900 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
33902 PR rtl-optimization/80233
33903 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
33904 as last_combined_insn. Do not test for BARRIER_P separately.
33906 2017-03-29 Andreas Schwab <schwab@suse.de>
33909 * calls.c (prepare_call_address): Convert funexp to Pmode before
33910 copying to temp reg.
33912 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33914 PR tree-optimization/80158
33915 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
33916 Handle possible future case of more than one alternate
33918 (replace_rhs_if_not_dup): Likewise.
33919 (replace_one_candidate): Likewise.
33921 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
33923 PR rtl-optimization/80193
33924 * ira.c (ira): Do not check allocation for LRA.
33926 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
33928 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
33929 (nvptx_output_simt_exit): Declare.
33930 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
33931 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
33932 (init_softstack_frame): Move initialization of crtl->is_leaf to...
33933 (nvptx_declare_function_name): ...here. Emit declaration of local
33934 memory space buffer for omp_simt_enter insn.
33935 (nvptx_output_unisimt_switch): New.
33936 (nvptx_output_softstack_switch): New.
33937 (nvptx_output_simt_enter): New.
33938 (nvptx_output_simt_exit): New.
33939 * config/nvptx/nvptx.h (struct machine_function): New fields
33940 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
33941 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
33942 (UNSPECV_SIMT_EXIT): Ditto.
33943 (omp_simt_enter_insn): New insn.
33944 (omp_simt_enter): New expansion.
33945 (omp_simt_exit): New insn.
33946 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
33948 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
33949 (expand_GOMP_SIMT_ENTER_ALLOC): New.
33950 (expand_GOMP_SIMT_EXIT): New.
33951 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
33952 (GOMP_SIMT_ENTER_ALLOC): Ditto.
33953 (GOMP_SIMT_EXIT): Ditto.
33954 * target-insns.def (omp_simt_enter): New insn.
33955 (omp_simt_exit): Ditto.
33956 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
33958 (lower_rec_simd_input_clauses): Implement SIMT privatization.
33959 (lower_rec_input_clauses): Likewise.
33960 (lower_lastprivate_clauses): Handle SIMT privatization.
33962 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
33963 (ompdevlow_adjust_simt_enter): New.
33964 (find_simtpriv_var_op): New.
33965 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
33966 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
33968 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
33969 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
33970 (copy_decl_for_dup_finish): Ditto.
33972 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
33974 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
33977 * config/i386/i386.c (ix86_option_override_internal): Always
33978 allow -mpreferred-stack-boundary=3 for 64-bit targets.
33980 2017-03-28 Bin Cheng <bin.cheng@arm.com>
33982 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
33984 2017-03-28 Bin Cheng <bin.cheng@arm.com>
33986 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
33987 mark new edge's irreducible flag accordign to it.
33988 (vect_do_peeling): Check loop preheader edge's irreducible flag
33989 and pass it to function slpeel_add_loop_guard.
33991 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
33993 PR tree-optimization/80218
33994 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33995 Update block frequencies and counts.
33997 2017-03-28 Richard Biener <rguenther@suse.de>
33999 PR tree-optimization/78644
34000 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
34001 of a simplification result we may not use it at all.
34003 2017-03-28 Richard Biener <rguenther@suse.de>
34006 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
34007 without arguments, generate default definition of a SSA name.
34009 2017-03-28 Richard Biener <rguenther@suse.de>
34011 PR middle-end/80222
34012 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
34013 TYPE_REF_CAN_ALIAS_ALL references.
34014 * fold-const.c (fold_indirect_ref_1): Likewise.
34016 2017-03-28 Martin Liska <mliska@suse.cz>
34019 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
34020 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
34022 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34023 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
34025 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
34026 (EXTRA_SPECS): Define.
34027 (SUBTARGET_EXTRA_SPECS): Likewise.
34028 (SUBTARGET_CPP_SPEC): Likewise.
34029 * config/arc/elf.h (EXTRA_SPECS): Renamed to
34030 SUBTARGET_EXTRA_SPECS.
34031 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
34033 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34035 * config/arc/simdext.md (vst64_insn): Update pattern.
34036 (vld32wh_insn): Likewise.
34037 (vld32wl_insn): Likewise.
34038 (vld64_insn): Likewise.
34039 (vld32_insn): Likewise.
34041 2017-03-28 Marek Polacek <polacek@redhat.com>
34044 * fold-const.c (fold_comparison): Use protected_set_expr_location
34045 instead of SET_EXPR_LOCATION.
34047 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
34049 * tree.c (add_expr): Avoid name lookup warning.
34051 2017-03-27 Jeff Law <law@redhat.com>
34053 PR tree-optimization/80216
34054 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
34055 function name. Limit recursion depth.
34056 (record_temporary_equivalences): Corresponding changes.
34058 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
34060 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
34063 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34066 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
34068 * cfgcleanup.c (reg_note_cfa_p): New array.
34069 (insns_have_identical_cfa_notes): New function.
34070 (old_insns_match_p): Don't cross-jump in between /f
34071 and non-/f instructions. If both i1 and i2 are frame related,
34072 verify all CFA notes, their order and content.
34074 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
34077 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
34078 HImode and SImode with zero extend to DImode to one insn.
34079 (bswap<mode>2_extenddi): Likewise.
34080 (bswapsi2_extenddi): Likewise.
34081 (bswaphi2_extendsi): Likewise.
34082 (bswaphi2): Combine bswap HImode and SImode into one insn.
34083 Separate memory insns from swapping register.
34084 (bswapsi2): Likewise.
34085 (bswap<mode>2): Likewise.
34086 (bswaphi2_internal): Delete, no longer used.
34087 (bswapsi2_internal): Likewise.
34088 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
34089 store, and gpr<-gpr swap insns.
34090 (bswap<mode>2_store): Likewise.
34091 (bswaphi2_reg): Register only splitter, combine with the splitter.
34092 (bswaphi2 splitter): Likewise.
34093 (bswapsi2_reg): Likewise.
34094 (bswapsi2 splitter): Likewise.
34095 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
34096 the insns into load, store, and register/register insns.
34097 (bswapdi2_ldbrx): Likewise.
34098 (bswapdi2_load): Likewise.
34099 (bswapdi2_store): Likewise.
34100 (bswapdi2_reg): Likewise.
34102 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
34104 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
34105 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
34107 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
34110 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
34112 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34113 special handling for target option conflicts between dform
34114 options (-mpower9-dform, -mpower9-dform-vector,
34115 -mpower9-dform-scalar) and -mno-direct-move.
34117 2017-03-27 Richard Biener <rguenther@suse.de>
34119 PR tree-optimization/80181
34120 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
34122 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34124 * config/arc/predicates.md (move_double_src_operand): Replace the
34125 call to move_double_src_operand with a call to address_operand.
34127 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34129 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
34130 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
34131 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
34133 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34135 * config/arc/predicates.md (long_immediate_loadstore_operand):
34136 Consider scaled addresses cases.
34138 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34140 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
34141 restored when in interrupt.
34142 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
34143 doesn't have delay slot.
34145 2017-03-27 Richard Biener <rguenther@suse.de>
34148 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
34149 inlined thunk clones.
34151 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34154 * asan.c (instrument_derefs): Copy over last operand from
34155 original COMPONENT_REF to the new COMPONENT_REF with
34156 DECL_BIT_FIELD_REPRESENTATIVE.
34157 * ubsan.c (instrument_object_size): Likewise.
34159 2017-03-27 Richard Biener <rguenther@suse.de>
34161 PR tree-optimization/80170
34162 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
34163 sure DR/SCEV didnt fold in constants we do not see when looking
34164 at the reference base alignment.
34166 2017-03-27 Richard Biener <rguenther@suse.de>
34168 PR middle-end/80171
34169 * gimple-fold.c (fold_ctor_reference): Properly guard against
34170 NULL return value from canonicalize_constructor_val.
34172 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
34175 * config/i386/i386.c (ix86_expand_builtin)
34176 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
34177 flags reg setting and flags reg using instructions.
34178 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
34179 clobbering instructions to zero extend op2.
34181 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
34183 * doc/install.texi (Configuration) <--with-aix-soname>:
34184 Update link to AIX ld.
34186 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
34188 PR rtl-optimization/80160
34189 PR rtl-optimization/80159
34190 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
34191 reg_alternate_class into account.
34193 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
34196 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
34197 to consider in curr_insn_transform.
34199 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34201 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
34202 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
34203 and emit_mode_inner.
34205 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34207 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
34208 argument to the overloaded builtin variants. Use the new flag to
34209 deprecate certain builtin variants.
34210 * config/s390/s390-builtin-types.def: Add new builtin types.
34211 * config/s390/s390-builtins.h: Support new flags field for
34212 overloaded builtins.
34213 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
34214 (s390_macro_to_expand): Enable vector float data type.
34215 (s390_cpu_cpp_builtins_internal): Indicate support of the new
34216 builtins by incrementing the __VEC__ version number.
34217 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
34219 (s390_resolve_overloaded_builtin): Emit error messages depending
34220 on the builtin flags.
34221 * config/s390/s390.c (s390_expand_builtin): Support additional
34222 flags argument. Change error message to match the messages
34223 emitted in s390-c.c.
34224 * config/s390/s390.md: New UNSPEC_* constants.
34225 (op_type): Add new instruction types.
34226 * config/s390/vecintrin.h: Add new builtins and test data class
34228 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
34229 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
34230 (VEC_INEXACT, VEC_NOINEXACT): New constants.
34231 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
34232 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
34233 ("vec_mergel<mode>"): V_HW -> VEC_HW.
34235 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
34236 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
34237 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
34238 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
34240 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
34241 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
34242 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
34243 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
34245 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
34246 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
34247 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
34248 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
34249 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
34250 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
34251 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
34253 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
34254 ("vec_scatter_element<V_HW_4:mode>_DI")
34255 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
34256 ("vec_fpint<mode>", "vflls")
34257 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
34258 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
34259 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
34260 ("*vec_cmphe<mode>_cc"): ... these.
34262 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
34263 mode constant instead of magic value.
34265 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34267 * config/s390/s390.c (s390_expand_vec_compare): Support other
34268 vector floating point modes than just V2DF.
34269 (s390_expand_vcond): Likewise.
34270 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
34271 (s390_cannot_change_mode_class): Prevent mode changes between TF
34272 and V1TF in vector registers.
34273 * config/s390/s390.md (DF, SF): New mode attributes.
34274 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
34275 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
34276 SFmode support for VRs.
34277 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
34279 (VFT, VF_HW): New mode iterators.
34280 (vw, sdx): New mode attributes.
34281 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
34282 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
34283 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
34284 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
34285 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
34286 also the new vector floating point modes. Renaming to ...
34288 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
34289 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
34290 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
34291 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
34292 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
34293 ("vec_unordered<mode>"): ... these.
34295 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
34296 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
34297 ("*vec_extendv2df"): New insn definitions.
34299 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34301 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
34302 ("mulditi3_2", "*muldi3_sign"): New patterns.
34303 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
34304 rename the pattern definition.
34306 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34308 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
34310 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
34312 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34314 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
34315 instruction if possible.
34316 * config/s390/vector.md (vec_halfnumelts): New mode
34318 ("*vec_vllezlf<mode>"): New pattern.
34320 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34322 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
34323 ("popcountv4si2", "popcountv2di2"): Rename to ...
34324 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
34325 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
34327 ("popcount<mode>2_vxe"): New pattern.
34329 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34331 * common/config/s390/s390-common.c (processor_flags_table): Add
34333 * config.gcc: Add arch12.
34334 * config/s390/driver-native.c (s390_host_detect_local_cpu):
34335 Default to arch12 for unknown CPU model numbers.
34336 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
34337 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
34338 PROCESSOR_max sanity check.
34339 * config/s390/s390-opts.h (enum processor_type): Add
34341 * config/s390/s390.c (processor_table): Add arch12.
34342 (s390_expand_builtin): Add check for B_VXE flag.
34343 (s390_issue_rate): Add PROCESSOR_ARCH12.
34344 (s390_get_sched_attrmask): Likewise.
34345 (s390_get_unit_mask): Likewise.
34346 (s390_sched_score): Enable z13 scheduling for arch12.
34347 (s390_sched_reorder): Likewise.
34348 (s390_sched_variable_issue): Likewise.
34349 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
34351 (s390_tune_attr): Use z13 scheduling also for arch12.
34352 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
34353 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
34354 (TARGET_VXE_P): New macros.
34355 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
34356 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
34357 * config/s390/s390.opt: Add arch12 as processor_type.
34359 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34361 * config/s390/s390.md
34362 ("fixuns_truncdddi2", "fixuns_trunctddi2")
34363 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
34364 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
34366 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
34367 Rename expanders to ...
34369 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
34370 ("fixuns_truncdddi2_emu"): ... these.
34372 ("fixuns_trunc<mode>si2_emu"): New expander.
34374 ("*fixuns_truncdfdi2_z13"): Rename to ...
34375 ("*fixuns_truncdfdi2_vx"): ... this.
34377 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34379 * config/s390/2964.md: Remove the single element vector compare
34380 instructions which are no longer used.
34381 * config/s390/s390.c (s390_select_ccmode): Remove handling of
34383 (s390_canonicalize_comparison): Remove handling of DFmode
34385 (s390_expand_vec_compare_scalar): Remove function.
34386 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
34387 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
34389 ("*cmp<mode>_ccs"): Add wfcdb instruction.
34391 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34393 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
34395 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
34396 will anyway by matched by mov<mode>_64dfp.
34398 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34400 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
34401 vlef/vstef. Add missing operand to vleif.
34403 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34405 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
34406 pair for all vector types with 64 bit elements.
34407 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
34408 * config/s390/vector.md (V_HW_64): ... here.
34409 (V_128_NOSINGLE): New mode iterator.
34410 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
34411 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
34412 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
34413 ("*vec_load_pairv2di"): Change to ...
34414 ("*vec_load_pair<mode>"): ... this one.
34416 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34418 * config/s390/constraints.md: Add comments.
34419 (jKK): Reject element sizes > 8 bytes.
34420 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
34422 * config/s390/s390.md: Add the s_operand checks formerly in
34423 s390_split_ok_p to various splitters where they are still
34425 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
34426 for 128 bit vectors. Plus two splitters.
34428 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34430 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
34433 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34436 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
34437 error if the boundary argument is not constant.
34439 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34441 PR rtl-optimization/80112
34442 * loop-doloop.c (doloop_condition_get): Don't check condition
34443 if cmp isn't SET with IF_THEN_ELSE src.
34445 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34447 PR tree-optimization/80158
34448 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
34449 replacing a candidate statement, also replace it for the
34450 candidate's alternate interpretation.
34451 (replace_rhs_if_not_dup): Likewise.
34452 (replace_one_candidate): Likewise.
34454 2017-03-24 Richard Biener <rguenther@suse.de>
34456 PR tree-optimization/80167
34457 * graphite-isl-ast-to-gimple.c
34458 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
34460 (translate_isl_ast_to_gimple::get_rename): Likewise.
34462 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34464 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
34465 handling of certain combinations of target options, including the
34466 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
34467 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
34469 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34472 * config/arm/arm.md (*load_multiple): Add reload_completed to
34473 matching condition.
34475 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34476 Richard Biener <rguenth@suse.de>
34478 PR tree-optimization/79908
34479 PR tree-optimization/80136
34480 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34481 been cast away, gimplify_and_add suffices.
34483 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
34485 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
34487 2017-03-23 Richard Biener <rguenther@suse.de>
34489 PR tree-optimization/80032
34490 * gimplify.c (gimple_push_cleanup): Forced unconditional
34491 cleanups still have to go to the conditional_cleanups
34494 2017-03-22 Jakub Jelinek <jakub@redhat.com>
34496 PR tree-optimization/80072
34497 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
34499 (next_operand_entry_id): Change type to unsigned int.
34500 (sort_by_operand_rank): Make sure to return the right return value
34501 even if unsigned fields are bigger than INT_MAX.
34502 (struct oecount): Change cnt and id type to unsigned int.
34503 (oecount_hasher::equal): Formatting fix.
34504 (oecount_cmp): Make sure to return the right return value
34505 even if unsigned fields are bigger than INT_MAX.
34506 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
34509 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
34510 TREE_READONLY on result if writing it more than once.
34513 * doc/invoke.texi (-fsanitize=thread): Document that with
34514 -fnon-call-exceptions atomics are not able to throw
34518 * tsan.c: Include tree-eh.h.
34519 (instrument_builtin_call): Call maybe_clean_eh_stmt or
34520 maybe_clean_or_replace_eh_stmt where needed.
34521 (instrument_memory_accesses): Add cfg_changed argument.
34522 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
34523 if it returned true.
34524 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
34526 PR rtl-optimization/63191
34527 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
34528 wrapper function, moved the whole old content into ...
34529 (ix86_delegitimize_address_1): ... this. New inline function.
34530 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
34531 true as last argument instead of ix86_delegitimize_address.
34533 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34535 * config/aarch64/aarch64.c (generic_branch_cost): Copy
34536 cortexa57_branch_cost.
34538 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34540 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
34542 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34545 * doc/md.texi (Constraints): Document wA constraint.
34546 * config/rs6000/constraints.md (wA): New.
34547 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
34548 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
34549 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
34550 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
34552 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
34555 * gimplify.c (is_oacc_declared): New function.
34556 (oacc_default_clause): Use it to set default flags for acc declared
34557 variables inside parallel regions.
34558 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
34559 declared variables.
34560 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
34561 declare attribute to any decl as necessary.
34563 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
34566 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
34567 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
34568 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
34569 (arm_arch_lpae): This.
34570 * config/arm/arm.c (arm_arch7ve): Rename into ...
34571 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
34572 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
34575 2017-03-22 Martin Liska <mliska@suse.cz>
34578 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
34579 error message instead of an ICE.
34581 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34583 * doc/extend.texi (6.11 Additional Floating Types): Revise.
34585 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
34587 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34589 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34592 2017-03-21 Martin Sebor <msebor@redhat.com>
34594 * doc/extend.texi: Use "cannot" instead of "can't."
34595 * doc/hostconfig.texi: Same.
34596 * doc/install.texi: Same.
34597 * doc/invoke.texi: Same.
34598 * doc/loop.texi: Same.
34599 * doc/md.texi: Same.
34600 * doc/objc.texi: Same.
34601 * doc/rtl.texi: Same.
34602 * doc/tm.texi: Same.
34603 * doc/tm.texi.in: Same.
34604 * doc/trouble.texi: Same.
34606 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
34609 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
34610 (collect_checksum_attributes): Set it.
34611 (die_checksum_ordered): Use it.
34613 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34615 PR tree-optimization/79908
34616 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
34617 change: For a VA_ARG whose LHS has been cast away, use
34618 force_gimple_operand to construct the side effects.
34620 2017-03-21 David Malcolm <dmalcolm@redhat.com>
34622 PR translation/80001
34623 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
34624 more amenable to translation.
34625 (oacc_loop_auto_partitions): Likewise.
34627 2017-03-21 Marek Polacek <polacek@redhat.com>
34628 Martin Sebor <msebor@redhat.com>
34630 PR tree-optimization/80109
34631 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
34632 on INTEGRAL_TYPE_P.
34634 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34635 Segher Boessenkool <segher@kernel.crashing.org>
34638 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
34639 check reg_used_between_p between insn and one of succ or succ2
34640 depending on if succ is artificial insn not inserted into insn
34643 2017-03-21 Martin Liska <mliska@suse.cz>
34645 PR gcov-profile/80081
34646 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
34647 * doc/gcc.texi: Include gcov-dump stuff.
34648 * doc/gcov-dump.texi: New file.
34650 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
34652 PR rtl-optimization/79150
34653 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
34654 conditional jump, if the jump is the last insn of the loop.
34656 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34657 Richard Biener <rguenth@suse.de>
34659 PR tree-optimization/79908
34660 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34661 been cast away, use force_gimple_operand to construct the side
34664 2017-03-21 Martin Liska <mliska@suse.cz>
34666 PR libfortran/79956
34667 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
34670 2017-03-21 Brad Spengler <spender@grsecurity.net>
34673 * plugin.c (htab_hash_plugin): New function.
34674 (add_new_plugin): Use it and adjust.
34675 (parse_plugin_arg_opt): Adjust.
34676 (init_one_plugin): Likewise.
34678 2017-03-21 Richard Biener <rguenther@suse.de>
34680 PR tree-optimization/80032
34681 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
34682 if set force the cleanup to happen unconditionally.
34683 (gimplify_target_expr): Push inserted clobbers with force_uncond
34684 to avoid them being removed by control-dependent DCE.
34686 2017-03-21 Richard Biener <rguenther@suse.de>
34688 PR tree-optimization/80122
34689 * tree-inline.c (copy_bb): Do not expans va-arg packs or
34690 va_arg_pack_len when the inlined call stmt requires pack
34692 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
34694 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34697 * tsan.c (instrument_builtin_call): If the memory model argument
34698 is not a constant, assume it is valid.
34701 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
34704 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34706 PR rtl-optimization/79910
34707 * combine.c (can_combine_p): Do not allow combining an I0 or I1
34708 if its dest is used by an insn before I2 (other than the combined
34709 insns themselves, which are properly handled already).
34711 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34714 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34716 * combine.c (record_used_regs): New static function.
34717 (try_combine): Handle situations where there is an additional
34718 instruction between I2 and I3 which needs to have a LOG_LINK
34722 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34724 * combine.c (try_combine): Delete redundant i1 test. Call
34725 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34727 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34730 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
34731 alternatives 13/14.
34733 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34735 PR tree-optimization/80054
34736 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
34737 the optimization if a PHI or any of its arguments is not dominated
34738 by the candidate's basis. Use gphi* rather than gimple* as
34740 (replace_profitable_candidates): Clean up a gimple* variable that
34741 should be a gphi* variable.
34743 2017-03-20 Martin Sebor <msebor@redhat.com>
34746 * doc/extend.texi (attribute constructor): Document present limitation.
34748 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
34751 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
34752 __POWER9_VECTOR__ #ifdef control, change template definition to
34753 use Power9-specific built-in function.
34754 (vec_any_eq): Likewise.
34755 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
34756 to control outcomes from this test.
34757 (vector_ae_<mode>p): For VEC_F modes, likewise.
34759 2017-03-20 Ian Lance Taylor <iant@google.com>
34761 * config/i386/i386.c (ix86_function_regparm): Save an extra
34762 register for -fsplit-stack with DECL_STATIC_CHAIN.
34764 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34767 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
34768 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
34770 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34772 * config/riscv/riscv.c (riscv_print_operand): Use "fence
34774 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
34777 2017-03-20 Marek Polacek <polacek@redhat.com>
34780 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
34782 2017-03-20 Richard Biener <rguenther@suse.de>
34784 PR tree-optimization/80113
34785 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
34786 allocate extra SSA name for PHI def.
34787 (add_close_phis_to_outer_loops): Likewise.
34788 (add_close_phis_to_merge_points): Likewise.
34789 (copy_loop_close_phi_args): Likewise.
34790 (copy_cond_phi_nodes): Likewise.
34792 2017-03-20 Martin Liska <mliska@suse.cz>
34794 PR middle-end/79753
34795 * tree-chkp.c (chkp_build_returned_bound): Do not build
34796 returned bounds for a LHS that's not a BOUNDED_P type.
34798 2017-03-20 Martin Liska <mliska@suse.cz>
34802 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
34803 COMPLEX_CST and VECTOR_CST.
34805 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34808 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
34809 target operand. A new splitter adds the clobber statement in case
34810 the target operand is dead anyway.
34812 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
34814 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
34815 to age-old versions of binutils and glibc.
34817 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
34819 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
34821 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34823 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
34825 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34827 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
34828 requirement for binutils 2.13.
34830 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34832 * combine.c (try_combine): Delete redundant i1 test. Call
34833 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34835 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
34837 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
34838 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
34840 <riscv64-*-elf>: Re-arrange section
34841 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
34842 <riscv32-*-linux>: Likewise.
34843 <riscv64-*-elf>: Likewise
34844 <riscv64-*-linux>: Likewise.
34846 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
34849 * aarch64.opt(verbose-cost-dump): Fix typo.
34851 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
34854 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
34855 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
34857 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34859 * reload.c (find_reloads): When reloading a nonoffsettable address,
34860 use RELOAD_OTHER for it and its address reloads.
34862 PR rtl-optimization/79910
34863 * combine.c (record_used_regs): New static function.
34864 (try_combine): Handle situations where there is an additional
34865 instruction between I2 and I3 which needs to have a LOG_LINK
34868 2017-03-17 Jeff Law <law@redhat.com>
34870 PR tree-optimization/71437
34871 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
34872 conditional in the hash table first.
34873 (vrp_dom_walker::before_dom_children): Extract condition from
34874 ASSERT_EXPR. Record condition, its inverion and any implied
34875 conditions as well.
34877 2017-03-17 Marek Polacek <polacek@redhat.com>
34878 Markus Trippelsdorf <markus@trippelsdorf.de>
34880 PR tree-optimization/80079
34881 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
34884 2017-03-17 Richard Biener <rguenther@suse.de>
34886 PR middle-end/80075
34887 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
34888 Properly verify the LHS before the RHS possibly claims to be
34890 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
34893 2017-03-17 Martin Jambor <mjambor@suse.cz>
34895 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
34896 (List of -O2 options): Likewise.
