1 2017-12-21 Jakub Jelinek <jakub@redhat.com>
4 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
5 if navail is >= dir.len.
7 2017-12-21 Steve Ellcey <sellcey@cavium.com>
9 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
12 2017-12-21 Jakub Jelinek <jakub@redhat.com>
14 PR rtl-optimization/80747
15 PR rtl-optimization/83512
16 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
17 succ edge from ENTRY, copy partition from e->dest to the newly
19 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
20 ENTRY, use BB_PARTITION of its successor block as current_partition.
21 Don't copy partition when splitting succ edge from ENTRY.
23 PR tree-optimization/83523
24 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
25 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
26 (convert_mult_to_fma): Likewise.
28 PR tree-optimization/83521
29 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
30 gimple_build_assign without code on result of
31 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
34 2017-12-21 Andrew Pinski <apinski@cavium.com>
35 Steve Ellcey <sellcey@cavium.com>
37 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
40 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
43 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
44 constraints to operand 2.
45 (*ashl<mode>3_mask_1): Ditto.
46 (*<shift_insn><mode>3_mask): Ditto.
47 (*<shift_insn><mode>3_mask_1): Ditto.
48 (*<rotate_insn><mode>3_mask): Ditto.
49 (*<rotate_insn><mode>3_mask_1): Ditto.
51 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
53 * reorg.c (make_return_insns): Reemit each insn with its own location.
55 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
57 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
58 generation for cases where splatting a value is not useful.
59 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
60 across a vec_duplicate and a paradoxical subreg forming a vector
63 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
65 * combine.c (simplify_set): Do not transform subregs to zero_extends
66 if the destination is not a scalar int mode.
68 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
71 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
72 shared zero if the input has overflowed.
74 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
76 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
77 support for Cortex-A55 and Cortex-A75.
79 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
81 * common/config/arm/arm-common.c (compare_opt_names): Add function
82 comment. Use strcmp instead of manual loop.
84 2017-12-21 Martin Liska <mliska@suse.cz>
87 * gcov-dump.c (dump_gcov_file): Do not read info about
88 support_unexecuted_blocks for gcda files.
90 2017-12-21 Jakub Jelinek <jakub@redhat.com>
92 PR rtl-optimization/82973
93 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
94 (valid_for_const_vector_p): ... this.
95 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
96 (valid_for_const_vector_p): ... this. Adjust function comment.
97 (gen_vec_duplicate): Adjust caller.
98 * optabs.c (expand_vector_broadcast): Likewise.
99 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
100 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
102 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
103 instead of GET_CODE == CONST_FIXED.
104 (simplify_subreg): Use CONST_FIXED_P macro instead of
105 GET_CODE == CONST_FIXED.
108 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
109 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
110 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
111 (ix86_option_override_internal): Adjust for
112 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
113 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
114 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
115 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
116 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
117 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
118 def_builtin2 for CET builtins.
119 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
120 up formatting in IX86_BUILTIN_RDPID code.
121 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
122 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
123 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
124 ix86_isa_flags2 to ix86_isa_flags.
125 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
127 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
128 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
129 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
131 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
132 Or in OPTION_MASK_ISA_AVX512F_SET.
133 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
134 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
135 (ix86_handle_option): Adjust for
136 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
137 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
139 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
140 Alan Hayward <alan.hayward@arm.com>
141 David Sherwood <david.sherwood@arm.com>
143 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
144 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
145 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
146 Track polynomial offsets.
148 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
149 Alan Hayward <alan.hayward@arm.com>
150 David Sherwood <david.sherwood@arm.com>
152 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
153 Treat drb->init as a poly_int. Fail if its misalignment wrt
154 vector_alignment isn't known.
156 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
157 Alan Hayward <alan.hayward@arm.com>
158 David Sherwood <david.sherwood@arm.com>
160 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
162 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
163 (vect_loop_versioning): Take the loop versioning threshold as a
165 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
166 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
167 versioning_threshold.
168 (vect_analyze_loop_2): Compute the loop versioning threshold
169 whenever loop versioning is needed, and store it in the new
170 field rather than combining it with the cost model threshold.
171 (vect_transform_loop): Update call to vect_loop_versioning.
172 Try to combine the loop versioning and cost thresholds here.
174 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
175 Alan Hayward <alan.hayward@arm.com>
176 David Sherwood <david.sherwood@arm.com>
178 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
179 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
180 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
181 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
182 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
183 (group_compare_offset): Update accordingly.
184 (split_small_address_groups_p): Likewise.
185 (record_use): Take addr_offset as a poly_uint64 rather than
186 an unsigned HOST_WIDE_INT.
187 (strip_offset): Return the offset as a poly_uint64 rather than
188 an unsigned HOST_WIDE_INT.
189 (record_group_use, split_address_groups): Track polynomial offsets.
190 (add_iv_candidate_for_use): Likewise.
191 (addr_offset_valid_p): Take the offset as a poly_int64 rather
192 than a HOST_WIDE_INT.
193 (strip_offset_1): Return the offset as a poly_int64 rather than
196 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
197 Alan Hayward <alan.hayward@arm.com>
198 David Sherwood <david.sherwood@arm.com>
200 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
201 rather than a HOST_WIDE_INT.
202 * tree.c (get_binfo_at_offset): Likewise.
204 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
205 Alan Hayward <alan.hayward@arm.com>
206 David Sherwood <david.sherwood@arm.com>
208 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
209 rather than a HOST_WIDE_INT.
210 * tree-sra.c (build_ref_for_offset): Likewise.
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 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
218 * tree.c (mem_ref_offset): Likewise.
219 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
220 * builtins.c (get_object_alignment_2): Likewise.
221 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
222 * gimple-fold.c (get_base_constructor): Likewise.
223 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
224 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
226 * ipa-polymorphic-call.c
227 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
228 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
229 (get_ancestor_addr_info): Likewise.
230 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
231 * match.pd: Likewise.
232 * tree-data-ref.c (dr_analyze_innermost): Likewise.
233 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
234 * tree-eh.c (tree_could_trap_p): Likewise.
235 * tree-object-size.c (addr_object_size): Likewise.
236 * tree-ssa-address.c (copy_ref_info): Likewise.
237 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
238 (indirect_refs_may_alias_p): Likewise.
239 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
240 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
241 (non_rewritable_mem_ref_base): Likewise.
242 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
243 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
244 * varasm.c (decode_addr_const): Likewise.
246 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
247 Alan Hayward <alan.hayward@arm.com>
248 David Sherwood <david.sherwood@arm.com>
250 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
251 offsets for MEM_REFs.
253 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
254 Alan Hayward <alan.hayward@arm.com>
255 David Sherwood <david.sherwood@arm.com>
257 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
258 as a poly_uint64 rather than an unsigned int.
259 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
261 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
262 Alan Hayward <alan.hayward@arm.com>
263 David Sherwood <david.sherwood@arm.com>
265 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
267 (decode_addr_const): Update accordingly.
269 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
270 Alan Hayward <alan.hayward@arm.com>
271 David Sherwood <david.sherwood@arm.com>
273 * tree.h (bit_field_size, bit_field_offset): New functions.
274 * hsa-gen.c (gen_hsa_addr): Use them.
275 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
276 (simplify_vector_constructor): Likewise.
277 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
278 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
279 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
280 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
283 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
284 Alan Hayward <alan.hayward@arm.com>
285 David Sherwood <david.sherwood@arm.com>
287 * expr.h (emit_group_load, emit_group_load_into_temps)
288 (emit_group_store): Take the size as a poly_int64 rather than an int.
289 * expr.c (emit_group_load_1, emit_group_load): Likewise.
290 (emit_group_load_into_temp, emit_group_store): Likewise.
292 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
293 Alan Hayward <alan.hayward@arm.com>
294 David Sherwood <david.sherwood@arm.com>
296 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
297 an unsigned int to a poly_uint64.
298 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
299 sizes as poly_uint64s rather than unsigned ints.
300 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
303 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
304 Alan Hayward <alan.hayward@arm.com>
305 David Sherwood <david.sherwood@arm.com>
307 * lra-constraints.c (emit_inc): Change inc_amount from an int
310 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
311 Alan Hayward <alan.hayward@arm.com>
312 David Sherwood <david.sherwood@arm.com>
314 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
316 (add_stack_var, stack_var_cmp, partition_stack_vars)
317 (dump_stack_var_partition): Update accordingly.
318 (alloc_stack_frame_space): Take the size as a poly_int64 rather
319 than a HOST_WIDE_INT.
320 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
321 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
322 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
323 rather than a HOST_WIDE_INT.
325 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
326 Alan Hayward <alan.hayward@arm.com>
327 David Sherwood <david.sherwood@arm.com>
329 * target.def (return_pops_args): Treat both the input and output
330 sizes as poly_int64s rather than HOST_WIDE_INTS.
331 * targhooks.h (default_return_pops_args): Update accordingly.
332 * targhooks.c (default_return_pops_args): Likewise.
333 * doc/tm.texi: Regenerate.
334 * emit-rtl.h (incoming_args): Change pops_args, size and
335 outgoing_args_size from int to poly_int64_pod.
336 * function.h (expr_status): Change x_pending_stack_adjust and
337 x_stack_pointer_delta from int to poly_int64.
338 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
339 (ARGS_SIZE_RTX): Update accordingly.
340 * calls.c (highest_outgoing_arg_in_use): Change from int to
342 (stack_usage_watermark, stored_args_watermark): New variables.
343 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
344 (emit_call_1): Change the stack_size and rounded_stack_size
345 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
347 (save_fixed_argument_area): Check stack_usage_watermark.
348 (initialize_argument_information): Change old_pending_adj from
349 a HOST_WIDE_INT * to a poly_int64_pod *.
350 (compute_argument_block_size): Return the size as a poly_int64
352 (finalize_must_preallocate): Track polynomial argument sizes.
353 (compute_argument_addresses): Likewise.
354 (internal_arg_pointer_based_exp): Track polynomial offsets.
355 (mem_overlaps_already_clobbered_arg_p): Rename to...
356 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
357 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
358 Check stored_args_used_watermark.
359 (load_register_parameters): Update accordingly.
360 (check_sibcall_argument_overlap_1): Likewise.
361 (combine_pending_stack_adjustment_and_call): Take the unadjusted
362 args size as a poly_int64 rather than an int. Return a bool
363 indicating whether the optimization was possible and return
364 the new adjustment by reference.
365 (check_sibcall_argument_overlap): Track polynomail argument sizes.
366 Update stored_args_watermark.
367 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
368 (expand_call): Likewise. Maintain stack_usage_watermark and
369 stored_args_watermark. Update calls to
370 combine_pending_stack_adjustment_and_call.
371 (emit_library_call_value_1): Handle polynomial argument sizes.
372 Call stack_region_maybe_used_p and mark_stack_region_used.
373 Maintain stack_usage_watermark.
374 (store_one_arg): Likewise. Update call to
375 mem_overlaps_already_clobbered_arg_p.
376 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
378 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
379 * config/microblaze/microblaze.c (microblaze_function_prologue):
381 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
382 TARGET_RETURN_POPS_ARGS interface.
383 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
385 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
386 * config/i386/i386.c (ix86_return_pops_args): Update for new
387 TARGET_RETURN_POPS_ARGS interface.
388 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
389 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
390 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
391 TARGET_RETURN_POPS_ARGS interface.
392 * config/vax/vax.c (vax_return_pops_args): Likewise.
393 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
394 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
395 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
396 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
397 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
398 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
399 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
400 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
401 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
402 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
403 and x_stack_pointer_delta from int to poly_int64.
404 * dojump.c (do_pending_stack_adjust): Update accordingly.
405 * explow.c (allocate_dynamic_stack_space): Handle polynomial
406 stack_pointer_deltas.
407 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
408 (pad_to_arg_alignment): Track polynomial offsets.
409 (assign_parm_find_stack_rtl): Likewise.
410 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
411 * toplev.c (output_stack_usage): Update reference to
412 current_function_pushed_stack_size.
414 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
415 Alan Hayward <alan.hayward@arm.com>
416 David Sherwood <david.sherwood@arm.com>
418 * function.c (in_arg_offset, var_offset, dynamic_offset)
419 (out_arg_offset, cfa_offset): Change from int to poly_int64.
420 (instantiate_new_reg): Return the new offset as a poly_int64_pod
421 rather than a HOST_WIDE_INT.
422 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
423 (instantiate_virtual_regs_in_insn): Likewise.
425 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
426 Alan Hayward <alan.hayward@arm.com>
427 David Sherwood <david.sherwood@arm.com>
429 * rtl.h (get_args_size, add_args_size_note): New functions.
430 (find_args_size_adjust): Return a poly_int64 rather than a
432 (fixup_args_size_notes): Likewise. Make the same change to the
433 end_args_size parameter.
434 * rtlanal.c (get_args_size, add_args_size_note): New functions.
435 * builtins.c (expand_builtin_trap): Use add_args_size_note.
436 * calls.c (emit_call_1): Likewise.
437 * explow.c (adjust_stack_1): Likewise.
438 * cfgcleanup.c (old_insns_match_p): Update use of
439 find_args_size_adjust.
440 * combine.c (distribute_notes): Track polynomial arg sizes.
441 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
442 end_true_args_size, beg_delay_args_size and end_delay_args_size
443 from HOST_WIDE_INT to poly_int64.
444 (add_cfi_args_size): Take the args_size as a poly_int64 rather
445 than a HOST_WIDE_INT.
446 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
447 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
448 polynomial arg sizes.
449 * emit-rtl.c (try_split): Use get_args_size.
450 * recog.c (peep2_attempt): Likewise.
451 * reload1.c (reload_as_needed): Likewise.
452 * expr.c (find_args_size_adjust): Return the adjustment as a
453 poly_int64 rather than a HOST_WIDE_INT.
454 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
455 to a poly_int64 and change the return type in the same way.
456 (emit_single_push_insn): Track polynomial arg sizes.
458 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
459 Alan Hayward <alan.hayward@arm.com>
460 David Sherwood <david.sherwood@arm.com>
462 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
463 from HOST_WIDE_INT to poly_int64.
464 * expr.c (push_block, emit_push_insn): Likewise.
466 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
467 Alan Hayward <alan.hayward@arm.com>
468 David Sherwood <david.sherwood@arm.com>
470 * function.h (frame_space): Change start and length from HOST_WIDE_INT
472 (get_frame_size): Return the size as a poly_int64 rather than a
474 (frame_offset_overflow): Take the offset as a poly_int64 rather
475 than a HOST_WIDE_INT.
476 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
477 (assign_stack_temp): Likewise for the size.
478 * function.c (get_frame_size): Return a poly_int64 rather than
480 (frame_offset_overflow): Take the offset as a poly_int64 rather
481 than a HOST_WIDE_INT.
482 (try_fit_stack_local): Take the start, length and size as poly_int64s
483 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
484 rather than a HOST_WIDE_INT.
485 (add_frame_space): Take the start and end as poly_int64s rather than
487 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
488 (assign_stack_temp): Likewise for the size.
489 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
491 (find_temp_slot_from_address): Handle polynomial offsets.
492 (combine_temp_slots): Likewise.
493 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
495 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
496 poly_int64 rather than a HOST_WIDE_INT.
497 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
498 than a HOST_WIDE_INT.
499 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
500 polynomial frame offsets.
501 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
502 as a poly_int64 rather than an int.
503 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
504 * config/v850/v850-protos.h (compute_frame_size): Likewise.
505 * config/v850/v850.c (compute_frame_size): Likewise.
506 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
507 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
508 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
509 * config/pa/pa.c (pa_compute_frame_size): Likewise.
510 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
511 rather than a HOST_WIDE_INT.
512 * explow.c (get_dynamic_stack_base): Likewise.
513 * final.c (final_start_function): Use the constant lower bound
514 of the frame size for -Wframe-larger-than.
515 * ira.c (do_reload): Adjust for new get_frame_size return type.
516 * lra.c (lra): Likewise.
517 * reload1.c (reload): Likewise.
518 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
519 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
520 * rtlanal.c (get_initial_register_offset): Return the offset as
521 a poly_int64 rather than a HOST_WIDE_INT.
523 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
524 Alan Hayward <alan.hayward@arm.com>
525 David Sherwood <david.sherwood@arm.com>
527 * reload1.c (elim_table): Change initial_offset, offset and
528 previous_offset from HOST_WIDE_INT to poly_int64_pod.
529 (offsets_at): Change the target array's element type from
530 HOST_WIDE_INT to poly_int64_pod.
531 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
532 (elimination_costs_in_insn, update_eliminable_offsets)
533 (verify_initial_elim_offsets, set_offsets_for_label)
534 (init_eliminable_invariants): Update after above changes.
536 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
537 Alan Hayward <alan.hayward@arm.com>
538 David Sherwood <david.sherwood@arm.com>
540 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
541 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
542 (decomposition): Change start and end from HOST_WIDE_INT
544 (decompose, immune_p): Update accordingly.
545 (find_inc_amount): Return a poly_int64 rather than an int.
546 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
549 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
550 Alan Hayward <alan.hayward@arm.com>
551 David Sherwood <david.sherwood@arm.com>
553 * tree.h (get_inner_reference): Return the bitsize and bitpos
554 as poly_int64_pods rather than HOST_WIDE_INT.
555 * fold-const.h (ptr_difference_const): Return the pointer difference
556 as a poly_int64_pod rather than a HOST_WIDE_INT.
557 * expr.c (get_inner_reference): Return the bitsize and bitpos
558 as poly_int64_pods rather than HOST_WIDE_INT.
559 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
561 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
562 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
563 (optimize_bit_field_compare): Update call to get_inner_reference.
564 (decode_field_reference): Likewise.
565 (fold_unary_loc): Track polynomial offsets and sizes.
566 (split_address_to_core_and_offset): Return the bitpos as a
567 poly_int64_pod rather than a HOST_WIDE_INT.
568 (ptr_difference_const): Likewise for the pointer difference.
569 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
570 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
571 * dbxout.c (dbxout_expand_expr): Likewise.
572 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
573 (loc_list_from_tree_1, fortran_common): Likewise.
574 * gimple-laddress.c (pass_laddress::execute): Likewise.
575 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
576 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
577 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
578 * tree-affine.c (tree_to_aff_combination): Likewise.
579 (get_inner_reference_aff): Likewise.
580 * tree-data-ref.c (split_constant_offset_1): Likewise.
581 (dr_analyze_innermost): Likewise.
582 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
583 * tree-sra.c (ipa_sra_check_caller): Likewise.
584 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
585 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
586 (instrument_bool_enum_load, instrument_object_size): Likewise.
587 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
588 to get_inner_reference.
589 * hsa-gen.c (gen_hsa_addr): Likewise.
590 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
591 * tsan.c (instrument_expr): Likewise.
592 * match.pd: Update call to ptr_difference_const.
594 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
595 Alan Hayward <alan.hayward@arm.com>
596 David Sherwood <david.sherwood@arm.com>
598 * fold-const.c (fold_comparison): Track sizes and offsets as
599 poly_int64s rather than HOST_WIDE_INTs when folding address
602 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
603 Alan Hayward <alan.hayward@arm.com>
604 David Sherwood <david.sherwood@arm.com>
606 * expr.h (get_bit_range): Return the bitstart and bitend as
607 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
608 as a poly_int64 rather than a HOST_WIDE_INT.
609 * expr.c (get_bit_range): Likewise.
610 (expand_assignment): Update call accordingly.
611 * fold-const.c (optimize_bit_field_compare): Likewise.
613 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
614 Alan Hayward <alan.hayward@arm.com>
615 David Sherwood <david.sherwood@arm.com>
617 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
618 (aarch64_legitimate_address_p): Use it instead of an rtx code,
619 as an optional final parameter.