34897 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
34900 2017-03-17 Tom de Vries <tom@codesourcery.com>
34902 * gcov-dump.c (print_usage): Print bug_report_url.
34904 2017-03-17 Richard Biener <rguenther@suse.de>
34906 PR middle-end/80050
34907 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
34908 (parser::peek): Likewise.
34910 2017-03-17 Richard Biener <rguenther@suse.de>
34912 PR tree-optimization/80048
34913 * sese.c (free_sese_info): Properly release rename_map and
34914 copied_bb_map elements.
34916 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
34918 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
34919 Add linked-list forward and backlinks. Insert on
34920 construction, remove on destruction.
34921 (class pass_store_merging): Add m_stores_head field.
34922 (pass_store_merging::terminate_and_process_all_chains):
34923 Iterate over m_stores_head list.
34924 (pass_store_merging::terminate_all_aliasing_chains):
34926 (pass_store_merging::execute): Check for debug stmts first.
34927 Push new chains onto the m_stores_head stack.
34929 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
34932 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
34933 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
34934 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
34936 2017-03-16 Jeff Law <law@redhat.com>
34938 PR tree-optimization/71437
34939 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
34940 member function. Implementation moved into after_dom_children
34941 member function and into the threader's thread_outgoing_edges
34943 (dom_opt_dom_walker::after_dom_children): Simplify by moving
34944 some code into new thread_outgoing_edges.
34945 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
34946 definition. Simplify marker handling (do it here). Assume we always
34947 have the available expression and the const/copies tables.
34948 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
34950 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
34951 * tree-vrp.c (equiv_stack): No longer file scoped.
34952 (vrp_dom_walker): New class.
34953 (vrp_dom_walker::before_dom_children): New member function.
34954 (vrp_dom_walker::after_dom_children): Likewise.
34955 (identify_jump_threads): Setup domwalker. Use it rather than
34956 walking edges in a random order by hand. Simplify setup/finalization.
34957 (finalize_jump_threads): Remove.
34958 (vrp_finalize): Do not call identify_jump_threads here.
34959 (execute_vrp): Do it here instead and call thread_through_all_blocks
34962 PR tree-optimization/71437
34963 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
34965 (simplify_stmt_for_jump_threading): Add basic_block argument. All
34967 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
34968 (dom_opt_dom_walker::thread_across_edge): Remove
34969 handle_dominating_asserts argument. All callers changed.
34970 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
34971 changes. Remove calls to lhs_of_dominating_assert. Other
34972 uses of handle_dominating_asserts turn into unconditional code
34973 (simplify_control_stmt_condition_1): Likewise.
34974 (simplify_control_stmt_condition): Likewise.
34975 (thread_through_normal_block, thread_across_edge): Likewise.
34976 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
34977 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
34978 object if it is not an SSA_NAME.
34979 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
34980 before calling into the VRP specific simplifiers.
34981 (identify_jump_threads): Remove handle_dominating_asserts
34984 2017-03-16 Jakub Jelinek <jakub@redhat.com>
34987 * tree-diagnostic.c (default_tree_printer): No longer static.
34988 * tree-diagnostic.h (default_tree_printer): New prototype.
34990 2017-03-16 Tamar Christina <tamar.christina@arm.com>
34992 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
34993 Change ins into fmov.
34995 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34997 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
34998 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
34999 Use h_con constraint for operand 1.
35000 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
35001 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
35003 2017-03-15 Jeff Law <law@redhat.com>
35005 PR tree-optimization/71437
35006 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
35007 (record_temporary_equivalences): Use it.
35009 PR tree-optimization/71437
35010 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
35011 tree-ssa-scopedtables.
35012 (lookup_avail_expr, build_and_record_new_cond): Likewise.
35013 (record_conditions, record_cond, vuse_eq): Likewise.
35014 (record_edge_info): Adjust to API tweak of record_conditions.
35015 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
35016 (record_temporary_equivalences, optimize_stmt): Likewise.
35017 (eliminate_redundant_computations): Likewise.
35018 (record_equivalences_from_stmt): Likewise.
35019 * tree-ssa-scopedtables.c: Include options.h and params.h.
35020 (vuse_eq): New function, moved from tree-ssa-dom.c
35021 (build_and_record_new_cond): Likewise.
35022 (record_conditions): Likewise. Accept vector of conditions rather
35023 than edge_equivalence structure for first argument.
35024 for the first argument.
35025 (avail_exprs_stack::lookup_avail_expr): New member function, moved
35026 from tree-ssa-dom.c.
35027 (avail_exprs_stack::record_cond): Likewise.
35028 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
35029 from tree-ssa-dom.c.
35030 (avail_exprs_stack): Add new member functions lookup_avail_expr
35032 (record_conditions): Declare.
35034 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
35037 * lra-constraints.c (process_alt_operands): Increase reject for
35038 reloading an input/output operand.
35040 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
35043 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
35044 insns to convert from signed/unsigned char/short to IEEE 128-bit
35046 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
35048 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
35051 * config/i386/i386.c (ix86_vector_duplicate_value): Create
35052 subreg of inner mode for values already in registers.
35054 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
35056 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
35057 iteration reg is used after the loop.
35059 2017-03-14 Martin Sebor <msebor@redhat.com>
35061 PR tree-optimization/79800
35062 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
35063 precision in negative-positive range.
35064 (format_floating): Call non-const overload with adjusted precision.
35066 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
35069 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
35072 2017-03-14 Martin Sebor <msebor@redhat.com>
35074 PR middle-end/80020
35075 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
35076 * builtins.def (aligned_alloc): Use it.
35079 * Makefile.in (GTFILES): Add calls.c.
35080 * calls.c: Include "gt-calls.h".
35082 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
35084 PR rtl-optimization/79728
35085 * regs.h (struct target_regs): New field
35086 x_contains_allocatable_regs_of_mode.
35087 (contains_allocatable_regs_of_mode): New macro.
35088 * reginfo.c (init_reg_sets_1): Initialize it, and change
35089 contains_reg_of_mode so it includes global regs as well.
35090 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
35091 rather than contains_regs_of_mode.
35093 2017-03-14 Martin Liska <mliska@suse.cz>
35095 * doc/invoke.texi: Document options that can't be combined with
35096 -fcheck-pointer-bounds.
35098 2017-03-14 Martin Liska <mliska@suse.cz>
35100 PR middle-end/79831
35101 * doc/invoke.texi (-Wchkp): Document the option.
35103 2017-03-14 Martin Liska <mliska@suse.cz>
35105 * Makefile.in: Install gcov-dump.
35107 2017-03-14 Martin Liska <mliska@suse.cz>
35109 * multiple_target.c (expand_target_clones): Bail out for
35110 an invalid attribute.
35112 2017-03-14 Richard Biener <rguenther@suse.de>
35114 * alias.c (struct alias_set_entry): Pack properly.
35115 * cfgloop.h (struct loop): Likewise.
35116 * cse.c (struct set): Likewise.
35117 * ipa-utils.c (struct searchc_env): Likewise.
35118 * loop-invariant.c (struct invariant): Likewise.
35119 * lra-remat.c (struct cand): Likewise.
35120 * recog.c (struct change_t): Likewise.
35121 * rtl.h (struct address_info): Likewise.
35122 * symbol-summary.h (function_summary): Likewise.
35123 * tree-loop-distribution.c (struct partition): Likewise.
35124 * tree-object-size.c (struct object_size_info): Likewise.
35125 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
35126 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
35127 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
35128 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
35129 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
35130 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
35131 (struct _stmt_vec_info): Likewise.
35133 2017-03-14 Martin Liska <mliska@suse.cz>
35136 * multiple_target.c (create_dispatcher_calls): Check that
35137 a target can create a function dispatcher.
35139 2017-03-14 Martin Liska <mliska@suse.cz>
35142 * multiple_target.c (expand_target_clones): Drop local.local
35143 flag for default implementation.
35145 2017-03-14 Richard Biener <rguenther@suse.de>
35147 PR tree-optimization/80030
35148 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
35150 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
35152 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
35153 gcc_fallthrough() instead of __attribute__((fallthrough));
35155 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35157 * doc/gcc.texi: Remove "up" link to (DIR).
35158 * doc/gccint.texi: Ditto.
35160 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35162 * doc/install.texi (Specific) <avr>: Remove reference to
35165 2017-03-13 Jeff Law <law@redhat.com>
35167 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
35168 attribute rather than comments.
35170 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
35171 match_scratch operand is highest.
35173 2017-03-13 Martin Liska <mliska@suse.cz>
35175 PR middle-end/78339
35176 * ipa-pure-const.c (warn_function_noreturn): If the declarations
35177 is a CHKP clone, use original declaration.
35179 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35181 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
35182 (arc_conditional_register_usage): Use a different allocation order
35183 when optimizing for size.
35184 * common/config/arc/arc-common.c (arc_option_optimization_table):
35185 Section anchors default on when optimizing for size.
35187 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35189 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
35191 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35193 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
35194 * config/arc/arc.md (cpu_facility): Add cd variant.
35195 (*movqi_insn): Add code density variant.
35196 (*movhi_insn): Likewise.
35197 (*movqi_insn): Likewise.
35198 (*addsi3_mixed): Likewise.
35199 (subsi3_insn): Likewise.
35201 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35203 * config/arc/arc.md (movsi_cond_exec): Update constraint.
35205 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35207 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
35208 expressions with MINUS and UNARY ops.
35210 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35213 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
35215 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
35216 between vec_select and vector argument.
35217 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
35218 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
35219 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
35220 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
35221 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
35222 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
35224 2017-03-13 Richard Biener <rguenther@suse.de>
35227 * params.def (vect-max-peeling-for-alignment): Fix typo.
35229 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35231 * doc/install.texi (Specific) <mips-*-*>: Remove description of
35232 issue that only occurred with binutils below 2.18.
35234 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35236 * doc/install.texi (Specific) <cris-axis-elf>: No longer
35237 refer to binutils 2.11/2.12 minimum.
35239 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35241 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
35242 ftp.kernel.org and simplify binutils requirement.
35244 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
35246 * doc/invoke.texi (Warning Options): Fix spelling of link-time
35248 (Optimize Options): Ditto. Also remove redundancy.
35250 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35252 PR translation/79848
35253 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
35255 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
35256 to G_ to avoid double translation.
35258 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35260 PR translation/79923
35261 * auto-profile.c (get_combined_location): Convert leading
35262 character of diagnostics to lower case and remove trailing period.
35263 (read_profile): Likewise for various diagnostics.
35264 * config/arm/arm.c (arm_option_override): Remove trailing period
35265 from various diagnostics.
35266 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
35267 (msp430_expand_delay_cycles): Likewise.
35269 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35272 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
35273 full command-line argument, rather than just "str".
35274 (aarch64_validate_march): Likewise.
35275 (aarch64_validate_mtune): Likewise.
35277 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
35279 PR rtl-optimization/78911
35280 * lra-assigns.c (must_not_spill_p): New function.
35281 (spill_for): Use it.
35283 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35285 PR tree-optimization/79981
35286 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
35287 ATOMIC_COMPARE_EXCHANGE ifn result.
35288 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
35289 IFN_ATOMIC_COMPARE_EXCHANGE.
35291 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35294 * opts.c (parse_sanitizer_options): Add missing question mark to
35295 "did you mean" message.
35297 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35299 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
35301 (VMULEUH_UNS): Likewise.
35302 (VMULOUB_UNS): Likewise.
35303 (VMULOUH_UNS): Likewise.
35304 * config/rs6000/rs6000.c (builtin_function_type): Remove
35305 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
35307 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35310 * read-rtl-function.c (function_reader::read_rtx_operand): Update
35311 x with result of extra_parsing_for_operand_code_0.
35312 (function_reader::extra_parsing_for_operand_code_0): Convert
35313 return type from void to rtx, returning x. When reading
35314 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
35315 larger size containing struct block_symbol.
35317 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
35319 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
35320 -mfloat128-hardware without -m64.
35322 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
35325 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
35326 entries to the case statement that marks unsigned arguments to
35327 overloaded functions.
35329 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
35331 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
35332 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
35334 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
35337 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
35338 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
35340 2017-03-10 Martin Liska <mliska@suse.cz>
35344 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
35345 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
35348 2017-03-10 Olivier Hainque <hainque@adacore.com>
35350 * tree-switch-conversion (array_value_type): Start by resetting
35351 candidate type to it's main variant.
35353 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35355 PR rtl-optimization/79909
35356 * combine.c (try_combine): Use simplify_replace_rtx on individual
35357 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
35358 of the whole CALL_INSN_FUNCTION_USAGE.
35360 PR tree-optimization/79972
35361 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
35362 get_range_info on SSA_NAMEs. Formatting fixes.
35364 2017-03-10 Richard Biener <rguenther@suse.de>
35365 Jakub Jelinek <jakub@redhat.com>
35367 PR tree-optimization/77975
35368 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
35369 edge to be constant.
35370 (get_val_for): For constant x return it. Formatting fix.
35371 (loop_niter_by_eval): Avoid pointless looping if the next iteration
35372 would use the same bases as the current one.
35374 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35376 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
35377 instead of vec_select for V1TImode.
35378 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
35380 (VSX_LE_128): Add V1TI to this mode iterator.
35381 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
35382 (*vsx_le_perm_store_<mode>): Likewise.
35383 (pre-reload splitter for VSX stores): Likewise.
35384 (post-reload splitter for VSX stores): Likewise.
35385 (*vsx_xxpermdi2_le_<mode>): Likewise.
35386 (*vsx_lxvd2x2_le_<mode>): Likewise.
35387 (*vsx_stxvd2x2_le_<mode>): Likewise.
35389 2017-03-09 Michael Eager <eager@eagercon.com>
35391 Correct failures with --enable-checking=yes,rtl.
35393 * config/microblaze/microblaze.c (microblaze_expand_shift):
35394 Replace GET_CODE test with CONST_INT_P and INTVAL test with
35395 test for const0_rtx.
35396 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
35397 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
35399 2017-03-09 Richard Biener <rguenther@suse.de>
35401 PR tree-optimization/79977
35402 * graphite-scop-detection.c (scop_detection::merge_sese):
35403 Handle the case of extra exits to blocks dominating the entry.
35405 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
35407 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
35410 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
35412 PR rtl-optimization/79949
35413 * lra-constraints.c (process_alt_operands): Check memory when
35414 trying to predict a cycle. Print about the overall increase.
35416 2017-03-09 Richard Biener <rguenther@suse.de>
35418 PR middle-end/79971
35419 * gimple-expr.c (useless_type_conversion_p): Preserve
35420 TYPE_SATURATING for fixed-point types.
35422 2017-03-09 Richard Biener <rguenther@suse.de>
35425 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
35426 alignment of BLKmode params.
35428 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35431 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
35432 (VALL_NO_V2Q): Likewise.
35434 * config/aarch64/aarch64-simd.md
35435 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
35437 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
35438 VALL_NO_V2Q mode iterator.
35439 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
35441 2017-03-09 Martin Liska <mliska@suse.cz>
35443 PR tree-optimization/79631
35444 * tree-chkp-opt.c (chkp_is_constant_addr): Call
35445 tree_int_cst_sign_bit just for INTEGER constants.
35447 2017-03-09 Martin Liska <mliska@suse.cz>
35451 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
35454 2017-03-09 Marek Polacek <polacek@redhat.com>
35457 * tree.c (inchash::add_expr): Handle TREE_VEC.
35459 2017-03-09 Martin Liska <mliska@suse.cz>
35462 (chkp_narrow_size_and_offset): New function.
35463 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
35464 (void chkp_parse_bit_field_ref): New function.
35465 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
35466 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
35468 2017-03-09 Martin Liska <mliska@suse.cz>
35471 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
35472 (chkp_find_bounds_1): Remove gcc_unreachable.
35474 2017-03-09 Jakub Jelinek <jakub@redhat.com>
35477 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
35478 BUILT_IN_SYNC*, determine the access type from the size suffix and
35479 always build a MEM_REF with that type. Handle forgotten
35480 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
35483 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
35484 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
35485 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
35486 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
35487 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
35488 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
35489 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
35490 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
35491 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
35492 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
35493 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
35494 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
35495 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
35496 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
35497 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
35498 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
35499 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
35500 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
35501 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
35502 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
35503 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
35504 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
35505 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
35506 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
35507 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
35508 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
35509 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
35510 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
35511 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
35512 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
35513 definitions outside of __OPTIMIZE__ guarded section.
35516 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
35517 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
35518 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
35519 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
35522 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35524 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
35525 ("vfenez<mode>"): Add missing constraints.
35527 2017-03-08 Martin Sebor <msebor@redhat.com>
35530 * config/nds32/nds32.c (nds32_option_override):
35531 Fix misspelled diagnostic.
35533 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35536 * gimplify.c (gimplify_omp_for): Replace index var in outer
35537 taskloop statement with an artificial variable and add
35538 OMP_CLAUSE_PRIVATE clause for it.
35540 2017-03-08 Richard Biener <rguenther@suse.de>
35542 PR tree-optimization/79955
35543 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
35544 for accesses that are completely outside of the variable.
35546 2017-03-08 Andrew Haley <aph@redhat.com>
35548 PR tree-optimization/79943
35549 * tree-ssa-loop-split.c (compute_new_first_bound): When
35550 calculating the new upper bound, (END-BEG) should be added, not
35553 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35555 * config/avr/avr.md (setmemhi): Make sure match_dup
35556 operand number comes before match_scratch.
35558 2017-03-08 Richard Biener <rguenther@suse.de>
35560 PR tree-optimization/79920
35561 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
35562 with ncopies == 1 to ...
35563 (vect_transform_slp_perm_load): ... here. Properly compute
35564 all element loads by iterating VF times over the group. Do
35565 not handle ncopies (computed in a broken way) in
35566 vect_create_mask_and_perm.
35568 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35571 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
35572 is a uniform vector, use uniform_vector_p return value instead of
35573 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
35575 2017-03-07 Marek Polacek <polacek@redhat.com>
35577 PR middle-end/79809
35578 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
35579 (alloca_call_type): Likewise.
35581 2017-03-07 Martin Liska <mliska@suse.cz>
35583 * gcov.c (process_args): Put comment to correct location.
35585 2017-03-07 Martin Liska <mliska@suse.cz>
35587 PR middle-end/68270
35588 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
35589 Use array_at_struct_end_p instead of DECL_CHAIN (field).
35590 (chkp_narrow_bounds_for_field): Likewise.
35591 (chkp_parse_array_and_component_ref): Pass one more argument to
35594 2017-03-07 Richard Biener <rguenther@suse.de>
35596 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
35599 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
35601 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
35602 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
35604 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35607 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
35608 to end of description.
35609 (PARAM_MAX_STORES_TO_MERGE): Likewise.
35611 2017-03-07 Jakub Jelinek <jakub@redhat.com>
35613 PR rtl-optimization/79901
35614 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
35616 (*avx512f_<code><mode>3<mask_name>): ... this.
35617 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
35618 iterator instead of VI8_AVX2_AVX512BW.
35620 PR rtl-optimization/79901
35621 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
35622 min/max expander, expand it using expand_vec_cond_expr.
35625 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
35628 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35631 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
35632 to void * for PCH reasons.
35633 * dwarf2out.c (output_loc_operands, output_die): Cast
35634 v.val_vec.array to unsigned char *.
35636 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
35639 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
35642 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
35644 PR rtl-optimization/79571
35645 * lra-constraints.c (process_alt_operands): Calculate static
35646 reject and subtract it from overall when only addresses will be
35649 2017-03-06 Julia Koval <julia.koval@intel.com>
35652 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
35653 incoming stack boundary to 128 for 64-bit targets.
35655 2017-03-06 Richard Biener <rguenther@suse.de>
35657 PR tree-optimization/79894
35658 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
35659 to NULL after folding it.
35661 2017-03-06 Richard Biener <rguenther@suse.de>
35663 PR tree-optimization/79824
35664 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
35665 check disabling peeling for gaps.
35667 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
35669 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
35670 attributes): Document gettimeofday.
35672 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35674 * config/s390/s390.c (s390_option_override_internal): Set
35675 PARAM_MIN_VECT_LOOP_BOUND
35677 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35679 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
35680 * config/s390/s390.md: Likewise.
35682 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35685 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
35686 (<avx2_avx512>_perm<mode>): Rename to ...
35687 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
35689 (<avx512>_perm<mode>_mask): Rename to ...
35690 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
35692 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
35693 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
35694 instead of VI8F_256_512.
35695 (avx512f_perm<mode>): New define_expand.
35696 (avx512f_perm<mode>_mask): Likewise.
35697 (avx512f_perm<mode>_1<mask_name>): New define_insn.
35698 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
35700 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35702 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
35703 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
35705 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
35707 2017-03-06 Martin Liska <mliska@suse.cz>
35710 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
35711 when having a SSA NAME w/o VAR_DECL assigned to it.
35713 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35715 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
35716 msa_dpsub_<su>_d): Fix MODE for vec_select.
35718 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35720 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
35722 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
35724 2017-03-06 Richard Biener <rguenther@suse.de>
35726 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
35727 * plugin.c (register_plugin_info): Likewise.
35728 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
35730 2017-03-05 Jakub Jelinek <jakub@redhat.com>
35732 * config/i386/sse.md (sse_storehps, sse_storelps,
35733 avx_<castmode><avxsizesuffix>_<castmode>,
35734 avx512f_<castmode><avxsizesuffix>_<castmode>,
35735 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
35736 in condition that at least one operand is not a MEM.
35738 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35740 PR middle-end/79805
35741 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
35742 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
35744 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
35745 gimple_call_nothrow_p flag based on whether original builtin can throw.
35746 If it can, emit following stmts on the fallthrough edge.
35747 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
35748 don't create new bb if inserting just debug stmts on the edge, try to
35749 insert them on the fallthru bb or just reset debug stmts.
35751 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
35754 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
35755 restore recog_data (including the operand rtxes inside it) around
35756 the call to get_insn_template.
35758 2017-03-03 Martin Sebor <msebor@redhat.com>
35760 PR tree-optimization/79699
35761 * context.c (context::~context): Free MPFR caches to avoid
35762 a memory leak on program exit.
35764 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35766 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
35767 Use wide_int::ulow () instead of .elt (0).
35769 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35771 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
35772 (*pushxf): Limit oF constraint to 32bit targets and add oC
35773 constraint for 64bit targets.
35774 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
35775 (*pushdf): Change rmF constraint to rmC.
35777 2017-03-03 Martin Liska <mliska@suse.cz>
35779 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
35780 Remove unused variable.
35782 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35785 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
35786 is a memory operand, increase num_memory.
35787 (ix86_expand_args_builtin): Likewise.
35789 2017-03-03 Jan Hubicka <jh@suse.cz>
35792 * ipa-devirt.c (maybe_record_node): Properly handle
35793 __cxa_pure_virtual visibility.
35795 2017-03-03 Martin Liska <mliska@suse.cz>
35797 PR tree-optimization/79803
35798 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
35800 (pass_loop_prefetch::execute): Disabled optimization if an
35801 assumption about L1 cache size is not met.
35803 2017-03-03 Martin Liska <mliska@suse.cz>
35805 PR rtl-optimization/79574
35806 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
35807 (hash_scan_set): Likewise.
35808 (dump_hash_table): Likewise.
35809 (hoist_code): Likewise.
35811 2017-03-03 Richard Biener <rguenther@suse.de>
35813 * fixed-value.c (fixed_from_string): Restore use of elt (1)
35814 in place of uhigh ().
35815 (fixed_convert_from_real): Likewise.
35817 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35820 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
35822 2017-03-03 Richard Biener <rguenther@suse.de>
35824 PR middle-end/79818
35825 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
35826 TYPE_OVERFLOW_UNDEFINED check.
35828 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35830 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
35832 (vector_ae_<mode>_p): Likewise.
35833 (vector_nez_<mode>_p): Likewise.
35834 (vector_ne_v2di_p): Likewise.
35835 (vector_ae_v2di_p): Likewise.
35836 (vector_ne_<mode>_p): Likewise.
35837 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
35839 (vsx_tsqrt<mode>2_fe): Likewise.
35841 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
35844 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
35846 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35848 PR rtl-optimization/79780
35849 * cprop.c (one_cprop_pass): When second and further conditional trap
35850 in a single basic block is turned into an unconditional trap, turn it
35851 into a deleted note to avoid RTL verification failures.
35853 2017-03-02 Richard Biener <rguenther@suse.de>
35855 * fold-const.c (const_binop): Use ulow () instead of elt (0).
35857 2017-03-02 Richard Biener <rguenther@suse.de>
35859 PR tree-optimization/79345
35861 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
35862 param and abort the walk, returning -1 if it is hit.
35863 (walk_aliased_vdefs): Take a limit param and pass it on.
35864 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
35865 defaulting to 0 and return a signed int.
35866 * tree-ssa-uninit.c (struct check_defs_data): New struct.
35867 (check_defs): New helper.
35868 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
35869 about uninitialized memory.