620 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
621 (aarch64_legitimate_address_p): Likewise.
622 (aarch64_print_address_internal): Take an aarch64_addr_query_type
623 instead of an rtx code.
624 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
625 (aarch64_legitimate_address_hook_p): Likewise.
626 (aarch64_print_ldpstp_address): Likewise.
627 (aarch64_print_operand_address): Likewise.
628 (aarch64_address_cost): Likewise.
629 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
630 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
631 (aarch64_mem_pair_lanes_operand): Likewise.
633 2017-12-20 Richard Biener <rguenther@suse.de>
635 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
636 update_stmt_if_modified.
638 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
640 PR tree-optimization/83491
641 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
642 before walking uses. Improve coding style and comments.
644 2017-12-20 Tom de Vries <tom@codesourcery.com>
646 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
648 2017-12-20 Jakub Jelinek <jakub@redhat.com>
651 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
653 (pass_ipa_free_fn_summary): Add small_p private data member,
654 initialize to false in the ctor.
655 (pass_ipa_free_fn_summary::clone,
656 pass_ipa_free_fn_summary::set_pass_param,
657 pass_ipa_free_fn_summary::gate): New methods.
658 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
659 | TODO_dump_symtab if small_p.
660 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
661 entry and add another instance of the pass with false parm after
663 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
664 ipa_free_fn_summary here.
666 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
668 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
670 2017-12-20 Martin Sebor <msebor@redhat.com>
673 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
674 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
675 (expand_builtin_strncmp): Same.
677 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
680 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
681 allowed debug stmts before labels.
682 (expand_gimple_basic_block): Likewise.
683 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
684 * gimple-iterator.h (gsi_after_labels): Likewise.
685 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
686 rename reused variable, and simplify using gsi_move_before.
687 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
688 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
689 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
690 (gimple_verify_flow_info, gimple_block_label): Likewise.
691 (make_blocks): Move debug markers after adjacent labels.
692 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
693 allowed debug insns outside blocks.
694 * df-scan.c (df_insn_delete): Likewise.
695 * lra-constraints.c (update_ebb_live_info): Likewise.
696 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
697 (vt_initialize, delete_vta_debug_insns): Likewise.
698 (reemit_marker_as_note): Drop BB parm. Adjust callers.
700 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
701 Alan Hayward <alan.hayward@arm.com>
702 David Sherwood <david.sherwood@arm.com>
704 * poly-int-types.h (round_down_to_byte_boundary): New macro.
705 (round_up_to_byte_boundary): Likewise.
706 * expr.h (get_bit_range): Add temporary shim.
707 * gimple-ssa-store-merging.c (store_operand_info): Change the
708 bitsize, bitpos, bitregion_start and bitregion_end fields from
709 unsigned HOST_WIDE_INT to poly_uint64.
710 (merged_store_group): Likewise load_align_base.
711 (compatible_load_p, compatible_load_p): Update accordingly.
712 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
713 (split_group, imm_store_chain_info::output_merged_store): Likewise.
714 (mem_valid_for_store_merging): Return the bitsize, bitpos,
715 bitregion_start and bitregion_end as poly_uint64s rather than
716 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
717 (handled_load): Take the bitsize, bitpos,
718 bitregion_start and bitregion_end as poly_uint64s rather than
719 unsigned HOST_WIDE_INTs.
720 (pass_store_merging::process_store): Update call to
721 mem_valid_for_store_merging.
723 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
724 Alan Hayward <alan.hayward@arm.com>
725 David Sherwood <david.sherwood@arm.com>
727 * builtins.c (get_object_alignment_2): Track polynomial offsets
728 and sizes. Update the alignment handling.
730 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
731 Alan Hayward <alan.hayward@arm.com>
732 David Sherwood <david.sherwood@arm.com>
734 * tree.h (get_inner_reference): Add a version that returns the
735 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
736 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
737 the case in which bitpos is not associated with the first byte.
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 * tree-affine.h (get_inner_reference_aff): Return the size as a
745 * tree-affine.c (get_inner_reference_aff): Likewise.
746 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
747 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
749 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
750 Alan Hayward <alan.hayward@arm.com>
751 David Sherwood <david.sherwood@arm.com>
753 * fold-const.c (pointer_may_wrap_p): Take the offset as a
754 HOST_WIDE_INT rather than a poly_int64.
756 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
757 Alan Hayward <alan.hayward@arm.com>
758 David Sherwood <david.sherwood@arm.com>
760 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
761 HOST_WIDE_INT to poly_int64_pod.
762 (perform_symbolic_merge): Update accordingly.
763 (bswap_replace): Likewise.
765 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
766 Alan Hayward <alan.hayward@arm.com>
767 David Sherwood <david.sherwood@arm.com>
769 * tree-affine.h (aff_tree::offset): Change from widest_int
771 (wide_int_ext_for_comb): Delete.
772 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
773 constants as poly_widest_int rather than widest_int.
774 (aff_combination_constant_multiple_p): Return the multiplier
775 as a poly_widest_int.
776 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
778 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
779 version static and add an overload for poly_widest_int.
780 (aff_combination_const, aff_combination_add_cst)
781 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
782 the constants as poly_widest_int rather than widest_int.
783 (tree_to_aff_combination): Generalize INTEGER_CST case to
785 (aff_combination_to_tree): Track offsets as poly_widest_ints.
786 (aff_combination_add_product, aff_combination_mult): Handle
788 (aff_combination_constant_multiple_p): Return the multiplier
789 as a poly_widest_int.
790 * tree-predcom.c (determine_offset): Return the offset as a
792 (split_data_refs_to_components, suitable_component_p): Update
794 (valid_initializer_p): Update call to
795 aff_combination_constant_multiple_p.
796 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
797 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
798 as a poly_int64 rather than a HOST_WIDE_INT.
799 (get_address_cost): Handle polynomial offsets.
800 (iv_elimination_compare_lt): Likewise.
801 (rewrite_use_nonlinear_expr): Likewise.
803 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
804 Alan Hayward <alan.hayward@arm.com>
805 David Sherwood <david.sherwood@arm.com>
807 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
808 as a poly_int64_pod rather than a HOST_WIDE_INT.
809 (get_addr_base_and_unit_offset): Likewise.
810 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
811 (get_addr_base_and_unit_offset): Likewise.
812 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
813 to poly_int64 in example.
814 * fold-const.c (fold_binary_loc): Update call to
815 get_addr_base_and_unit_offset.
816 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
817 (maybe_canonicalize_mem_ref_addr): Likewise.
818 (gimple_fold_stmt_to_constant_1): Likewise.
819 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
821 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
822 * match.pd: Likewise.
823 * omp-low.c (lower_omp_target): Likewise.
824 * tree-sra.c (build_ref_for_offset): Likewise.
825 (build_debug_ref_for_model): Likewise.
826 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
827 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
828 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
829 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
830 (constant_pointer_difference): Likewise.
831 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
832 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
833 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
834 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
835 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
836 (set_ssa_val_to): Likewise.
837 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
838 (maybe_diag_stxncpy_trunc): Likewise.
839 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
840 * tree.c (build_simple_mem_ref_loc): Likewise.
841 (array_at_struct_end_p): Likewise.
843 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
844 Alan Hayward <alan.hayward@arm.com>
845 David Sherwood <david.sherwood@arm.com>
847 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
848 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
849 (get_ref_base_and_extent_hwi): Declare.
850 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
851 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
852 (get_ref_base_and_extent_hwi): New function.
853 * cfgexpand.c (expand_debug_expr): Update call to
854 get_ref_base_and_extent.
855 * dwarf2out.c (add_var_loc_to_decl): Likewise.
856 * gimple-fold.c (get_base_constructor): Return the offset as a
857 poly_int64_pod rather than a HOST_WIDE_INT.
858 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
859 * ipa-polymorphic-call.c
860 (ipa_polymorphic_call_context::set_by_invariant)
861 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
862 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
863 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
864 rather than get_ref_base_and_extent.
865 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
866 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
867 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
869 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
870 call to get_ref_base_and_extent.
871 * tree-sra.c (create_access, get_access_for_expr): Likewise.
872 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
873 (stmt_kills_ref_p): Likewise.
874 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
875 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
877 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
878 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
879 when calling native_encode_expr.
880 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
881 call to get_ref_base_and_extent.
882 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
883 get_ref_base_and_extent.
884 * var-tracking.c (track_expr_p): Likewise.
886 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
887 Alan Hayward <alan.hayward@arm.com>
888 David Sherwood <david.sherwood@arm.com>
890 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
891 HOST_WIDE_INT to poly_int64_pod.
892 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
893 polynomail parameter offsets.
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 * gengtype.c (main): Handle poly_int64_pod.
900 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
901 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
902 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
903 from HOST_WIDE_INT to poly_int64_pod.
904 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
905 (copy_cfa): New function.
906 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
907 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
908 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
909 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
910 (initial_return_save): Treat offsets as poly_ints.
911 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
912 offset is nonconstant.
913 (reg_save): Take the offset as a poly_int64. Fall back to
914 DW_CFA_expression for nonconstant offsets.
915 (queue_reg_save): Take the offset as a poly_int64.
916 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
918 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
919 Alan Hayward <alan.hayward@arm.com>
920 David Sherwood <david.sherwood@arm.com>
922 * rtl.h (operand_subword, operand_subword_force): Take the offset
923 as a poly_uint64 an unsigned int.
924 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
926 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
927 Alan Hayward <alan.hayward@arm.com>
928 David Sherwood <david.sherwood@arm.com>
930 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
931 'p' format code. Use INT_LIST rather than SUBREG as the example of
932 a code with an XINT and an XEXP. Remove the implication that
933 accessing an rtx field using XINT is expected to work.
934 * rtl.def (SUBREG): Change format from "ei" to "ep".
935 * rtl.h (rtunion::rt_subreg): New field.
936 (XCSUBREG): New macro.
937 (SUBREG_BYTE): Use it.
938 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
939 Update constructor accordingly.
940 (subreg_shape::operator ==): Update accordingly.
941 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
942 than an unsigned int.
943 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
944 a poly_uint64 rather than an unsigned int.
945 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
946 than an unsigned int.
947 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
948 (subreg_size_highpart_offset): Return a poly_uint64 rather than
949 an unsigned int. Take the sizes as poly_uint64s.
950 (subreg_offset_from_lsb): Return a poly_uint64 rather than
951 an unsigned int. Take the shift as a poly_uint64 rather than
953 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
954 as a poly_uint64 rather than an unsigned int.
955 (simplify_subreg_regno): Likewise.
956 (byte_lowpart_offset): Return the memory offset as a poly_int64
958 (subreg_memory_offset): Likewise. Take the subreg offset as a
959 poly_uint64 rather than an unsigned int.
960 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
961 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
962 poly_uint64 rather than an unsigned int.
963 * rtl.c (rtx_format): Describe 'p' in comment.
964 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
965 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
966 offset as a poly_uint64 rather than an unsigned int.
967 (byte_lowpart_offset): Return the memory offset as a poly_int64
969 (subreg_memory_offset): Likewise. Take the subreg offset as a
970 poly_uint64 rather than an unsigned int.
971 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
972 mode sizes as poly_uint64s rather than unsigned ints. Return a
973 poly_uint64 rather than an unsigned int.
974 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
975 (copy_insn_1): Handle 'p'.
976 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
977 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
978 an unsigned int. Return the shift in the same way.
979 (subreg_lsb): Return the shift as a poly_uint64 rather than an
981 (subreg_size_offset_from_lsb): Take the sizes and shift as
982 poly_uint64s rather than unsigned ints. Return the offset as
984 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
985 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
987 * rtlhash.c (add_rtx): Handle 'p'.
988 * genemit.c (gen_exp): Likewise.
989 * gengenrtl.c (type_from_format, gendef): Likewise.
990 * gensupport.c (subst_pattern_match, get_alternatives_number)
991 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
992 (subst_dup): Likewise.
993 * gengtype.c (adjust_field_rtx_def): Likewise.
994 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
995 (match_pattern_2): Likewise.
996 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
997 (rtx_test::subreg_field): New function.
998 (operator ==, safe_to_hoist_p, transition_parameter_type)
999 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
1000 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
1001 * genpeep.c (match_rtx): Likewise.
1002 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
1003 (rtx_writer::print_rtx_operand): Handle 'p'.
1004 (print_value): Handle SUBREG.
1005 * read-rtl.c (apply_int_iterator): Likewise.
1006 (rtx_reader::read_rtx_operand): Handle 'p'.
1007 * alias.c (rtx_equal_for_memref_p): Likewise.
1008 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
1009 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
1011 * calls.c (expand_call): Likewise.
1012 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
1013 (make_extraction, gen_lowpart_for_combine): Likewise.
1014 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
1016 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
1017 rather than an unsigned int. Treat subreg offsets as poly_ints.
1018 (exp_equiv_p): Handle 'p'.
1019 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
1020 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
1021 * dse.c (find_shift_sequence): Likewise.
1022 * dwarf2out.c (rtl_for_decl_location): Likewise.
1023 * expmed.c (extract_low_bits): Likewise.
1024 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
1025 (expand_expr_real_2): Likewise.
1026 * final.c (alter_subreg): Likewise.
1027 (leaf_renumber_regs_insn): Handle 'p'.
1028 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
1029 Treat subreg offsets as poly_ints.
1030 * fwprop.c (forward_propagate_and_simplify): Likewise.
1031 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
1032 * ira.c (get_subreg_tracking_sizes): Likewise.
1033 * ira-conflicts.c (go_through_subreg): Likewise.
1034 * ira-lives.c (process_single_reg_class_operands): Likewise.
1035 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
1036 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
1037 as a poly_uint64 rather than an unsigned int.
1038 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
1039 subreg offsets as poly_ints.
1040 * lra-constraints.c (operands_match_p): Handle 'p'.
1041 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
1042 * lra-spills.c (assign_mem_slot): Likewise.
1043 * postreload.c (move2add_valid_value_p): Likewise.
1044 * recog.c (general_operand, indirect_operand): Likewise.
1045 * regcprop.c (copy_value, maybe_mode_change): Likewise.
1046 (copyprop_hardreg_forward_1): Likewise.
1047 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
1048 (record_subregs_of_mode): Likewise.
1049 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
1050 * reload.c (operands_match_p): Handle 'p'.
1051 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
1052 * reload1.c (alter_reg, choose_reload_regs): Likewise.
1053 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
1054 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
1055 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
1056 (simplify_const_poly_int_tests<N>::run): Likewise.
1057 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
1058 a poly_uint64 rather than an unsigned int.
1059 * valtrack.c (debug_lowpart_subreg): Likewise.
1060 * var-tracking.c (var_lowpart): Likewise.
1061 (loc_cmp): Handle 'p'.
1063 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1064 Alan Hayward <alan.hayward@arm.com>
1065 David Sherwood <david.sherwood@arm.com>
1067 * ira.c (get_subreg_tracking_sizes): New function.
1068 (init_live_subregs): Take an integer size rather than a register.
1069 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
1070 to init_live_subregs.
1072 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1073 Alan Hayward <alan.hayward@arm.com>
1074 David Sherwood <david.sherwood@arm.com>
1076 * expr.c (store_constructor_field): Change bitsize from a
1077 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
1078 HOST_WIDE_INT to a poly_int64.
1079 (store_constructor): Change size from a HOST_WIDE_INT to
1081 (store_field): Likewise bitsize and bitpos.
1083 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1084 Alan Hayward <alan.hayward@arm.com>
1085 David Sherwood <david.sherwood@arm.com>
1087 * expmed.h (store_bit_field): Change bitregion_start and
1088 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
1089 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
1090 (store_bit_field_1, store_integral_bit_field, store_bit_field)
1091 (store_fixed_bit_field, store_split_bit_field): Likewise.
1092 * expr.c (store_constructor_field, store_field): Likewise.
1093 (optimize_bitfield_assignment_op): Likewise. Make the same change
1094 to bitsize and bitpos.
1095 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
1096 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
1097 same change in the constructor arguments.
1098 (get_best_mode): Change bitregion_start and bitregion_end from
1099 unsigned HOST_WIDE_INT to poly_uint64.
1100 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
1101 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
1103 (bit_field_mode_iterator::next_mode): Update for new types
1104 of m_bitregion_start and m_bitregion_end.
1105 (get_best_mode): Change bitregion_start and bitregion_end from
1106 unsigned HOST_WIDE_INT to poly_uint64.
1108 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1109 Alan Hayward <alan.hayward@arm.com>
1110 David Sherwood <david.sherwood@arm.com>
1112 * rtl.h (simplify_gen_subreg): Add a temporary overload that
1113 accepts poly_uint64 offsets.
1114 * expmed.h (extract_bit_field): Take bitsize and bitnum as
1115 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1116 * expmed.c (lowpart_bit_field_p): Likewise.
1117 (extract_bit_field_as_subreg): New function, split out from...
1118 (extract_bit_field_1): ...here. Take bitsize and bitnum as
1119 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
1120 extractions, check that BITSIZE matches the size of the extracted
1121 value and that BITNUM is an exact multiple of that size.
1122 If all else fails, try forcing the value into memory if
1123 BITNUM is variable, and adjusting the address so that the
1124 offset is constant. Split the part that can only handle constant
1125 bitsize and bitnum out into...
1126 (extract_integral_bit_field): ...this new function.
1127 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
1128 rather than unsigned HOST_WIDE_INTs.
1130 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1131 Alan Hayward <alan.hayward@arm.com>
1132 David Sherwood <david.sherwood@arm.com>
1134 * expmed.h (store_bit_field): Take bitsize and bitnum as
1135 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1136 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
1137 that returns the byte size.
1138 (store_bit_field_1): Take bitsize and bitnum as
1139 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
1140 to simple_mem_bitfield_p. Split the part that can only handle
1141 constant bitsize and bitnum out into...
1142 (store_integral_bit_field): ...this new function.
1143 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
1144 than unsigned HOST_WIDE_INTs.
1145 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
1147 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1148 Alan Hayward <alan.hayward@arm.com>
1149 David Sherwood <david.sherwood@arm.com>
1151 * lra-int.h (lra_reg): Change offset from int to poly_int64.
1152 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
1154 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1155 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
1156 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
1157 offset as a poly_int64 rather than an int.
1158 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
1159 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1160 * lra-constraints.c (equiv_address_substitution): Track offsets
1162 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
1163 (curr_insn_transform): Handle the new form of sp_offset.
1164 * lra-eliminations.c (lra_elim_table): Change previous_offset
1165 and offset from HOST_WIDE_INT to poly_int64.
1166 (print_elim_table, update_reg_eliminate): Update accordingly.
1167 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
1168 (get_elimination): Update accordingly.
1169 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
1170 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1171 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
1172 poly_int64 offsets generally.
1173 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
1174 (mark_not_eliminable, init_elimination): Update accordingly.
1175 (remove_reg_equal_offset_note): Return a bool and pass the new
1176 offset back by pointer as a poly_int64.
1177 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
1178 rather than a HOST_WIDE_INT.
1179 (do_remat): Track offsets poly_int64s.
1180 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
1182 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1183 Alan Hayward <alan.hayward@arm.com>
1184 David Sherwood <david.sherwood@arm.com>
1186 * rtl.h (mem_attrs): Add a default constructor. Change size and
1187 offset from HOST_WIDE_INT to poly_int64.
1188 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
1189 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
1190 (widen_memory_access): Take the sizes and offsets as poly_int64s
1191 rather than HOST_WIDE_INTs.