35870 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
35871 bogus uninitialized warning.
35872 (fixed_convert_from_real): Likewise.
35874 2017-03-02 Bin Cheng <bin.cheng@arm.com>
35876 PR tree-optimization/66768
35877 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
35878 iv_use if base object can't be determined.
35880 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35882 PR tree-optimization/79345
35883 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
35884 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
35885 (get_pattern_stats): Initialize it.
35886 * genemit.c (gen_expand): Verify match_scratch numbers come after
35887 match_operand/match_dup numbers.
35888 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
35889 match_scratch numbers.
35890 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
35892 * config/s390/s390.md (trunctdsd2): Likewise.
35894 2017-03-02 Richard Biener <rguenther@suse.de>
35896 * wide-int.h (wide_int_storage::operator=): Implement in terms
35899 2017-03-02 Richard Biener <rguenther@suse.de>
35901 PR tree-optimization/79777
35902 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
35903 the to insert expression to sth existing.
35905 2017-03-01 Martin Sebor <msebor@redhat.com>
35907 PR middle-end/79692
35908 * gimple-ssa-sprintf.c
35909 (directive::known_width_and_precision): New function.
35910 (format_integer): Use it.
35911 (get_mpfr_format_length): Consider the full range of precision
35912 when computing %g output with the # flag. Set the likely byte
35913 count to 3 rather than 1 when precision is indeterminate.
35914 (format_floating): Correct the lower bound of precision.
35916 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35918 * doc/invoke.texi: Document default code model for 64-bit Linux.
35920 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35923 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
35924 udiv rather than div since input pattern is unsigned.
35926 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
35928 * config/i386/i386.c (print_reg): Warn for values of
35929 unsupported size in integer register.
35931 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
35934 * config/rs6000/predicates.md (current_file_function_operand): Do
35935 not allow self calls to be local if the function is replaceable.
35937 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
35940 * config/rs6000/altivec.h (vec_ctz and others): Change the
35941 preprocessor macro that controls conditional compilation from
35942 _ARCH_PWR9 to __POWER9_VECTOR__.
35943 (vec_all_ne): Change parameterization of __altivec_scalar_pred
35944 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
35945 control (instead of _ARCH_PWR9 control) so that template
35946 definition uses power9-specific function.
35947 (vec_any_eq): Likewise.
35948 (vec_all_ne): Change macro definition to use a power9-specific
35949 expansion under #ifdef __POWER9_VECTOR__ control (instead of
35950 _ARCH_PWR9 control).
35951 (vec_any_eq) Likewise.
35952 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
35953 expansion for CMPNEF to remove support for xvcmpnesp instruction.
35954 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
35955 support for xvcmpnedp instruction.
35956 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
35957 macro expansion so that Power9 implementation of vec_all_ne does
35958 not use the AltiVec predicate framework.
35959 (VCMPNEH_P): Likewise.
35960 (VCMPNEW_P): Likewise.
35961 (VCMPNED_P): Likewise.
35962 (VCMPNEFP_P): Likewise.
35963 (VCMPNEDP_P): Likewise.
35964 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
35965 implementation of vec_any_eq to not use AltiVec predicate
35967 (VCMPAEH_P): Likewise.
35968 (VCMPAEW_P): Likewise.
35969 (VCMPAED_P): Likewise.
35970 (VCMPAEFP_P): Likewise.
35971 (VCMPAEDP_P): Likewise.
35972 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
35973 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
35974 not use the AltiVec predicate framework.
35975 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
35976 of vec_any_eq to not use AltiVec predicate framework.
35977 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35978 support for predefined __POWER9_VECTOR__ macro to indicate that
35979 Power9 instruction selection is enabled.
35980 (altivec_overloaded_builtins): Remove extraneous
35981 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
35982 function argument types RS6000_BTI_bool_V16QI and
35983 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
35984 entry for overloaded function argument types RS6000_BTI_bool_V4SI
35985 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
35986 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
35987 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
35988 Power9 for implementations of vec_cmpne. Change the signature for
35989 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
35990 (representing vec_all_ne) to remove the previously described first
35991 argument of type RS6000_BTI_INTSI, as this was an artifact of
35992 reliance on the AltiVec predicate framework, which is no longer
35993 used in the implementation of these functions. Add
35994 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
35995 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
35996 since, unlike the AltiVec predicate framework implementation, we
35997 do not share function descriptors between vec_alle and vec_anyeq.
35998 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
35999 set of modes that receive special treatment even when
36000 TARGET_P9_VECTOR is true. The special treatment emits code that
36001 does not depend on Power9 instructions.
36002 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
36003 define_expand to not rely on AltiVec predicate framework.
36004 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36006 (vector_ne_v2di_p): Change this define_expand to not rely on
36007 AltiVec predicate framework.
36008 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
36010 (vector_ne_<mode>_p): Change this define_expand to not rely on
36011 AltiVec predicate framework.
36012 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36014 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
36015 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
36016 define_insn pattern.
36017 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
36018 define_insn pattern because the xvcmpne<VSs>. instruction is not
36020 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
36021 instruction is not supported.
36023 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36025 * config/nvptx/nvptx.c: Include intl.h.
36027 2017-03-01 Martin Jambor <mjambor@suse.cz>
36030 * ipa-prop.h (ipa_bits): Removed field known.
36031 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
36032 to pointers. Adjusted their comments to warn about their sharing.
36033 (ipcp_transformation_summary): Change bits to a vector of pointers.
36034 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
36035 (ipa_get_ipa_bits_for_value): Declare.
36036 * tree-vrp.h (value_range): Mark as GTY((for_user)).
36037 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
36038 (ipa_bits_hash_table): Likewise.
36039 (ipa_vr_ggc_hash_traits): Likewise.
36040 (ipa_vr_hash_table): Likewise.
36041 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
36042 being pointers and vr_known being removed.
36043 (ipa_set_jf_unknown): Likewise.
36044 (ipa_get_ipa_bits_for_value): New function.
36045 (ipa_set_jfunc_bits): Likewise.
36046 (ipa_get_value_range): New overloaded functions.
36047 (ipa_set_jfunc_vr): Likewise.
36048 (ipa_compute_jump_functions_for_edge): Use the above functions to
36049 construct bits and vr parts of jump functions.
36050 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
36051 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36053 (ipcp_grow_transformations_if_necessary): Also allocate
36054 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36056 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
36057 them. Fix too long lines.
36058 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
36059 vr_known being removed.
36060 (ipa_read_jump_function): Use new setter functions to construct bits
36061 and vr parts of jump functions or set them to NULL.
36062 (write_ipcp_transformation_info): Adjust for bits being pointers.
36063 (read_ipcp_transformation_info): Likewise.
36064 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
36066 Include gt-ipa-prop.h.
36067 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
36069 (ipcp_store_bits_results): Likewise.
36070 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
36071 Do not write to existing jump functions but use a temporary instead.
36073 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36076 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
36077 attempt to use its first operand as BIT_FIELD_REF base.
36079 2017-03-01 Richard Biener <rguenther@suse.de>
36081 PR middle-end/79721
36082 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
36083 interpolating formula in wrapping arithmetic.
36084 (chrec_apply): Convert chrec_evaluate return value to wanted type.
36086 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36088 PR tree-optimization/79734
36089 * tree-vect-generic.c (expand_vector_condition): Optimize
36090 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
36091 Handle VEC_COND_EXPR where comparison has different inner width from
36092 type's inner width.
36094 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
36096 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
36097 markup, and similar issues. Remove @opindex entries for things
36098 that aren't options. Add missing -mmpy-option entries.
36100 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36102 PR tree-optimization/79737
36103 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
36104 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
36105 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
36106 instead of byte_size. Formatting fix.
36107 (shift_bytes_in_array_right): Formatting fix.
36109 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
36112 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
36113 condition on optimize for the leaf function test.
36115 2017-02-28 Martin Liska <mliska@suse.cz>
36118 * read-rtl-function.c (function_reader::handle_unknown_directive):
36119 Bail out when one uses -flto.
36121 2017-02-28 Martin Liska <mliska@suse.cz>
36123 * common.opt: Replace space with tabular for options of <number>
36125 * config/i386/i386.opt: Show <number> value for
36126 -mlarge-data-threshold.
36127 * opts.c (print_filtered_help): Do not display number in hexadecimal
36130 2017-02-28 Martin Liska <mliska@suse.cz>
36132 * common.opt: Fix --help=option -Q for options which are of
36135 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
36137 * config/i386/i386.c (print_reg): Error out for values
36138 of 8-bit size in invalid integer register.
36140 2017-02-28 Martin Sebor <msebor@redhat.com>
36142 PR tree-optimization/79691
36143 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
36145 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36148 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
36149 gcc_unreachable with output_operand_lossage.
36151 2017-02-28 Richard Biener <rguenther@suse.de>
36153 PR tree-optimization/79740
36154 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
36156 (visit_nary_op): Insert the nary into the hashtable if we
36157 pattern-matched sth.
36158 * tree-ssa-pre.c (eliminate_insert): Robustify.
36160 2017-02-28 Richard Biener <rguenther@suse.de>
36162 PR middle-end/79731
36163 * fold-const.c (decode_field_reference): Reject out-of-bound
36166 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36168 * config/i386/i386.c: Include intl.h.
36169 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
36170 instead of just cond ? "..." : "...".
36171 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
36172 * coverage.c (read_counts_file): Likewise.
36173 * omp-offload.c: Include intl.h.
36174 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
36175 of just cond ? "..." : "...".
36176 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
36177 of just cond ? "..." : "...".
36179 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
36182 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
36184 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
36185 'tune for' CPU name.
36186 * config/arm/arm-cpu-data.h: Regenerated.
36188 2017-02-28 Richard Biener <rguenther@suse.de>
36190 PR tree-optimization/79732
36191 * tree-inline.c (expand_call_inline): Do not shadow var.
36193 2017-02-28 Richard Biener <rguenther@suse.de>
36195 PR tree-optimization/79723
36196 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
36197 address-space properly.
36199 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
36201 * doc/optinfo.texi (Optimization groups): Fix option used for
36203 * doc/invoke.texi (-fopt-info): Document "omp".
36204 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
36205 (OPTGROUP_ALL): Add OPTGROUP_OMP.
36206 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
36207 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
36208 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
36210 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
36212 * dumpfile.c (optgroup_options): Instead of "openmp", associate
36213 OPTGROUP_OMP with "omp".
36215 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
36218 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
36219 for arithmetic shift of unsigned V2DI.
36221 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
36223 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
36224 arc/linux.h headers.
36225 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
36226 (LINK_SPEC): Likewise.
36227 (ARC_TLS_EXTRA_START_SPEC): Likewise.
36228 (EXTRA_SPECS): Likewise.
36229 (STARTFILE_SPEC): Likewise.
36230 (ENDFILE_SPEC): Likewise.
36231 (LIB_SPEC): Likewise.
36232 (TARGET_SDATA_DEFAULT): Likewise.
36233 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
36234 (MULTILIB_DEFAULTS): Likewise.
36235 (DWARF2_UNWIND_INFO): Likewise.
36236 * config/arc/big.h: New file.
36237 * config/arc/elf.h: Likewise.
36238 * config/arc/linux.h: Likewise.
36239 * config/arc/t-uClibc: Remove.
36241 2017-02-27 Bin Cheng <bin.cheng@arm.com>
36243 PR tree-optimization/77536
36244 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
36245 (tree_transform_and_unroll_loop): Use above function to compute the
36246 estimated niter of unrolled loop and use it when scaling profile.
36247 Also use count info rather than frequency if it's non-zero.
36248 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
36249 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
36250 (vect_transform_loop): Call above function.
36252 2017-02-27 Richard Biener <rguenther@suse.de>
36254 PR tree-optimization/45397
36255 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
36256 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
36257 (visit_nary_op): Add pattern matching for CSEing sign-changed
36258 or truncated operations with wider ones.
36260 2017-02-27 Richard Biener <rguenther@suse.de>
36262 PR tree-optimization/79690
36263 * tree-vect-stmts.c (vectorizable_store): Use vector type
36264 built from the DR with address-space.
36266 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
36268 * doc/invoke.texi (Optimize Options): Refine the description
36269 of asan-use-after-return.
36271 2017-02-25 Alan Modra <amodra@gmail.com>
36273 PR rtl-optimization/79584
36274 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
36275 base, not ad->base_term, the reg within base. Remove assertion
36276 that ad->base == ad->base_term. Replace gen_int_mode using
36277 bogus mode with const0_rtx.
36279 2017-02-25 Jakub Jelinek <jakub@redhat.com>
36281 PR middle-end/79396
36282 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
36283 FMA_EXPR like tcc_binary or tcc_unary.
36285 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
36288 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
36290 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
36291 (output_loc_operands): Handle DW_OP_call_ref and
36292 DW_OP_GNU_variable_value.
36293 (struct variable_value_struct): New type.
36294 (struct variable_value_hasher): Likewise.
36295 (variable_value_hash): New variable.
36296 (string_types): Remove.
36297 (copy_loc_descr): New function.
36298 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
36299 (prepend_loc_descr_to_each): New function.
36300 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
36301 instead of add_loc_descr_to_each if the first argument is single
36302 location list and the second has multiple.
36303 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
36304 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
36305 when looking for variable value which doesn't have other location info.
36306 (loc_list_from_tree): Formatting fix.
36307 (gen_array_type_die): Simplify DW_AT_string_length handling.
36308 (adjust_string_types): Remove.
36309 (gen_subprogram_die): Don't call adjust_string_types nor test/set
36310 string_types. Call resolve_variable_values.
36311 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
36312 (resolve_addr_in_expr): Likewise. Add A argument.
36313 (copy_deref_exprloc): Remove deref argument. Adjust for the
36314 original expression being DW_OP_GNU_variable_value with optionally
36315 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
36316 optionally after it.
36317 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
36318 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
36319 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
36320 (variable_value_hasher::hash, variable_value_hasher::equal): New
36322 (resolve_variable_value_in_expr, resolve_variable_value,
36323 resolve_variable_values, note_variable_value_in_expr,
36324 note_variable_value): New functions.
36325 (dwarf2out_early_finish): Call note_variable_value on all toplevel
36328 2017-02-24 Jakub Jelinek <jakub@redhat.com>
36331 * opts.h (handle_generated_option): Add GENERATED_P argument.
36332 * opts-common.c (handle_option): Adjust function comment.
36333 (handle_generated_option): Add GENERATED_P argument, pass it to
36335 (control_warning_option): Pass false to handle_generated_option
36337 * opts.c (maybe_default_option): Pass true to handle_generated_option
36339 * optc-gen.awk: Likewise.
36341 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36343 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
36344 a REG, look at the REG it is a SUBREG of.
36345 (splitter for cmpeqsi_t): Ditto.
36347 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36349 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
36350 the special USEs with the pattern of the insn, not the insn itself.
36352 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
36355 * doc/invoke.texi: Document -mload-store-pairs.
36357 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36358 Sandra Loosemore <sandra@codesourcery.com>
36360 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
36361 argument isn't a CONST_INT.
36362 (nios2_alternate_compare_const): Assert op is a CONST_INT.
36363 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
36364 (nios2_validate_compare): Bypass alternate compare logic if *op2
36365 is not a CONST_INT.
36366 (ldstwm_operation_p): Return false if first_base is not a REG or
36367 if first_offset is not a CONST_INT.
36369 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36371 * config/cris/cris.md: Use correct operand in a define_peephole2.
36373 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36375 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
36377 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36379 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
36380 this_insn if it is an INSN or JUMP_INSN.
36381 (force_offsettable): Look at base, not at addr.
36382 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
36383 on things that aren't necessarily CONST_INTs.
36385 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
36387 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
36388 -mfpmath=sse is the default also for x86-32 targets with SSE2
36389 instruction set when @option{-ffast-math} is enabled
36391 2017-02-24 Jeff Law <law@redhat.com>
36393 PR rtl-optimizatoin/79286
36394 * ira.c (update_equiv_regs): Drop may_trap_p exception to
36397 2017-02-24 Richard Biener <rguenther@suse.de>
36399 PR tree-optimization/79389
36400 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
36403 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
36405 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
36406 function comment to reflect reality.
36407 (loop_exits_before_overflow): Fix typo in function description.
36409 2017-02-24 Richard Biener <rguenther@suse.de>
36411 PR tree-optimization/79389
36412 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
36413 properly that a threading opportunity exists. Detect conditional
36414 copy/constant propagation opportunities.
36416 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
36418 * config/visium/visium.md (type): Add trap.
36419 (b): New mode attribute.
36420 (*btst): Rename into...
36421 (*btst<mode>): ...this and adjust.
36422 (*cbranchsi4_btst_insn): Rename into...
36423 (*cbranch<mode>4_btst_insn): ...this and adjust.
36424 (trap): New define_insn.
36426 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36428 PR tree-optimization/79389
36429 * ifcvt.c (struct noce_if_info): Add rev_cond field.
36430 (noce_reversed_cond_code): New function.
36431 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
36432 reversed_comparison_code. Formatting fix.
36433 (noce_try_store_flag): Test rev_cond != NULL in addition to
36434 reversed_comparison_code.
36435 (noce_try_store_flag_constants): Likewise.
36436 (noce_try_store_flag_mask): Likewise.
36437 (noce_try_addcc): Use rev_cond if non-NULL instead of
36438 reversed_comparison_code.
36439 (noce_try_cmove_arith): Likewise. Formatting fixes.
36440 (noce_try_minmax, noce_try_abs): Clear rev_cond.
36441 (noce_find_if_block): Initialize rev_cond.
36442 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
36443 instead of false as last argument never attempt to reverse it
36446 2017-02-23 Bin Cheng <bin.cheng@arm.com>
36448 PR tree-optimization/79663
36449 * tree-predcom.c (combine_chains): Process refs in reverse order
36450 only for ZERO length chains, and add explaining comment.
36452 2017-02-23 Jeff Law <law@redhat.com>
36454 PR tree-optimization/79578
36455 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
36456 in call to operand_equal_p.
36458 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
36461 * config/i386/cpuid.h: Fix another undefined behavior.
36463 2017-02-23 Richard Biener <rguenther@suse.de>
36465 PR tree-optimization/79683
36466 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
36467 vector types for data-refs.
36469 2017-02-23 Martin Liska <mliska@suse.cz>
36471 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
36473 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36475 PR middle-end/79665
36476 * internal-fn.c (get_range_pos_neg): Moved to ...
36477 * tree.c (get_range_pos_neg): ... here. No longer static.
36478 * tree.h (get_range_pos_neg): New prototype.
36479 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
36480 are known to be in between 0 and signed maximum inclusive, try to
36481 expand both unsigned and signed divmod and use the cheaper one from
36484 2017-02-22 Jeff Law <law@redhat.com>
36486 PR tree-optimization/79578
36487 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
36488 to compare base operands.
36490 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
36493 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
36494 gpc_reg_operand instead of fpr_reg_operand.
36496 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
36498 * config/mips/mips.c (mips_return_in_memory): Force FP
36499 vector types to be returned in memory for o32 ABI.
36501 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36503 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
36504 instead of DW_TAG_member for static data member declarations and don't
36505 set no_linkage_name for static inline data members.
36506 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
36509 2017-02-22 Martin Liska <mliska@suse.cz>
36511 * doc/invoke.texi: Replace inequality signs with square brackets
36514 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
36517 * lra-constraints.c (simplify_operand_subreg): Handle
36518 WORD_REGISTER_OPERATIONS targets.
36520 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36523 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
36524 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
36525 elimination by swapping fld*.
36527 2017-02-22 Richard Biener <rguenther@suse.de>
36529 PR tree-optimization/79673
36530 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
36531 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
36532 irrelevant address-space qualifiers and avoiding a
36533 ADDR_SPACE_CONVERT_EXPR from fold_convert.
36535 2017-02-22 Richard Biener <rguenther@suse.de>
36537 PR tree-optimization/79666
36538 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
36539 to not symbolically negate if that may introduce undefined
36542 2017-02-22 Martin Liska <mliska@suse.cz>
36545 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
36546 * data-streamer-out.c (streamer_write_gcov_count_stream):
36548 * value-prof.c (stream_out_histogram_value): Make assert more
36549 precise based on type of counter.
36551 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
36554 * config/i386/i386.md (standard_x87sse_constant_load splitter):
36555 Use nonimmediate_operand instead of memory_operand for operand 1.
36556 (float-extend standard_x87sse_constant_load splitter): Ditto.
36558 2017-02-21 Jeff Law <law@redhat.com>
36560 PR tree-optimization/79621
36561 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
36562 blocks with edges to themselves.
36564 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36567 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
36568 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
36569 Use gimple_call_builtin_p.
36572 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
36573 on temporarily removed DEBUG_INSNs.
36575 PR tree-optimization/79649
36576 * tree-loop-distribution.c (classify_partition): Give up on
36577 non-generic address space loads/stores.
36579 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
36581 * doc/loop.texi (Loop manipulation): Remove nonexistent
36582 tree_ssa_loop_version from the documentation.
36583 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
36585 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36588 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
36589 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
36590 * config/rs6000/rs6000.c: Include except.h.
36591 (rs6000_expand_split_stack_prologue): Call
36592 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
36594 2017-02-21 Martin Jambor <mjambor@suse.cz>
36597 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
36598 have been analyzed.
36600 2017-02-21 Martin Jambor <mjambor@suse.cz>
36602 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
36603 for backward compatibility only.
36604 * doc/invoke.texi (Option Summary): Remove all references to
36605 -fipa-cp-alignment.
36607 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
36611 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36613 * lra-constraints.c (curr_insn_transform): Handle
36614 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36616 2017-02-21 Martin Liska <mliska@suse.cz>
36618 * config/i386/i386.opt: Replace -masm-dialect with -masm.
36620 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
36622 PR translation/79638
36623 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
36625 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
36628 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
36629 (arm_function_ok_for_sibcall): Return false for an indirect call by
36630 descriptor if all the argument registers are used.
36631 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
36632 alignment of the function.
36634 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36636 PR tree-optimization/61441
36637 * simplify-rtx.c (simplify_const_unary_operation): For
36638 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
36639 the sNaN unmodified.
36641 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
36643 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
36644 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
36645 instead of SYSTEM_HEADER_DIR.
36647 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
36648 Martin Liška <mliska@suse.cz>
36650 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
36651 Fix typos and grammar, use active voice, and clarify.
36653 2017-02-20 Marek Polacek <polacek@redhat.com>
36655 PR middle-end/79537
36656 * gimplify.c (gimplify_expr): Handle unused *&&L;.
36659 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
36661 2017-02-20 Jakub Jelinek <jakub@redhat.com>
36664 * config/i386/i386.c (ix86_expand_builtin): Handle
36665 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
36666 ix86_builtins_isa[fcode].isa as a requirement of those
36667 flags and any other flag in the bitmask.
36668 (ix86_init_mmx_sse_builtins): Use 0 instead of
36669 ~OPTION_MASK_ISA_64BIT as mask.
36670 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
36671 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
36672 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
36673 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
36675 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36678 * lra-constraints.c (split_reg): Check requested split mode
36679 is supported by the register.
36681 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36683 * lra-constraints.c (simplify_operand_subreg): Remove early
36686 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36689 * lra-constraints.c (curr_insn_transform): Tighten condition
36690 for converting SUBREG reloads from OP_OUT to OP_INOUT.
36692 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36695 * lra-constraints.c (curr_insn_transform): Handle
36696 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36698 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
36701 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
36703 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
36705 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
36708 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
36711 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36713 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
36716 2017-02-18 Jakub Jelinek <jakub@redhat.com>
36718 * final.c (last_columnnum, override_columnnum): New variables.
36719 (final_start_function): Set last_columnnum, pass it to begin_prologue
36720 hook and pass 0 to dwarf2out_begin_prologue.
36721 (final_scan_insn): Update override_columnnum. Pass last_columnnum
36722 to source_line debug hook.
36723 (notice_source_line): Compute last_columnnum and for debug_column_info
36724 return true on column changes.
36725 * debug.h (struct gcc_debug_hooks): Add column argument to
36726 source_line and begin_prologue hooks.
36727 (debug_nothing_int_charstar_int_bool): Remove prototype.
36728 (debug_nothing_int_int_charstar,
36729 debug_nothing_int_int_charstar_int_bool): New prototypes.
36730 (dwarf2out_begin_prologue): Add column argument.
36731 * debug.c (do_nothing_debug_hooks): Adjust source_line and
36732 begin_prologue hooks.
36733 (debug_nothing_int_charstar_int_bool): Remove.
36734 (debug_nothing_int_int_charstar,
36735 debug_nothing_int_int_charstar_int_bool): New functions.
36736 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
36737 through to dwarf2out_source_line.
36738 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
36739 (dwarf2out_source_line): Add column argument, emit it if requested.
36740 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
36742 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36743 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36744 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
36745 through to dwarf2out_begin_prologue.
36746 (vmsdbgout_source_line): Add column argument, pass it through to
36747 dwarf2out_source_line.
36748 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
36749 dbxout_source_line caller.
36750 (dbxout_source_line): Add column argument.
36752 * common.opt (gno-column-info, gcolumn-info): New options.