1192 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
1193 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
1195 (adjust_offset_for_component_ref): Change the offset from a
1196 HOST_WIDE_INT to a poly_int64.
1197 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
1198 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
1199 * dce.c (find_call_stack_args): Likewise.
1200 * dse.c (record_store): Likewise.
1201 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
1202 * print-rtl.c (rtx_writer::print_rtx): Likewise.
1203 * read-rtl-function.c (test_loading_mem): Likewise.
1204 * rtlanal.c (may_trap_p_1): Likewise.
1205 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1206 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
1207 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
1208 (mem_attrs::mem_attrs): New function.
1209 (set_mem_attributes_minus_bitpos): Change bitpos from a
1210 HOST_WIDE_INT to poly_int64.
1211 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
1212 (clear_mem_offset, clear_mem_size, change_address)
1213 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
1214 initialize mem_attrs.
1215 (set_mem_offset, set_mem_size, adjust_address_1)
1216 (adjust_automodify_address_1, offset_address, widen_memory_access):
1217 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
1219 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1220 Alan Hayward <alan.hayward@arm.com>
1221 David Sherwood <david.sherwood@arm.com>
1223 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
1224 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
1225 rather than 0 to represent an unknown size. Assert that the size
1226 is known when the mode isn't BLKmode.
1227 (may_trap_p_1): Use -1 for unknown sizes.
1228 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
1230 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1231 Alan Hayward <alan.hayward@arm.com>
1232 David Sherwood <david.sherwood@arm.com>
1234 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
1235 to poly_int64. Update commentary for positions_needed.large.
1236 (read_info_type): Change offset and width from HOST_WIDE_INT
1238 (set_usage_bits): Likewise.
1239 (canon_address): Return the offset as a poly_int64 rather than
1240 a HOST_WIDE_INT. Use strip_offset_and_add.
1241 (set_all_positions_unneeded, any_positions_needed_p): Use
1242 positions_needed.large to track stores with non-constant widths.
1243 (all_positions_needed_p): Likewise. Take the offset and width
1244 as poly_int64s rather than ints. Assert that rhs is nonnull.
1245 (record_store): Cope with non-constant offsets and widths.
1246 Nullify the rhs of an earlier store if we can't tell which bytes
1248 (find_shift_sequence): Take the access_size and shift as poly_int64s
1250 (get_stored_val): Take the read_offset and read_width as poly_int64s
1251 rather than HOST_WIDE_INTs.
1252 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
1253 non-constant offsets and widths.
1255 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1256 Alan Hayward <alan.hayward@arm.com>
1257 David Sherwood <david.sherwood@arm.com>
1259 * inchash.h (inchash::hash::add_poly_int): New function.
1260 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
1261 Use poly_int64 rather than HOST_WIDE_INT.
1262 (ao_ref::max_size_known_p): New function.
1263 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
1264 rather than HOST_WIDE_INT.
1265 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
1266 to temporaries until its interface is adjusted to match.
1267 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
1268 (aliasing_component_refs_p, decl_refs_may_alias_p)
1269 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
1270 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
1271 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
1273 * alias.c (ao_ref_from_mem): Likewise.
1274 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1275 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
1276 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
1277 (maybe_trim_complex_store, maybe_trim_constructor_store)
1278 (live_bytes_read, dse_classify_store): Likewise.
1279 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
1280 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
1281 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
1282 (vn_reference_lookup_3): Likewise.
1283 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
1285 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1286 Alan Hayward <alan.hayward@arm.com>
1287 David Sherwood <david.sherwood@arm.com>
1289 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
1290 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
1291 instead of ranges_overlap_p.
1293 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1294 Alan Hayward <alan.hayward@arm.com>
1295 David Sherwood <david.sherwood@arm.com>
1297 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
1298 sizes as poly_int64s rather than HOST_WIDE_INTs.
1300 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1301 Alan Hayward <alan.hayward@arm.com>
1302 David Sherwood <david.sherwood@arm.com>
1304 * gimple-fold.h (fold_ctor_reference): Take the offset and size
1305 as poly_uint64 rather than unsigned HOST_WIDE_INT.
1306 * gimple-fold.c (fold_ctor_reference): Likewise.
1308 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1309 Alan Hayward <alan.hayward@arm.com>
1310 David Sherwood <david.sherwood@arm.com>
1312 * target.def (dwarf_poly_indeterminate_value): New hook.
1313 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
1314 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
1315 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
1316 * doc/tm.texi: Regenerate.
1317 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
1318 offset as a poly_int64.
1319 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
1320 offset as a poly_int64.
1321 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
1322 Take the offset as a poly_int64.
1323 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
1324 (frame_pointer_fb_offset): Change to a poly_int64.
1325 (int_loc_descriptor): Take the offset as a poly_int64. Use
1326 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
1327 (based_loc_descr): Take the offset as a poly_int64.
1328 Use strip_offset_and_add to handle (plus X (const)).
1329 Use new_reg_loc_descr instead of an open-coded version of the
1330 previous implementation.
1331 (mem_loc_descriptor): Handle CONST_POLY_INT.
1332 (compute_frame_pointer_to_fb_displacement): Take the offset as a
1333 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
1335 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1336 Alan Hayward <alan.hayward@arm.com>
1337 David Sherwood <david.sherwood@arm.com>
1339 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
1340 (gen_rtx_REG_offset): Take the offset as a poly_int64.
1341 * inchash.h (inchash::hash::add_poly_hwi): New function.
1342 * gengtype.c (main): Register poly_int64.
1343 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
1344 offset as a poly_int.
1345 (reg_attr_hasher::equal): Use must_eq to compare offsets.
1346 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
1347 offset as a poly_int64.
1348 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
1349 * print-rtl.c (print_poly_int): New function.
1350 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
1352 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
1354 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
1355 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
1356 (vt_get_decl_and_offset): Return the offset as a poly_int64.
1357 Enforce track_offset_p for parts of a PARALLEL.
1358 (vt_add_function_parameter): Use const_offset for the final
1359 offset to track. Use get_tracked_reg_offset for the parts
1362 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1363 Alan Hayward <alan.hayward@arm.com>
1364 David Sherwood <david.sherwood@arm.com>
1366 * target.def (truly_noop_truncation): Take poly_uint64s instead of
1367 unsigned ints. Change default to hook_bool_puint64_puint64_true.
1368 * doc/tm.texi: Regenerate.
1369 * hooks.h (hook_bool_uint_uint_true): Delete.
1370 (hook_bool_puint64_puint64_true): Declare.
1371 * hooks.c (hook_bool_uint_uint_true): Delete.
1372 (hook_bool_puint64_puint64_true): New function.
1373 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
1374 instead of unsigned ints.
1375 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
1376 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
1378 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1379 Alan Hayward <alan.hayward@arm.com>
1380 David Sherwood <david.sherwood@arm.com>
1382 * optabs.h (expand_operand): Add an int_value field.
1383 (create_expand_operand): Add an int_value parameter and use it
1384 to initialize the new expand_operand field.
1385 (create_integer_operand): Replace with a declaration of a function
1386 that accepts poly_int64s. Move the implementation to...
1387 * optabs.c (create_integer_operand): ...here.
1388 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
1389 the mode preserves the value of int_value, instead of calling
1390 const_int_operand on the rtx. Use gen_int_mode to generate
1393 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1394 Alan Hayward <alan.hayward@arm.com>
1395 David Sherwood <david.sherwood@arm.com>
1397 * dumpfile.h (dump_dec): Declare.
1398 * dumpfile.c (dump_dec): New function.
1399 * pretty-print.h (pp_wide_integer): Turn into a function and
1400 declare a poly_int version.
1401 * pretty-print.c (pp_wide_integer): New function for poly_ints.
1403 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1404 Alan Hayward <alan.hayward@arm.com>
1405 David Sherwood <david.sherwood@arm.com>
1407 * doc/generic.texi (POLY_INT_CST): Document.
1408 * tree.def (POLY_INT_CST): New tree code.
1409 * treestruct.def (TS_POLY_INT_CST): New tree layout.
1410 * tree-core.h (tree_poly_int_cst): New struct.
1411 (tree_node): Add a poly_int_cst field.
1412 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
1413 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
1414 instead of a wide_int_ref.
1415 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
1417 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1418 instead of an unsigned HOST_WIDE_INT.
1419 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
1420 (ptrdiff_tree_p): Declare.
1421 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
1422 extern inline implementations if the target doesn't use POLY_INT_CST.
1423 (poly_int_tree_p): New function.
1424 (wi::unextended_tree): New class.
1425 (wi::int_traits <unextended_tree>): New override.
1426 (wi::extended_tree): Add a default constructor.
1427 (wi::extended_tree::get_tree): New function.
1428 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
1429 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
1430 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
1431 (wi::tree_to_poly_wide_ref): New typedefs.
1432 (wi::ints_for): Provide overloads for extended_tree and
1434 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
1435 (wi::to_wide): New functions.
1436 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
1437 * tree.c (poly_int_cst_hasher): New struct.
1438 (poly_int_cst_hash_table): New variable.
1439 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
1440 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
1442 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
1443 (init_ttree): Initialize poly_int_cst_hash_table.
1444 (build_int_cst, build_int_cst_type, build_invariant_address): Take
1445 a poly_int64 instead of a HOST_WIDE_INT.
1446 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1447 instead of an unsigned HOST_WIDE_INT.
1448 (wide_int_to_tree): Rename to...
1449 (wide_int_to_tree_1): ...this.
1450 (build_new_poly_int_cst, build_poly_int_cst): New functions.
1451 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
1452 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
1453 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
1454 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
1455 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
1457 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
1458 (streamer_read_tree_body): Likewise.
1459 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
1460 (streamer_write_tree_body): Likewise.
1461 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
1462 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
1463 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
1464 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
1465 * gimple-expr.h (is_gimple_constant): Likewise.
1466 * gimplify.c (maybe_with_size_expr): Likewise.
1467 * print-tree.c (print_node): Likewise.
1468 * tree-data-ref.c (data_ref_compare_tree): Likewise.
1469 * tree-pretty-print.c (dump_generic_node): Likewise.
1470 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1471 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
1472 * tree-vrp.c (compare_values_warnv): Likewise.
1473 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
1474 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
1475 (force_expr_to_var_cost): Likewise.
1476 * tree-ssa-loop.c (for_each_index): Likewise.
1477 * fold-const.h (build_invariant_address, size_int_kind): Take a
1478 poly_int64 instead of a HOST_WIDE_INT.
1479 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
1480 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
1482 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
1483 (int_const_binop_2): New function, split out from...
1484 (int_const_binop_1): ...here. Handle POLY_INT_CST.
1485 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
1486 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
1487 * gimple-ssa-strength-reduction.c (slsr_process_add)
1488 (slsr_process_mul): Check for INTEGER_CSTs before using them
1490 * stor-layout.c (bits_from_bytes): New function.
1491 (bit_from_pos): Use it.
1492 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
1493 by BITS_PER_UNIT to get the TYPE_SIZE.
1494 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
1495 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
1497 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1498 Alan Hayward <alan.hayward@arm.com>
1499 David Sherwood <david.sherwood@arm.com>
1501 * doc/rtl.texi (const_poly_int): Document. Also document the
1502 rtl sharing behavior.
1503 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
1504 * rtl.h (const_poly_int_def): New struct.
1505 (rtx_def::u): Add a cpi field.
1506 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
1507 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
1508 (wi::rtx_to_poly_wide_ref): New typedef
1509 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
1510 (poly_int_rtx_p): New functions.
1511 (trunc_int_for_mode): Declare a poly_int64 version.
1512 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1513 (immed_wide_int_const): Take a poly_wide_int_ref rather than
1515 (strip_offset): Declare.
1516 (strip_offset_and_add): New function.
1517 * rtl.def (CONST_POLY_INT): New rtx code.
1518 * rtl.c (rtx_size): Handle CONST_POLY_INT.
1519 (shared_const_p): Use poly_int_rtx_p.
1520 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
1522 (gen_int_shift_amount): Likewise.
1523 * emit-rtl.c (const_poly_int_hasher): New class.
1524 (const_poly_int_htab): New variable.
1525 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
1526 (const_poly_int_hasher::hash): New function.
1527 (const_poly_int_hasher::equal): Likewise.
1528 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
1529 (immed_wide_int_const): Rename to...
1530 (immed_wide_int_const_1): ...this and make static.
1531 (immed_wide_int_const): New function, taking a poly_wide_int_ref
1532 instead of a wide_int_ref.
1533 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
1534 (gen_lowpart_common): Handle CONST_POLY_INT.
1535 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
1536 * cselib.c (cselib_hash_rtx): Likewise.
1537 * dwarf2out.c (const_ok_for_output_1): Likewise.
1538 * expr.c (convert_modes): Likewise.
1539 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
1540 * rtlhash.c (add_rtx): Likewise.
1541 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
1542 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1543 Handle existing CONST_POLY_INT rtxes.
1544 * expmed.h (expand_shift): Take a poly_int64 instead of a
1546 * expmed.c (expand_shift): Likewise.
1547 * rtlanal.c (strip_offset): New function.
1548 (commutative_operand_precedence): Give CONST_POLY_INT the same
1549 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
1551 * rtl-tests.c (const_poly_int_tests): New struct.
1552 (rtl_tests_c_tests): Use it.
1553 * simplify-rtx.c (simplify_const_unary_operation): Handle
1555 (simplify_const_binary_operation): Likewise.
1556 (simplify_binary_operation_1): Fold additions of symbolic constants
1557 and CONST_POLY_INTs.
1558 (simplify_subreg): Handle extensions and truncations of
1560 (simplify_const_poly_int_tests): New struct.
1561 (simplify_rtx_c_tests): Use it.
1562 * wide-int.h (storage_ref): Add default constructor.
1563 (wide_int_ref_storage): Likewise.
1564 (trailing_wide_ints): Use GTY((user)).
1565 (trailing_wide_ints::operator[]): Add a const version.
1566 (trailing_wide_ints::get_precision): New function.
1567 (trailing_wide_ints::extra_size): Likewise.
1569 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1570 Alan Hayward <alan.hayward@arm.com>
1571 David Sherwood <david.sherwood@arm.com>
1573 * emit-rtl.h (gen_int_shift_amount): Declare.
1574 * emit-rtl.c (gen_int_shift_amount): New function.
1575 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
1577 * calls.c (shift_return_value): Likewise.
1578 * cse.c (fold_rtx): Likewise.
1579 * dse.c (find_shift_sequence): Likewise.
1580 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
1581 (expand_shift, expand_smod_pow2): Likewise.
1582 * lower-subreg.c (shift_cost): Likewise.
1583 * optabs.c (expand_superword_shift, expand_doubleword_mult)
1584 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
1585 (expand_vec_perm_var): Likewise.
1586 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1587 (simplify_binary_operation_1): Likewise.
1588 * combine.c (try_combine, find_split_point, force_int_to_mode)
1589 (simplify_shift_const_1, simplify_shift_const): Likewise.
1590 (change_zero_ext): Likewise. Use simplify_gen_binary.
1592 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1594 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
1596 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1598 * doc/invoke.texi (ARM Options): Document accepted extension options
1599 for -march=armv8.3-a.
1601 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
1604 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
1605 or --with-float=softfp, set the default CPU to arm10e.
1607 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
1609 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
1610 * config/visium/predicates.md (const_shift_operand): Likewise.
1611 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
1612 (visium_legitimize_reload_address): Likewise.
1614 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1617 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
1619 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1622 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
1624 2017-12-20 Julia Koval <julia.koval@intel.com>
1626 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
1627 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
1628 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
1629 * config.gcc: Include vpclmulqdqintrin.h.
1630 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
1631 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
1633 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
1634 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
1635 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
1636 * config/i386/i386.c (isa2_opts): Add -mcx16.
1637 (isa_opts): Add -mpclmulqdq, remove -mcx16.
1638 (ix86_option_override_internal): Move mcx16 to flags2.
1639 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
1640 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
1641 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
1642 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
1643 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
1644 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
1645 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
1646 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
1647 * doc/invoke.texi: Add -mvpclmulqdq.
1649 2017-12-20 Tom de Vries <tom@codesourcery.com>
1652 * config/i386/i386.c (ix86_static_chain): Move
1653 DECL_STATIC_CHAIN test ...
1654 * calls.c (rtx_for_static_chain): ... here. New function.
1655 * calls.h (rtx_for_static_chain): Declare.
1656 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
1657 instead of targetm.calls.static_chain.
1658 * df-scan.c (df_get_entry_block_def_set): Same.
1660 2017-12-19 Tom de Vries <tom@codesourcery.com>
1662 PR tree-optimization/83493
1663 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
1666 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1668 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
1669 inform with hardcoded english plural handling.
1671 2017-12-18 Jeff Law <law@redhat.com>
1673 PR tree-optimization/83477
1674 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
1675 a non-virtual PHI, always push a new range.
1677 2017-12-19 Martin Sebor <msebor@redhat.com>
1680 * builtins.c (compute_objsize): Handle non-constant offsets.
1682 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1684 PR tree-optimization/83444
1685 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
1686 character load case, if get_stridx on MEM_REF's operand doesn't
1687 look usable, retry with get_addr_stridx.
1689 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
1692 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
1693 (variable_tracking_main_1): Keep markers even when VTA fails.
1696 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
1697 even if there are markers before it.
1698 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
1700 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1702 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
1703 typical order conditions.
1704 * sel-sched.c (extract_new_fences_from): Likewise.
1705 * config/visium/constraints.md (J, K, L): Likewise.
1706 * config/visium/predicates.md (const_shift_operand): Likewise.
1707 * config/visium/visium.c (visium_legitimize_address,
1708 visium_legitimize_reload_address): Likewise.
1709 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
1710 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
1711 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
1712 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
1713 SET_DUMP_DETAIL): Likewise.
1714 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
1715 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
1716 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
1717 avr_set_core_architecture, avr_set_current_function,
1718 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
1719 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
1720 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
1721 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
1722 avr_map_decompose, avr_fold_builtin): Likewise.
1723 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
1724 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
1725 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
1726 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
1727 * config/m32c/m32c.c (m32c_conditional_register_usage,
1728 m32c_address_cost): Likewise.
1729 * config/m32c/predicates.md (shiftcount_operand,
1730 longshiftcount_operand): Likewise.
1731 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
1732 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
1733 can_use_cdx_ldstw): Likewise.
1734 * config/nios2/nios2.h (CDX_REG_P): Likewise.
1735 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
1737 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
1738 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
1739 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
1740 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
1741 * config/vax/vax.c (adjacent_operands_p): Likewise.
1742 * config/ft32/constraints.md (L, b, KA): Likewise.
1743 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
1745 * cfgexpand.c (expand_stack_alignment): Likewise.
1746 * gcse.c (insert_expr_in_table): Likewise.
1747 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
1748 * cgraphunit.c (cgraph_node::expand): Likewise.
1749 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
1750 * emit-rtl.c (add_insn): Likewise.
1751 * input.c (dump_location_info): Likewise.
1752 * passes.c (NEXT_PASS): Likewise.
1753 * read-rtl-function.c (parse_note_insn_name,
1754 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
1756 * sched-rgn.c (sched_rgn_init): Likewise.
1757 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
1758 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
1759 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
1761 * explow.c (eliminate_constant_term): Likewise.
1762 * final.c (leaf_renumber_regs_insn): Likewise.
1763 * cfgrtl.c (print_rtl_with_bb): Likewise.
1764 * genhooks.c (emit_init_macros): Likewise.
1765 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
1766 * tree-data-ref.c (conflict_fn): Likewise.
1767 * selftest.c (assert_streq): Likewise.
1768 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
1769 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
1770 fold_binary_loc, multiple_of_p): Likewise.