36753 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
36754 (check_die): Also test for multiple DW_AT_decl_column attributes.
36755 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
36756 DW_AT_decl_column if requested.
36757 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
36759 (gen_variable_die): Likewise.
36760 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
36761 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
36764 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
36765 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
36766 (ix86_handle_option): Handle OPT_m3dnowa.
36767 * doc/invoke.texi (-m3dnowa): Document.
36768 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
36769 -m3dnowa instead of -m3dnow -march=athlon.
36772 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
36773 instead of gcc_assert for K, r and R code checks. Formatting fixes.
36775 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36778 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
36779 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
36780 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
36781 generator for vsx_xxpermdi_<mode>_be.
36782 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
36783 force big-endian semantics.
36784 (vsx_xxpermdi_<mode>_be): New define_expand with same
36785 implementation as previous version of vsx_xxpermdi_<mode>.
36787 2017-02-17 Jakub Jelinek <jakub@redhat.com>
36789 PR tree-optimization/79327
36790 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
36791 variable, its initialization and use.
36793 2017-02-17 Julia Koval <julia.koval@intel.com>
36795 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
36796 (OPTION_MASK_ISA_PKU_UNSET): New.
36797 (ix86_handle_option): Handle -mrdpid.
36798 * config/i386/cpuid.h (bit_RDPID): New.
36799 * config/i386/driver-i386.c (host_detect_local_cpu):
36800 Detect RDPID feature.
36801 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
36802 * config/i386/i386-c.c (ix86_target_macros_internal):
36804 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
36805 (ix86_valid_target_attribute_inner_p): Add "rdpid".
36806 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
36807 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
36808 * config/i386/i386.md (define_insn "rdpid"): New.
36809 * config/i386/i386.opt Add -mrdpid.
36810 * config/i386/immintrin.h (_rdpid_u32): New.
36812 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
36814 PR rtl-optimization/79541
36815 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
36816 instead of transforming it into USE.
36818 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
36820 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
36821 If HONOR_SNANS (SFmode) force the input to a register.
36822 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
36823 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
36824 an frsp or similar insn.
36826 2017-02-17 Martin Liska <mliska@suse.cz>
36828 PR rtl-optimization/79577
36829 * params.def (selsched-max-sched-times): Increase minimum to 1.
36831 2017-02-17 Martin Liska <mliska@suse.cz>
36833 PR rtl-optimization/79574
36834 * gcse.c (want_to_gcse_p): Prevent integer overflow.
36836 2017-02-17 Martin Liska <mliska@suse.cz>
36838 PR tree-optimization/79529
36839 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
36840 ssa_defined_default_def_p to handle cases which are implicitly
36842 * tree-ssa.c (ssa_defined_default_def_p): New function.
36843 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
36844 which are implicitly defined.
36845 * tree-ssa.h (ssa_defined_default_def_p): Declare.
36847 2017-02-17 Richard Biener <rguenther@suse.de>
36849 PR middle-end/79576
36850 * params.def (max-ssa-name-query-depth): Limit to 10.
36852 2017-02-17 Richard Biener <rguenther@suse.de>
36854 PR tree-optimization/79552
36855 * tree-ssa-structalias.c (visit_loadstore): Properly verify
36858 2017-02-17 Richard Biener <rguenther@suse.de>
36861 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
36863 2017-02-17 Marek Polacek <polacek@redhat.com>
36865 PR middle-end/79536
36866 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
36867 (fold_negate_expr): New wrapper.
36869 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
36871 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
36872 Correct terminology and de-emphasize pre-standard behavior.
36874 2017-02-16 Alan Modra <amodra@gmail.com>
36876 PR rtl-optimization/79286
36877 * ira.c (def_dominates_uses): New function.
36878 (update_equiv_regs): Don't create an equivalence for insns that
36879 may trap where the register def does not dominate the use.
36881 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
36883 PR rtl-optimization/78127
36884 * lra.c (lra): Call lra_eliminate before finish the loop after
36887 2017-02-16 Richard Biener <rguenther@suse.de>
36889 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
36891 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
36892 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
36893 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
36894 (isl_val_int_from_wi): New function.
36895 (extract_affine_gmp): Rename to ...
36896 (extract_affine_wi): ... this, take a widest_int.
36897 (extract_affine_int): Just wrap extract_affine_wi.
36898 (add_param_constraints): Use isl_val_int_from_wi.
36899 (add_loop_constraints): Likewise, and extract_affine_wi.
36901 2017-02-15 Jeff Law <law@redhat.com>
36903 PR middle-end/79521
36904 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
36905 ira_init_register_move_cost_if_necessary.
36907 2017-02-15 Martin Sebor <msebor@redhat.com>
36909 PR middle-end/32003
36910 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
36911 removed in a prior commit.
36913 2017-02-15 Bin Cheng <bin.cheng@arm.com>
36915 PR tree-optimization/79347
36916 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
36917 counters during peeling.
36919 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
36921 * Makefile.in (site.exp): Remove "set ISLVER".
36923 2017-02-15 Jakub Jelinek <jakub@redhat.com>
36926 * real.c (real_from_integer): Call real_convert even for decimal.
36928 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
36931 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
36933 2017-02-14 Andrew Pinski <apinski@cavium.com>
36935 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
36936 cores and change the partno/implementer to be correct.
36937 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
36938 the 'B" as the implementer.
36939 * config/aarch64/aarch64-tune.md: Regenerate.
36941 2017-02-14 Carl Love <cel@us.ibm.com>
36943 * config/rs6000/rs6000.c: Add case statement entry to make the
36944 xvcvuxdsp built-in argument unsigned.
36945 * config/rs6000/vsx.md: Fix the source and return operand types so they
36946 match the instruction definitions from the ISA document. Fix typo
36947 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
36950 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
36953 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
36954 member early_clobber_alts.
36955 * lra-lives.c (reg_early_clobber_p): New.
36956 (process_bb_lives): Use it.
36957 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
36958 (debug_operand_data): Initialize early_clobber_alts.
36959 (setup_operand_alternative): Set up early_clobber_alts.
36960 (collect_non_operand_hard_regs): Ditto. Pass early clobber
36961 alternatives to new_insn_reg.
36962 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
36964 (lra_update_insn_regno_info): Pass the new arg.
36966 2017-02-14 Jakub Jelinek <jakub@redhat.com>
36968 PR middle-end/79505
36969 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
36970 (new_oacc_loop_raw): Don't clear already cleared fields.
36973 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
36974 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
36975 _mm512_prefetch_i64gather_ps): New inline functions and macros.
36977 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
36980 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
36982 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
36985 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
36986 the extra instruction to the right place to store 128-bit constant
36989 2017-02-14 Martin Sebor <msebor@redhat.com>
36991 PR middle-end/79448
36992 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
36993 warning for strings of unknown length.
36995 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
36997 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
36999 2017-02-14 Jeff Law <law@redhat.com>
37002 * ira-costs.c (scan_one_insn): Initialize register move costs
37003 for pseudos seen in USE/CLOBBER insns.
37005 PR tree-optimization/79095
37006 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
37007 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
37008 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
37009 if the operands are known to be not equal, then the resulting range
37011 (intersect_ranges): If the new range is ~[0,0] and the old range is
37012 wide, then prefer ~[0,0].
37013 * tree-vrp.c (overflow_comparison_p_1): New function.
37014 (overflow_comparison_p): New function.
37015 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
37016 if NAME is used in an overflow test.
37017 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
37018 overflow check that can be expressed as an equality test, then adjust
37019 ops to be that equality test.
37021 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37023 * config/s390/s390-builtin-types.def: Remove flags argument.
37024 * config/s390/s390.c (s390_init_builtins): Likewise.
37026 2017-02-14 Martin Liska <mliska@suse.cz>
37028 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
37029 vector. Fix trailing white spaces.
37031 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
37033 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
37036 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37038 PR rtl-optimization/68664
37039 * config/arm/arm.c (arm_sched_can_speculate_insn):
37040 New function. Declare prototype.
37041 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37043 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37045 PR rtl-optimization/68664
37046 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
37048 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37050 2017-02-14 Amit Pawar <amit.pawar@amd.com>
37052 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
37053 max skip bytes for function, loop and jump.
37055 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37057 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
37058 ABS_EXPR for gimple dump.
37060 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37063 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
37065 PR tree-optimization/79408
37066 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
37067 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
37068 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
37069 also if rhs1 is INTEGER_CST.
37071 2017-02-14 Richard Biener <rguenther@suse.de>
37073 PR middle-end/79432
37074 * tree-into-ssa.c (insert_phi_nodes): When the function can
37075 have abnormal edges rewrite SSA names with broken use-def
37076 dominance out of SSA and register them for PHI insertion.
37078 2017-02-13 Martin Sebor <msebor@redhat.com>
37080 PR middle-end/79496
37081 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
37082 clearing info.nowrite flag when snprintf size argument is a range.
37084 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37086 * cprop.c (cprop_jump): Add missing space in string literal.
37087 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
37088 (get_constraint_for_component_ref): Likewise.
37089 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
37090 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
37091 * lra-constraints.c (process_alt_operands): Likewise.
37092 * ipa-inline.c (inline_small_functions): Likewise.
37093 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
37094 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
37095 * trans-mem.c (diagnose_tm_1_op): Likewise.
37096 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
37097 (grid_parallel_clauses_gridifiable): Likewise.
37099 * config/nvptx/mkoffload.c (process): Add space in between
37102 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
37103 "MOD4_SSE_REGS" and "ALL_REGS".
37105 * spellcheck.c (test_data): Add , in between "foo" and "food".
37107 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37110 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
37111 boundary crossing check and subsequent code generation agree.
37113 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37115 * config/aarch64/aarch64.c (has_memory_op): Delete.
37116 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
37119 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37121 PR rtl-optimization/79388
37122 PR rtl-optimization/79450
37123 * combine.c (distribute_notes): When removing TEM_INSN for which
37124 corresponding dest has last value recorded, invalidate that last
37127 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37129 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
37130 of explicit '@'. Add missing assembly comment marker on branch costs
37133 2017-02-13 Nathan Sidwell <nathan@acm.org>
37135 * gengtype-lex.l (<in_struct>): Add '/'.
37137 2017-02-13 Martin Liska <mliska@suse.cz>
37140 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
37142 2017-02-13 Richard Biener <rguenther@suse.de>
37144 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
37146 * configure: Re-generate.
37147 * config.in: Likewise.
37148 * graphite-dependences.c: Simplify as if
37149 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
37150 * graphite-isl-ast-to-gimple.c: Likewise.
37151 * graphite-optimize-isl.c: Likewise.
37152 * graphite-poly.c: Likewise.
37153 * graphite-sese-to-poly.c: Likewise.
37154 * graphite.h: Likewise.
37155 * toplev.c: Include isl/version.h and use isl_version () for
37156 printing the ISL version.
37157 * doc/install.texi: Update ISL requirement.
37159 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37161 * doc/standards.texi (Standards): Update reference to
37164 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37166 * doc/extend.texi (Named Address Spaces): sourceware.org now
37168 * doc/install.texi (Binaries): Ditto.
37171 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37173 * doc/cpp.texi: Replace "stringify"/"stringification" with C
37174 standard terminology "stringize"/"stringizing" throughout.
37175 * doc/cppinternals.texi: Likewise.
37177 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37179 * doc/extend.texi: Fix some spelling mistakes and typos.
37180 * doc/invoke.texi: Likewise.
37182 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37185 * params.def (inline-min-speedup) Change from 10 to 8.
37187 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37189 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
37192 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37195 * ipa-inline-analysis.c (get_minimal_bb): New function.
37196 (record_modified): Use it.
37197 (remap_edge_change_prob): Handle also ancestor functions.
37199 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
37201 * doc/contrib.texi (Contributors): Remove broken link into
37202 the Mauve CVS repository.
37204 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37206 PR middle-end/79454
37207 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
37208 result computation whenever lhs doesn't have vector mode, not
37209 just when it has BLKmode.
37211 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37213 * doc/makefile.texi (profiledbootstrap): Refer to the
37214 installation instructions only in textual form.
37216 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37219 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
37221 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37223 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
37224 (Specific): Update mingw-w64 reference.
37226 (Specific): Remove broken link to Renesas RX processor.
37228 2017-02-10 Richard Biener <rguenther@suse.de>
37230 * toplev.c (process_options): Do not mention obsolete graphite
37231 options when printing sorry message about missing graphite support.
37232 Mention -floop-nest-optimize.
37234 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
37236 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
37237 (vtst_p16): Likewise.
37238 (vtstq_p8): Likewise.
37239 (vtstq_p16): Likewise.
37241 (vtstq_p64): Likewise.
37242 * config/arm/arm_neon.h (vgetq_lane_p64): New.
37243 (vset_lane_p64): New.
37244 (vsetq_lane_p64): New.
37246 2017-02-10 Jakub Jelinek <jakub@redhat.com>
37248 PR tree-optimization/79411
37249 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
37250 stmt operands are SSA_NAMEs used in abnormal phis.
37251 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
37254 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37257 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
37260 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37262 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
37264 2017-02-09 Jakub Jelinek <jakub@redhat.com>
37266 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
37270 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
37271 not arbitrary TREE_CONSTANT.
37274 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
37275 "omp declare target link" attribute unless is_global_var.
37276 * omp-offload.c (find_link_var_op): Likewise.
37278 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
37279 Chung-Lin Tang <cltang@codesourcery.com>
37281 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
37283 (gimplify_adjust_omp_clauses): Don't delete TILE.
37284 (gimplify_omp_for): Deal with TILE.
37285 * internal-fn.c (expand_GOACC_TILE): New function.
37286 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
37288 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
37289 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
37291 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
37292 avoid DIV for outermost collapse var.
37293 (expand_oacc_for): Insert tile element loop as needed. Adjust.
37294 Remove out of date comments, fix whitespace.
37295 * omp-general.c (omp_extract_for_data): Deal with tiling.
37296 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
37297 adjust OLF_DIM_BASE value.
37298 (struct omp_for_data): Add tiling field.
37299 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
37300 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
37301 for auto loops. Remove default auto determining, moved to
37302 oacc_loop_fixed_partitions.
37303 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
37304 stmts, add e_mask field.
37305 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
37306 (oacc_thread_numbers): Use oacc_dim_call.
37307 (oacc_xform_tile): New.
37308 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
37309 (finish_oacc_loop): Adjust for ifns vector.
37310 (oacc_loop_discover_walk): Append loop abstraction sites to list,
37311 add case for GOACC_TILE fns.
37312 (oacc_loop_xform_loop): Delete.
37313 (oacc_loop_process): Iterate over call list directly, and add
37314 handling for GOACC_TILE fns.
37315 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
37317 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
37318 vector partitioning to outer loops. Assign 2 partitions to loops
37319 when available. Add TILE handling.
37320 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
37321 (execite_oacc_device_lower): Process GOACC_TILE fns,
37322 ignore unknown specs.
37323 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
37324 * tree.c (omp_clause_num_ops): Adjust TILE ops.
37325 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
37327 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
37329 * configure.ac (ACX_BUGURL): Update.
37330 * configure: Regenerate.
37332 2017-02-09 Richard Biener <rguenther@suse.de>
37334 PR tree-optimization/69823
37335 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
37336 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
37338 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37340 * config/arc/arc-c.def: Add __NPS400__ definition.
37341 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
37342 (TARGET_NPS400): Define.
37344 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37346 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
37348 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
37349 pointer, arch_info.
37350 (arc_cpu_types): Fill the arch_info field with a pointer into the
37351 arc_arch_types table.
37352 (arc_selected_cpu): Declare.
37353 * config/arc/arc.c (arc_selected_cpu): Make global.
37354 (arc_selected_arch): Delete.
37355 (arc_base_cpu): Delete.
37356 (arc_override_options): Remove references to deleted variables,
37357 update access to arch information.
37358 (ARC_OPT): Update access to arch information.
37359 (ARC_OPTX): Likewise.
37360 * config/arc/arc.h (arc_base_cpu): Remove declaration.
37361 (TARGET_ARC600): Update access to arch information.
37362 (TARGET_ARC601): Likewise.
37363 (TARGET_ARC700): Likewise.
37364 (TARGET_EM): Likewise.
37365 (TARGET_HS): Likewise.
37366 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
37369 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
37372 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
37373 condition/operands for integer GE/LE/GEU/LEU operations.
37375 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
37377 PR translation/79397
37378 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
37381 2017-02-08 Martin Jambor <mjambor@suse.cz>
37384 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
37385 whether allocation happened.
37386 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
37387 nothing was allocated.
37389 2017-02-08 Jakub Jelinek <jakub@redhat.com>
37391 PR tree-optimization/79408
37392 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
37393 constant, but SSA_NAME with a known integer range, use the minimum
37394 of that range instead of op1 to determine if modulo can be replaced
37395 with its first operand.
37397 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37399 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
37401 2017-02-08 Richard Biener <rguenther@suse.de>
37403 PR tree-optimization/71824
37404 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37405 Check all loops contained in the merged region.
37407 2017-02-07 Andrew Pinski <apinski@cavium.com>
37409 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
37411 2017-02-07 Andrew Pinski <apinski@cavium.com>
37413 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
37414 (thunderxt88): Likewise.
37415 (thunderxt81): Disable LSE and change v8.1 to v8.
37416 (thunderxt83): Likewise.
37418 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37419 Richard Biener <rguenther@suse.de>
37421 PR middle-end/79399
37422 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
37423 type from int to size_t.
37424 * ira-costs.c (struct_costs_size): Change type from int to size_t.
37426 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37428 PR rtl-optimization/79386
37429 * cprop.c (bypass_conditional_jumps): Initialize
37430 bypass_last_basic_block already before splitting bbs after
37431 unconditional traps...
37432 (bypass_conditional_jumps): ... rather than here.
37435 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
37436 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
37437 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
37438 fix -masm=intel patterns.
37440 2017-02-07 Richard Biener <rguenther@suse.de>
37442 PR tree-optimization/79256
37443 PR middle-end/79278
37444 * builtins.c (get_object_alignment_2): Use min_align_of_type
37445 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
37446 and ADJUST_FIELD_ALIGN.
37448 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
37450 * doc/tm.texi: Regenerate.
37451 * stor-layout.c (layout_decl): Adjust.
37452 (update_alignment_for_field): Likewise.
37453 (place_field): Likewise.
37454 (min_align_of_type): Likewise.
37455 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
37456 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
37457 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
37458 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
37459 * config/frv/frv.c (frv_adjust_field_align): Likewise.
37460 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
37461 * config/i386/i386.c (x86_field_alignment): Likewise.
37462 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
37463 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
37464 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
37465 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
37466 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
37467 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
37471 2017-01-30 Richard Biener <rguenther@suse.de>
37473 PR tree-optimization/79256
37474 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37475 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37478 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
37480 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
37481 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
37482 builtins to SImode and emit a zero-extend, if necessary.
37484 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37486 * docs/invoke.texi (RISC-V Options): Alphabetize.
37488 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37490 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
37493 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37495 * config/riscv/riscv.c: New file.
37496 * common/config/riscv/riscv-common.c: Likewise.
37497 * config.gcc: Likewise.
37498 * config/riscv/constraints.md: Likewise.
37499 * config/riscv/elf.h: Likewise.
37500 * config/riscv/generic.md: Likewise.
37501 * config/riscv/linux.h: Likewise.
37502 * config/riscv/multilib-generator: Likewise.
37503 * config/riscv/peephole.md: Likewise.
37504 * config/riscv/pic.md: Likewise.
37505 * config/riscv/predicates.md: Likewise.
37506 * config/riscv/riscv-builtins.c: Likewise.
37507 * config/riscv/riscv-c.c: Likewise.
37508 * config/riscv/riscv-ftypes.def: Likewise.
37509 * config/riscv/riscv-modes.def: Likewise.
37510 * config/riscv/riscv-opts.h: Likewise.
37511 * config/riscv/riscv-protos.h: Likewise.
37512 * config/riscv/riscv.h: Likewise.
37513 * config/riscv/riscv.md: Likewise.
37514 * config/riscv/riscv.opt: Likewise.
37515 * config/riscv/sync.md: Likewise.
37516 * config/riscv/t-elf-multilib: Likewise.
37517 * config/riscv/t-linux: Likewise.
37518 * config/riscv/t-linux-multilib: Likewise.
37519 * config/riscv/t-riscv: Likewise.
37520 * configure.ac: Likewise.
37521 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
37522 Waterman as RISC-V maintainers.
37523 * doc/install.texi: Add RISC-V entries.
37524 * doc/invoke.texi: Add RISC-V options section.
37525 * doc/md.texi: Add RISC-V constraints section.
37526 * configure: Regenerated.
37528 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
37531 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
37532 false values to be constant vectors with all 0 or all 1 bits set.
37533 (vcondu<mode><mode>): Likewise.
37534 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
37536 (fpmask_comparison_operator): Update comment.
37537 (vecint_comparison_operator): New predicate.
37538 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
37539 vector conditionals when the true and false values are constant
37540 vectors with all 0 bits or all 1 bits set.
37542 2017-02-06 Martin Sebor <msebor@redhat.com>
37544 PR tree-optimization/79376
37545 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
37547 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
37549 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
37550 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
37551 to simplify split condition.
37553 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37555 * omp-expand.c (oxpand_omp_atomic_fetch_op,
37556 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
37559 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
37561 PR rtl-optimization/68664
37562 * target.def (can_speculate_insn): New hook.
37563 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
37564 * doc/tm.texi: Regenerate.
37565 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
37566 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
37567 (rs6000_sched_can_speculate_insn): New function.
37569 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37571 PR tree-optimization/79284
37572 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
37573 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
37574 vectorizable_mask_load_store, vectorizable_operation,
37575 vect_is_simple_cond, get_same_sized_vectype): Use it instead
37576 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
37577 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
37578 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
37579 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
37580 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
37581 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
37582 is_gimple_assign (stmt). Replace another such test with
37583 is_gimple_assign (stmt).
37585 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37588 * config/avr/avr.c (rtl-iter.h): Include it.
37589 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
37590 (avr_legitimate_combined_insn): ...and implementation.
37592 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
37594 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
37595 * config/s390/s390.c (s390_const_operand_ok)
37596 (s390_canonicalize_comparison, s390_extract_part)
37597 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
37598 (s390_contiguous_bitmask_p, s390_rtx_costs)
37599 (legitimize_pic_address): Likewise.
37600 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
37601 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
37602 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
37603 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
37604 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
37606 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37608 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
37609 REGNO($0) == REGNO($1).
37611 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37613 * config/s390/linux.h(SIZE_TYPE): Add comment.
37615 2017-02-06 Julian Brown <julian@codesourcery.com>
37616 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37617 Virendra Pathak <virendra.pathak@broadcom.com>
37619 * config/aarch64/aarch64-cores.def: Change the scheduler
37621 * config/aarch64/aarch64.md: Include thunderx2t99.md.
37622 * config/aarch64/thunderx2t99.md: New file.
37624 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37626 * doc/standards.texi (Go Language): Update link to language
37629 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37631 * tree-eh.c (lower_resx): Sanitize profile.
37632 (cleanup_empty_eh_move_lp): Likewise.
37634 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37637 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
37639 * cfgloopmanip.h (loop_version): Update prototype.
37640 * modulo-sched.c (sms_schedule): Update call of loop_version.
37641 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
37642 * tree-parloops.c (gen_parallel_loop): Likewise.
37643 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
37644 * tree-ssa-loop-split.c (split_loop): Likewise.
37645 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
37646 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
37648 2017-02-05 Martin Liska <mliska@suse.cz>
37651 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
37653 (print_operand_address): Initialize a struct to zero.
37655 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37657 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
37658 garbage collector only in textual form.
37660 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37662 * doc/extend.texi (x86 specific memory model extensions for
37663 transactional memory): Simplify a phrase.
37665 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
37668 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
37669 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
37670 (atomic_storedi_1): Likewise.
37672 2017-02-04 Jakub Jelinek <jakub@redhat.com>
37674 PR tree-optimization/79338
37675 * tree-parloops.c (gather_scalar_reductions): Don't call
37676 vect_analyze_loop_form for loop->inner before destroying loop's
37679 2017-02-03 Martin Sebor <msebor@redhat.com>
37681 PR tree-optimization/79327
37682 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
37683 when precision has resulted in leading zeros.
37684 (format_integer): Adjust the likely counter to assume an unknown
37685 argument that may be zero is non-zero.
37687 2017-02-03 Jason Merrill <jason@redhat.com>
37690 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
37691 avoid copying non-taken branch.
37693 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37695 PR tree-optimization/79340
37696 * tree-vect-loop.c (vectorizable_reduction): Release
37697 vec_defs elements after safe_splicing them into other vectors.
37700 PR tree-optimization/79327
37701 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
37702 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
37704 (format_integer): Use wide_int_to_tree instead of build_int_cst
37705 + to_?hwi. If argmin is NULL, just set argmin and argmax to
37706 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
37707 of shortest and longest sequence.
37709 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
37711 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
37712 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
37714 2017-02-03 Walter Lee <walt@tilera.com>
37717 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
37718 after initial stackframe link reg save.
37719 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
37721 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37724 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
37725 wu for stxssp alternative.
37727 2017-02-03 Martin Sebor <msebor@redhat.com>
37729 PR tree-optimization/79352
37730 * gimple-fold.c (get_range_strlen): Add argument.
37731 (get_range_strlen): Change return type to bool.
37732 (get_maxval_strlen): Pass in a dummy argument.
37733 * gimple-fold.h (get_range_strlen): Change return type to bool.
37734 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
37735 * tree.h (array_at_struct_end_p): Add argument.
37736 * tree.c (array_at_struct_end_p): Handle it.
37738 2017-02-03 Martin Liska <mliska@suse.cz>
37741 * multiple_target.c (create_dispatcher_calls): Redirect edge
37742 from a caller of a dispatcher.
37743 (expand_target_clones): Make the clones local.
37744 (ipa_target_clone): Do both target clones and resolvers.
37745 (ipa_dispatcher_calls): Remove the pass.
37746 (pass_dispatcher_calls::gate): Likewise.
37747 (make_pass_dispatcher_calls): Likewise.
37748 * passes.def (pass_target_clone): Put as very first IPA early
37751 2017-02-03 Martin Liska <mliska@suse.cz>
37753 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
37754 in case of a function with ifunc attribute.
37756 2017-02-03 Martin Liska <mliska@suse.cz>
37758 * cgraph.c (cgraph_node::dump): Dump function version info.
37759 * symtab.c (symtab_node::dump_base): Add missing new line.
37761 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37763 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
37764 (ifcombine_ifandif): Use it.
37766 2017-02-03 Martin Liska <mliska@suse.cz>
37768 * doc/invoke.texi: Document default value for
37769 use-after-scope-direct-emission-threshold.
37771 2017-02-03 Martin Liska <mliska@suse.cz>
37773 PR tree-optimization/79339
37774 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
37775 (format_floating): Likewise.
37777 2017-02-03 Martin Liska <mliska@suse.cz>
37780 * ipa-prop.c (ipa_node_params_t::insert): Remove current
37782 (ipa_node_params_t::remove): Likewise.
37783 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
37784 initialization from removed ipa_node_params_t::insert.
37785 (ipa_node_params::~ipa_node_params): Move from removed
37786 ipa_node_params_t::release.
37787 * symbol-summary.h (symbol_summary::m_released): New member.
37788 Do not release a summary twice. Do not allow to call finalizer
37789 for types of a summary that live in GGC memory.
37791 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37793 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
37796 2017-02-02 Martin Sebor <msebor@redhat.com>
37798 PR middle-end/79275
37799 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
37800 (format_string): Tighten up the range of output for non-constant
37801 strings and correct the expected range for wide non-constant strings.
37803 2017-02-02 Martin Sebor <msebor@redhat.com>
37805 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
37807 PR middle-end/32003
37808 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
37810 (-fdump-tree-@var): Add to index and document how to come up
37811 with pass-specific option and dump file names.
37812 (-fdump-passes): Clarify where to look for output.
37814 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37816 PR middle-end/77445
37817 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
37818 statistics of the analyzed path; allow threading for speed when
37819 any of BBs along the path are optimized for speed.
37821 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
37823 PR middle-end/78468
37824 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
37825 settings of the virtual registers.
37828 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
37830 * explow.c (get_dynamic_stack_size): Take known alignment of stack
37831 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
37834 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37836 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
37837 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
37839 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37841 * config/s390/s390.md: Add missing comments with the expanded
37843 * config/s390/vector.md: Likewise.
37844 * config/s390/vx-builtins.md: Likewise.
37846 2017-02-02 Jakub Jelinek <jakub@redhat.com>
37849 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
37850 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
37851 conditions on a single line.
37853 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37855 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
37856 __S390_VX__ to __VX__.
37858 2017-02-01 Andrew Pinski <apinski@cavium.com>
37860 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
37861 stmt_info to record_stmt_cost.
37862 (vect_get_known_peeling_cost): Pass stmt_info if known to
37864 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
37865 cpu_vector_cost field into
37866 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
37867 field into vec_int_stmt_cost and vec_fp_stmt_cost.
37868 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
37869 splitting of scalar_stmt_cost and vec_stmt_cost.
37870 (thunderx_vector_cost): Likewise.
37871 (cortexa57_vector_cost): LIkewise.
37872 (exynosm1_vector_cost): Likewise.
37873 (xgene1_vector_cost): Likewise.
37874 (thunderx2t99_vector_cost): Improve after the splitting of the two
37876 (aarch64_builtin_vectorization_cost): Update for the splitting of
37877 scalar_stmt_cost and vec_stmt_cost.
37879 2017-02-01 Torvald Riegel <triegel@redhat.com>
37880 Richard Henderson <rth@redhat.com>
37882 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
37883 conditional on existance of a fast atomic load.
37884 * optabs-query.c (can_atomic_load_p): New function.
37885 * optabs-query.h (can_atomic_load_p): Declare it.
37886 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
37887 no fast atomic load is available for the particular size of access.
37888 (expand_atomic_compare_and_swap): Likewise.
37889 (expand_atomic_load): Likewise.
37890 (expand_atomic_store): Likewise.
37891 (expand_atomic_fetch_op): Likewise.
37892 * testsuite/lib/target-supports.exp
37893 (check_effective_target_sync_int_128): Remove x86 because it provides
37894 no fast atomic load.
37895 (check_effective_target_sync_int_128_runtime): Likewise.
37897 2017-02-01 Richard Biener <rguenther@suse.de>
37899 * graphite.c: Include tree-vectorizer.h for find_loop_location.
37900 (graphite_transform_loops): Provide opt-info for optimized nests.
37901 * tree-parloop.c (parallelize_loops): Provide opt-info for
37902 parallelized loops.
37904 2017-02-01 Richard Biener <rguenther@suse.de>
37906 PR middle-end/79315
37907 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
37908 was not set before.
37910 2017-02-01 Richard Biener <rguenther@suse.de>
37912 PR tree-optimization/71824
37913 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37914 Verify the loops are valid in the merged SESE region.
37915 (scop_detection::can_represent_loop_1): Check analyzing the
37916 evolution of the number of iterations in the region succeeds.
37918 2017-01-31 Ian Lance Taylor <iant@golang.org>
37920 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
37921 REG_ARGS_SIZE note to 32-bit push insns and call insn.
37923 2017-01-31 David Malcolm <dmalcolm@redhat.com>
37925 PR preprocessor/79210
37926 * input.c (get_substring_ranges_for_loc): Replace line_width
37927 assertion with error-handling.
37929 2017-01-31 Richard Biener <rguenther@suse.de>
37931 PR tree-optimization/77318
37932 * graphite-sese-to-poly.c (extract_affine): Fix assert.
37933 (create_pw_aff_from_tree): Take loop parameter.
37934 (add_condition_to_pbb): Pass loop of the condition to
37935 create_pw_aff_from_tree.
37937 2017-01-31 Jakub Jelinek <jakub@redhat.com>
37939 * config/s390/s390.c (s390_asan_shadow_offset): New function.
37940 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
37942 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
37946 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
37948 (convert_int_to_float128): Likewise.
37949 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
37950 (convert_int_to_float128): Likewise.
37951 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
37952 (UNSPEC_IEEE128_CONVERT): Likewise.
37953 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
37954 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
37955 Use local variables for IBM extended format.
37956 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
37957 (fix_trunc<mode>si2_fprs): Likewise.
37958 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
37959 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
37960 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
37961 to know that we can now have integers of all sizes in vector
37963 (fix<uns>_<mode>di2_hw): Likewise.
37964 (float<uns>_<mode>si2_hw): Likewise.
37965 (fix_<mode>si2_hw): Likewise.
37966 (fixuns_<mode>si2_hw): Likewise.
37967 (float<uns>_<mode>di2_hw): Likewise.
37968 (float_<mode>di2_hw): Likewise.
37969 (float_<mode>si2_hw): Likewise.
37970 (floatuns_<mode>di2_hw): Likewise.
37971 (floatuns_<mode>si2_hw): Likewise.
37972 (xscvqp<su>wz_<mode>): Delete, no longer used.
37973 (xscvqp<su>dz_<mode>): Likewise.
37974 (xscv<su>dqp_<mode>): Likewise.
37975 (ieee128_mfvsrd_64bit): Likewise.
37976 (ieee128_mfvsrd_32bit): Likewise.
37977 (ieee128_mfvsrwz): Likewise.
37978 (ieee128_mtvsrw): Likewise.
37979 (ieee128_mtvsrd_64bit): Likewise.
37980 (ieee128_mtvsrd_32bit): Likewise.
37982 2017-01-31 Martin Liska <mliska@suse.cz>
37985 * ipa-prop.c (ipa_free_all_node_params): Call release method
37986 instead of ~sumbol_summary to not to trigger double times
37989 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
37991 PR tree-optimization/71691
37992 * bitmap.h (class auto_bitmap): New.
37993 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
37994 is_maybe_undefined instead of ssa_undefined_value_p.
37996 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37998 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
37999 __S390_ARCH_LEVEL__ to __ARCH__.
38001 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38003 PR tree-optimization/79267
38004 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
38005 if should_remove_lhs_p is true.
38007 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
38010 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
38011 (add_alignment_attribute): New.
38012 (base_type_die): Add alignment attribute.
38013 (subrange_type_die): Likewise.
38014 (modified_type_die): Likewise.
38015 (gen_array_type_die): Likewise.
38016 (gen_descr_array_type_die: Likewise.
38017 (gen_enumeration_type_die): Likewise.
38018 (gen_subprogram_die): Likewise.
38019 (gen_variable_die): Likewise.
38020 (gen_field_die): Likewise.
38021 (gen_ptr_to_mbr_type_die): Likewise.
38022 (gen_struct_or_union_type_die): Likewise.
38023 (gen_subroutine_type_die): Likewise.
38024 (gen_typedef_die): Likewise.
38025 (base_type_cmp): Compare alignment attribute.
38027 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38030 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
38031 (setb_unsigned) New pattern for setb with CCUNS.
38032 * config/rs6000/rs6000.c (expand_block_compare): Use a different
38033 subfc./subfe sequence to avoid overflow problems. Generate a
38034 shorter sequence with cmpld/setb for power9.
38035 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
38036 for generating subfc. instruction.
38037 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
38038 now uses this instruction.
38040 2017-01-30 Ian Lance Taylor <iant@google.com>
38043 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
38044 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
38046 2017-01-30 Martin Sebor <msebor@redhat.com>
38048 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
38049 Move constant to the right of a relational operator.
38050 (get_mpfr_format_length, format_character, format_string): Ditto.
38051 (should_warn_p, maybe_warn): Same.
38053 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
38055 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
38058 * asan.c (get_translation_unit_decl): Remove function.
38059 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
38061 2017-01-30 Martin Liska <mliska@suse.cz>
38063 PR gcov-profile/79259
38064 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
38065 -fprofile-generate.
38067 2017-01-30 Martin Liska <mliska@suse.cz>
38070 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
38071 Initialize variables with NULL value.
38073 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
38076 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
38078 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
38080 2017-01-30 Richard Biener <rguenther@suse.de>
38082 PR tree-optimization/79276
38083 * tree-vrp.c (process_assert_insertions): Properly adjust common
38084 when removing a duplicate.
38086 2017-01-30 Richard Biener <rguenther@suse.de>
38088 PR tree-optimization/79256
38089 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38090 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38092 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
38094 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
38097 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
38098 ("*r<noxa>sbg_<mode>_sll_bitmask")
38099 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
38100 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
38101 Use contiguous_bitmask_nowrap_operand.
38103 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38106 * config/rs6000/altivec.h (vec_xl): Revise #define.
38107 (vec_xst): Likewise.
38109 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
38111 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
38113 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
38115 PR rtl-optimization/79194
38116 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
38117 traps before call to bypass_conditional_jumps.
38119 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38121 PR tree-optimization/71374
38122 * lra-constraints.c (check_conflict_input_operands): New.
38123 (match_reload): Use it.
38125 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38128 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
38129 account to calculate conflict_set.
38131 2017-01-27 Bin Cheng <bin.cheng@arm.com>
38133 PR rtl-optimization/78559
38134 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
38135 other_insn in combine.
38137 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
38139 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
38140 uint16_type_node for BT_UINT16.
38142 2017-01-27 David Malcolm <dmalcolm@redhat.com>
38144 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
38145 "RTL Tests" to menu.
38146 (GIMPLE Tests): New node.
38147 (RTL Tests): New node.
38149 2017-01-27 Richard Biener <rguenther@suse.de>
38151 PR tree-optimization/79245
38152 * tree-loop-distribution.c (distribute_loop): Apply cost
38153 modeling also to detected patterns.
38155 2017-01-27 Richard Biener <rguenther@suse.de>
38157 PR tree-optimization/71433
38158 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
38159 (compare_assert_loc): New function.
38160 (process_assert_insertions): Sort and optimize assert locations
38161 to remove duplicates and push down identical assertions on
38162 edges to their destination block.
38164 2017-01-27 Richard Biener <rguenther@suse.de>
38166 PR tree-optimization/79244
38167 * tree-vrp.c (remove_range_assertions): Forcefully propagate
38168 out SSA names even if abnormal.
38170 2017-01-27 Jakub Jelinek <jakub@redhat.com>
38172 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
38173 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
38174 instead of MPFR_RNDN.
38176 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
38179 * arm.c (arm_option_override): Don't call build_target_option_node
38180 until after doing all option overrides.
38181 (arm_valid_target_attribute_tree): Likewise.
38183 2017-01-27 Martin Liska <mliska@suse.cz>
38185 * doc/invoke.texi (-fprofile-arcs): Document profiling support
38186 for {cd}tors and C++ {cd}tors.
38188 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
38190 * config/s390/s390.md ("*setmem_long_and")
38191 ("*setmem_long_and_31z"): Use zero_extend instead of and.
38193 2017-01-26 Martin Sebor <msebor@redhat.com>
38195 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
38198 2017-01-26 Martin Sebor <msebor@redhat.com>
38200 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
38201 HAVE_DFmode before using XFmode or DFmode.
38202 (parse_directive): Avoid using the z length modifier to avoid
38203 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
38205 PR middle-end/78703
38206 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
38207 to accept adjustment as an array.
38208 (get_int_range): New function.
38209 (struct directive): Make width and prec arrays.
38210 (directive::set_width, directive::set_precision): Call get_int_range.
38211 (format_integer, format_floating): Handle width and precision ranges.
38212 (format_string, parse_directive): Same.
38214 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38217 * dwarf2out.c (generate_skeleton_bottom_up): For children with
38218 comdat_type_p set, just clone them, but keep the children in the
38222 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
38223 which have direct callers with -fvar-tracking-assignments enabled
38225 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
38226 inside of type units.
38228 2017-01-26 Martin Sebor <msebor@redhat.com>
38230 PR middle-end/78703
38231 * gimple-ssa-sprintf.c (struct result_range): Add likely and
38233 (struct format_result): Replace number_chars, number_chars_min,
38234 and number_chars_max with a single member of struct result_range.
38236 (format_result::operator+=): Adjust.
38237 (struct fmtresult): Remove bounded. Handle likely and unlikely
38239 (fmtresult::adjust_for_width_or_precision): New function.
38240 (fmtresult:type_max_digits): New function.
38241 (bytes_remaining): Handle likely and unlikely counters.
38242 (min_bytes_remaining): Remove.
38243 (format_percent): Simplify.
38244 (format_integer, format_floating): Set likely and unlikely counters.
38245 (get_string_length, format_character, format_string): Same.
38246 (format_plain, should_warn_p): New function.
38247 (maybe_warn): Call should_warn_p. Update diagnostic messages
38248 and handle those for all directives, including plain strings.
38249 (format_directive): Handle likely and unlikely counters.
38250 Remove unnecessary quoting from diagnostics. Add an informational
38252 (add_bytes): Remove.
38253 (pass_sprintf_length::compute_format_length): Simplify.
38254 (try_substitute_return_value): Handle likely and unlikely counters.
38256 2017-01-26 Carl Love <cel@us.ibm.com>
38258 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
38259 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
38261 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
38264 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
38265 endianess for subregs into account.
38266 * lra-constraints.c (lra_constraints): Do risky transformations
38267 always on the first iteration.
38268 * lra-lives.c (check_pseudos_live_through_calls): Add arg
38269 last_call_used_reg_set.
38270 (process_bb_lives): Define and use last_call_used_reg_set.
38271 * lra.c (lra): Always continue after lra_constraints on the first
38274 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
38276 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
38278 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
38280 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38282 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
38283 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
38284 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
38285 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
38286 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
38287 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
38288 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
38289 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
38290 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
38292 2017-01-26 Marek Polacek <polacek@redhat.com>
38295 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
38296 for the third operand.
38298 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38300 PR middle-end/79236
38301 * omp-low.c (struct omp_context): Add simt_stmt field.
38302 (scan_omp_for): Return omp_context *.
38303 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
38304 context to the _simt_ SIMD stmt.
38305 (lower_omp_for): For combined SIMD with sibling _simt_
38306 SIMD, make sure to use the same decls in _looptemp_
38307 clauses as in the sibling.
38309 2017-01-26 David Sherwood <david.sherwood@arm.com>
38311 PR middle-end/79212
38312 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
38315 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38318 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
38319 emit fld b; fld a; if possible.
38321 * brig-builtins.def: Update copyright years.
38322 * config/arm/arm_acle_builtins.def: Update copyright years.
38324 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
38327 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
38328 constraint instead of o for the stxsd instruction.
38330 2017-01-25 Carl Love <cel@us.ibm.com>
38332 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
38333 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
38335 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
38337 * doc/invoke.texi (C++ Dialect Options): Fix typo.
38339 2017-01-25 Richard Biener <rguenther@suse.de>
38341 PR tree-optimization/69264
38342 * target.def (vector_alignment_reachable): Improve documentation.
38343 * doc/tm.texi: Regenerate.
38344 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
38346 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
38347 earlier changes with respect to TYPE_USER_ALIGN.
38348 (vector_alignment_reachable_p): Likewise. Improve dumping.
38350 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38353 * config/arm/arm.md (xordi3): Force constant operand into a register
38356 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38358 * doc/invoke.texi (-fstore-merging): Correct default optimization
38359 levels at which it is enabled.
38360 (-O): Move -fstore-merging from list to...
38363 2017-01-25 Richard Biener <rguenther@suse.de>
38366 * omp-expand.c: Include debug.h.
38367 (expand_omp_taskreg): Make sure to generate early debug before
38368 outlining anything from a function.
38369 (expand_omp_target): Likewise.
38370 (grid_expand_target_grid_body): Likewise.
38372 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
38375 * asan.c (get_translation_unit_decl): New function.
38376 (asan_add_global): Extract modules file name from globals
38377 TRANSLATION_UNIT_DECL name.
38379 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
38382 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
38383 for long calls with APCS frame and VFP.
38385 2017-01-24 David Malcolm <dmalcolm@redhat.com>
38387 * cfg.c (original_copy_tables_initialized_p): New function.
38388 * cfg.h (original_copy_tables_initialized_p): New decl.
38389 * cfgrtl.c (relink_block_chain): Guard the call to
38390 free_original_copy_tables with a call to
38391 original_copy_tables_initialized_p.
38392 * cgraph.h (symtab_node::native_rtl_p): New decl.
38393 * cgraphunit.c (symtab_node::native_rtl_p): New function.
38394 (symtab_node::needed_p): Don't assert for early assembly output
38395 for __RTL functions.
38396 (cgraph_node::finalize_function): Set "force_output" for __RTL
38398 (cgraph_node::analyze): Bail out early for __RTL functions.
38399 (analyze_functions): Update assertion to support __RTL functions.
38400 (cgraph_node::expand): Bail out early for __RTL functions.
38401 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
38403 * function.h (struct function): Update comment for field
38405 * gimple-expr.c: Include "tree-pass.h".
38406 (gimple_has_body_p): Return false for __RTL functions.
38407 * Makefile.in (OBJS): Add run-rtl-passes.o.
38408 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
38410 (gcc::pass_manager::get_clean_slate): New accessor.
38411 * passes.c: Include "insn-addr.h".
38412 (should_skip_pass_p): Add logging. Update logic for running
38413 "expand" to be compatible with both __GIMPLE and __RTL. Guard
38414 property-provider override so it is only done for gimple passes.
38416 (skip_pass): New function.
38417 (execute_one_pass): Call skip_pass when skipping passes.
38418 * read-md.c (md_reader::read_char): Support filtering
38419 the input to a subset of line numbers.
38420 (md_reader::md_reader): Initialize fields
38421 m_first_line and m_last_line.
38422 (md_reader::read_file_fragment): New function.
38423 * read-md.h (md_reader::read_file_fragment): New decl.
38424 (md_reader::m_first_line): New field.
38425 (md_reader::m_last_line): New field.
38426 * read-rtl-function.c (function_reader::create_function): Only
38427 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
38428 curr_properties. Set DECL_INITIAL to a dummy block.
38429 (read_rtl_function_body_from_file_range): New function.
38430 * read-rtl-function.h (read_rtl_function_body_from_file_range):
38432 * run-rtl-passes.c: New file.
38433 * run-rtl-passes.h: New file.
38435 2017-01-24 Jeff Law <law@redhat.com>
38437 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
38440 2017-01-24 Bin Cheng <bin.cheng@arm.com>
38442 PR tree-optimization/79159
38443 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
38444 (record_nonwrapping_iv): Improve boundary using above function if no
38445 value range information.
38447 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
38448 Martin Jambor <mjambor@suse.cz>
38450 * brig-builtins.def: New file.
38451 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
38452 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
38453 (DEF_HSAIL_SAT_BUILTIN): Likewise.
38454 (DEF_HSAIL_INTR_BUILTIN): Likewise.
38455 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
38456 * builtin-types.def (BT_INT8): New.
38457 (BT_INT16): Likewise.
38458 (BT_UINT8): Likewise.
38459 (BT_UINT16): Likewise.
38460 (BT_FN_ULONG): Likewise.
38461 (BT_FN_UINT_INT): Likewise.
38462 (BT_FN_UINT_ULONG): Likewise.
38463 (BT_FN_UINT_LONG): Likewise.
38464 (BT_FN_UINT_PTR): Likewise.
38465 (BT_FN_ULONG_PTR): Likewise.
38466 (BT_FN_INT8_FLOAT): Likewise.
38467 (BT_FN_INT16_FLOAT): Likewise.
38468 (BT_FN_UINT32_FLOAT): Likewise.
38469 (BT_FN_UINT16_FLOAT): Likewise.
38470 (BT_FN_UINT8_FLOAT): Likewise.
38471 (BT_FN_UINT64_FLOAT): Likewise.
38472 (BT_FN_UINT16_UINT32): Likewise.
38473 (BT_FN_UINT32_UINT16): Likewise.
38474 (BT_FN_UINT16_UINT16_UINT16): Likewise.
38475 (BT_FN_INT_PTR_INT): Likewise.
38476 (BT_FN_UINT_PTR_UINT): Likewise.
38477 (BT_FN_LONG_PTR_LONG): Likewise.
38478 (BT_FN_ULONG_PTR_ULONG): Likewise.
38479 (BT_FN_VOID_UINT64_UINT64): Likewise.
38480 (BT_FN_UINT8_UINT8_UINT8): Likewise.
38481 (BT_FN_INT8_INT8_INT8): Likewise.
38482 (BT_FN_INT16_INT16_INT16): Likewise.
38483 (BT_FN_INT_INT_INT): Likewise.
38484 (BT_FN_UINT_FLOAT_UINT): Likewise.
38485 (BT_FN_FLOAT_UINT_UINT): Likewise.
38486 (BT_FN_ULONG_UINT_UINT): Likewise.
38487 (BT_FN_ULONG_UINT_PTR): Likewise.
38488 (BT_FN_ULONG_ULONG_ULONG): Likewise.
38489 (BT_FN_UINT_UINT_UINT): Likewise.
38490 (BT_FN_VOID_UINT_PTR): Likewise.
38491 (BT_FN_UINT_UINT_PTR: Likewise.
38492 (BT_FN_UINT32_UINT64_PTR): Likewise.
38493 (BT_FN_INT_INT_UINT_UINT): Likewise.
38494 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
38495 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
38496 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
38497 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
38498 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
38499 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
38500 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
38501 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
38502 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
38503 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
38504 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
38505 * doc/frontends.texi: List BRIG FE.
38506 * doc/install.texi (Testing): Add BRIG tesring requirements.
38507 * doc/invoke.texi (Overall Options): Mention BRIG.
38508 * doc/standards.texi (Standards): Doucment BRIG HSA version.
38510 2017-01-24 Richard Biener <rguenther@suse.de>
38512 PR translation/79208
38513 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
38515 2017-01-24 Martin Jambor <mjambor@suse.cz>
38518 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
38519 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
38520 and known_contexts.
38522 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
38524 PR middle-end/79123
38525 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
38526 casts from signed to unsigned really don't have a range.
38528 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
38530 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
38531 GMP_RNDx for compatiblity.
38533 2017-01-24 Martin Liska <mliska@suse.cz>
38536 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
38537 that would prevent us to call alloca with -1 as argument.