1771 * reload.c (push_reload, find_equiv_reg): Likewise.
1772 * et-forest.c (et_nca, et_below): Likewise.
1773 * dbxout.c (dbxout_symbol_location): Likewise.
1774 * reorg.c (relax_delay_slots): Likewise.
1775 * dojump.c (do_compare_rtx_and_jump): Likewise.
1776 * gengtype-parse.c (type): Likewise.
1777 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
1778 simplify_const_relational_operation): Likewise.
1779 * reload1.c (do_output_reload): Likewise.
1780 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
1781 * gengtype.c (type_for_name): Likewise.
1782 * gimple-ssa-sprintf.c (format_directive): Likewise.
1784 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1787 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
1788 accessing it. Adjust comment.
1790 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1793 * predict.c (zero_one_minusone): New function.
1794 (apply_return_prediction): Avoid return prediction for functions
1795 returning only -1, 0 and 1 values, unless they only return -1 and 0
1798 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1800 * config/arc/arc.c (legitimate_scaled_address_p): Clean
1801 fall-through warning.
1802 (arc_compute_frame_size): Remove unused variables.
1803 (arc_print_operand): Fix fprintif format.
1804 (arc_can_follow_jump): Clean fall-through warning.
1806 2017-12-19 Marek Polacek <polacek@redhat.com>
1809 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
1812 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1814 * config/arc/arc.c (overriderregs): New variable.
1815 (arc_override_options): Track fixed/call saved/call options.
1816 (arc_conditional_register_usage): Check against overrideregs
1817 variable whenever we change register properties.
1819 2017-12-19 Nathan Sidwell <nathan@acm.org>
1821 * opts.c (finish_options): Don't prefix dump_base_name if it
1822 already contains directories.
1824 2017-12-19 Martin Liska <mliska@suse.cz>
1826 PR rtl-optimization/82675
1827 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
1828 more element in sbitmap.
1830 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
1832 * gimplify.c (gimplify_expr): Use error_operand_p.
1834 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
1837 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
1840 2017-12-19 Daniel Cederman <cederman@gaisler.com>
1842 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
1843 the jump is to a label.
1845 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1847 PR tree-optimization/83444
1848 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
1853 * ipa-inline.c (flatten_remove_node_hook): New function.
1854 (ipa_inline): Keep only nodes with flatten attribute at the end of
1855 the array in the order from ipa_reverse_postorder, only walk that
1856 portion of array for flattening, if there is more than one such
1857 node, temporarily register a removal hook and ignore removed nodes.
1859 PR tree-optimization/80631
1860 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
1861 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
1862 IFN_REDUC_MAX or IFN_REDUC_MIN.
1864 2017-12-18 Jeff Law <law@redhat.com>
1866 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
1867 record symbolic equivalences from backedges in the CFG.
1870 2017-11-19 Jeff Law <law@redhat.com>
1872 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1873 of degenerates resulting from ignoring an edge.
1875 2017-12-18 Martin Sebor <msebor@redhat.com>
1878 PR tree-optimization/78450
1879 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
1880 (handle_builtin_strlen): Call it.
1882 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
1884 PR rtl-optimization/83424
1885 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
1887 2017-12-18 Marek Polacek <polacek@redhat.com>
1890 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1891 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
1894 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1896 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
1897 (umaddsidi4, umaddsidi_split): Likewise.
1899 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1901 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
1904 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1907 * config/s390/s390.c (s390_option_override): Avoid strncpy.
1909 2017-12-18 Richard Biener <rguenther@suse.de>
1911 PR tree-optimization/81877
1912 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
1913 (outermost_indep_loop): Adjust.
1914 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
1915 (can_sm_ref_p): Adjust.
1917 2017-12-18 Richard Biener <rguenther@suse.de>
1920 * tree.c (array_at_struct_end_p): Return true if the underlying
1921 object has space for at least one element in excess of what
1922 the array domain specifies.
1924 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1926 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
1927 example to avoid overfull hbox.
1928 * doc/invoke.texi (Option Summary): Add missing @gol.
1929 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
1932 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1933 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1935 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
1936 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
1937 (AArch64 Options): Document -mverbose-cost-dump.
1938 (ARM Options): Likewise, plus -mflip-thumb.
1940 2017-12-17 Martin Sebor <msebor@redhat.com>
1943 * gimple-ssa-warn-restrict.c
1944 (builtin_memref::offset_out_of_bounds): Correct the handling of
1947 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1949 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
1951 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
1953 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
1955 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
1957 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
1960 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
1962 * doc/invoke.texi: Fix some typos.
1964 2017-12-16 Martin Sebor <msebor@redhat.com>
1966 PR tree-optimization/78918
1967 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
1968 * builtins.c (check_sizes): Rename...
1969 (check_access): ...to this. Rename function arguments for clarity.
1970 (check_memop_sizes): Adjust names.
1971 (expand_builtin_memchr, expand_builtin_memcpy): Same.
1972 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
1973 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
1974 (check_strncat_sizes, expand_builtin_strncat): Same.
1975 (expand_builtin_strncpy, expand_builtin_memset): Same.
1976 (expand_builtin_bzero, expand_builtin_memcmp): Same.
1977 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
1978 (maybe_emit_sprintf_chk_warning): Same.
1979 (expand_builtin_strcpy): Adjust.
1980 (expand_builtin_stpcpy): Same.
1981 (expand_builtin_with_bounds): Detect out-of-bounds accesses
1982 in pointer-checking forms of memcpy, memmove, and mempcpy.
1983 (gcall_to_tree_minimal, max_object_size): Define new functions.
1984 * builtins.h (max_object_size): Declare.
1985 * calls.c (alloc_max_size): Call max_object_size instead of
1986 hardcoding ssizetype limit.
1987 (get_size_range): Handle new argument.
1988 * calls.h (get_size_range): Add a new argument.
1989 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
1990 * doc/invoke.texi (-Wrestrict): Adjust, add example.
1991 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
1993 (gimple_fold_builtin_memory_chk): Same.
1994 (gimple_fold_builtin_stxcpy_chk): New function.
1995 * gimple-ssa-warn-restrict.c: New source.
1996 * gimple-ssa-warn-restrict.h: New header.
1997 * gimple.c (gimple_build_call_from_tree): Propagate location.
1998 * passes.def (pass_warn_restrict): Add new pass.
1999 * tree-pass.h (make_pass_warn_restrict): Declare.
2000 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
2002 (handle_builtin_strcat): Same.
2003 (strlen_optimize_stmt): Rename...
2004 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
2005 stpncpy, strncpy, and their checking forms.
2007 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
2009 PR rtl-optimization/82849
2010 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
2011 and get_max_loop_iterations_int.
2013 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2014 Alan Hayward <alan.hayward@arm.com>
2015 David Sherwood <david.sherwood@arm.com>
2017 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
2018 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
2019 size as a poly_uint64.
2020 (mode_for_vector, mode_for_int_vector): Take the number of vector
2021 elements as a poly_uint64.
2022 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
2023 size as a poly_uint64.
2024 (mode_for_vector, mode_for_int_vector): Take the number of vector
2025 elements as a poly_uint64.
2027 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2028 Alan Hayward <alan.hayward@arm.com>
2029 David Sherwood <david.sherwood@arm.com>
2031 * machmode.h (MACRO_MODE): New macro.
2032 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
2033 * targhooks.c (default_libcall_value, default_secondary_reload)
2034 (default_memory_move_cost, default_register_move_cost)
2035 (default_class_max_nregs): Likewise.
2037 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2038 Alan Hayward <alan.hayward@arm.com>
2039 David Sherwood <david.sherwood@arm.com>
2041 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
2042 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
2044 * genautomata.c (main): Likewise.
2045 * genconditions.c (write_header): Likewise.
2046 * genemit.c (main): Likewise.
2047 * genextract.c (print_header): Likewise.
2048 * genopinit.c (main): Likewise.
2049 * genoutput.c (output_prologue): Likewise.
2050 * genpeep.c (main): Likewise.
2051 * genpreds.c (write_insn_preds_c): Likewise.
2052 * genrecog.c (writer_header): Likewise.
2053 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
2054 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
2055 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
2056 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
2057 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
2058 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
2059 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
2060 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
2061 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
2062 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
2063 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
2064 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
2065 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
2066 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
2067 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
2068 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
2069 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
2070 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
2071 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
2072 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
2073 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
2074 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
2075 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
2076 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
2077 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
2078 * config/darwin.c (IN_TARGET_CODE): Likewise.
2079 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
2080 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
2081 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
2082 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
2083 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
2084 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
2085 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
2086 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
2087 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
2088 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
2089 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
2090 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
2091 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
2092 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
2093 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
2094 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
2095 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
2096 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
2097 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
2098 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
2099 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
2100 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
2101 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
2102 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
2103 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
2104 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
2105 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
2106 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
2107 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
2108 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
2109 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
2110 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
2111 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
2112 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
2113 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
2114 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
2115 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
2116 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
2117 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
2118 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
2119 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
2120 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
2121 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
2122 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
2123 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
2124 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
2125 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
2126 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
2127 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
2128 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
2129 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
2130 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
2131 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
2132 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
2133 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
2134 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
2135 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
2136 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
2137 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
2138 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
2139 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2140 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
2141 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
2142 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
2143 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
2144 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
2145 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
2146 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
2147 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
2148 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
2149 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
2150 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2151 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
2152 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
2153 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
2154 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
2155 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
2156 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
2157 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
2158 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
2159 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
2160 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
2161 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
2162 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
2163 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
2164 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
2165 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
2166 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
2167 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
2168 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
2169 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
2170 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
2171 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
2172 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
2173 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
2174 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
2175 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
2176 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
2177 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
2178 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
2179 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
2180 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
2181 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
2182 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
2183 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
2184 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
2186 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2187 Alan Hayward <alan.hayward@arm.com>
2188 David Sherwood <david.sherwood@arm.com>
2190 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
2193 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2194 Alan Hayward <alan.hayward@arm.com>
2195 David Sherwood <david.sherwood@arm.com>
2197 * doc/generic.texi (VEC_SERIES_EXPR): Document.
2198 * doc/md.texi (vec_series@var{m}): Document.
2199 * tree.def (VEC_SERIES_EXPR): New tree code.
2200 * tree.h (build_vec_series): Declare.
2201 * tree.c (build_vec_series): New function.
2202 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
2203 * tree-pretty-print.c (dump_generic_node): Likewise.
2204 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
2205 * tree-inline.c (estimate_operator_cost): Likewise.
2206 * expr.c (expand_expr_real_2): Likewise.
2207 * optabs-tree.c (optab_for_tree_code): Likewise.
2208 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
2209 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
2210 * expmed.c (make_tree): Handle VEC_SERIES.
2211 * optabs.def (vec_series_optab): New optab.
2212 * optabs.h (expand_vec_series_expr): Declare.
2213 * optabs.c (expand_vec_series_expr): New function.
2214 * tree-vect-generic.c (expand_vector_operations_1): Check that
2215 the operands also have vector type.
2217 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2218 Alan Hayward <alan.hawyard@arm.com>
2219 David Sherwood <david.sherwood@arm.com>
2221 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
2222 (VEC_COND_EXPR): Add missing @tindex.
2223 * doc/md.texi (vec_duplicate@var{m}): Document.
2224 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
2225 * tree.c (build_vector_from_val): Add stubbed-out handling of
2226 variable-length vectors, using VEC_DUPLICATE_EXPR.
2227 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
2228 * cfgexpand.c (expand_debug_expr): Likewise.
2229 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2230 * tree-inline.c (estimate_operator_cost): Likewise.
2231 * tree-pretty-print.c (dump_generic_node): Likewise.
2232 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
2233 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
2234 (test_vec_duplicate_folding): New function.
2235 (fold_const_c_tests): Call it.
2236 * optabs.def (vec_duplicate_optab): New optab.
2237 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
2238 * optabs.h (expand_vector_broadcast): Declare.
2239 * optabs.c (expand_vector_broadcast): Make non-static. Try using
2240 vec_duplicate_optab.
2241 * expr.c (store_constructor): Try using vec_duplicate_optab for
2243 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
2245 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
2248 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
2249 div/mod latencies a bit.
2251 2017-12-15 Jeff Law <law@redhat.com>
2253 PR tree-optimization/36550
2254 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
2255 (mark_threaded_blocks): Rewrite code to avoid block copying when
2256 optimizing for size. Don't pessimize blocks which will be
2257 copied, but all the statements will be dead.
2259 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
2261 PR tree-optimization/81165
2262 * tree-ssa-threadupdate.c (uses_in_bb): New.
2263 (estimate_threading_killed_stmts): New.
2264 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
2265 * tree-ssa-threadedge.c
2266 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
2269 2017-12-15 Jeff Law <law@redhat.com>
2271 PR tree-optimization/83410
2272 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
2273 threads when parallelizing loops.
2275 2017-12-15 Jakub Jelinek <jakub@redhat.com>
2277 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
2279 * config/alpha/alpha.c (vms_attribute_table): Swap
2280 affects_type_identity and handler fields, adjust comments.
2281 * config/mips/mips.c (mips_attribute_table): Likewise.
2282 * config/visium/visium.c (visium_attribute_table): Likewise.
2283 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
2284 * config/microblaze/microblaze.c (microblaze_attribute_table):
2286 * config/spu/spu.c (spu_attribute_table): Likewise.
2287 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
2288 * config/arc/arc.c (arc_attribute_table): Likewise.
2289 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
2290 * config/v850/v850.c (v850_handle_interrupt_attribute,
2291 v850_handle_data_area_attribute): Formatting fixes.
2292 (v850_attribute_table): Swap affects_type_identity and handler
2293 fields, adjust comments.
2294 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
2295 * config/arm/arm.c (arm_attribute_table): Likewise.
2296 * config/avr/avr.c (avr_attribute_table): Likewise.
2297 * config/s390/s390.c (s390_attribute_table): Likewise.
2298 * config/sh/sh.c (sh_attribute_table): Likewise.
2299 * config/i386/i386.c (ix86_handle_cconv_attribute,
2300 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
2301 (ix86_attribute_table): Swap affects_type_identity and handler
2302 fields, adjust comments.
2303 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2304 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
2305 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
2306 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
2307 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
2308 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2309 * config/rx/rx.c (rx_attribute_table): Likewise.
2310 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
2311 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
2312 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
2313 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
2314 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2315 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
2316 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
2317 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
2318 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2319 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2320 * doc/plugins.texi (user_attr): Likewise. Add NULL for
2322 * attribs.c (empty_attribute_table): Swap affects_type_identity and
2324 (register_scoped_attributes, decl_attributes): Formatting fixes.
2326 PR tree-optimization/83269
2327 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
2328 subtraction in arg0's type if type is signed and arg0 is unsigned.
2332 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
2333 convert? on @0 instead of convert. Check type of @1, not @0.
2334 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
2335 convert? on @0 instead of convert. Check type of @1, not @0.
2336 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
2337 only optimize if either both @1 and @2 types are narrower
2338 precision, or both are wider or equal precision, and in the former
2339 case only if both have undefined overflow.
2341 2017-12-15 Richard Biener <rguenther@suse.de>
2344 * internal-fn.def (IFN_NOP): Add.
2345 * internal-fn.c (expand_NOP): Do nothing.
2346 * lto-streamer-in.c (input_function): Instead of removing
2347 sanitizer calls replace them with IFN_NOP calls.
2349 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
2350 Alan Hayward <alan.hayward@arm.com>
2351 David Sherwood <david.sherwood@arm.com>
2353 * dse.c (store_info, read_info_type): Replace begin and end with
2355 (print_range): New function.
2356 (set_all_positions_unneeded, any_positions_needed_p)
2357 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
2359 (record_store): Likewise. Optimize the case in which all positions
2361 (get_stored_val): Replace read_begin and read_end with read_offset
2363 (replace_read): Update call accordingly.
2365 2017-12-15 Bin Cheng <bin.cheng@arm.com>
2367 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
2368 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
2369 (loop_cand::loop_cand): Initialize above members.
2370 (loop_cand::supported_operations): Delete.
2371 (loop_cand::can_interchange_p): Inline above function.
2372 (loop_cand::classify_simple_reduction): Record number of constant
2373 initialized simple reductions.
2374 (should_interchange_loops): New parameters. Check stmt cost of loops
2376 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
2377 Update call to should_interchange_loops.
2378 (should_interchange_loop_nest): Update call to
2379 should_interchange_loops.
2381 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
2384 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
2385 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
2386 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
2388 2017-12-15 Julia Koval <julia.koval@intel.com>
2390 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
2391 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
2393 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
2394 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
2395 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
2397 2017-12-15 Julia Koval <julia.koval@intel.com>
2399 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
2400 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
2401 * config/i386/sse.md (vaesenc_<mode>): New pattern.
2402 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
2403 _mm_aesenc_epi128): New intrinsics.
2405 2017-12-15 Julia Koval <julia.koval@intel.com>
2407 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
2408 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
2410 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
2411 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
2412 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
2414 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2416 * gimple-ssa-strength-reduction.c (analyze_increments):
2417 Distinguish replacement costs for constant strides from those for
2420 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2422 * var-tracking.c (variable_tracking_main_1): Formatting fix.
2424 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2426 * doc/invoke.texi: Document -Wcast-function-type.
2427 * recog.h (stored_funcptr): Change signature.
2428 * tree-dump.c (dump_node): Avoid warning.
2429 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
2431 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
2434 * gimple-fold.c (get_range_strlen): Add the handling of non-member
2437 2017-12-14 David Malcolm <dmalcolm@redhat.com>
2439 PR tree-optimization/83312
2440 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
2441 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
2442 "val" param, and to cope with arbitrary basic blocks.
2443 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
2444 handle NULL_TREE for "val", dropping "bb" param.
2445 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
2446 drop "bb" param. Handle NULL_TREE for "val".
2447 (find_case_label_for_value): Make "switch_stmt" param const.
2448 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
2450 (vrp_prop::check_all_array_refs): Reimplement as...
2451 (check_array_bounds_dom_walker::before_dom_children): ...this new
2452 vfunc. Replace linear search through BB block list, excluding
2453 those with non-executable in-edges via dominator walk.
2455 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2457 * config/arm/arm.opt (mverbose-cost-dump): New option.
2458 * config/arm/arm.c (arm_rtx_costs): Use it.
2460 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
2463 * reload1.c (emit_input_reload_insns): Skip debug markers.
2465 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
2467 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
2471 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
2472 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
2473 * config/sh/sh.c: Skip debug insns besides notes.
2474 * config/sh/sh.md: Likewise.
2475 * config/sh/sh_treg_combine.cc: Likewise.
2476 * config/sh/sync.md: Likewise.
2478 2017-12-14 Tom de Vries <tom@codesourcery.com>
2480 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2481 Add item for weak_undefined.
2483 2017-12-14 Richard Biener <rguenther@suse.de>
2485 PR tree-optimization/67842
2486 * sese.h (bb_in_region): Remove #if 0'ed code.
2488 2017-12-14 Richard Biener <rguenther@suse.de>
2490 PR tree-optimization/83326
2491 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2492 may_be_zero parameter and handle it by not marking the first
2493 peeled copy as not exiting the loop.
2494 (try_peel_loop): Likewise.
2495 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
2496 to handle the case of constant or zero iterations and perform
2497 loop header copying on-the-fly.
2499 2017-12-14 Richard Biener <rguenther@suse.de>
2501 PR tree-optimization/83418
2502 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
2503 Instead of asserting we don't get unfolded comparisons deal with them.
2505 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2508 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
2509 basic blocks. Assert debug bind insns don't appear outside of bbs,
2510 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
2513 PR tree-optimization/83198
2514 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
2515 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
2516 value if arg is a REAL_CST with incompatible type.
2518 2017-12-14 Sudakshina Das <sudi.das@arm.com>
2519 Bin Cheng <bin.cheng@arm.com>
2522 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
2524 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
2527 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2529 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
2530 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
2532 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
2533 Alan Hayward <alan.hayward@arm.com>
2534 David Sherwood <david.sherwood@arm.com>
2536 * poly-int.h: New file.
2537 * poly-int-types.h: Likewise.
2538 * coretypes.h: Include them.
2539 (POLY_INT_CONVERSION): Define.
2540 * target.def (estimated_poly_value): New hook.
2541 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
2542 * doc/tm.texi: Regenerate.
2543 * doc/poly-int.texi: New file.
2544 * doc/gccint.texi: Include it.
2545 * doc/rtl.texi: Describe restrictions on subreg modes.
2546 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
2547 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
2548 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
2549 * targhooks.h (default_estimated_poly_value): Declare.
2550 * targhooks.c (default_estimated_poly_value): New function.
2551 * target.h (estimated_poly_value): Likewise.
2552 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
2553 (wi::unary_traits): Delete.
2554 (wi::binary_traits::signed_shift_result_type): Define for
2555 offset_int << HOST_WIDE_INT, etc.
2556 (generic_wide_int::operator <<=): Define for all types and use
2557 wi::lshift instead of <<.
2558 (wi::hwi_with_prec): Add a default constructor.
2559 (wi::ints_for): New class.
2560 (operator <<): Define for all wide-int types.
2561 (operator /): New function.
2562 (operator %): Likewise.
2563 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
2564 (ASSERT_MAYBE_NE_AT): New macros.
2566 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
2567 Dominik Vogt <vogt@linux.vnet.ibm.com>
2570 * emit-rtl.c (init_emit): Remove ??? comment.
2571 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2572 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
2573 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
2574 alignment of 3 virtual registers to BITS_PER_WORD.
2576 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
2578 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
2580 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
2581 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
2583 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2586 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
2588 2017-12-13 Jakub Jelinek <jakub@redhat.com>
2590 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
2591 with the exception of debug begin stmt markers appear before
2595 * final.c (rest_of_handle_final): Call variable_tracking_main only
2596 if !flag_var_tracking.
2598 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2599 Jakub Jelinek <jakub@redhat.com>
2603 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
2604 labels debug stmts that can only appear after labels.
2606 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
2608 PR rtl-optimization/82398
2609 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
2610 EXPR_USEFULNESS in priority comparison.
2612 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
2614 PR rtl-optimization/83393
2615 * combine.c (move_deaths): If reg_stat points to a too new insn in
2616 last_death, do not use it: find the proper insn instead.
2618 2017-12-12 Jeff Law <law@redhat.com>
2620 PR tree-optimization/83298
2621 PR tree-optimization/83362
2622 PR tree-optimization/83383
2623 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
2624 push_value_range a public interface. Add new argument to
2625 record_ranges_from_stmt.
2626 * gimple-ssa-evrp-analyze.c
2627 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
2628 Update comments. Handle recording temporary equivalences.
2629 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
2630 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
2631 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
2632 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
2633 gimple-ssa-evrp-analyze.h.
2634 (record_temporary_equivalences_from_phis): Add new argument. When
2635 the PHI arg is an SSA_NAME, set the result's range to the range
2637 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
2638 from statements too.
2639 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
2640 Pass it down to children as needed.
2641 (thread_outgoing_edges): Likewise.
2642 (thread_across_edge): Likewise. Push/pop range state as needed.
2643 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
2645 2017-12-12 Julia Koval <julia.koval@intel.com>
2647 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
2648 (PTA_CANNONLAKE): Remove PTA_CLWB.
2650 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2653 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
2654 (vcondv2div2df): Likewise.
2655 (vconduv2dfv2di): Likewise.
2657 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2659 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
2660 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
2661 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
2662 ATTR_MATHFN_FPROUNDING_ERRNO.
2664 2017-12-12 Richard Biener <rguenther@suse.de>
2666 PR tree-optimization/83385
2667 * graphite-scop-detection.c (get_order, order): Remove.
2668 (bb_to_rpo): New global.
2670 (build_scops): Sort pbbs in RPO order.
2672 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
2674 * combine.c (simplify_set): Do not transform subregs to zero_extends
2675 if the destination mode is a vector mode.
2677 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2679 PR tree-optimization/83359
2680 * tree-cfg.h (fold_loop_internal_call): Declare.
2681 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
2682 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
2683 (find_loop_dist_alias): New function.
2684 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
2685 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
2688 PR tree-optimization/80631
2689 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
2690 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
2691 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
2692 hardcoding zero as the value if COND_EXPR is never true. For
2693 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
2694 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
2695 hardcoding MAX_EXPR as the reduction operation.
2696 (is_nonwrapping_integer_induction): Allow negative step.
2697 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
2698 vect_create_epilog_for_reduction, if no value is suitable, don't
2699 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
2701 2017-12-12 Richard Biener <rguenther@suse.de>
2703 PR tree-optimization/81889
2704 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
2705 range info from the non-wrapping IV instead of just the range
2708 2017-12-12 Julia Koval <julia.koval@intel.com>
2710 * config.gcc: Add vaesintrin.h.
2711 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2713 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2714 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2716 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2717 * config/i386/immintrin.h: Include vaesintrin.h.
2718 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2719 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2720 _mm_aesdec_epi128): New intrinsics.
2722 2017-12-12 Julia Koval <julia.koval@intel.com>
2724 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
2725 OPTION_MASK_ISA_VAES_UNSET): New.
2726 (ix86_handle_option): Handle -mvaes.
2727 * config/i386/cpuid.h: Define bit_VAES.
2728 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
2729 * config/i386/i386-c.c (__VAES__): New.
2730 * config/i386/i386.c (ix86_target_string): Add -mvaes.
2731 (ix86_valid_target_attribute_inner_p): Ditto.
2732 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
2733 * config/i386/i386.opt: Add -mvaes.
2734 * doc/invoke.texi: Ditto.
2736 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
2738 * debug.h (gcc_debug_hooks): Add inline_entry.
2739 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
2740 * debug.c (do_nothing_debug_hooks): Likewise.
2741 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2742 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
2743 (dwarf2_lineno_debug_hooks): Likewise.
2745 * common.opt (gstatement-frontiers): New, setting
2746 debug_nonbind_markers_p.
2747 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
2748 * toplev.c (process_options): Autodetect value for debug statement
2750 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
2751 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
2753 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
2754 markers. Integrate source bind into debug stmt expand loop.
2755 (pass_expand::execute): Check debug marker limit. Avoid deep
2756 TER and expand debug locations for debug bind insns only.
2757 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
2759 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
2760 * final.c (reemit_insn_block_notes): Take current block from
2761 nonbind markers. Declare note where it's first set.
2762 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
2763 begin stmt markers if enabled.
2764 (notice_source_line): Handle nonbind markers. Fail if their
2765 location is unknown or that of builtins.
2766 (rest_of_handle_final): Convert begin stmt markers to notes if
2767 var-tracking didn't run.
2768 (rest_of_clean_state): Skip begin stmt markers.
2769 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
2771 * function.c (allocate_struct_function): Set begin_stmt_markers.
2772 * function.h (struct function): Add debug_marker_count counter
2773 and debug_nonbind_markers flag.
2774 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
2775 * gimple-low.c (lower_function_body): Adjust
2776 debug_nonbind_markers.
2777 (lower_stmt): Drop or skip gimple debug stmts.
2778 (lower_try_catch): Skip debug stmts.
2779 * gimple.c (gimple_build_debug_begin_stmt): New.
2780 (gimple_copy): Increment debug_marker_count if copying one.
2781 * gimple.h (gimple_build_debug_begin_stmt): Declare.
2782 * gimplify.c (rexpr_location): New.
2783 (rexpr_has_location): New.
2784 (warn_switch_unreachable_r): Handle gimple debug stmts.
2785 (shortcut_cond_r): Call expr_location.
2787 (find_goto_label): New.
2788 (shortcut_cond_expr): Call expr_has_location, expr_location, and
2790 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
2792 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
2793 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
2794 (LANG_HOOKS_INITIALIZER): ... this.
2795 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
2796 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
2798 (update_ebb_live_info): Skip debug insn markers.
2799 * lra.c (debug_insn_static_data): Rename to...
2800 (debug_bind_static_data): ... this.
2801 (debug_marker_static_data): New.
2802 (lra_set_insn_recog_data): Select one of the above depending
2804 (lra_update_isn_regno_info): Don't assume debug insns have
2806 (push_insns): Skip debug insns.
2807 * lto-streamer-in.c (input_function): Drop debug stmts
2808 depending on active options. Adjust debug_nonbind_markers.
2809 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
2810 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
2811 begin stmt marker notes.
2812 (print_insn): Likewise.
2813 * recog.c (extract_insn): Recognize rtl for debug markers.
2814 * rtl.def (DEBUG_MARKER): New.
2815 * tree-inline.c: Include params.h.
2816 (remap_gimple_stmt): Handle nonbind markers.
2817 (maybe_move_debug_stmts_to_successors): Likewise.
2818 (copy_debug_stmt): Likewise.
2819 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
2820 markers regardless of no side effects.
2821 (tsi_link_before): Don't update container's side effects when adding
2822 a begin stmt marker.
2823 (tsi_link_after): Likewise.
2824 (expr_first): Skip begin stmt markers.
2825 (expr_last): Likewise.
2826 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
2827 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2828 Disregard nonbind markers.
2829 * tree.c (make_node_stat): Don't set side effects for begin stmt
2831 (build1_stat): Likewise.
2832 * tree.def (DEBUG_BEGIN_STMT): New.
2833 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
2834 * var-tracking.c (delete_debug_insns): Renamed to...
2835 (delete_vta_debug_insns): ... this.
2836 (reemit_marker_as_note): New.
2837 (vt_initialize): Reemit markers.
2838 (delete_vta_debug_insns): Likewise.
2839 (vt_debug_insns_local): Reemit or delete markers.
2840 (variable_tracking_main_1): Likewise.
2841 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
2842 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
2843 (gimple_debug_nonbind_marker_p): New.
2844 (gimple_build_debug_bind): Adjust.
2845 (gimple_build_debug_begin_stmt): New.
2846 * doc/invoke.texi (max-debug-marker-count): New param.
2847 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
2848 (debug_parameter_ref, debug_marker): New.
2849 (NOTE_INSN_BEGIN_STMT): New.
2850 (DEBUG_INSN): Describe begin stmt markers.
2852 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
2853 without debug insns, we wouldn't, but clean up debug insns
2854 after a control flow insn nevertheless.
2856 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
2857 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
2858 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
2859 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
2860 (rtl_tidy_fallthru_edge): Likewise.
2861 (rtl_verify_fallthru): Likewise.
2862 (rtl_verify_bb_layout): Likewise.
2863 (skip_insns_after_block): Likewise.
2864 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
2865 * dwarf2out.c: Include print-rtl.h.
2866 (dwarf2out_next_real_insn): New.
2867 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
2868 Dump debug binds in asm comments.
2869 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
2870 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
2871 callers to use gsi_start_nondebug_bb instead.
2872 (gsi_after_labels): Skip gimple debug stmts.
2873 (gsi_start_nondebug): New.
2874 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
2875 (proper_loop_form_for_interchange): Adjust.
2876 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
2877 * gimple.h (gimple_seq_last_nondebug_stmt): New.
2878 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
2879 (collect_fallthrough_labels): Likewise.
2880 (should_warn_for_implicit_fallthrough): Likewise.
2881 (warn_implicit_fallthrough_r): Likewise.
2882 (expand_FALLTHROUGH_r): Likewise.
2883 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
2884 (graphite_copy_stmts_from_block): Skip nonbind markers.
2885 * haifa-sched.c (sched_extend_bb): Skip debug insns.
2886 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
2887 * jump.c (clean_barriers): Skip debug insns.
2888 * omp-expand.c (expand_parallel_call): Skip debug insns.
2889 (expand_task_call): Likewise.
2890 (remove_exit_barrier): Likewise.
2891 (expand_omp_taskreg): Likewise.
2892 (expand_omp_for_init_counts): Likewise.
2893 (expand_omp_for_generic): Likewise.
2894 (expand_omp_for_static_nochunk): Likewise.
2895 (expand_omp_for_static_chunk): Likewise.
2896 (expand_omp_simd): Likewise.
2897 (expand_omp_taskloop_for_outer): Likewise.
2898 (expand_omp_taskloop_for_inner): Likewise.
2899 (expand_oacc_for): Likewise.
2900 (expand_omp_sections): Likewise.
2901 (expand_omp_single): Likewise.
2902 (expand_omp_synch): Likewise.
2903 (expand_omp_atomic_load): Likewise.
2904 (expand_omp_atomic_store): Likewise.
2905 (expand_omp_atomic_fetch_op): Likewise.
2906 (expand_omp_atomic_pipeline): Likewise.
2907 (expand_omp_atomic_mutex): Likewise.
2908 (expand_omp_target): Likewise.
2909 (grid_expand_omp_for_loop): Likewise.
2910 (grid_expand_target_grid_body): Likewise.
2911 (build_omp_regions_1): Likewise.
2912 * omp-low.c (check_combined_parallel): Skip debug stmts.
2913 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
2914 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
2915 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
2917 * tree-cfg.c (make_blobs_1): Skip debug stmts.
2918 (make_edges): Likewise.
2919 (cleanup_dead_labels): Likewise.
2920 (gimple_can_merge_blocks_p): Likewise.
2921 (stmt_starts_bb_p): Likewise.
2922 (gimple_block_label): Likewise.
2923 (gimple_redirect_edge_and_branch): Likewise.
2924 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
2926 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
2928 * tree-pretty-print (print_declaration): Omit initializer in slim
2930 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
2932 (eliminate_unnecessary_stmts): Stabilize block removal order.
2933 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
2934 * var-tracking.c (get_first_insn): New.
2935 (vt_emit_notes): Call it.
2936 (vt_initialize): Walk any insns before the first BB.
2937 (delete_debug_insns): Likewise.
2939 * gimple.h (enum gimple_debug_subcode): Add
2940 GIMPLE_DEBUG_BEGIN_STMT.
2941 (gimple_debug_begin_stmt_p): New.
2942 (gimple_debug_nonbind_marker_p): New.
2943 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
2944 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
2945 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
2946 * insn-notes.def (BEGIN_STMT): New.
2947 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
2948 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
2949 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
2950 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
2951 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
2952 (INSN_DEBUG_MARKER_KIND): New.
2953 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
2954 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
2955 (INSN_VAR_LOCATION_PTR): New.
2956 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
2958 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
2959 deps only if debug bind insns are enabled.
2960 (pass_expand::execute): Avoid deep TER and expand
2961 debug locations for debug bind insns only.
2962 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
2963 debug stmts special handling down to debug bind stmts.
2964 * combine.c (try_combine): Narrow debug insns special handling
2965 down to debug bind insns.
2966 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
2967 Narrow debug insns preexisting special handling down to debug
2969 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
2970 handling down to debug bind insns.
2971 * function.c (instantiate_virtual_regs): Skip debug markers,
2972 adjust handling of debug binds.
2973 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
2974 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
2975 * haifa-sched.c (schedule_insn): Narrow special handling of debug
2976 insns to debug bind insns.
2977 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
2978 special handling of debug stmts to debug bind stmts.
2979 * ipa-split.c (split_function): Likewise.
2980 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
2981 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
2983 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
2984 * regrename.c (build_def_use): Likewise.
2985 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
2986 (pass_cprop_hardreg): Narrow special casing of debug insns to
2988 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
2989 * reload1.c (reload): Likewise.
2990 * sese.c (sese_insert_phis_for_liveouts): Narrow special
2991 casing of debug stmts to debug bind stmts.
2992 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
2993 * ssa-iterators.h (num_imm_uses): Likewise.
2994 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
2995 debug stmts to debug bind stmts.
2996 * tree-inline.c (tree_function_versioning): Narrow special casing
2997 of debug stmts to debug bind stmts.
2998 * tree-loop-distribution.c (generate_loops_for_partition):
2999 Narrow special casing of debug stmts to debug bind stmts.
3000 * tree-sra.c (analyze_access_subtree): Narrow special casing
3001 of debug stmts to debug bind stmts.
3002 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
3003 stmts to debug bind stmts.
3004 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
3005 casing of debug stmts to debug bind stmts.
3006 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
3007 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
3008 casing of debug stmts to debug bind stmts.
3009 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3011 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
3012 debug stmts to debug bind stmts.
3013 (gimple_replace_ssa_lhs): Likewise.
3014 (insert_debug_temp_for_var_def): Likewise.
3015 (insert_debug_temps_for_defs): Likewise.
3016 (reset_debug_uses): Likewise.
3017 * tree-ssanames.c (release_ssa_name_fn): Likewise.
3018 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
3019 (adjust_debug_stmts): Likewise.
3020 (adjust_phi_and_debug_stmts): Likewise.
3021 (vect_do_peeling): Likewise.
3022 * tree-vect-loop.c (vect_transform_loop): Likewise.
3023 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
3024 * var-tracking.c (adjust_mems): Narrow special casing of debug
3025 insns to debug bind insns.
3026 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
3027 (compute_bb_dataflow, vt_find_locations): Likewise.
3028 (vt_expand_loc, emit_notes_for_changes): Likewise.
3029 (vt_init_cfa_base): Likewise.
3030 (vt_emit_notes): Likewise.
3031 (vt_initialize): Likewise.
3032 (vt_finalize): Likewise.
3034 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
3035 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
3036 (next_nonnote_nondebug_insn_bb): New.
3037 (prev_nonnote_nondebug_insn_bb): New.
3038 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
3039 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
3040 (prev_nonnote_nondebug_insn_bb): Declare.
3041 (next_nonnote_nondebug_insn_bb): Declare.
3042 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
3043 * cfgrtl.c (get_last_bb_insn): Likewise.
3044 * lra.c (push_insns): Likewise.
3046 2017-12-11 David Malcolm <dmalcolm@redhat.com>
3049 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
3051 * selftest.c (selftest::test_runner::test_runner): New ctor.
3052 (selftest::test_runner::~test_runner): New dtor.
3053 * selftest.h (class selftest::test_runner): New class.
3055 2017-12-11 Carl Love <cel@us.ibm.com>
3057 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
3058 vec_extract_fp32_from_shortl]): Add #defines.
3059 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
3060 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
3061 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
3062 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
3063 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
3064 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
3065 * doc/extend.texi: Add documentation for the added builtins.
3067 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
3069 PR rtl-optimization/80693
3070 PR rtl-optimization/81019
3071 PR rtl-optimization/81020
3072 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
3073 are not mentioned in i3. Place the REG_UNUSED note on i2,
3074 possibly modified to REG_DEAD, if it did not originate in i3.
3076 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3078 * recog.c (store_data_bypass_p_1): New function.
3079 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
3080 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
3082 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
3084 PR rtl-optimization/83361
3085 * ifcvt.c (if_convert): Call fixup_partitions.
3087 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
3089 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3090 early folding of splat_u{8,16,32}.
3092 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3094 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
3095 output_operand_lossage first argument with capital letter.
3096 (aarch64_override_options): Don't start error and sorry first argument
3097 with capital letter.
3099 2017-12-11 Andi Kleen <ak@linux.intel.com>
3101 PR gcov-profile/83355
3102 * auto-profile.c (string_table::get_index_by_decl): Don't
3103 recurse when abstract origin points to itself.