38539 2017-01-24 Jakub Jelinek <jakub@redhat.com>
38541 * dwarf2out.c (output_compilation_unit_header, output_file_names):
38542 Avoid -Wformat-security warning.
38544 2017-01-23 Andrew Pinski <apinski@cavium.com>
38546 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
38549 2017-01-23 Martin Sebor <msebor@redhat.com>
38551 PR middle-end/78703
38552 * gimple-ssa-sprintf.c (warn_level): New global.
38553 (format_integer): Use it here and throughout the rest of the file.
38554 Use the same switch to compute sign as base.
38555 (maybe_warn): New function.
38556 (format_directive): Factor out warnings into maybe_warn.
38557 Add debugging output. Use warn_level.
38558 (add_bytes): Use warn_level.
38559 (pass_sprintf_length::compute_format_length): Add debugging output.
38560 (try_substitute_return_value): Same.
38561 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
38563 PR middle-end/78703
38564 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
38565 (struct fmtresult, format_integer, format_floating): Adjust.
38566 (fmtresult::fmtresult): Set max correctly in two argument ctor.
38567 (get_string_length, format_string,format_directive): Same.
38568 (pass_sprintf_length::compute_format_length): Same.
38569 (try_substitute_return_value): Simplify slightly.
38571 PR middle-end/78703
38572 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
38573 (fmtresult::operator+=): Outlined.
38574 (struct fmtresult): Add ctors.
38575 (struct conversion_spec): Rename...
38576 (struct directive): ...to this. Add and remove data members.
38577 (directive::set_width, directive::set_precision): New functions.
38578 (format_percent): Use fmtresult ctor.
38579 (get_width_and_precision): Remove.
38580 (format_integer): Make naming changes. Avoid computing width and
38582 (format_floating): Same. Adjust indentation.
38583 (format_character, format_none): New functions.
38584 (format_string): Moved character handling to format_character.
38585 (format_directive): Remove arguments, change return type.
38586 (parse_directive): New function.
38587 (pass_sprintf_length::compute_format_length): Move directive
38588 parsing to parse_directive.
38590 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38592 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
38593 (assign_assembler_name_if_needed): ... this.
38594 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
38595 (assign_assembler_name_if_needed): ... this.
38596 (free_lang_data_in_cgraph): Adjust callers.
38597 * cgraphunit.c (cgraph_node::analyze): Likewise.
38598 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
38601 2017-01-23 Richard Biener <rguenther@suse.de>
38603 PR tree-optimization/79088
38604 PR tree-optimization/79188
38605 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
38606 resetting loop bounds after last path deletion. Reset loop
38607 bounds of the target loop, make code match the comments.
38608 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
38609 Make sure loops need no fixups.
38611 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
38613 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
38614 exponent support with double type for first argument.
38615 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
38616 type returned by __builtin_vec_extract_sig,
38617 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
38618 functions from "vector int" to "vector unsigned int" or from
38619 "vector long long int" to "vector unsigned long long int".
38620 Changed type returned by __builtin_vec_extract_exp,
38621 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
38622 functions from "vector int" to "vector unsigned int" or from
38623 "vector long long int" to "vector unsigned long long int".
38624 Changed return type of __builtin_vec_test_data_class,
38625 __builtin_vec_test_data_class_sp, and
38626 __builtin_vec_test_data_class_dp from "vector int" to
38627 "vector bool int" or from "vector long long int" to "vector bool
38628 long long int" and changed second argument type from "unsigned
38629 int" to "int". Added new overloaded function forms "vector float
38630 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
38631 "vector float __builtin_vec_insert_exp_sp (vector float, vector
38632 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
38633 double, vector unsigned long long int)" and "vector double
38634 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
38635 long int)". Changed return type of
38636 __builtin_scalar_test_data_class and
38637 __builtin_scalar_test_data_class_sp and
38638 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
38639 int" and changed second argument from "unsigned int" to "int".
38640 Changed type returned by __builtin_scalar_test_neg,
38641 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
38642 from "int" to "bool int". Added new overloaded function form
38643 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
38644 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
38645 exponent double-precision with floating point first argument.
38646 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
38647 documentation of scalar_test_data_class, scalar_test_neg,
38648 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
38649 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
38650 vec_test_data_class built-in functions to reflect refinements in
38651 their type signatures.
38653 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
38655 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
38657 (aarch64_elf_asm_destructor): Likewise.
38659 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
38661 PR rtl-optimization/78634
38662 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
38663 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
38664 * ifcvt.c (noce_try_cmove): Add missing cost check.
38666 PR rtl-optimization/71724
38667 * combine.c (if_then_else_cond): Look for situations where it is
38668 beneficial to undo the work of one of the recursive calls.
38670 2017-01-23 Bin Cheng <bin.cheng@arm.com>
38672 PR tree-optimization/70754
38673 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
38674 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
38675 combined stmt before it if not NULL.
38676 (combine_chains): Process refs reversely and compute dominance point
38679 2017-01-23 Martin Liska <mliska@suse.cz>
38681 PR tree-optimization/79196
38682 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
38683 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
38685 (strlen_optimize_stmt): Call the renamed function.
38687 2017-01-23 Michael Matz <matz@suse.de>
38689 PR tree-optimization/78384
38690 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
38692 2017-01-23 Richard Biener <rguenther@suse.de>
38694 PR tree-optimization/79186
38695 * tree-vrp.c (register_new_assert_for): Make sure we've seen
38696 both incoming edges before moving an assert.
38698 2017-01-23 Martin Jambor <mjambor@suse.cz>
38700 * ipa-prop.c (load_from_param_1): Removed.
38701 (load_from_unmodified_param): Bits from load_from_param_1 put back
38703 (load_from_param): Removed.
38704 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
38705 with stmt. Reverted back to use of load_from_unmodified_param.
38707 2017-01-23 Martin Jambor <mjambor@suse.cz>
38710 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
38711 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
38712 field a pointer to garbage collected vector, mark lattices and
38713 ipcp_orig_node with GTY((skip)).
38714 (ipa_get_param_count): Adjust to descriptors being a pointer.
38715 (ipa_get_param): Likewise.
38716 (ipa_get_type): Likewise.
38717 (ipa_get_param_move_cost): Likewise.
38718 (ipa_set_param_used): Likewise.
38719 (ipa_get_controlled_uses): Likewise.
38720 (ipa_set_controlled_uses): Likewise.
38721 (ipa_is_param_used): Likewise.
38722 (ipa_node_params_t): Move into garbage collector. New methods insert
38724 (ipa_node_params_sum): Annotate wth GTY(()).
38725 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
38727 (ipa_load_from_parm_agg): Adjust declaration.
38728 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
38729 * ipa-profile.c (ipa_profile): Likewise.
38730 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
38731 (ipa_populate_param_decls): Make descriptors parameter garbage
38733 (ipa_dump_param): Adjust to descriptors being a pointer.
38734 (ipa_alloc_node_params): Likewise.
38735 (ipa_initialize_node_params): Likewise.
38736 (load_from_param_1): Make descriptors parameter garbage collected.
38737 (load_from_unmodified_param): Likewise.
38738 (load_from_param): Likewise.
38739 (ipa_load_from_parm_agg): Likewise.
38740 (ipa_node_params::~ipa_node_params): Removed.
38741 (ipa_free_all_node_params): Remove call to delete operator.
38742 (ipa_node_params_t::insert): New.
38743 (ipa_node_params_t::remove): Likewise.
38744 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
38745 copy known_csts and known_contexts vectors.
38746 (ipa_read_node_info): Adjust to descriptors being a pointer.
38747 (ipcp_modif_dom_walker): Make m_descriptors field garbage
38749 (ipcp_transform_function): Make descriptors variable garbage
38752 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
38754 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
38755 * config/i386/avx512dqintrin.h: Ditto.
38756 * config/i386/avx512fintrin.h: Ditto.
38757 * config/i386/i386.c: Handle new builtins.
38758 * config/i386/i386-builtin.def: Add new builtins.
38759 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
38760 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
38762 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38763 Martin Liska <mliska@suse.cz>
38765 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
38766 * asan.c (asan_expand_poison_ifn): Support stores and use
38767 appropriate ASAN report function.
38768 * internal-fn.c (expand_ASAN_POISON_USE): New function.
38769 * internal-fn.def (ASAN_POISON_USE): Declare.
38770 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
38771 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
38772 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
38773 ASAN_POISON calls w/o LHS.
38774 * tree-ssa.c (execute_update_addresses_taken): Create clobber
38775 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
38776 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
38777 * gimplify.c (asan_poison_variables): Add attribute
38778 use_after_scope_memory to variables that really needs to live
38780 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
38781 having the attribute.
38783 2017-01-23 Martin Liska <mliska@suse.cz>
38785 * asan.c (create_asan_shadow_var): New function.
38786 (asan_expand_poison_ifn): Likewise.
38787 * asan.h (asan_expand_poison_ifn): New declaration.
38788 * internal-fn.c (expand_ASAN_POISON): Likewise.
38789 * internal-fn.def (ASAN_POISON): New builtin.
38790 * sanopt.c (pass_sanopt::execute): Expand
38791 asan_expand_poison_ifn.
38792 * tree-inline.c (copy_decl_for_dup_finish): Make function
38794 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
38795 * tree-ssa.c (is_asan_mark_p): New function.
38796 (execute_update_addresses_taken): Rewrite local variables
38797 (identified just by use-after-scope as addressable) into SSA.
38799 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38801 * doc/install.texi (Specific): opensource.apple.com uses https
38802 now. Remove trailing slash.
38804 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38806 * README.Portability: Remove note on an Irix compatibility issue.
38808 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
38810 * gcov.c (INCLUDE_ALGORITHM): Define.
38811 (INCLUDE_VECTOR): Define.
38812 No longer include <vector> and <algorithm> directly.
38814 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
38816 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
38818 * doc/invoke.texi (Code Gen Options): Ditto.
38820 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
38823 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
38825 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
38827 rtl-optimization/79125
38828 * cprop.c (local_cprop_pass): Handle cases where we make an
38829 unconditional trap.
38831 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
38835 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
38836 read from, for big endian.
38838 2017-01-20 Jiong Wang <jiong.wang@arm.com>
38840 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
38841 register pauth builtins for LP64 only.
38843 2017-01-20 Marek Polacek <polacek@redhat.com>
38846 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
38849 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38851 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
38853 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
38854 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
38855 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
38857 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38860 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
38861 in big-endian mode when they are not a single duplicated value.
38863 2017-01-20 Richard Biener <rguenther@suse.de>
38865 * BASE-VER: Bump to 7.0.1.
38867 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38869 * omp-low.c (omplow_simd_context): New struct. Use it...
38870 (lower_rec_simd_input_clauses): ...here and...
38871 (lower_rec_input_clauses): ...here to hold common data. Adjust all
38872 references to idx, lane, max_vf, is_simt.
38874 2017-01-20 Graham Markall <graham.markall@embecosm.com>
38876 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
38879 2017-01-20 Martin Jambor <mjambor@suse.cz>
38881 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
38882 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
38884 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
38885 (GTFILES): Rename hsa.c to hsa-common.c.
38886 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
38887 * hsa-dump.c: Likewise.
38888 * hsa-gen.c: Likewise.
38889 * hsa-regalloc.c: Likewise.
38890 * ipa-hsa.c: Likewise.
38891 * omp-expand.c: Likewise.
38892 * omp-low.c: Likewise.
38893 * toplev.c: Likewise.
38895 2017-01-20 Marek Polacek <polacek@redhat.com>
38898 * doc/invoke.texi: Document -Wduplicated-branches.
38899 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
38900 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
38901 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
38902 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
38903 return 0 only when not OEP_LEXICOGRAPHIC.
38904 (fold_build_cleanup_point_expr): Use the expression
38905 location when building CLEANUP_POINT_EXPR.
38906 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
38907 * tree.c (add_expr): Handle error_mark_node.
38909 2017-01-20 Martin Liska <mliska@suse.cz>
38912 * tree-profile.c (init_ic_make_global_vars): Do not call
38914 (gimple_init_gcov_profiler): Likewise.
38916 2017-01-20 Martin Liska <mliska@suse.cz>
38919 * cgraph.h (maybe_create_reference): Remove argument and
38921 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
38923 * ipa-cp.c (create_specialized_node): Likewise.
38924 * symtab.c (symtab_node::maybe_create_reference): Handle
38925 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
38927 2017-01-20 Martin Liska <mliska@suse.cz>
38929 * read-rtl-function.c (function_reader::create_function): Use
38930 build_decl instread of build_decl_stat.
38932 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
38934 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
38935 * config/i386/avx512dqintrin.h: Ditto.
38936 * config/i386/avx512fintrin.h: Ditto.
38937 * config/i386/i386-builtin-types.def: Add new types.
38938 * config/i386/i386.c: Handle new types.
38939 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
38940 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
38941 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
38942 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
38943 (__builtin_ia32_kshiftridi): New.
38944 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
38946 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
38950 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
38951 define to rs6000_init_stack_protect_guard.
38952 (rs6000_init_stack_protect_guard): New function.
38954 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
38955 Yunqiang Su <yunqiang.su@imgtec.com>
38957 * config.gcc (supported_defaults): Add madd4.
38958 (with_madd4): Add validation.
38959 (all_defaults): Add madd4.
38960 * config/mips/mips.opt (mmadd4): New option.
38961 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
38963 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
38965 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
38966 (ISA_HAS_FUSED_MADD4): Likewise.
38967 * doc/invoke.texi (-mmadd4): Document the new option.
38968 * doc/install.texi (--with-madd4): Document the new option.
38970 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38972 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
38973 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
38974 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
38975 (aarch64_init_pauth_hint_builtins): New.
38976 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
38977 (aarch64_expand_builtin): Expand new builtins.
38979 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38981 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
38982 * combine-stack-adj.c (no_unhandled_cfa): Handle
38983 REG_CFA_TOGGLE_RA_MANGLE.
38984 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
38985 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
38986 info for return address signing.
38987 (aarch64_expand_epilogue): Likewise.
38989 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38991 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
38992 * config/aarch64/aarch64-protos.h
38993 (aarch64_return_address_signing_enabled): New declaration.
38994 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
38996 (aarch64_expand_prologue): Sign return address before it's pushed onto
38998 (aarch64_expand_epilogue): Authenticate return address fetched from
39000 (aarch64_override_options): Sanity check for ILP32 and ISA level.
39001 (aarch64_attributes): New function attributes for "sign-return-address".
39002 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
39003 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
39004 ("*do_return"): Generate combined instructions according to key index.
39005 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
39006 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
39008 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
39009 * config/aarch64/aarch64.opt (msign-return-address=): New.
39010 * doc/extend.texi (AArch64 Function Attributes): Documents
39011 "sign-return-address=".
39012 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
39014 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39016 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
39017 overall option summary.
39019 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39021 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
39022 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
39023 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
39024 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
39026 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
39028 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
39029 -mpower9-minmax by default for -mcpu=power9.
39030 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
39031 128-bit floating point.
39033 2017-01-20 Alan Modra <amodra@gmail.com>
39035 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
39036 optimizing for size.
39038 2017-01-20 Alan Modra <amodra@gmail.com>
39041 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
39042 for strcmp and strncmp from corresponding builtin decl.
39044 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39046 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
39047 instead of i386/rtems-64.h.
39048 * config/i386/rtems-64.h: Remove.
39050 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39054 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
39056 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
39058 2017-01-19 Tamar Christina <tamar.christina@arm.com>
39060 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
39061 Change int to HOST_WIDE_INT.
39062 * config/aarch64/aarch64-protos.h
39063 (aarch64_simd_gen_const_vector_dup): Likewise.
39064 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
39066 2017-01-19 David Malcolm <dmalcolm@redhat.com>
39068 * langhooks-def.h (lhd_type_for_size): New decl.
39069 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
39070 * langhooks.c (lhd_type_for_size): New function, taken from
39073 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
39075 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
39076 define_bypass for CR latency.
39077 (power9-cracked-alu): Update bypass latency and remove power9-branch.
39078 (power9-alu2): Add define_bypass for CR latency.
39080 (power9-mul): Update insn latency.
39081 (power9-mul-compare): Update insn latency, bypass latency and remove
39084 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39086 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
39088 * config/aarch64/aarch64.md
39089 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
39090 aarch64_nopcrelative_literal_loads.
39091 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
39093 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
39095 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
39096 TARGET_LOONGSON_3A.
39097 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
39099 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
39102 * config.gcc (supported_defaults): Add lxc1-sxc1.
39103 (with_lxc1_sxc1): Add validation.
39104 (all_defaults): Add lxc1-sxc1.
39105 * config/mips/mips.opt (mlxc1-sxc1): New option.
39106 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39108 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39109 __mips_no_lxc1_sxc1.
39110 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
39111 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
39112 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
39114 2017-01-19 Richard Biener <rguenther@suse.de>
39116 PR tree-optimization/72488
39117 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
39118 sure to restore SSA info.
39119 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
39121 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
39123 PR rtl-optimization/79121
39124 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
39125 of the inner type when shifting an extended value.
39127 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39130 * symtab.c (symtab_node::equal_address_to): Fix comparing of
39131 interposable aliases.
39133 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
39136 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
39137 Use the evmergelohi instruction.
39138 (mov_si<mode>_e500_subreg4_2_le): Likewise.
39139 (mov_sitf_e500_subreg8_2_be): Likewise.
39140 (mov_sitf_e500_subreg12_2_le): Likewise.
39141 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
39142 (mov_si<mode>_e500_subreg4_2_be): Likewise.
39143 (mov_sitf_e500_subreg8_2_le): Likewise.
39144 (mov_sitf_e500_subreg12_2_be): Likewise.
39146 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39148 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
39149 attribute from vecsimple to vecperm.
39150 (altivec_vbpermq2): Likewise.
39152 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39155 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
39157 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39158 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
39159 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
39160 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
39161 case where N arg is SIZE_MAX.
39162 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
39163 (cmpstrsi): Add pattern.
39165 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
39167 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
39168 __builtin_vec_revb builtins.
39169 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
39170 built-in functions to support generation of the ISA 3.0 XXBR<x>
39171 vector byte reverse instructions.
39172 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
39173 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
39174 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
39175 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
39176 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
39177 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
39178 (P9V_BUILTIN_VEC_REVB): Likewise.
39179 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
39180 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
39181 (p9_xxbrq_v16qi): Likewise.
39182 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
39183 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
39184 (p9_xxbrh_v8hi): Likewise.
39185 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
39186 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
39187 vec_revb built-in functions.
39189 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
39191 PR rtl-optimization/78952
39192 * config/i386/i386.md (any_extract): New code iterator.
39193 (*insvqi_2): Use any_extract for source operand.
39194 (*insvqi_3): Use any_shiftrt for source operand.
39196 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
39198 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
39200 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
39202 2017-01-18 Matthias Klose <doko@ubuntu.com>
39204 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
39206 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39208 * config/rs6000/altivec.h (vec_bperm): Change #define.
39209 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
39210 (altivec_vbpermq2): New define_insn.
39211 (altivec_vbpermd): Likewise.
39212 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
39213 function interface.
39214 (VBPERMD): Likewise.
39215 (VBPERM): New polymorphic function interface.
39216 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
39217 Add entries for P9V_BUILTIN_VEC_VBPERM.
39218 * doc/extend.texi: Add interfaces for vec_bperm.
39220 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39222 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
39223 first letter of error messages.
39224 (s390_resolve_overloaded_builtin): Likewise.
39225 * config/s390/s390.c (s390_expand_builtin): Likewise.
39226 (s390_invalid_arg_for_unprototyped_fn): Likewise.
39227 (s390_valid_target_attribute_inner_p): Likewise.
39228 * config/s390/s390.md ("tabort"): Likewise.
39230 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
39232 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
39233 (ISA_AVOID_DIV_HILO): New macro.
39234 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
39235 (ISA_HAS_DDIV): Likewise.
39237 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39239 * doc/invoke.texi (fabi-version): Correct number of occurrences.
39241 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39243 * doc/invoke.texi (fabi-version): Spelling fix.
39245 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39248 * doc/invoke.texi (fabi-version): Mention mangling fix for
39251 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39254 * doc/invoke.texi (fabi-version): Document discriminator mangling.
39256 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
39259 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
39260 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
39262 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
39264 (stack_protect_test): Ditto.
39265 * config/rs6000/rs6000.opt (mstack-protector-guard=,
39266 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
39268 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
39269 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
39270 -mstack-protector-guard-offset=.
39271 (RS/6000 and PowerPC Options): Ditto.
39273 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39275 * config/i386/i386.h (MASK_CLASS_P): New define.
39276 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
39277 there are no registers from different register sets also when
39278 mask registers are used. Update function comment.
39279 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
39280 to (*k/*r) and (*k/*km) alternatives.
39282 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
39284 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
39285 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
39286 (EH_RETURN_HANDLER_RTX): New define.
39287 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
39288 Force frame pointer in EH return functions.
39289 (aarch64_expand_epilogue): Add barrier for eh_return.
39290 (aarch64_final_eh_return_addr): Remove.
39291 (aarch64_eh_return_handler_rtx): New function.
39292 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
39294 (aarch64_eh_return_handler_rtx): New prototype.
39296 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39298 * config/rs6000/altivec.h (vec_rlmi): New #define.
39299 (vec_vrlnm): Likewise.
39300 (vec_rlnm): Likewise.
39301 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
39302 (UNSPEC_VRLNM): Likewise.
39303 (VIlong): New mode iterator.
39304 (altivec_vrl<VI_char>mi): New define_insn.
39305 (altivec_vrl<VI_char>nm): Likewise.
39306 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
39308 (VRLDNM): Likewise.
39309 (RLNM): New polymorphic function entry.
39310 (VRLWMI): New monomorphic function entry.
39311 (VRLDMI): Likewise.
39312 (RLMI): New polymorphic function entry.
39313 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
39314 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
39315 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
39318 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39321 * dwarf2out.c (field_byte_offset): Restore the
39322 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
39323 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
39324 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
39327 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39330 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
39332 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39335 * dwarf2out.c (add_data_member_location_attribute): For constant
39336 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
39337 instead of DW_AT_data_member_location, DW_AT_bit_offset and
39338 DW_AT_byte_size attributes.
39340 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39342 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
39343 after forcing to constant memory when the code model is medium.
39345 2017-01-17 Julia Koval <julia.koval@intel.com>
39348 * config/i386/avx512fintrin.h
39349 (_mm512_i32gather_ps): Change __addr type to void const*.
39350 (_mm512_mask_i32gather_ps): Ditto.
39351 (_mm512_i32gather_pd): Ditto.
39352 (_mm512_mask_i32gather_pd): Ditto.
39353 (_mm512_i64gather_ps): Ditto.
39354 (_mm512_mask_i64gather_ps): Ditto.
39355 (_mm512_i64gather_pd): Ditto.
39356 (_mm512_mask_i64gather_pd): Ditto.
39357 (_mm512_i32gather_epi32): Ditto.
39358 (_mm512_mask_i32gather_epi32): Ditto.
39359 (_mm512_i32gather_epi64): Ditto.
39360 (_mm512_mask_i32gather_epi64): Ditto.
39361 (_mm512_i64gather_epi32): Ditto.
39362 (_mm512_mask_i64gather_epi32): Ditto.
39363 (_mm512_i64gather_epi64): Ditto.
39364 (_mm512_mask_i64gather_epi64): Ditto.
39365 (_mm512_i32scatter_ps): Change __addr type to void*.
39366 (_mm512_mask_i32scatter_ps): Ditto.
39367 (_mm512_i32scatter_pd): Ditto.
39368 (_mm512_mask_i32scatter_pd): Ditto.
39369 (_mm512_i64scatter_ps): Ditto.
39370 (_mm512_mask_i64scatter_ps): Ditto.
39371 (_mm512_i64scatter_pd): Ditto.
39372 (_mm512_mask_i64scatter_pd): Ditto.
39373 (_mm512_i32scatter_epi32): Ditto.
39374 (_mm512_mask_i32scatter_epi32): Ditto.
39375 (_mm512_i32scatter_epi64): Ditto.
39376 (_mm512_mask_i32scatter_epi64): Ditto.
39377 (_mm512_i64scatter_epi32): Ditto.
39378 (_mm512_mask_i64scatter_epi32): Ditto.
39379 (_mm512_i64scatter_epi64): Ditto.
39380 (_mm512_mask_i64scatter_epi64): Ditto.
39381 * config/i386/avx512pfintrin.h
39382 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
39383 (_mm512_mask_prefetch_i32gather_ps): Ditto.
39384 (_mm512_mask_prefetch_i64gather_pd): Ditto.
39385 (_mm512_mask_prefetch_i64gather_ps): Ditto.
39386 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
39387 (_mm512_prefetch_i32scatter_ps): Ditto.
39388 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
39389 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
39390 (_mm512_prefetch_i64scatter_pd): Ditto.
39391 (_mm512_prefetch_i64scatter_ps): Ditto.
39392 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
39393 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
39394 * config/i386/avx512vlintrin.h
39395 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
39396 (_mm_mmask_i32gather_ps): Ditto.
39397 (_mm256_mmask_i32gather_pd): Ditto.