3105 2017-12-11 Bin Cheng <bin.cheng@arm.com>
3107 PR tree-optimization/83320
3108 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
3109 (prune_datarefs_not_in_loop): Ditto.
3111 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
3113 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
3115 2017-12-10 Jakub Jelinek <jakub@redhat.com>
3117 PR tree-optimization/83337
3118 * gimple-loop-interchange.cc (compute_access_stride): Handle
3119 bitfield DRs properly.
3121 2017-12-09 Jakub Jelinek <jakub@redhat.com>
3123 PR tree-optimization/83338
3124 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
3125 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
3126 vector of unsigned integers to vector of signed integers.
3128 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
3130 PR rtl-optimization/83317
3131 * lra-constraints.c (process_address_1): Add insn code check.
3133 2017-12-08 Michael Matz <matz@suse.de>
3135 Fix PR tree-optimization/83323
3136 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
3137 head-controlled loops and loop BBs.
3138 * common.opt (funroll-and-jam): Remove, instead ...
3139 (floop-unroll-and-jam): ... reuse this option.
3140 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
3141 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
3142 (-floop-unroll-and-jam): ... this option.
3144 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3146 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
3147 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
3148 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
3150 2017-12-08 Richard Biener <rguenther@suse.de>
3153 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
3154 handle accesses outside of zero-sized vars.
3156 2017-12-08 Martin Jambor <mjambor@suse.cz>
3158 PR tree-optimization/83141
3159 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
3160 test for MEM_REFs implicitely changing types with padding. Remove
3162 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
3164 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3166 * config/arc/arc.c (arc_attribute_table): Add exclusions to
3168 * config/avr/avr.c (avr_attribute_table): Likewise.
3169 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3170 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3171 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3172 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
3173 of struct attribute_spec.
3174 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3176 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
3179 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
3181 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
3184 * profile-count.c (profile_count::from_gcov_type): Move from
3185 profile-count.h; handle overflow.
3186 * profile-count.h (profile_count::from_gcov_type): Move offline.
3188 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
3190 PR rtl-optimization/83304
3191 * combine.c (move_deaths): If we do not know where a register died,
3194 2017-12-08 Richard Biener <rguenther@suse.de>
3196 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
3197 Provide -fopt-info-loop feedback when we interchange in a nest.
3199 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3201 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
3202 for armv6 ARM CPU IDs.
3204 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3206 * common/config/arm/arm-common.c: Include <algorithm>.
3207 (INCLUDE_VECTOR): Define.
3208 (compare_opt_names): New function.
3209 (arm_rewrite_selected_arch): Only strip out extensions that can be
3210 expressed through -mfpu. Sort the remaining extensions
3213 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3215 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
3216 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
3217 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
3218 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
3220 2017-06-08 Tristan Gingold <gindold@adacore.com>
3223 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
3224 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
3226 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3228 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
3229 of struct attribute_spec.
3231 2017-12-08 Julia Koval <julia.koval@intel.com>
3233 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
3234 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
3235 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
3236 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
3237 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
3238 _mm_maskz_dpwssds_epi32): Ditto.
3240 2017-12-08 Richard Biener <rguenther@suse.de>
3242 PR tree-optimization/81303
3243 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
3244 conditions try to create a comparison vector type matching
3245 the data vector type.
3246 (vectorizable_condition): Adjust.
3247 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
3248 Leave invariant conditions alone in case we can vectorize those.
3250 2017-12-08 Julia Koval <julia.koval@intel.com>
3252 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
3253 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
3254 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
3255 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
3256 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
3258 2017-12-08 Julia Koval <julia.koval@intel.com>
3260 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
3261 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
3262 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
3263 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
3264 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
3265 _mm_maskz_dpbusds_epi32): New intrinsics.
3267 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3269 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
3271 (ld<bh>io_signed): Likewise.
3272 (st<bhw>io): Likewise.
3273 * config/nios2/predicates.md (ldstio_memory_operand): Allow
3274 SMALL_INT12 constant integer operand.
3276 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3278 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
3280 (nios2_large_constant_allowed): ...this. Adjust uses.
3281 (nios2_plus_symbolic_constant_p): Rename to...
3282 (nios2_plus_large_constant_p): ...this. Adjust uses.
3283 (nios2_legitimate_address_p): Correct CONST_INT handling.
3284 (nios2_symbolic_memory_operand_p): Rename to...
3285 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
3286 (nios2_large_constant_p): Check for large constant integers too.
3287 (nios2_split_large_constant): Handle constant integers.
3288 (nios2_split_symbolic_memory_operand): Rename to...
3289 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
3290 (nios2_legitimize_constant_address): Handle constant integers.
3291 (r0rel_constant_p): Handle small constant integers.
3292 (nios2_print_operand_address): Handle r0-relative integer addresses.
3293 * config/nios2/nios2-protos.h: Adjust for renamed functions.
3294 * config/nios2/nios2.md: Adjust for renamed functions.
3296 2017-12-07 Andrew Waterman <andrew@sifive.com>
3298 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
3299 (TARGET_HAVE_SRODATA_SECTION): New define.
3300 (riscv_select_section): New function.
3302 2017-12-08 Joseph Myers <joseph@codesourcery.com>
3303 Alexander Monakov <amonakov@ispras.ru>
3304 Jakub Jelinek <jakub@redhat.com>
3307 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
3309 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3311 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
3312 a bool success value. Don't call output_operand_lossage here.
3313 (aarch64_print_ldpstp_address): Return a bool success value.
3314 (aarch64_print_operand_address): Call output_addr_const if
3315 aarch64_print_address_internal fails.
3316 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
3317 'y'; call output_operand_lossage instead. Call output_operand_lossage
3318 if aarch64_print_ldpstp_address fails.
3320 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3322 * tree-vector-builder.h
3323 (tree_vector_builder::binary_encoded_nelts): Declare.
3324 * tree-vector-builder.c
3325 (tree_vector_builder::binary_encoded_nelts): New function.
3326 * fold-const.c (negate_expr_p): Likewise.
3327 (operand_equal_p, fold_checksum_tree): Likewise.
3328 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
3329 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
3330 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
3331 (uniform_vector_p): Likewise.
3332 * varasm.c (const_hash_1, compare_constant): Likewise.
3333 * tree-ssa-ccp.c: Include tree-vector-builder.h.
3334 (valid_lattice_transition): Operate directly on the VECTOR_CST
3336 * ipa-icf.c: Include tree-vector-builder.h.
3337 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
3338 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
3340 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3342 * tree.c (build_vector): Delete.
3343 * tree.h (build_vector): Make static and move into the self-testing
3346 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3348 * vector-builder.h (vector_builder::derived): New const overload.
3349 (vector_builder::elt): New function.
3350 * tree-vector-builder.h (tree_vector_builder::type): New function.
3351 (tree_vector_builder::apply_step): Declare.
3352 * tree-vector-builder.c (tree_vector_builder::apply_step): New
3354 * gimple-fold.h (tree_vector_builder): Declare.
3355 (gimple_build_vector): Take a tree_vector_builder instead of a
3356 type and vector of elements.
3357 * gimple-fold.c (gimple_build_vector): Likewise.
3358 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
3360 (get_initial_defs_for_reduction): Likewise.
3361 (vectorizable_induction): Likewise.
3363 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3365 * tree-vector-builder.h
3366 (tree_vector_builder::new_binary_operation): Declare.
3367 * tree-vector-builder.c
3368 (tree_vector_builder::new_binary_operation): New function.
3369 * fold-const.c (fold_relational_const): Use it.
3370 (const_binop): Likewise. Check that both input vectors have
3371 the same number of elements, thus excluding things like WIDEN_SUM.
3372 Check whether it is possible to operate directly on the encodings
3375 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3377 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
3378 new_unary_operation, operating only on the encoded elements.
3379 (const_unop): Likewise.
3380 (exact_inverse): Likewise.
3381 (distributes_over_addition_p): New function.
3382 (const_binop): Use tree_vector_builder and new_unary_operation
3383 for combinations of VECTOR_CST and INTEGER_CST. Operate only
3384 on the encoded elements unless the encoding is strided and the
3385 operation does not distribute over addition.
3386 (fold_convert_const): Use tree_vector_builder and
3387 new_unary_operation. Operate only on the encoded elements
3388 for truncating integer conversions, or for non-stepped encodings.
3390 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3392 * config/sparc/sparc.c: Include tree-vector-builder.h.
3393 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
3394 * expmed.c: Include tree-vector-builder.h.
3395 (make_tree): Use tree_vector_builder instead of build_vector.
3396 * fold-const.c: Include tree-vector-builder.h.
3397 (const_binop): Use tree_vector_builder instead of build_vector.
3398 (const_unop): Likewise.
3399 (native_interpret_vector): Likewise.
3400 (fold_vec_perm): Likewise.
3401 (fold_ternary_loc): Likewise.
3402 * gimple-fold.c: Include tree-vector-builder.h.
3403 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
3405 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
3406 (simplify_vector_constructor): Use tree_vector_builder instead
3408 * tree-vect-generic.c: Include tree-vector-builder.h.
3409 (add_rshift): Use tree_vector_builder instead of build_vector.
3410 (expand_vector_divmod): Likewise.
3411 (optimize_vector_constructor): Likewise.
3412 * tree-vect-loop.c: Include tree-vector-builder.h.
3413 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
3414 of build_vector. Explicitly use a stepped encoding for
3416 * tree-vect-slp.c: Include tree-vector-builder.h.
3417 (vect_get_constant_vectors): Use tree_vector_builder instead
3419 (vect_transform_slp_perm_load): Likewise.
3420 (vect_schedule_slp_instance): Likewise.
3421 * tree-vect-stmts.c: Include tree-vector-builder.h.
3422 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
3423 (vect_gen_perm_mask_any): Likewise.
3424 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
3425 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
3427 (build_vector_from_val): Likewise. Explicitly use a duplicate
3430 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
3432 * doc/generic.texi (VECTOR_CST): Describe new representation of
3434 * vector-builder.h: New file.
3435 * tree-vector-builder.h: Likewise.
3436 * tree-vector-builder.c: Likewise.
3437 * Makefile.in (OBJS): Add tree-vector-builder.o.
3438 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
3439 * tree-core.h (tree_base): Add a vector_cst field to the u union.
3440 (tree_vector): Change the number of elements to
3441 vector_cst_encoded_nelts.
3442 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
3443 (VECTOR_CST_ELTS): Delete.
3444 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
3445 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
3446 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
3447 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
3448 (VECTOR_CST_ENCODED_ELT): Likewise.
3449 (vector_cst_encoded_nelts): New function.
3450 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3451 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3452 (vector_cst_int_elt, vector_cst_elt): Declare.
3453 * tree.c: Include tree-vector-builder.h.
3454 (tree_code_size): Abort if passed VECTOR_CST.
3455 (tree_size): Update for new VECTOR_CST layout.
3456 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3457 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3458 (build_vector): Use tree_vector_builder.
3459 (vector_cst_int_elt, vector_cst_elt): New functions.
3460 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
3461 encoded elements and then create the vector in the canonical form.
3462 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
3463 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
3464 (tree_c_tests): Call test_vector_cst_patterns.
3465 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
3467 (hash_tree): Likewise.
3468 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
3469 (streamer_write_tree_header): Likewise.
3470 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
3471 (streamer_alloc_tree): Likewise. Update call to make_vector.
3472 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
3474 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3476 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
3477 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
3478 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
3480 2017-12-07 Bin Cheng <bin.cheng@arm.com>
3481 Richard Biener <rguenther@suse.de>
3483 PR tree-optimization/81303
3484 * Makefile.in (gimple-loop-interchange.o): New object file.
3485 * common.opt (floop-interchange): Reuse the option from graphite.
3486 * doc/invoke.texi (-floop-interchange): Ditto. New document for
3487 -floop-interchange and mention it for -O3.
3488 * opts.c (default_options_table): Enable -floop-interchange at -O3.
3489 * gimple-loop-interchange.cc: New file.
3490 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
3491 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
3492 * passes.def (pass_linterchange): New pass.
3493 * timevar.def (TV_LINTERCHANGE): New time var.
3494 * tree-pass.h (make_pass_linterchange): New declaration.
3495 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
3496 interchange. Record IV before/after increment in new parameters.
3497 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
3498 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
3500 (check_reduction_path): ...New function here.
3501 * tree-vectorizer.h (check_reduction_path): New declaration.
3503 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
3506 PR rtl-optimization/80818
3507 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
3508 always early clobbered.
3509 * lra-lives.c (process_bb_lives): Check input hard regs for early
3510 clobbered non-operand hard reg.
3512 2017-12-07 Jakub Jelinek <jakub@redhat.com>
3515 * tree-cfg.c (verify_gimple_assign_binary): Don't require
3516 types_compatible_p, just that TYPE_MODE is the same.
3518 2017-12-07 Martin Sebor <msebor@redhat.com>
3521 * attribs.c (empty_attribute_table): Initialize new member of
3522 struct attribute_spec.
3523 (decl_attributes): Add argument. Handle mutually exclusive
3524 combinations of attributes.
3525 (selftests::test_attribute_exclusions): New function.
3526 (selftests::attribute_c_tests): Ditto.
3527 * attribs.h (decl_attributes): Add default argument.
3528 * selftest.h (attribute_c_tests): Declare.
3529 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
3530 * tree-core.h (attribute_spec::exclusions, exclude): New type and
3532 * doc/extend.texi (Common Function Attributes): Update const and pure.
3533 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
3534 of struct attribute_spec.
3535 * config/arc/arc.c (arc_attribute_table): Same.
3536 * config/arm/arm.c (arm_attribute_table): Same.
3537 * config/avr/avr.c ( avr_attribute_table): Same.
3538 * config/bfin/bfin.c (bfin_attribute_table): Same.
3539 * config/cr16/cr16.c (cr16_attribute_table): Same.
3540 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
3541 * config/h8300/h8300.c (h8300_attribute_table): Same.
3542 * config/i386/i386.c (ix86_attribute_table): Same.
3543 * config/ia64/ia64.c (ia64_attribute_table): Same.
3544 * config/m32c/m32c.c (m32c_attribute_table): Same.
3545 * config/m32r/m32r.c (m32r_attribute_table): Same.
3546 * config/m68k/m68k.c (m68k_attribute_table): Same.
3547 * config/mcore/mcore.c (mcore_attribute_table): Same.
3548 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
3549 * config/mips/mips.c (mips_attribute_table): Same.
3550 * config/msp430/msp430.c (msp430_attribute_table): Same.
3551 * config/nds32/nds32.c (nds32_attribute_table): Same.
3552 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
3553 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
3554 * config/rl78/rl78.c (rl78__attribute_table): Same.
3555 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
3556 * onfig/rx/rx.c (rx_attribute_table): Same.
3557 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
3558 * config/sh/sh.c (sh_attribute_table): Same.
3559 * config/sparc/sparc.c (sparc_attribute_table): Same.
3560 * config/spu/spu.c (spu_attribute_table): Same.
3561 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
3562 * config/v850/v850.c (v850_attribute_table): Same.
3563 * config/visium/visium.c (visium_attribute_table): Same.
3565 2017-12-07 Tamar Christina <tamar.christina@arm.com>
3568 * config/arm/arm.c (INCLUDE_STRING): Define.
3569 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
3570 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
3573 2017-12-07 Michael Matz <matz@suse.de>
3575 Add unroll and jam pass
3577 * gimple-loop-jam.c: New file.
3578 * Makefile.in (OBJS): Add gimple-loop-jam.o.
3579 * common.opt (funroll-and-jam): New option.
3580 * opts.c (default_options_table): Add unroll-and-jam at -O3.
3581 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
3582 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
3583 * passes.def: Add pass_loop_jam.
3584 * timevar.def (TV_LOOP_JAM): Add.
3585 * tree-pass.h (make_pass_loop_jam): Declare.
3586 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
3587 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
3588 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
3589 to flow_loop_tree_node_add.
3590 (duplicate_subloops, copy_loops_to): Append to sibling list.
3591 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
3592 * doc/invoke.texi (-funroll-and-jam): Document new option.
3593 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
3595 2017-12-07 Richard Biener <rguenther@suse.de>
3597 PR tree-optimization/83296
3598 PR tree-optimization/67769
3599 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
3600 flow sensitive info in an unrelated BB.
3601 (value_replacement): Use reset_flow_sensitive_info.
3602 (minmax_replacement): Reset flow sensitive info on the def
3603 we move. Do not reset flow sensitive info in the whole BB
3604 we move the stmt to.
3605 (abs_replacement): Likewise.
3607 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3610 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3611 rs6000_cpu to the given -mcpu=, or to the default processor.
3613 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3615 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
3616 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
3617 instead of rs6000_cpu_attr.
3618 (rs6000_adjust_cost): Ditto.
3619 (is_microcoded_insn): Ditto.
3620 (rs6000_adjust_priority): Ditto.
3621 (rs6000_issue_rate): Ditto.
3622 (rs6000_use_sched_lookahead): Ditto.
3623 (rs6000_use_sched_lookahead_guard): Ditto.
3624 (rs6000_sched_reorder): Ditto.
3625 (force_new_group): Ditto.
3626 * config/rs6000/rs6000.md (cpu attribute): Ditto.
3627 (group_ending_nop): Ditto.
3629 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3631 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
3632 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
3633 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
3634 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
3636 (rs6000_reassociation_width): Ditto.
3637 (rs6000_emit_epilogue): Ditto.
3638 (rs6000_adjust_cost): Ditto.
3639 (is_microcoded_insn): Ditto.
3640 (is_cracked_insn): Ditto.
3641 (rs6000_adjust_priority): Ditto.
3642 (rs6000_sched_reorder): Ditto.
3643 (rs6000_sched_reorder2): Ditto.
3644 (insn_must_be_first_in_group): Ditto.
3645 (insn_must_be_last_in_group): Ditto.
3646 (rs6000_register_move_cost): Ditto.
3647 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
3650 2017-12-07 Julia Koval <julia.koval@intel.com>
3652 * config.gcc: Add vaesintrin.h.
3653 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3655 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3656 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3658 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3659 * config/i386/immintrin.h: Include vaesintrin.h.
3660 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3661 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3662 _mm_aesdec_epi128): New intrinsics.
3664 2017-12-06 David Malcolm <dmalcolm@redhat.com>
3666 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
3667 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
3668 c-family/c-spellcheck.cc.
3669 (best_macro_match::best_macro_match): Likewise.
3671 (struct edit_distance_traits<cpp_hashnode *>): Move to
3672 c-family/c-spellcheck.h.
3673 (class best_macro_match): Likewise.
3675 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3677 PR tree-optimization/83293
3678 * gimple-ssa-strength-reduction.c (insert_initializers): Use
3679 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
3680 might insert into empty bb.
3683 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
3684 simplify for plus with :c added, and pointer_plus without that.
3685 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
3686 with undefined overflow and the conversion is not widening,
3687 perform negation in utype and only convert to type afterwards.
3688 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
3689 simplify for plus with :c added, and pointer_plus without that.
3690 If type is integral with undefined overflow and the conversion is
3691 not widening, perform minus in utype and only convert to type
3692 afterwards. Move the last pointer_diff_expr simplify into the
3695 2017-12-06 Martin Sebor <msebor@redhat.com>
3697 PR tree-optimization/82646
3698 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
3699 strncpy, not maxlen.
3701 2017-12-06 Martin Sebor <msebor@redhat.com>
3703 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
3706 PR tree-optimization/83075
3707 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
3708 strncat/strncpy don't change length of source string.
3710 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
3713 2017-11-29 Martin Aberg <maberg@gaisler.com>
3715 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3716 to prevent b2bst errata sequence.