39398 (_mm_mmask_i32gather_pd): Ditto.
39399 (_mm256_mmask_i64gather_ps): Ditto.
39400 (_mm_mmask_i64gather_ps): Ditto.
39401 (_mm256_mmask_i64gather_pd): Ditto.
39402 (_mm_mmask_i64gather_pd): Ditto.
39403 (_mm256_mmask_i32gather_epi32): Ditto.
39404 (_mm_mmask_i32gather_epi32): Ditto.
39405 (_mm256_mmask_i32gather_epi64): Ditto.
39406 (_mm_mmask_i32gather_epi64): Ditto.
39407 (_mm256_mmask_i64gather_epi32): Ditto.
39408 (_mm_mmask_i64gather_epi32): Ditto.
39409 (_mm256_mmask_i64gather_epi64): Ditto.
39410 (_mm_mmask_i64gather_epi64): Ditto.
39411 (_mm256_i32scatter_ps): Change __addr type to void*.
39412 (_mm256_mask_i32scatter_ps): Ditto.
39413 (_mm_i32scatter_ps): Ditto.
39414 (_mm_mask_i32scatter_ps): Ditto.
39415 (_mm256_i32scatter_pd): Ditto.
39416 (_mm256_mask_i32scatter_pd): Ditto.
39417 (_mm_i32scatter_pd): Ditto.
39418 (_mm_mask_i32scatter_pd): Ditto.
39419 (_mm256_i64scatter_ps): Ditto.
39420 (_mm256_mask_i64scatter_ps): Ditto.
39421 (_mm_i64scatter_ps): Ditto.
39422 (_mm_mask_i64scatter_ps): Ditto.
39423 (_mm256_i64scatter_pd): Ditto.
39424 (_mm256_mask_i64scatter_pd): Ditto.
39425 (_mm_i64scatter_pd): Ditto.
39426 (_mm_mask_i64scatter_pd): Ditto.
39427 (_mm256_i32scatter_epi32): Ditto.
39428 (_mm256_mask_i32scatter_epi32): Ditto.
39429 (_mm_i32scatter_epi32): Ditto.
39430 (_mm_mask_i32scatter_epi32): Ditto.
39431 (_mm256_i32scatter_epi64): Ditto.
39432 (_mm256_mask_i32scatter_epi64): Ditto.
39433 (_mm_i32scatter_epi64): Ditto.
39434 (_mm_mask_i32scatter_epi64): Ditto.
39435 (_mm256_i64scatter_epi32): Ditto.
39436 (_mm256_mask_i64scatter_epi32): Ditto.
39437 (_mm_i64scatter_epi32): Ditto.
39438 (_mm_mask_i64scatter_epi32): Ditto.
39439 (_mm256_i64scatter_epi64): Ditto.
39440 (_mm256_mask_i64scatter_epi64): Ditto.
39441 (_mm_i64scatter_epi64): Ditto.
39442 (_mm_mask_i64scatter_epi64): Ditto.
39443 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
39444 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
39445 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
39446 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
39447 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
39448 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
39449 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
39450 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
39451 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
39452 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
39453 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
39454 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
39455 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
39456 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
39457 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
39458 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
39459 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
39460 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
39461 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
39462 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
39463 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
39464 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
39465 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
39466 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
39467 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
39468 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
39469 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
39470 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
39471 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
39472 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
39473 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
39474 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
39475 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
39476 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
39477 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
39478 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
39479 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
39480 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
39481 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
39482 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
39483 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
39484 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
39485 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
39486 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
39487 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
39488 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
39489 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
39490 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
39491 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
39492 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
39493 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
39494 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
39495 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
39496 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
39497 definitions accordingly.
39499 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
39500 Kuan-Lin Chen <kuanlinchentw@gmail.com>
39503 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
39506 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
39509 * ira-conflicts.c (ira_build_conflicts): Update total conflict
39510 hard regs for inner regno.
39512 2017-01-17 Martin Liska <mliska@suse.cz>
39515 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
39516 assumption and add comment.
39518 2017-01-17 Nathan Sidwell <nathan@acm.org>
39520 * ipa-visibility.c (localize_node): New function, broken out of ...
39521 (function_and_variable_visibility): ... here. Call it.
39523 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39525 PR middle-end/77445
39526 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
39527 correctly set frequency of oudgoing edge.
39528 (duplicate_thread_path): Fix profile updating.
39530 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39533 * configure.ac: Add GCC_BASE_VER.
39534 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
39535 version from BASE-VER file.
39536 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
39537 (gcc.o): Depend on $(BASEVER).
39538 * common.opt (dumpfullversion): New option.
39539 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
39540 * doc/invoke.texi: Document -dumpfullversion.
39541 * doc/install.texi: Document --with-gcc-major-version-only.
39542 * configure: Regenerated.
39544 2017-01-17 Richard Biener <rguenther@suse.de>
39546 PR tree-optimization/71433
39547 * tree-vrp.c (register_new_assert_for): Merge same asserts
39548 on all incoming edges.
39549 (process_assert_insertions_for): Handle insertions at the
39552 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
39554 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
39555 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
39557 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
39560 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
39563 2017-01-17 Alan Modra <amodra@gmail.com>
39566 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
39567 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
39568 symbolic stack limit when pic.
39570 2017-01-16 Martin Sebor <msebor@redhat.com>
39572 PR tree-optimization/78608
39573 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
39575 2017-01-16 Jeff Law <law@redhat.com>
39578 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
39579 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
39580 for several include directories that may be relative to sysroot.
39581 * config/i386/x-mingw32 (gplus_includedir): Define.
39582 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
39583 (native_system_includedir): Likewise.
39584 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
39585 override if TARGET_SYSTEM_ROOT is defined.
39586 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
39588 PR tree-optimization/79090
39589 PR tree-optimization/33562
39590 PR tree-optimization/61912
39591 PR tree-optimization/77485
39592 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
39593 and computed trims into the dump file.
39595 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39597 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
39599 2017-01-16 Jakub Jelinek <jakub@redhat.com>
39602 * gimplify.c (gimplify_init_constructor): If want_value and
39603 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
39607 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
39608 sequence. Formatting fixes.
39609 (doloop_optimize): Formatting fixes.
39612 * gcc.c (debug_level_greater_than_spec_func): New function.
39613 (static_spec_functions): Add debug-level-gt spec function.
39614 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
39616 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39617 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39618 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
39619 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
39620 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
39621 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
39623 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
39625 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
39626 QImode fixups to general and mask registers only.
39628 2017-01-16 Carl Love <cel@us.ibm.com>
39630 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
39631 for built-in functions
39632 vector signed char vec_nabs (vector signed char)
39633 vector signed short vec_nabs (vector signed short)
39634 vector signed int vec_nabs (vector signed int)
39635 vector signed long long vec_nabs (vector signed long long)
39636 vector float vec_nabs (vector float)
39637 vector double vec_nabs (vector double)
39638 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
39640 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
39641 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
39642 * doc/extend.texi: Update the documentation file for the new built-in
39645 2017-01-16 Martin Sebor <msebor@redhat.com>
39647 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
39650 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39652 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
39653 UNSPEC_VSX__XXSPLTD to require special splat handling.
39655 2017-01-16 David Malcolm <dmalcolm@redhat.com>
39658 * system.h: Poison strndup.
39660 2017-01-16 Alan Modra <amodra@gmail.com>
39663 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
39666 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
39668 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
39670 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
39672 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
39673 call recog here. Assert that INSN_CODE (insn) is non-negative.
39675 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39678 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
39680 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
39681 in the currently scheduled RTL fragment.
39683 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39685 PR rtl-optimization/78751
39686 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
39689 2017-01-14 Jeff Law <law@redhat.com>
39691 PR tree-optimization/79090
39692 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
39693 variable length stores.
39694 (compute_trims): Delete dead assignment to *trim_tail.
39695 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
39698 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
39700 PR rtl-optimization/78626
39701 PR rtl-optimization/78727
39702 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
39703 of a block, and split such blocks after everything else is finished.
39705 2017-01-14 Alan Modra <amodra@gmail.com>
39708 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
39709 target legitimate_combined_insn.
39710 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
39711 (rs6000_legitimate_combined_insn): New function.
39712 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
39714 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
39715 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
39716 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
39718 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
39720 * doc/frontends.texi (G++ and GCC): Remove references to Java.
39722 2017-01-13 Jeff Law <law@redhat.com>
39724 PR tree-optimization/33562
39725 PR tree-optimization/61912
39726 PR tree-optimization/77485
39727 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
39729 (delete_dead_assignment): Likewise.
39730 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
39731 statement to delete_dead_call and delete_dead_assignment.
39733 2017-01-13 David Malcolm <dmalcolm@redhat.com>
39736 * substring-locations.c (format_warning_va): Strengthen case 1 so
39737 that both endpoints of the substring must be within the format
39738 range for just the substring to be printed.
39740 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
39742 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
39743 * config/i386/i386.c (ix86_target_string): Add missing options
39744 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
39745 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
39746 flags_other and ix86_target_other to flags2_other. Display unknown
39748 (ix86_valid_target_attribute_inner_p): Add missing options and
39749 reorder options by implied ISAs, as in ix86_target_string.
39751 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39753 * hash-table.h (hash_table::too_empty_p): New function.
39754 (hash_table::expand): Use it.
39755 (hash_table::traverse): Likewise.
39756 (hash_table::empty_slot): Use sizeof (value_type) instead of
39757 sizeof (PTR) to convert bytes to elements. Shrink the table
39758 if the current size is excessive for the current number of
39761 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39763 * ira-costs.c (record_reg_classes): Break from the inner loop
39764 early once alt_fail is known to be true. Update outer loop
39765 handling accordingly.
39767 2017-01-13 Jeff Law <law@redhat.com>
39769 * tree-ssa-dse.c (decrement_count): New function.
39770 (increment_start_addr, maybe_trim_memstar_call): Likewise.
39771 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
39772 when we know the partially dead statement is a mem* function.
39774 PR tree-optimization/61912
39775 PR tree-optimization/77485
39776 * tree-ssa-dse.c: Include expr.h.
39777 (maybe_trim_constructor_store): New function.
39778 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
39780 PR tree-optimization/33562
39781 PR tree-optimization/61912
39782 PR tree-optimization/77485
39783 * doc/invoke.texi: Document new dse-max-object-size param.
39784 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
39785 * tree-ssa-dse.c: Include params.h.
39786 (dse_store_status): New enum.
39787 (initialize_ao_ref_for_dse): New, partially extracted from
39789 (valid_ao_ref_for_dse, normalize_ref): New.
39790 (setup_live_bytes_from_ref, compute_trims): Likewise.
39791 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
39792 (maybe_trim_partially_dead_store): Likewise.
39793 (maybe_trim_complex_store): Likewise.
39794 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
39795 Track what bytes live from the original store. Return tri-state
39796 for dead, partially dead or live.
39797 (dse_dom_walker): Add constructor, destructor and new private members.
39798 (delete_dead_call, delete_dead_assignment): New extracted from
39800 (dse_optimize_stmt): Make a member of dse_dom_walker.
39801 Use initialize_ao_ref_for_dse.
39803 PR tree-optimization/33562
39804 PR tree-optimization/61912
39805 PR tree-optimization/77485
39806 * sbitmap.h (bitmap_count_bits): Prototype.
39807 (bitmap_clear_range, bitmap_set_range): Likewise.
39808 * sbitmap.c (bitmap_clear_range): New function.
39809 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
39811 2017-01-13 Martin Liska <mliska@suse.cz>
39814 * function.c (set_cfun): Add new argument force.
39815 * function.h (set_cfun): Likewise.
39816 * ipa-inline-transform.c (inline_call): Use the function when
39817 strict alising from is dropped for function we inline to.
39819 2017-01-13 Richard Biener <rguenther@suse.de>
39821 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
39822 for dumping GIMPLE INTEGER_CSTs.
39824 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
39826 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
39827 to 201112L since C++17.
39829 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
39832 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
39833 if -fsanitize=kernel-address is present.
39835 2017-01-13 Richard Biener <rguenther@suse.de>
39837 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
39838 as _Literal ( type ) number in case usual suffixes do not
39839 preserve all information.
39841 2017-01-13 Richard Biener <rguenther@suse.de>
39843 PR tree-optimization/77283
39844 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
39845 and ssa-iterators.h.
39846 (is_feasible_trace): Implement a cost model based on joiner
39849 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
39852 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
39853 char or short to __float128/_Float128 directly.
39855 2017-01-12 Martin Sebor <msebor@redhat.com>
39857 to -Wformat-overflow.
39858 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
39859 (min_bytes_remaining): Same.
39860 (get_string_length): Same.
39861 (format_string): Same.
39862 (format_directive): Same.
39864 (pass_sprintf_length::handle_gimple_call): Same.
39866 2017-01-12 Jakub Jelinek <jakub@redhat.com>
39868 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
39869 info.nowrite calls with no lhs that can't throw. Return bool
39870 whether gsi_remove has been called or not.
39871 (pass_sprintf_length::handle_gimple_call): Return bool whether
39872 try_substitute_return_value called gsi_remove. Formatting fix.
39873 (pass_sprintf_length::execute): Don't use gsi_remove if
39874 handle_gimple_call returned true.
39877 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
39878 be removed due to side-effects, don't remove following barrier nor
39879 turn the successor edge into fallthru edge.
39881 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39884 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
39885 element-reversing loads and stores as not swappable.
39887 2017-01-12 Nathan Sidwell <nathan@acm.org>
39888 Nicolai Stange <nicstange@gmail.com>
39890 * combine.c (try_combine): Don't ignore result of overlap checking
39891 loop. Combine overlap & asm check into single loop.
39893 2017-01-12 Richard Biener <rguenther@suse.de>
39895 * tree-pretty-print.c (dump_generic_node): Provide -gimple
39896 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
39898 2017-01-12 Richard Biener <rguenther@suse.de>
39900 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
39901 and TS_TARGET_OPTION directly derive from TS_BASE.
39902 * tree-core.h (tree_optimization_option): Derive from tree_base.
39903 (tree_target_option): Likewise.
39905 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
39907 * config/i386/i386.c (memory_address_length): Increase len
39908 only when rip_relative_addr_p returns false.
39910 2017-01-11 Julia Koval <julia.koval@intel.com>
39912 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
39913 (OPTION_MASK_ISA_SGX_SET): New.
39914 (ix86_handle_option): Handle OPT_msgx.
39915 * config.gcc: Added sgxintrin.h.
39916 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
39917 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
39918 * config/i386/i386.c (ix86_target_string): Add -msgx.
39920 (ix86_option_override_internal): Handle new options.
39921 (ix86_valid_target_attribute_inner_p): Add sgx.
39922 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
39923 * config/i386/i386.opt: Add msgx.
39924 * config/i386/sgxintrin.h: New file.
39925 * config/i386/x86intrin.h: Add sgxintrin.h.
39927 2017-01-11 Jakub Jelinek <jakub@redhat.com>
39930 * fold-const.c (maybe_nonzero_address): Return 1 for function
39932 (tree_single_nonzero_warnv_p): Don't handle function local objects
39936 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
39939 2017-01-11 David Malcolm <dmalcolm@redhat.com>
39942 * opts.c: Include "spellcheck.h"
39943 (struct string_fragment): New struct.
39944 (struct edit_distance_traits<const string_fragment &>): New
39946 (get_closest_sanitizer_option): New function.
39947 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
39949 2017-01-11 Jakub Jelinek <jakub@redhat.com>
39951 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
39953 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
39954 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
39955 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
39956 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
39957 for initial die_offset if dwarf_split_debug_info.
39958 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
39959 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
39961 (output_skeleton_debug_sections): Formatting fix. Use
39962 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
39963 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
39965 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
39967 * config/arm/cortex-a53.md: Add bypasses for
39968 cortex_a53_r2f_cvt.
39969 (cortex_a53_r2f): Only use for transfers.
39970 (cortex_a53_f2r): Likewise.
39971 (cortex_a53_r2f_cvt): Add reservation for conversions.
39972 (cortex_a53_f2r_cvt): Likewise.
39974 2017-01-11 Tamar Christina <tamar.christina@arm.com>
39976 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
39977 to all inlined functions, change static to extern.
39979 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
39982 * config/arm/arm.c (legitimize_pic_address): Handle reference to
39984 (arm_assemble_integer): Likewise.
39986 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
39988 * config.gcc: Use new awk script to check CPU, FPU and architecture
39989 parameters for --with-... options.
39990 * config/arm/parsecpu.awk: New file
39991 * config/arm/arm-cpus.in: New file.
39992 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
39994 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
39996 * config/arm/t-arm: Update dependency rules.
39997 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
39998 of processing .def files.
39999 * config/arm/genopt.sh: Deleted.
40000 * config/arm/gentune.sh: Deleted.
40001 * config/arm/arm-cores.def: Deleted.
40002 * config/arm/arm-arches.def: Deleted.
40003 * config/arm/arm-fpus.def: Deleted.
40004 * config/arm/arm-tune.md: Regenerated.
40005 * config/arm/arm-tables.opt: Regenerated.
40006 * config/arm/arm-cpu.h: New generated file.
40007 * config/arm/arm-cpu-data.h: New generated file.
40008 * config/arm/arm-cpu-cdata.h: New generated file.
40010 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
40013 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
40015 (input_varpool_node): Unpack dynamically_initialized bit.
40017 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
40019 PR rtl-optimization/79032
40020 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
40021 the alignment of the adjusted memory reference against that of MODE,
40022 instead of the alignment of the original memory reference.
40024 2017-01-11 Martin Jambor <mjambor@suse.cz>
40026 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
40028 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
40029 decorated functions.
40031 2017-01-11 Richard Biener <rguenther@suse.de>
40033 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
40034 set range/nonnull info for PHI results. Do not set it on
40035 stmts marked for removal.
40037 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
40039 * expr.c (store_field): In the bitfield case, fetch the return value
40040 from the registers before applying a single big-endian adjustment.
40041 Always do a final load for a BLKmode value not larger than a word.
40043 2017-01-10 David Malcolm <dmalcolm@redhat.com>
40046 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40047 that we correctly handle column numbers greater than
40048 LINE_MAP_MAX_COLUMN_NUMBER.
40050 2017-01-10 Martin Sebor <msebor@redhat.com>
40052 PR middle-end/78245
40053 * gimple-ssa-sprintf.c (get_destination_size): Call
40054 {init,fini}object_sizes.
40055 * tree-object-size.c (addr_object_size): Adjust.
40056 (pass_through_call): Adjust.
40057 (pass_object_sizes::execute): Adjust.
40058 * tree-object-size.h (fini_object_sizes): Declare.
40060 2017-01-10 Martin Sebor <msebor@redhat.com>
40062 PR tree-optimization/78775
40063 * builtins.c (get_size_range): Move...
40064 * calls.c: ...to here.
40065 (alloc_max_size): Accept zero argument.
40066 (operand_signed_p): Remove.
40067 (maybe_warn_alloc_args_overflow): Call get_size_range.
40068 * calls.h (get_size_range): Declare.
40070 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
40072 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
40073 from TI's devices.csv file as of September 2016.
40074 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
40076 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
40078 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
40079 * doc/invoke.texi: Likewise.
40080 * doc/md.texi: Likewise.
40081 * doc/objc.texi: Likewise.
40083 2017-01-10 Joshua Conner <joshconner@google.com>
40085 * config/arm/fuchsia-elf.h: New file.
40086 * config/fuchsia.h: New file.
40087 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
40088 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
40090 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
40092 2016-01-10 Richard Biener <rguenther@suse.de>
40094 PR tree-optimization/79034
40095 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
40096 Propagate out degenerate PHIs in the joiner.
40098 2017-01-10 Martin Liska <mliska@suse.cz>
40100 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
40101 (sort_congruence_classes_by_decl_uid): Likewise.
40102 (sort_congruence_class_groups_by_decl_uid): Likewise.
40103 (sem_item_optimizer::merge_classes): Sort class, groups in these
40104 classes and members in the groups by DECL_UID of declarations.
40105 This would make merge operations stable.
40107 2017-01-10 Martin Liska <mliska@suse.cz>
40109 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
40110 usage of m_classes_vec.
40111 (sem_item_optimizer::~sem_item_optimizer): Likewise.
40112 (sem_item_optimizer::get_group_by_hash): Likewise.
40113 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
40114 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
40115 (sem_item_optimizer::verify_classes): Likewise.
40116 (sem_item_optimizer::process_cong_reduction): Likewise.
40117 (sem_item_optimizer::dump_cong_classes): Likewise.
40118 (sem_item_optimizer::merge_classes): Likewise.
40119 * ipa-icf.h (congruence_class_hash): Rename from
40120 congruence_class_group_hash. Remove declaration of m_classes_vec.
40122 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
40124 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
40125 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
40126 * config.gcc: Add avx512vpopcntdqintrin.h.
40127 * config/i386/avx512vpopcntdqintrin.h: New.
40128 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
40129 * config/i386/i386-builtin-types.def: Add new types.
40130 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
40131 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
40132 __builtin_ia32_vpopcountq_v8di_mask): New.
40133 * config/i386/i386-c.c (ix86_target_macros_internal): Define
40134 __AVX512VPOPCNTDQ__.
40135 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
40136 (PTA_AVX512VPOPCNTDQ): Define.
40137 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
40138 TARGET_AVX512VPOPCNTDQ_P): Define.
40139 * config/i386/i386.opt: Add mavx512vpopcntdq.
40140 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
40141 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
40143 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40145 PR middle-end/77484
40146 * predict.def (PRED_CALL): Set to 67.
40148 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
40150 * expr.c (store_field): In the bitfield case, if the value comes from
40151 a function call and is of an aggregate type returned in registers, do
40152 not modify the field mode; extract the value in all cases if the mode
40153 is BLKmode and the size is not larger than a word.
40155 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
40158 * config/i386/cpuid.h: Fix undefined behavior.
40160 2017-01-04 Jeff Law <law@redhat.com>
40162 PR tree-optimization/79007
40163 PR tree-optimization/67955
40164 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
40165 conservative for pt.null when flag_non_call_exceptions is on.
40167 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40169 PR translation/79019
40170 PR translation/79020
40171 * params.def (PARAM_INLINE_MIN_SPEEDUP,
40172 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
40173 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
40175 * config/avr/avr.opt (maccumulate-args): Likewise.
40176 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
40177 * common.opt (freport-bug): Likewise.
40178 * cif-code.def (CIF_FINAL_ERROR): Likewise.
40179 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
40180 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
40181 translatable string.
40182 * config/i386/i386.c (function_value_32): Likewise.
40183 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
40184 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
40186 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
40187 * common/config/msp430/msp430-common.c (msp430_handle_option):
40189 * symtab.c (symtab_node::verify_base): Likewise.
40190 * opts.c (set_debug_level): Likewise.
40191 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
40192 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
40193 missing whitespace to translatable strings.
40194 * config/avr/avr.md (bswapsi2): Fix typo in comment.
40195 * config/sh/superh.h: Likewise.
40196 * config/i386/xopintrin.h: Likewise.
40197 * config/i386/znver1.md: Likewise.
40198 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
40199 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
40200 * double-int.h (struct double_int): Likewise.
40201 * double-int.c (div_and_round_double): Likewise.
40202 * wide-int.cc: Likewise.
40203 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
40204 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
40205 * cfgcleanup.c (crossjumps_occured): Renamed to ...
40206 (crossjumps_occurred): ... this.
40207 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
40210 PR tree-optimization/78899
40211 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
40212 returning bool return struct loop *, NULL for failure and the new
40214 (versionable_outer_loop_p): Don't version outer loop if it has
40215 dont_vectorized bit set.
40216 (tree_if_conversion): When versioning outer loop, ensure
40217 tree_if_conversion is performed also on the inner loop of the
40218 non-vectorizable outer loop copy.
40219 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
40220 LOOP_VECTORIZED in inner loop of the scalar outer loop and
40221 prevent vectorization of it.
40222 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
40223 the outer loop vectorization of the non-scalar version is attempted
40224 before vectorization of the inner loop in scalar version. If
40225 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
40226 vectorization of its inner loop.
40227 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
40228 has 2 inner loops, rename also on edges from bb whose single pred
40229 is outer_loop->header. Fix typo in function comment.
40231 2017-01-09 Martin Sebor <msebor@redhat.com>
40234 * asan.c (asan_emit_stack_protection): Increase local buffer size
40235 to avoid snprintf truncation warning.
40237 2017-01-09 Andrew Pinski <apinski@cavium.com>
40239 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
40240 to reference thunderx2t99 for the tuning structure
40241 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
40243 (thunderx2t99_extra_costs): This.
40244 * config/aarch64/aarch64-tune.md: Regenerate.
40245 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
40246 (vulcan_addrcost_table): This.
40247 (vulcan_regmove_cost): Rename to ...
40248 (thunderx2t99_regmove_cost): This.
40249 (vulcan_vector_cost): Rename to ...
40250 (thunderx2t99_vector_cost): this.
40251 (vulcan_branch_cost): Rename to ...
40252 (thunderx2t99_branch_cost): This.
40253 (vulcan_tunings): Rename to ...
40254 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
40255 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
40257 2017-01-09 Martin Jambor <mjambor@suse.cz>
40261 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
40262 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
40263 (propagate_vr_accross_jump_function): Use the above function for all
40264 value range computations for pass-through jump functions and type
40265 converasion from explicit value range values.
40266 (ipcp_propagate_stage): Do not attempt to deduce types of formal
40267 parameters from TYPE_ARG_TYPES.
40268 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
40269 (ipa_write_node_info): Stream type of the actual argument.
40270 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
40272 2017-01-09 Martin Liska <mliska@suse.cz>
40275 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
40276 (lookup_compiler): Do not show error message with have_E.
40278 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40280 PR tree-optimization/78938
40281 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
40282 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
40283 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
40284 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
40287 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40289 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
40292 2017-01-09 Richard Biener <rguenther@suse.de>
40294 PR tree-optimization/78997
40295 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
40296 name condition properly.
40298 2017-01-09 Richard Biener <rguenther@suse.de>
40301 * dwarf2out.c (is_cxx): New overload with context.
40302 (is_naming_typedef_decl): Use it.
40304 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40306 * invoke.texi (Option Summary): Correct spacing in option lists
40307 and add line breaks to fix over-long lines.
40309 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40311 PR middle-end/17660
40313 * extend.texi (Common Variable Attributes): Add xref to GCC
40314 Internals manual to explain mode attribute keywords.
40316 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40319 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
40320 and Preprocessor Options.
40321 (Options for Linking): Document -pthread here....
40322 (RS/6000 and PowerPC Options): ...not here.
40323 (Solaris 2 Options): ...or here.
40324 * doc/cppopts.texi: Document -pthread.
40326 2017-01-08 Martin Sebor <msebor@redhat.com>
40328 PR middle-end/77708
40329 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
40330 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
40331 New member functions.
40332 (format_directive): Used them.
40334 (pass_sprintf_length::handle_gimple_call): Same.
40335 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
40336 to avoid truncation for any argument.
40337 (extract_affine_mul): Same.
40338 * tree.c (get_file_function_name): Same.
40340 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40342 PR middle-end/77484
40343 * predict.def (PRED_INDIR_CALL): Set to 86.
40345 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40347 PR preprocessor/54124
40348 * doc/cppopts.texi: Reformat -d subtable to list the full name
40349 of the options. Add cross-reference to the docs for the general
40350 compiler -d options.
40351 * doc/invoke.texi (Developer Options): Add cross-reference to the
40352 preprocessor-specific -d option documentation.
40354 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40356 PR preprocessor/13498
40357 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
40358 redudant material, and reflect new command-line options.
40359 (System Headers): Likewise.
40361 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40363 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
40364 -isystem, and -idirafter. Copy-edit.
40365 * doc/cppopts.texi: Copy-edit. Remove contradiction about
40366 default for -ftrack-macro-expansion. Delete obsolete and
40367 badly-formatted implementation details about -fdebug-cpp output.
40368 * doc/cppwarnopts.texi: Copy-edit.
40370 2017-01-07 David Malcolm <dmalcolm@redhat.com>
40373 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40374 that the transition from a max line width >= 1<<10 to narrower
40375 lines works correctly.
40377 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
40379 * doc/options.texi (PerFunction): New.
40380 * opt-functions.awk (switch_flags): Map both Optimization and
40381 PerFunction to CL_OPTIMIZATION.
40382 * opth-gen.awk: Test for PerFunction flag along with
40384 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
40385 it only when the latter is present. Skip those that don't in
40386 the hash function generator.
40387 * common.opt (fvar-tracking): Mark as PerFunction instead of
40389 (fvar-tracking-assignments): Likewise.
40390 (fvar-tracking-assignments-toggle): Likewise.
40391 (fvar-tracking-uninit): Likewise.
40393 2017-01-07 Jakub Jelinek <jakub@redhat.com>
40395 PR translation/79018
40396 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
40399 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
40402 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
40403 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
40404 TARGET_LONG_JUMP_TABLE_OFFSETS.
40405 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
40406 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
40407 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
40408 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
40409 * config/m68k/m68k.md (tablejump expander): Likewise.
40410 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
40411 TARGET_LONG_JUMP_TABLE_OFFSETS.
40412 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
40413 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
40415 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
40416 David Holsgrove <david.holsgrove@xilinx.com>
40418 * common/config/microblaze/microblaze-common.c
40419 (TARGET_EXCEPT_UNWIND_INFO): Remove.
40420 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
40422 * config/microblaze/microblaze.c (microblaze_must_save_register)
40423 (microblaze_expand_epilogue, microblaze_return_addr): Handle
40425 (microblaze_eh_return): New function.
40426 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
40427 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
40428 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
40429 * config/microblaze/microblaze.md (eh_return): New pattern.
40431 2017-01-06 Jakub Jelinek <jakub@redhat.com>
40433 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
40434 GCC_DIAGNOSTIC_STRINGIFY): Define.
40436 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
40438 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40440 * config/arm/arm.md (<mcrr>): New.
40442 * config/arm/arm.c (arm_arch5te): New.
40443 (arm_option_override): Set arm_arch5te.
40444 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
40446 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
40447 (arm_mcrr_qualifiers): ... this. New.
40448 (MRRC_QUALIFIERS): Define to...
40449 (arm_mrrc_qualifiers): ... this. New.
40450 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
40452 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
40453 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
40454 (MRRCI, mrrc, MRRC): New.
40455 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
40456 VUNSPEC_MRRC2): New.
40458 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40460 * config/arm/arm.md (<mcr>): New.
40462 * config/arm/arm.c (arm_coproc_builtin_available): Add
40463 support for mcr, mrc, mcr2 and mrc2.
40464 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
40465 (arm_mcr_qualifiers): ... this. New.
40466 (MRC_QUALIFIERS): Define to ...
40467 (arm_mrc_qualifiers): ... this. New.
40468 (MCR_QUALIFIERS): Define to ...
40469 (arm_mcr_qualifiers): ... this. New.
40470 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
40472 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
40473 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
40474 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
40475 VUNSPEC_MRC2): New.
40477 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40479 * config/arm/arm.md (*ldc): New.
40483 * config/arm/arm.c (arm_coproc_builtin_available): Add
40484 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
40485 (arm_coproc_ldc_stc_legitimate_address): New.
40486 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
40487 'qualifier_const_pointer'.
40488 (LDC_QUALIFIERS): Define to...
40489 (arm_ldc_qualifiers): ... this. New.
40490 (STC_QUALIFIERS): Define to...
40491 (arm_stc_qualifiers): ... this. New.
40492 * config/arm/arm-protos.h
40493 (arm_coproc_ldc_stc_legitimate_address): New.
40494 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
40495 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
40496 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
40497 stc2, stcl, stc2l): New.
40498 * config/arm/constraints.md (Uz): New.
40499 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
40500 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
40501 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
40502 VUNSPEC_STC2L): New.
40504 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40506 * config/arm/arm.md (<cdp>): New.
40507 * config/arm/arm.c (neon_const_bounds): Rename this ...
40508 (arm_const_bounds): ... this.
40509 (arm_coproc_builtin_available): New.
40510 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
40511 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
40512 (CDP_QUALIFIERS): Define to...
40513 (arm_cdp_qualifiers): ... this. New.
40515 (arm_expand_builtin_args): Add case for 6 arguments.
40516 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
40517 (arm_const_bounds): ... this.
40518 (arm_coproc_builtin_available): New.
40519 * config/arm/arm_acle.h (__arm_cdp): New.
40521 * config/arm/arm_acle_builtins.def (cdp): New.
40523 * config/arm/iterators.md (CDPI,CDP,cdp): New.
40524 * config/arm/neon.md: Rename all 'neon_const_bounds' to
40525 'arm_const_bounds'.
40526 * config/arm/types.md (coproc): New.
40527 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
40528 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
40529 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
40530 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
40532 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40534 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
40535 (UBINOP_QUALIFIERS): New.
40537 (acle_builtin_data): New. Change comment.
40538 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
40539 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
40540 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
40541 arm_acle_builtins.def.
40542 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
40543 (arm_init_acle_builtins): New.
40544 (CRC32_BUILTIN): Remove.
40545 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
40546 crc32cb, crc32ch and crc32cw.
40547 (arm_init_crc32_builtins): Remove.
40548 (arm_init_builtins): Use arm_init_acle_builtins rather
40549 than arm_init_crc32_builtins.
40550 (arm_expand_acle_builtin): New.
40551 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
40552 * config/arm/arm_acle_builtins.def: New.
40554 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40556 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
40557 (arm_builtin_datum): ... this.
40558 (arm_init_neon_builtin): Rename to ...
40559 (arm_init_builtin): ... this. Add a new parameters PREFIX
40560 and USE_SIG_IN_NAME.
40561 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
40562 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
40563 'arm_builtin_datum'.
40564 (arm_init_vfp_builtins): Likewise.
40565 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
40566 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
40567 (arm_expand_neon_args): Rename to ...
40568 (arm_expand_builtin_args): ... this. Rename builtin_arg
40569 enum values and differentiate between ARG_BUILTIN_MEMORY
40570 and ARG_BUILTIN_NEON_MEMORY.
40571 (arm_expand_neon_builtin_1): Rename to ...
40572 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
40573 values, arm_expand_builtin_args and add bool parameter NEON.
40574 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
40575 (arm_expand_vfp_builtin): Likewise.
40576 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
40578 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40580 PR middle-end/77484
40581 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
40582 * predict.c (tree_estimate_probability_bb): Reverse direction of
40583 polymorphic call predictor.
40585 2017-01-06 David Malcolm <dmalcolm@redhat.com>
40587 * passes.c (execute_one_pass): Split out pass-skipping logic into...
40588 (determine_pass_name_match): ...this new function and...
40589 (should_skip_pass_p): ...this new function.
40591 2017-01-06 Nathan Sidwell <nathan@acm.org>
40593 * ipa-visibility.c (function_and_variable_visibility): Reformat
40594 comments and long lines. Remove extrneous if.
40595 * symtab.c (symtab_node::make_decl_local): Fix code format.
40596 (symtab_node::set_section_for_node): Fix comment typo.
40598 2017-01-06 Martin Liska <mliska@suse.cz>
40601 * lra-constraints.c: Rename invariant to lra_invariant.
40602 * predict.c (set_even_probabilities): Initialize e to NULL.
40604 2017-01-05 Martin Sebor <msebor@redhat.com>
40606 PR tree-optimization/78910
40607 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
40608 (format_integer): Correct off-by-one error in the handling
40609 of precision with negative numbers in signed conversions..
40611 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
40613 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
40615 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40617 PR tree-optimization/71016
40618 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
40619 factor_out_conditional_conversion. Formatting fix.
40620 (factor_out_conditional_conversion): Add cond_stmt argument.
40621 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
40622 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
40625 2017-01-05 David Malcolm <dmalcolm@redhat.com>
40627 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
40628 read-rtl-function.o, and selftest-rtl.o.
40629 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
40630 (selftest::aarch64_test_loading_full_dump): New function.
40631 (selftest::aarch64_run_selftests): New function.
40632 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
40633 selftest::aarch64_run_selftests.
40634 * config/i386/i386.c
40635 (selftest::ix86_test_loading_dump_fragment_1): New function.
40636 (selftest::ix86_test_loading_call_insn): New function.
40637 (selftest::ix86_test_loading_full_dump): New function.
40638 (selftest::ix86_test_loading_unspec): New function.
40639 (selftest::ix86_run_selftests): Call the new functions.
40640 * emit-rtl.c (maybe_set_max_label_num): New function.
40641 * emit-rtl.h (maybe_set_max_label_num): New decl.
40642 * function.c (instantiate_decls): Guard call to
40643 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
40644 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
40646 * gensupport.c (gen_reader::gen_reader): Pass "false"
40647 for new "compact" param of rtx_reader.
40648 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
40649 rather than an empty string for NULL strings.
40650 * read-md.c: Potentially include config.h rather than bconfig.h.
40651 Wrap include of errors.h with #ifdef GENERATOR_FILE.
40652 (have_error): New global, copied from errors.c.
40653 (md_reader::read_name): Rename to...
40654 (md_reader::read_name_1): ...this, adding "out_loc" param,
40655 and converting "missing name or number" to returning false, rather
40657 (md_reader::read_name): Reimplement in terms of read_name_1.
40658 (md_reader::read_name_or_nil): New function.
40659 (md_reader::read_string): Handle "(nil)" by returning NULL.
40660 (md_reader::md_reader): Add new param "compact".
40661 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
40662 (md_reader::read_file): New method.
40663 * read-md.h (md_reader::md_reader): Add new param "compact".
40664 (md_reader::read_file): New method.
40665 (md_reader::is_compact): New accessor.
40666 (md_reader::read_name): Convert return type from void to file_location.
40667 (md_reader::read_name_or_nil): New decl.
40668 (md_reader::read_name_1): New decl.
40669 (md_reader::m_compact): New field.
40670 (noop_reader::noop_reader): Pass "false" for new "compact" param
40672 (rtx_reader::rtx_reader): Add new "compact" param.
40673 (rtx_reader::read_rtx_operand): Make virtual and convert return
40674 type from void to rtx.
40675 (rtx_reader::read_until): New decl.
40676 (rtx_reader::handle_any_trailing_information): New virtual function.
40677 (rtx_reader::postprocess): New virtual function.
40678 (rtx_reader::finalize_string): New virtual function.
40679 (rtx_reader::m_in_call_function_usage): New field.
40680 (rtx_reader::m_reuse_rtx_by_id): New field.
40681 * read-rtl-function.c: New file.
40682 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
40683 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
40684 (selftest::verify_three_block_rtl_cfg): New decl.
40685 * read-rtl-function.h: New file.
40686 * read-rtl.c: Potentially include config.h rather than bconfig.h.
40687 For host, include function.h, memmodel.h, and emit-rtl.h.
40688 (one_time_initialization): New function.
40689 (struct compact_insn_name): New struct.
40690 (compact_insn_names): New array.
40691 (find_code): Handle insn codes in compact dumps.
40692 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
40693 (bind_subst_iter_and_attr): Likewise.
40694 (add_condition_to_string): Likewise.
40695 (add_condition_to_rtx): Likewise.
40696 (apply_attribute_uses): Likewise.
40697 (add_current_iterators): Likewise.
40698 (apply_iterators): Likewise.
40699 (initialize_iterators): Guard usage of apply_subst_iterator with
40700 #ifdef GENERATOR_FILE.
40701 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
40702 (md_reader::read_mapping): Likewise.
40703 (add_define_attr_for_define_subst): Likewise.
40704 (add_define_subst_attr): Likewise.
40705 (read_subst_mapping): Likewise.
40706 (check_code_iterator): Likewise.
40707 (rtx_reader::read_rtx): Likewise. Move one-time initialization
40709 (one_time_initialization): New function.
40710 (rtx_reader::read_until): New method.
40711 (read_flags): New function.
40712 (parse_reg_note_name): New function.
40713 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
40714 Handle reuse_rtx ids.
40715 Wrap iterator lookup within #ifdef GENERATOR_FILE.
40716 Add parsing support for RTL dumps, mirroring the special-cases in
40717 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
40718 values, and calling handle_any_trailing_information.
40719 (rtx_reader::read_rtx_operand): Convert return type from void
40720 to rtx, returning return_rtx. Handle case 'e'. Call
40721 finalize_string on XSTR and XTMPL fields.
40722 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
40723 "(nil)" values were omitted. Call the postprocess vfunc on the
40725 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
40726 class ctor. Initialize m_in_call_function_usage. Call
40727 one_time_initialization.
40728 * rtl-tests.c (selftest::test_uncond_jump): Call
40729 set_new_first_and_last_insn.
40730 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
40731 * selftest-rtl.c: New file.
40732 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
40733 (selftest::get_insn_by_uid): New decl.
40734 * selftest-run-tests.c (selftest::run_tests): Call
40735 read_rtl_function_c_tests.
40736 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
40737 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
40740 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
40742 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
40743 operands in a special way. Assert that pos+len <= mode precision.
40745 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40747 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
40748 3 argument Alias with unlimited for the negative form.
40749 (fno-vect-cost-model): Removed.
40751 2017-01-05 Martin Liska <mliska@suse.cz>
40753 * hsa-gen.c (gen_hsa_divmod): New function.
40754 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
40756 2017-01-05 Martin Liska <mliska@suse.cz>
40759 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
40762 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40764 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
40765 small constant length operands.
40767 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40769 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
40770 between loop iterations.
40772 2017-01-05 Martin Liska <mliska@suse.cz>
40775 * gimplify.c (gimplify_decl_expr): Compare to
40776 asan_poisoned_variables instread of checking flags.
40777 (gimplify_target_expr): Likewise.
40778 (gimplify_expr): Likewise.
40779 (gimplify_function_tree): Conditionally initialize
40780 asan_poisoned_variables.
40782 2017-01-04 Jeff Law <law@redhat.com>
40784 PR tree-optimizatin/78812
40785 * rtl.h (contains_mem_rtx_p): Prototype.
40786 * ifcvt.c (containts_mem_rtx_p): Move from here to...
40787 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
40788 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
40789 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
40790 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
40792 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40794 * input.c (assert_char_at_range): Default-initialize actual_range.
40796 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40798 * df-scan.c (df_ref_create_structure): Make regno unsigned,
40799 to match the caller.
40801 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40803 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
40804 insns after final jump in test to emit dummy move.
40806 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40808 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
40809 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
40811 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40813 * multiple_target.c (create_dispatcher_calls): Init e_next.
40814 * tree-ssa-loop-split.c (split_loop): Init border.
40815 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
40818 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
40823 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
40824 (altivec_register_operand): Do not return true if the operand
40825 contains a SUBREG mixing SImode and SFmode.
40826 (vsx_register_operand): Likewise.
40827 (vsx_reg_sfsubreg_ok): New predicate.
40828 (vfloat_operand): Do not return true if the operand contains a
40829 SUBREG mixing SImode and SFmode.
40830 (vint_operand): Likewise.
40831 (vlogical_operand): Likewise.
40832 (gpc_reg_operand): Likewise.
40833 (int_reg_operand): Likewise.
40834 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
40835 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
40836 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
40838 (rs6000_emit_move_si_sf_subreg): New helper function.
40839 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
40840 fixup SUBREGs involving SImode and SFmode.
40841 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
40842 numbers for the new peephole2 optimization.
40843 (peephole2 for SFmode unions): New peephole2 to optimize cases in
40844 the GLIBC math library that do AND/IOR/XOR operations on single
40845 precision floating point.
40846 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
40847 target macros to say whether we need to avoid SUBREGs mixing
40849 (TARGET_ALLOW_SF_SUBREG): Likewise.
40850 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
40851 (UNSPEC_SI_FROM_SF): Likewise.
40852 (iorxor): Change spacing.
40853 (and_ior_xor): New iterator for AND, IOR, and XOR.
40854 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
40855 (movdi_from_sf_zero_ext): Likewise.
40856 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
40857 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
40858 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
40859 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
40860 (fms<mode>4): Likewise.
40861 (fnma<mode>4): Likewise.
40862 (fnms<mode>4): Likewise.
40863 (nfma<mode>4): Likewise.
40864 (nfms<mode>4): Likewise.
40866 2017-01-04 Marek Polacek <polacek@redhat.com>
40869 * doc/invoke.texi: Document -Wpointer-compare.
40871 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40873 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
40876 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
40877 descriptions for -gdwarf-5 and emit them as uleb128 instead of
40880 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
40883 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
40884 documentation of the powerpc_popcntb_ok attribute.
40885 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
40886 code to issue warning messages if a requested CPU configuration is
40887 not supported by the binary (assembler and loader) toolchain.
40888 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
40889 made to define a built-in function that has been disabled.
40890 (paired_init_builtins): Add assertion to prevent ICE if attempt is
40891 made to define a built-in function that has been disabled.
40892 (altivec_init_builtins): Add comment explaining why definition
40893 of the DST built-in functions is not preceded by an assertion
40894 check. Add assertions to prevent ICE if attempts are made to
40895 define an altivec predicate or an abs* built-in function that has
40897 (htm_init_builtins): Add comment explaining why definition of the
40898 htm built-in functions is not preceded by an assertion check.
40900 2017-01-04 Jeff Law <law@redhat.com>
40902 PR tree-optimizatin/67955
40903 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
40904 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
40905 the points-to solution does not include pt_null. Use DECL_PT_UID
40908 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
40910 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
40911 Use gen_int_mode instead of gen_lopwart for const_int operands.
40913 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40915 PR tree-optimization/71563
40916 * match.pd: Simplify X << Y into X if Y is known to be 0 or
40917 out of range value - has low bits known to be zero.
40919 2017-01-04 Alan Modra <amodra@gmail.com>
40921 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
40922 * configure: Regenerate.
40923 * config.in: Regenerate.
40925 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40928 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
40929 a substring of the message, but strcmp with the whole message. Ifdef
40930 ENABLE_NLS, translate the message first using dgettext.
40932 2017-01-03 Jeff Law <law@redhat.com>
40934 PR tree-optimizatin/78856
40935 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
40936 (mark_threaded_blocks): Remove code to truncate thread paths that
40937 cross multiple loop headers. Instead invalidate the cached loop
40938 iteration information and handle case of a thread path walking
40939 into an irreducible region.
40941 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
40944 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
40945 assertions. Add support for doing the signbit if the IEEE 128-bit
40946 floating point value is in a GPR.
40947 * config/rs6000/rs6000.md (Fsignbit): Delete.
40948 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
40949 Update the length attribute if the value is in a GPR.
40950 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
40951 the sign or zero extension instruction, since the value is always 0/1.
40952 (signbit<mode>2_dm2): Delete using <Fsignbit>.
40955 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
40956 extracting SImode to a GPR register so that we can generate a
40957 store, limit the vector to be in a traditional Altivec register
40958 for the vextuwrx instruction.
40960 2017-01-03 Ian Lance Taylor <iant@google.com>
40962 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
40964 2017-01-03 Martin Sebor <msebor@redhat.com>
40966 PR tree-optimization/78696
40967 * gimple-ssa-sprintf.c (format_floating): Correct handling of
40968 precision. Use MPFR for %f for greater fidelity. Correct handling
40970 (pass_sprintf_length::compute_format_length): Set width and precision
40971 specified by asrerisk to void_node for vararg functions.
40972 (try_substitute_return_value): Adjust dump output.
40974 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
40976 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
40978 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
40980 * doc/invoke.texi (SPARC options): Document -mlra as the default.
40981 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
40982 -mlra/-mno-lra was passed to the compiler.
40984 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
40986 PR rtl-optimization/65618
40987 * emit-rtl.c (try_split): Move initialization of "before" and
40988 "after" to just before the call to emit_insn_after_setloc.
40990 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
40992 * doc/md.texi (Standard Names): Remove reference to Java frontend.
40994 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
40996 * dwarf2out.c (gen_enumeration_type_die): When
40997 -gno-strict-dwarf, add a DW_AT_encoding attribute.
40999 2017-01-03 Jakub Jelinek <jakub@redhat.com>
41001 PR tree-optimization/78965
41002 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
41003 Change first argument from const call_info & to call_info &. For %n
41004 set info.nowrite to false.
41006 PR middle-end/78901
41007 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
41008 possibly throwing calls.
41010 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
41011 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
41012 and fns handling, rather than in a separate case SSA_NAME.
41014 2017-01-02 Jeff Law <law@redhat.com>
41016 * config/darwin-driver.c (darwin_driver_init): Const-correctness
41017 fixes for first_period and second_period variables.
41019 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
41022 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
41023 (*insvqi_1): New insn pattern.
41024 (*insvqi_1_mem_rex64): Ditto.
41025 (*insvqi_2): Ditto.
41026 (*insvqi_3): Rename from *insvqi.
41028 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
41030 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
41032 * doc/cfg.texi (Edges): Remove reference to Java.
41033 (Maintaining the CFG): Ditto.
41035 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41037 PR middle-end/77674
41038 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
41039 transparent aliases.
41041 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41043 PR middle-end/77484
41044 * predict.def (PRED_CALL): Update hitrate.
41045 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
41046 * predict.c (tree_estimate_probability_bb): Split CALL predictor
41047 into direct/indirect/polymorphic variants.
41049 2017-01-01 Jakub Jelinek <jakub@redhat.com>
41051 Update copyright years.
41053 * gcc.c (process_command): Update copyright notice dates.
41054 * gcov-dump.c (print_version): Ditto.
41055 * gcov.c (print_version): Ditto.
41056 * gcov-tool.c (print_version): Ditto.
41057 * gengtype.c (create_file): Ditto.
41058 * doc/cpp.texi: Bump @copying's copyright year.
41059 * doc/cppinternals.texi: Ditto.
41060 * doc/gcc.texi: Ditto.
41061 * doc/gccint.texi: Ditto.
41062 * doc/gcov.texi: Ditto.
41063 * doc/install.texi: Ditto.
41064 * doc/invoke.texi: Ditto.
41066 Copyright (C) 2017 Free Software Foundation, Inc.
41068 Copying and distribution of this file, with or without modification,
41069 are permitted in any medium without royalty provided the copyright
41070 notice and this notice are preserved.