3717 (sqrtdf2_fix): Likewise.
3719 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3721 PR tree-optimization/81945
3722 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
3723 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
3724 to dest_cfun has orig_loop_num set, either remap it to the new
3725 loop number if the loop got moved too, or clear it.
3727 2017-12-05 Steve Ellcey <sellcey@cavium.com>
3729 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
3731 (thunderx2t99_nothing): New insn reservation.
3732 (thunderx2t99_mrs): New insn reservation.
3733 (thunderx2t99_multiple): New insn reservation.
3734 (thunderx2t99_alu_basi): Add bfx to reservation.
3735 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
3737 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3740 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
3742 2017-12-05 Bin Cheng <bin.cheng@arm.com>
3744 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
3745 tree-ssa-pre.c::remove_dead_inserted_code.
3746 * tree-ssa-dce.h: New file.
3747 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
3748 (remove_dead_inserted_code): Move and rename to function
3749 tree-ssa-dce.c::simple_dce_from_worklist.
3750 (pass_pre::execute): Update use.
3752 2017-12-05 Richard Biener <rguenther@suse.de>
3754 PR tree-optimization/83277
3755 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
3756 to code-gen liveout vars.
3758 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
3760 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
3761 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
3762 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
3765 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
3767 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
3769 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
3770 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
3773 2017-12-05 Richard Biener <rguenther@suse.de>
3775 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
3776 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
3777 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
3778 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
3780 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3782 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
3784 2017-12-05 Martin Liska <mliska@suse.cz>
3785 Jakub Jelinek <jakub@redhat.com>
3787 * doc/invoke.texi: Document the options.
3788 * flag-types.h (enum sanitize_code): Add
3789 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3790 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
3791 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3792 * opts.c: Define new sanitizer options.
3793 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
3794 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
3796 2017-12-05 Julia Koval <julia.koval@intel.com>
3798 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
3799 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
3800 (ix86_handle_option): Handle -mavx512vnni.
3801 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
3802 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3803 * config/i386/i386-c (__AVX512VNNI__): New.
3804 * config/i386/i386.c (ix86_target_string): Handle new option.
3805 (ix86_valid_target_attribute_inner_p): Handle new option.
3806 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
3807 * config/i386/i386.opt (mavx512vnni): New option.
3809 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3812 * athlon.md: Disable for generic.
3813 * haswell.md: Enable for generic.
3814 * i386.c (ix86_sched_init_global): Add core hooks for generic.
3815 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
3817 (ix86_adjust_cost): Move generic to haswell path.
3819 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
3821 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
3822 instead of MEM_P in a couple more places. Fix formatting issues.
3824 2017-12-04 Jim Wilson <jimw@sifive.com>
3826 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
3827 instead of GP_REG_LAST-1.
3828 (riscv_adjust_libcall_cfi_prologue): Likewise.
3829 (riscv_adjust_libcall_cri_epilogue): Likewise.
3830 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
3833 2017-12-04 Luis Machado <luis.machado@linaro.org>
3835 * ipa-pure-const.c (check_decl): Add missing newline.
3836 (state_from_flags): Likewise.
3838 2017-12-04 Jeff Law <law@redhat.com>
3840 PR tree-optimizatin/78496
3841 * gimple-ssa-evrp-analyze.h
3842 (evrp_range_analyzer::get_vr_values): Simplify.
3843 * gimple-ssa-evrp-analyze.c: Corresponding changes.
3844 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
3845 and gimple-ssa-evrp-analyze.h.
3846 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
3847 (simplify_stmt_for_jump_threading): Copy a blob of code from
3848 tree-vrp.c to use ranges to simplify statements.
3849 (dom_opt_dom_walker::before_dom_children): Call
3850 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
3851 (dom_opt_dom_walker::after_dom_children): Similarly for
3852 evrp_range_analyzer::leave.
3853 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
3856 * gimple-ssa-evrp-analyze.c
3857 (evrp_range_analyzer::extract_range_from_stmt): Always use
3858 vr_values::update_value_range so preexisting range info is
3859 medged with new range info, even if the new range is VR_VARYING.
3861 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3863 * combine.c: Adjust comment.
3864 (use_crosses_set_p): Delete.
3865 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
3866 (try_combine): Ditto.
3868 2017-12-04 Richard Biener <rguenther@suse.de>
3870 PR tree-optimization/83255
3871 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
3872 Re-add zero-iteration check.
3874 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3876 PR rtl-optimization/83245
3877 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
3878 hard registers as earlyclobber, also if not in an asm.
3880 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3884 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3887 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3888 rs6000_cpu based on cpu_index, not tune_index.
3890 2017-12-04 Richard Biener <rguenther@suse.de>
3892 PR tree-optimization/83238
3893 * graphite-scop-detection.c (scop_detection::merge_sese): Make
3894 code match comment, rejecting invalid SESE regions.
3896 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
3898 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
3899 require base operand is a REG_POINTER prior to reload on targets
3900 with non-equivalent space registers.
3902 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3904 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
3905 (update_specialized_profile): Fix updating of counts.
3906 (perhaps_add_new_callers): Likewise.
3908 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3911 * x86-tune.def: Remove obsolette FIXMEs.
3912 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
3913 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
3914 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
3916 (X86_TUNE_PAD_RETURNS): Disable for generic.
3918 2017-12-02 Jakub Jelinek <jakub@redhat.com>
3920 PR tree-optimization/83170
3921 PR tree-optimization/83241
3922 * gimple-ssa-store-merging.c
3923 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
3924 gimple_vuse (ins_stmt) in case it has changed.
3925 (imm_store_chain_info::output_merged_store): Likewise.
3927 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
3931 * tree-cfg.c (pass_warn_function_return::execute): Handle
3932 __builtin_ubsan_handle_missing_return like __builtin_unreachable
3933 with BUILTINS_LOCATION.
3937 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
3938 is BLKmode for vector field with vector raw mode, use TYPE_MODE
3939 instead of DECL_MODE.
3941 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
3942 last argument to rtx pointer.
3943 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
3944 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
3945 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
3946 depending on the chosen ISAs.
3947 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
3948 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
3949 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
3951 * config/i386/sse.md (mov<mode>_internal): Likewise.
3952 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
3954 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3956 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
3958 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3960 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
3961 parameters from prototype.
3962 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
3963 parameters. Don't print a comment.
3964 (emit_fusion_gpr_load): Adjust.
3965 (emit_fusion_load_store): Adjust.
3966 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
3967 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
3968 comment on the second line.
3970 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3973 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3974 rs6000_cpu based on cpu_index, not tune_index.
3976 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3978 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
3979 which_alternative instead of which_alternative + 1.
3980 (output_asm_insn): Print an extra tab if the template is short.
3982 2017-12-01 Jim Wilson <jimw@sifive.com>
3984 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
3986 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
3987 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
3989 * doc/tm.texi: Regenerate.
3991 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
3994 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
3995 whether we can allocate pseudos before trying to fix an address.
3996 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
3997 memory address is indexed or indirect.
3998 (floatuns_<mode>si2_hw2): Likewise.
4000 2017-12-01 Jason Merrill <jason@redhat.com>
4002 * Makefile.in (TAGS): Add c-family/*.cc.
4004 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
4006 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
4007 (call_value_insn): Likewise.
4008 (sibcall_insn): Likewise.
4009 (sibcall_value_insn): Likewise.
4010 (movsi_aarch64): Likewise.
4011 (movdi_aarch64): Likewise.
4012 (add_losym_): Likewise.
4013 (ldr_got_small_): Likewise.
4014 (ldr_got_small_sidi): Likewise.
4015 (ldr_got_small_28k_): Likewise.
4016 (ldr_got_small_28k_sidi): Likewise.
4017 * config/aarch64/aarch64.c (aarch64_print_address_internal):
4018 Move output_addr_const to symbolic case. Add error check.
4020 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4022 * config/s390/predicates.md (plus16_Q_operand): New predicate.
4023 * config/s390/s390.md: Disable MVC merging peephole if it would
4024 disable operand forwarding.
4025 (new peephole2): Split MVCs if it would turn them into up to 2
4028 2017-12-01 Richard Biener <rguenther@suse.de>
4030 PR tree-optimization/83232
4031 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
4032 detection of same access. Instead of breaking the group here
4033 do not consider the duplicate. Add comment explaining real fix.
4035 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
4037 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
4039 2017-12-01 Sudakshina Das <sudi.das@arm.com>
4041 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
4043 2017-12-01 Jakub Jelinek <jakub@redhat.com>
4045 * function.h (struct function): Remove cilk_frame_decl,
4046 is_cilk_function and calls_cilk_spawn fields.
4047 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
4049 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
4050 cilk_elemental field.
4051 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
4052 * target.def: Adjust comment.
4053 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
4054 Don't test cilk_elemental.
4056 PR tree-optimization/83233
4057 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
4058 bswap_stat name for the struct.
4061 * tree.h (SWITCH_BREAK_LABEL_P): Define.
4062 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
4063 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
4064 SWITCH_BREAK_LABEL_P set on the label.
4065 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
4066 added for default case if it was missing and not all cases covered.
4067 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
4068 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
4070 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
4071 to avoid -Wimplicit-fallthrough warning.
4072 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
4073 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
4075 PR tree-optimization/83221
4076 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
4078 (init_reassoc): Formatting fix.
4081 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
4082 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
4086 * tree-cfg.c: Include asan.h.
4087 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
4089 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4091 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
4093 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4095 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
4096 vpcmpeqd instruction.
4098 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4100 * config/i386/i386.c (standard_sse_constant_opcode): Fix
4101 registers type for 128bit mode.
4103 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4105 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
4107 * gengtype-state.c (read_state_pair): Likewise.
4108 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
4109 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
4110 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
4111 * ifcvt.c (noce_try_store_flag_constants): Likewise.
4112 * tree-ssa-ccp.c (ccp_finalize): Likewise.
4113 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
4114 * builtins.c (fold_builtin_3): Likewise.
4115 * graphite-scop-detection.c
4116 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
4117 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
4120 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
4123 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
4124 parameter. Return true if ignore_decl_rtl_set_p is true and other
4125 conditions are satisfied.
4126 * asan.h (asan_protect_global): Add new parameter.
4127 * varasm.c (categorize_decl_for_section): Pass true as second parameter
4128 to asan_protect_global calls.
4130 2017-11-30 Jim Wilson <jimw@sifive.com>
4132 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
4133 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
4134 -mstrict-align, add info on default value. Delete redundant lines for
4135 -mabi. Add missing -mexplicit-relocs docs.
4137 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4139 * config/arc/arc.md (trap): New pattern.
4141 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4143 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
4144 instruction to end into a delay slot.
4145 * config/arc/arc.md (cond_delay_insn): Check if the instruction
4146 can be placed into a delay slot against reg_note.
4148 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4150 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
4151 labels number of usages.
4153 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4155 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
4157 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
4159 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4162 * internal-fn.c (expand_mul_overflow): Optimize unsigned
4163 multiplication by power of 2 constant into two shifts + comparison.
4165 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
4168 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
4170 2017-11-30 Richard Biener <rguenther@suse.de>
4172 PR tree-optimization/83202
4173 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
4174 allow_peel argument and guard peeling.
4175 (canonicalize_loop_induction_variables): Likewise.
4176 (canonicalize_induction_variables): Pass false.
4177 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
4178 peeling from cunrolli.
4180 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4182 * combine.c (try_combine): Print a message to dump file whenever
4183 I0, I1, or I2 cannot be combined into I3.
4185 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4187 PR rtl-optimization/83156
4188 PR rtl-optimization/82621
4189 * combine.c (try_combine): Don't split an I2 if one of the dests is
4190 set again before I3. Allow unused dests.
4192 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4194 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
4196 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
4198 PR rtl-optimization/80818
4199 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
4200 recursively. Use insn code for clobber.
4201 (lra_set_insn_recog_data): Pass the new arg to
4202 collect_non_operand_hard_regs.
4203 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
4205 (lra_update_insn_regno_info): Pass insn to
4206 add_regs_to_insn_regno_info.
4208 2017-11-29 Jim Wilson <jimw@sifive.com>
4209 Andrew Waterman <andrew@sifive.com>
4211 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
4212 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
4213 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
4214 use SINGLE_SHIFT_COST.
4215 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
4216 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
4218 2017-11-29 Julia Koval <julia.koval@intel.com>
4220 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
4221 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
4222 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
4223 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
4224 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
4225 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
4226 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
4227 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
4228 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
4229 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
4230 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
4231 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
4232 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
4233 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
4234 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
4235 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
4236 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
4237 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
4238 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
4239 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
4240 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
4241 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
4242 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
4243 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
4244 __builtin_ia32_vpshldv_v2di_mask,
4245 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
4246 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
4247 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
4249 2017-11-29 Julia Koval <julia.koval@intel.com>
4251 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
4252 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
4253 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
4254 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
4255 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
4256 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
4257 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
4258 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
4259 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
4260 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
4261 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
4262 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
4263 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
4264 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
4265 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
4266 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
4267 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
4268 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
4269 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4270 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
4271 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
4273 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4275 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
4276 movsi_pic_gotdata_op instruction as a load for the UT699 errata
4279 2017-11-29 Martin Aberg <maberg@gaisler.com>
4281 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4282 to prevent b2bst errata sequence.
4283 (sqrtdf2_fix): Likewise.
4285 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4287 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
4288 (div_sqrt_insn_p): New function.
4289 (sparc_do_work_around_errata): Insert NOP instructions to
4290 prevent sequences that could trigger the TN-0013 errata for
4291 certain LEON3 processors.
4292 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
4293 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
4294 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
4295 (in_branch_delay): Prevent div and sqrt in delay slot if
4297 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
4299 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4301 * config/sparc/sparc.c (atomic_insn_p): New function.
4302 (sparc_do_work_around_errata): Insert NOP instructions to
4303 prevent sequences that could trigger the TN-0010 errata for
4305 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
4306 instruction referable in atomic_insns_p.
4308 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4310 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
4311 (atomic_compare_and_swap_leon3_1): Likewise.
4314 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4316 * config/sparc/sparc.c (fpop_insn_p): New function.
4317 (sparc_do_work_around_errata): Insert NOP instructions to
4318 prevent sequences that could trigger the TN-0012 errata for
4320 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
4321 * config/sparc/sparc.md (fix_gr712rc): New attribute.
4322 (in_branch_annul_delay): Prevent floating-point instructions
4323 in delay slot of annulled integer branch.
4325 2017-11-29 Richard Biener <rguenther@suse.de>
4327 PR tree-optimization/83202
4328 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
4330 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
4331 nodes vectorized to the same stmts multiple times.
4332 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
4333 (vect_analyze_slp_instance): Adjust.
4334 (scalar_stmts_to_slp_tree_map_t): New typedef.
4335 (vect_schedule_slp_instance): Add a map recording the SLP node
4336 representing the vectorized stmts for a set of scalar stmts.
4337 Avoid code-generating redundancies.
4338 (vect_schedule_slp): Allocate map and pass it down.
4340 2017-11-29 Nathan Sidwell <nathan@acm.org>
4343 * tree.c (build_complex_type): Fix canonicalization. Only fill in
4346 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
4348 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
4349 cases for printing LDP/STP memory addresses.
4350 (aarch64_print_address_internal): Renamed from
4351 aarch64_print_operand_address, added parameter, add Pmode check.
4352 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
4353 (aarch64_print_operand_address): Indirect to
4354 aarch64_print_address_internal.
4355 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
4356 'y' operand output specifier.
4358 2017-11-29 Jakub Jelinek <jakub@redhat.com>
4361 * tree.c (build_simple_mem_ref_loc): Handle
4362 get_addr_base_and_unit_offset returning a MEM_REF.
4365 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
4368 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
4371 2017-11-28 David Malcolm <dmalcolm@redhat.com>
4373 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
4375 (layout::move_to_column): Likewise.
4377 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4380 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
4381 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
4383 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4384 Martin Jambor <mjambor@suse.cz>
4387 * tree.h (expr_type_first_operand_type_p): Declare
4388 * tree.c (expr_type_first_operand_type_p): New function.
4389 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
4390 (ipa_value_from_jfunc): Adjust declaration.
4391 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
4392 Use it as result type for arithmetics, unless it is NULL in which case
4393 be more conservative.
4394 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
4395 ipa_get_jf_pass_through_result.
4396 (propagate_vals_across_pass_through): Likewise.
4397 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
4398 is to propagate_vals_across_pass_through.
4399 (propagate_constants_across_call): Pass PARM_TYPE to
4400 propagate_scalar_across_jump_function.
4401 (find_more_scalar_values_for_callers_subset): Pass parameter type to
4402 ipa_value_from_jfunc.
4403 (cgraph_edge_brings_all_scalars_for_node): Likewise.
4404 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
4405 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
4406 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
4407 target_type, pass it to ipa_value_from_jfunc.
4408 (update_indirect_edges_after_inlining): Pass parameter type to
4409 try_make_edge_direct_simple_call.
4411 2017-11-28 Jeff Law <law@redhat.com>
4413 * gimple-ssa-evrp-analyze.c
4414 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
4415 refine ranges if scev_initialized_p returns true.
4416 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
4418 2017-11-28 Julia Koval <julia.koval@intel.com>
4420 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
4421 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
4422 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
4423 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
4424 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
4425 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
4426 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
4427 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
4428 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
4429 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
4430 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
4431 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
4432 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
4433 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
4434 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
4435 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
4436 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
4437 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
4438 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
4439 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
4440 __builtin_ia32_vpshrd_v2di_mask): New builtins.
4441 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
4443 2017-11-28 Julia Koval <julia.koval@intel.com>
4445 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
4446 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
4447 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
4448 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
4449 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
4450 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
4451 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
4452 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
4453 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
4454 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
4455 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
4456 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
4457 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
4458 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
4459 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
4460 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
4461 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
4462 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
4463 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
4464 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
4465 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
4466 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
4467 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
4468 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
4469 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
4470 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
4471 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
4472 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
4473 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
4474 __builtin_ia32_vpshld_v2di_mask): New builtins.
4475 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4476 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
4478 2017-11-28 Richard Biener <rguenther@suse.de>
4480 PR tree-optimization/80776
4481 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
4483 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
4485 (evrp_range_analyzer::record_ranges_from_incoming_edges):
4486 If the incoming edge is an effective fallthru because the other
4487 edge only reaches a __builtin_unreachable () then record ranges
4488 derived from the controlling condition in SSA info.
4489 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
4490 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4492 2017-11-28 Olivier Hainque <hainque@adacore.com>
4494 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
4495 on mingw build hosts.
4497 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4500 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
4502 2017-11-28 Richard Biener <rguenther@suse.de>
4505 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
4506 copies generated from memcpy use a character array as reference
4509 2017-11-28 Julia Koval <julia.koval@intel.com>
4510 Sebastian Peryt <sebastian.peryt@intel.com>
4512 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
4513 c-family/c-cilkplus.o, c-family/array-notation-common.o,
4514 cilk-common.o, cilk.h, cilk-common.c): Remove.
4516 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
4517 * builtins.c (is_builtin_name): Remove cilkplus condition.
4518 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
4519 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
4520 cilk-builtins.def, cilkplus.def): Remove.
4521 * cif-code.def (CILK_SPAWN): Remove.
4522 * cilk-builtins.def: Delete.
4523 * cilk-common.c: Ditto.
4525 * cilkplus.def: Ditto.
4526 * config/darwin.h (fcilkplus): Delete.
4527 * cppbuiltin.c: Ditto.
4528 * doc/extend.texi: Remove cilkplus doc.
4529 * doc/generic.texi: Ditto.
4530 * doc/invoke.texi: Ditto.
4531 * doc/passes.texi: Ditto.
4532 * gcc.c (fcilkplus): Remove.
4533 * gengtype.c (cilk.h): Remove.
4534 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
4536 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
4538 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
4540 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
4541 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
4542 cilkplus conditions.
4543 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
4544 inline_read_section): Ditto.
4545 * ipa-inline-analysis.c (cilk.h): Remove.
4546 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
4547 * lto-wrapper.c (merge_and_complain, append_compiler_options,
4548 append_linker_options): Remove condition for fcilkplus.
4549 * lto/lto-lang.c (cilk.h): Remove.
4550 (lto_init): Remove condition for fcilkplus.
4551 * omp-expand.c (expand_cilk_for_call): Delete.
4552 (expand_omp_taskreg, expand_omp_for_static_chunk,
4553 expand_omp_for): Remove cilkplus
4555 (expand_cilk_for): Delete.
4556 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
4557 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
4558 execute_lower_omp, diagnose_sb_0): Ditto.
4559 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
4560 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4561 * tree-nested.c: Ditto.
4562 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
4563 (dump_generic_node): Ditto.
4564 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4565 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
4567 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
4569 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4571 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
4572 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
4573 for vector mode and !TARGET_SIMD.
4575 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4577 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
4579 * tree.h (SWITCH_LABELS): Remove.
4580 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
4581 assert SWITCH_BODY is non-NULL.
4582 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
4584 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
4586 PR tree-optimization/80788
4587 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
4588 has TREE_OVERFLOW set, call drop_tree_overflow.
4590 2017-11-28 Richard Biener <rguenther@suse.de>
4592 PR tree-optimization/83158
4593 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
4595 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
4598 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
4599 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
4601 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
4603 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
4604 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
4606 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4609 * varasm.c (compare_constant): Take the mode of the constants into
4610 account when comparing floating point constants.
4612 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
4614 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
4615 from explicit instantiation of debug_helper.
4616 * vec.h (DEFINE_DEBUG_VEC): Ditto.
4618 2017-11-27 Richard Biener <rguenther@suse.de>
4620 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
4623 2017-11-27 Richard Biener <rguenther@suse.de>
4625 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
4626 (type_hash_canon): Also clear int_cst_hash_table entry for
4628 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
4630 2017-11-27 Tamar Christina <tamar.christina@arm.com>
4632 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4633 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4634 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4636 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4638 * hash-map.h (gt_cleare_cache): Avoid UB.
4640 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
4642 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
4643 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
4644 (decide_unroll_runtime_iterations): Likewise.
4645 (decide_unroll_stupid): Likewise.
4647 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4650 * config/i386/i386.md: Add a loop with incssp.
4652 2017-11-27 Martin Jambor <mjambor@suse.cz>
4654 PR tree-optimization/81248
4655 * tree-sra.c (splice_param_accesses): Remove size check.
4656 (decide_one_param_reduction): Fix size check.
4657 * gimple-pretty-print.c (dump_profile): Silence warning.
4658 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
4660 2017-11-27 Jakub Jelinek <jakub@redhat.com>
4663 * dbxout.c (lastlineno): New variable.
4664 (dbx_debug_hooks): Use dbxout_switch_text_section as
4665 switch_text_section debug hook.
4666 (dbxout_function_end): Switch to current_function_section
4667 rather than function_section. If crtl->has_bb_partition,
4668 output just one N_FUN, depending on in_cold_section_p.
4669 (dbxout_source_line): Remember last lineno in lastlineno.
4670 (dbxout_switch_text_section): New function.
4671 (dbxout_function_decl): Adjust dbxout_block caller.
4672 (dbx_block_with_cold_children): New function.
4673 (dbxout_block): Return true if any LBRAC/RBRAC have been
4674 emitted. Use dbx_block_with_cold_children at depth == 0
4675 in second partition. Add PARENT_BLOCKNUM argument, pass
4676 it optionally adjusted to children. Output LBRAC/RBRAC
4677 around recursive call only if the block is in the current
4678 partition, if not and anything was output, emit empty
4680 * final.c (final_scan_insn): Compute cold_function_name
4681 before calling switch_text_section debug hook. Call
4682 that hook even if dwarf2out_do_frame if not emitting
4686 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
4687 TREE_READONLY decls.
4689 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4691 PR rtl-optimization/82488
4692 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
4694 2017-11-26 Julia Koval <julia.koval@intel.com>
4696 * config/i386/i386.c (processor_target_table): Add skylake_cost for
4698 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
4701 2017-11-26 Julia Koval <julia.koval@intel.com>
4703 * config/i386/driver-i386.c (host_detect_local_cpu):
4704 Detect skylake-avx512.
4706 2017-11-26 Julia Koval <julia.koval@intel.com>
4708 * config.gcc: Add -march=cannonlake.
4709 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
4710 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
4711 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
4712 (PTA_CANNONLAKE): New.
4713 (processor_target_table): Add cannonlake.
4714 (ix86_option_override_internal): Ditto.
4715 (fold_builtin_cpu): Ditto.
4716 (get_builtin_code_for_version): Handle cannonlake.
4717 (M_INTEL_COREI7_CANNONLAKE): New.
4718 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
4719 * doc/invoke.texi: Add -march=cannonlake.
4721 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
4723 * plugin.c (add_new_plugin): Use platform-specific library extensions.
4724 (try_init_one_plugin): Alternative implementation for MinGW.
4725 * Makefile.in (plugin_implib): New.
4726 (gengtype-lex.c): Fix broken AIX workaround.
4727 * configure: Regenerate.
4728 * doc/plugins.texi: Document support for MinGW.
4730 2017-11-25 Jakub Jelinek <jakub@redhat.com>
4732 PR rtl-optimization/81553
4733 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
4734 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
4735 is a shift where C1 has different mode than the whole shift, use C1's
4736 mode for MULT rather than the shift's mode.
4739 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
4740 builtins not enabled in the currently selected ISA.
4742 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
4744 PR tree-optimization/71026
4745 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
4746 (insert_reciprocals): Change to insert reciprocals before a division
4747 by a square and to insert the square of a reciprocal.
4748 (execute_cse_reciprocals_1): Change to consider division by a square.
4749 (register_division_in): Add importance parameter.
4751 2017-11-24 Richard Biener <rguenther@suse.de>
4753 PR tree-optimization/82402
4754 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
4755 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
4757 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
4759 * match.pd (0-ptr): New transformation.
4761 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
4764 * ipa-inline.c (inline_small_functions): Set current badnes correctly
4765 when skipping checking.
4767 2017-11-24 Richard Biener <rguenther@suse.de>
4769 PR tree-optimization/83128
4770 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
4771 (vn_reference_lookup_3): Likewise.
4773 2017-11-24 Jakub Jelinek <jakub@redhat.com>
4776 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
4777 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
4780 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
4781 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
4782 check for the latter with gimple_call_builtin_p. Do not handle
4783 BUILT_IN_STPNCPY_CHK which is not a pass through call.
4785 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
4787 * config/arm/arm_neon.h: Fix pragma GCC push_options before
4790 2017-11-23 Julia Koval <julia.koval@intel.com>
4792 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
4793 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
4794 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
4795 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
4796 _mm512_maskz_expandloadu_epi16): New intrinsics.
4797 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
4798 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
4799 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
4800 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
4801 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
4802 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
4803 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
4804 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
4805 _mm256_maskz_expandloadu_epi8): New intrinsics.
4806 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
4807 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
4808 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
4809 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
4810 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
4811 * config/i386/sse.md (VI248_VLBW): New iterator.
4812 (expand<mode>_mask, expand<mode>_maskz): New patterns.
4814 2017-11-23 Julia Koval <julia.koval@intel.com>
4816 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
4817 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
4818 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
4819 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
4820 _mm512_mask_compressstoreu_epi16): New.
4821 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
4822 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
4823 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
4824 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
4825 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
4826 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
4827 _mm256_mask_compressstoreu_epi8): New.
4828 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
4829 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
4830 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
4831 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
4832 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
4833 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
4834 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
4835 __builtin_ia32_compresshi128_mask,
4836 __builtin_ia32_compressstoreuqi512_mask,
4837 __builtin_ia32_compressstoreuhi512_mask,
4838 __builtin_ia32_compressstoreuqi256_mask,
4839 __builtin_ia32_compressstoreuqi128_mask,
4840 __builtin_ia32_compressstoreuhi256_mask,
4841 __builtin_ia32_compressstoreuhi128_mask): New builtins.
4842 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
4844 (ix86_expand_special_args_builtin): Handle new types.
4845 (s4fma_expand): Handle new builtin array.
4846 * config/i386/immintrin.h: Include new headers.
4847 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
4848 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
4850 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4852 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
4853 fixes. Declare temp and g variables at the top in order to avoid
4854 {} in most of the cases.
4856 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
4858 * match.pd (ptr-0): New transformation.
4860 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4862 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
4863 (struct addr_mode_cost_table): New.
4864 (struct tune_params): Add field addr_mode_costs.
4865 * config/arm/arm.c (generic_addr_mode_costs): New.
4866 (arm_slowmul_tune): Initialise addr_mode_costs field.
4867 (arm_fastmul_tune): Likewise.
4868 (arm_strongarm_tune): Likewise.
4869 (arm_xscale_tune): Likewise.
4870 (arm_9e_tune): Likewise.
4871 (arm_marvell_pj4_tune): Likewise.
4872 (arm_v6t2_tune): Likewise.
4873 (arm_cortex_tune): Likewise.
4874 (arm_cortex_a8_tune): Likewise.
4875 (arm_cortex_a7_tune): Likewise.
4876 (arm_cortex_a15_tune): Likewise.
4877 (arm_cortex_a35_tune): Likewise.
4878 (arm_cortex_a53_tune): Likewise.
4879 (arm_cortex_a57_tune): Likewise.
4880 (arm_exynosm1_tune): Likewise.
4881 (arm_xgene1_tune): Likewise.
4882 (arm_cortex_a5_tune): Likewise.
4883 (arm_cortex_a9_tune): Likewise.
4884 (arm_cortex_a12_tune): Likewise.
4885 (arm_cortex_a73_tune): Likewise.
4886 (arm_v7m_tune): Likewise.
4887 (arm_cortex_m7_tune): Likewise.
4888 (arm_v6m_tune): Likewise.
4889 (arm_fa726te_tune): Likewise.
4890 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
4892 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4894 * config/arm/arm.c (arm_mem_costs): New function.
4895 (arm_rtx_costs_internal): Use arm_mem_costs.
4897 2017-11-23 Mark Wielaard <mark@klomp.org>
4899 * dwarf2out.c (init_sections_and_labels): Use generation to create
4900 unique ranges_section_label and ranges_base_label. Return generation.
4901 (output_rnglists): Add generation argument. Use generation to create
4902 unique ranges labels.
4903 (dwarf2out_finish): Get generation from init_sections_and_labels
4904 and pass generation to output_rnglists.
4906 2017-11-23 Mike Stump <mikestump@comcast.net>
4907 Eric Botcazou <ebotcazou@adacore.com>
4909 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
4910 * cfgloop.h (struct loop): Add unroll field.
4911 * function.h (struct function): Add has_unroll bitfield.
4912 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
4913 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
4914 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
4915 (pass_rtl_unroll_loops::gate): Likewise.
4916 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
4917 for which loop->unroll==1.
4918 (decide_unroll_constant_iterations): Use note for consistency and
4919 take loop->unroll into account. Return early if loop->unroll is set.
4920 Fix thinko in existing test.
4921 (decide_unroll_runtime_iterations): Use note for consistency and
4922 take loop->unroll into account.
4923 (decide_unroll_stupid): Likewise.
4924 * lto-streamer-in.c (input_cfg): Read loop->unroll.
4925 * lto-streamer-out.c (output_cfg): Write loop->unroll.
4926 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
4928 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
4929 (print_loop): Print loop->unroll if set.
4930 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
4931 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
4932 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
4934 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
4935 loop->unroll is set and smaller than the trip count. Otherwise bypass
4936 entirely the heuristics if loop->unroll is set. Remove dead note.
4937 Fix off-by-one bug in other note.
4938 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
4939 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
4941 (tree_unroll_loops_completely): Make static.
4942 (pass_complete_unroll::execute): Use correct type for variable.
4943 (pass_complete_unrolli::execute): Fix formatting.
4944 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
4946 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
4948 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
4949 enable when TARGET_PREFER_AVX128 is set.
4951 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4953 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
4955 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
4956 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
4958 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4960 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
4961 frequencies. Fix estimation of aggregate parameters.
4963 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4965 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
4968 2017-11-23 Tom de Vries <tom@codesourcery.com>
4970 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
4972 2017-11-23 Tom de Vries <tom@codesourcery.com>
4974 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
4977 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
4980 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
4981 sibcall_value_pcrel_fdpic): Use local variable instead of
4983 (calli_tbr_rel): Add missing operand 2.
4984 (call_valuei_tbr_rel): Add missing operand 3.
4986 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4989 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
4990 bitpos/bitsize covering the whole destination, use store_expr only if
4991 the complex mode is the same. Otherwise, use expand_normal and if
4992 it returns CONCAT, subreg each part separately instead of trying to
4993 subreg the whole result.
4995 2017-11-23 Richard Biener <rguenther@suse.de>
4997 PR tree-optimization/23094
4998 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
4999 come here from walking over backedges in the first iteration.
5000 (vn_reference_lookup_3): Skip clobbers that store the same value.
5002 2017-11-23 Richard Biener <rguenther@suse.de>
5004 PR tree-optimization/81403
5005 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
5006 a block we need a leader relative to.
5007 (phi_translate_1): For nary processing require a leader from
5008 get_representative_for given we run expression simplification
5009 using match-and-simplify. Remove previous fix.
5011 2017-11-22 Jeff Law <law@redhat.com>
5013 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
5014 Use new method allocate_value_range rather than accessing the
5015 vrp_value_range_pool data member directly.
5016 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
5017 to use extract_range_from_stmt method to avoid need for
5018 extract_range_from_assignment method.
5019 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
5020 method rather than setting values_propgated data member directly.
5021 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
5022 and values propagated data members and extract_range_from_assignment
5023 method. Reorder private data members to conform to standards.
5024 Add new methods set_lattice_propagation_complete and
5025 allocate_value_range.
5027 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
5029 PR rtl-optimization/83030
5030 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
5031 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
5032 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
5033 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
5034 CROSSING_JUMP_P flag is already set before setting it.
5035 * cfgrtl.c (fixup_partition_crossing): Likewise.
5036 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
5039 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5041 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
5042 Handle the case where both arguments are using gen_const_vec_series.
5044 2017-11-22 David Malcolm <dmalcolm@redhat.com>
5047 * pretty-print.c (pp_format): Move quoting implementation to
5048 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
5049 to pass address of "quote" local.
5050 (pp_begin_quote): New function.
5051 (pp_end_quote): New function.
5052 * pretty-print.h (printer_fn): Convert penultimate param from bool
5054 (pp_begin_quote): New decl.
5055 (pp_end_quote): New decl.
5056 * tree-diagnostic.c (default_tree_printer): Convert penultimate
5057 param from bool to bool *.
5058 * tree-diagnostic.h (default_tree_printer): Likewise.
5060 2017-11-22 Jeff Law <law@redhat.com>
5062 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
5063 Initialize vr_values.
5064 (evrp_range_analyzer::try_find_new_range): Call methods attached to
5065 vr_values via vr_values class instance rather than delegators.
5066 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
5067 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5068 (evrp_range_analyzer::push_value_range): Likewise.
5069 (evrp_range_analyzer::pop_value_range): Likewise.
5070 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
5071 most delegators. Those remaining are exposed as public interfaces.
5072 Make vr_values a pointer and private.
5073 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
5075 (evrp_range_analyzer::get_vr_value): New method.
5076 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
5077 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
5078 (class evrp_dom_walker): Attach evrp_folder class, initialize
5079 it in the ctor. Remove temporary delegators.
5080 (evrp_dom_walker::before_dom_children): Call methods in attached
5081 evrp_range_analyzer class via class instance pointer. Use
5082 free value_range_constant_singleton to remove need for
5083 op_with_constant_singleton_value delegator method. Do not
5084 create a vrp_prop class instance for every call! Narrow
5085 scope of a couple variables.
5086 (evrp_dom_walker::cleanup): Call methods in attached
5087 evrp_range_analyzer class via class instance pointer.
5088 * vr-values.h (class vr_values): Privatize many methods and
5091 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5093 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
5094 * cfgexpand.c (expand_debug_expr): Remove handling for them.
5095 * expr.c (expand_expr_real_2): Likewise.
5096 * fold-const.c (const_unop): Likewise.
5097 * optabs-tree.c (optab_for_tree_code): Likewise.
5098 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
5099 * tree-inline.c (estimate_operator_cost): Likewise.
5100 * tree-pretty-print.c (dump_generic_node): Likewise.
5101 (op_code_prio): Likewise.
5102 (op_symbol_code): Likewise.
5103 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
5104 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
5105 * internal-fn.c (direct_internal_fn_optab): New function.
5106 (direct_internal_fn_array, direct_internal_fn_supported_p
5107 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
5108 * fold-const-call.c (fold_const_reduction): New function.
5109 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
5111 * tree-vect-loop.c: Include internal-fn.h.
5112 (reduction_code_for_scalar_code): Rename to...
5113 (reduction_fn_for_scalar_code): ...this and return an internal
5115 (vect_model_reduction_cost): Take an internal_fn rather than
5117 (vect_create_epilog_for_reduction): Likewise. Build calls rather
5119 (vectorizable_reduction): Use internal functions rather than tree
5120 codes for the reduction operation. Update calls to the functions
5122 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
5123 Use calls to internal functions rather than REDUC tree codes.
5124 * config/aarch64/aarch64-simd.md: Update comment accordingly.
5126 2017-11-22 Olivier Hainque <hainque@adacore.com>
5128 * config/vxworks.c (vxworks_override_options): Pick default
5129 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
5130 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
5131 DWARF_GNAT_ENCODINGS_DEFAULT.
5132 * config/vxworksae.h: Likewise.
5134 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
5136 PR tree-optimization/83104
5137 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
5140 2017-11-22 Marek Polacek <polacek@redhat.com>
5141 H.J. Lu <hongjiu.lu@intel.com>
5142 Jason Merrill <jason@redhat.com>
5147 * calls.c (initialize_argument_information): Call
5148 warn_parameter_passing_abi target hook.
5149 (store_one_arg): Use 0 for empty record size. Don't push 0 size
5150 argument onto stack.
5151 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
5152 * common.opt: Update -fabi-version description.
5153 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
5154 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
5156 (ix86_is_empty_record): New function.
5157 (ix86_warn_parameter_passing_abi): New function.
5158 (TARGET_EMPTY_RECORD_P): Redefine.
5159 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
5160 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
5161 * doc/tm.texi: Regenerated.
5162 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
5163 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
5164 * dwarf2out.c (get_ultimate_context): Move to tree.c.
5165 * explow.c (hard_function_value): Call arg_int_size_in_bytes
5166 instead of int_size_in_bytes.
5167 * expr.c (copy_blkmode_to_reg): Likewise.
5168 * function.c (aggregate_value_p): Return 0 for empty types.
5169 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
5171 (locate_and_pad_parm): Call arg size_in_bytes instead
5173 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
5174 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
5175 * target.def (empty_record_p, warn_parameter_passing_abi): New target
5177 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
5178 (std_gimplify_va_arg_expr): Skip empty records. Call
5179 arg_size_in_bytes instead size_in_bytes.
5180 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
5181 * tree-core.h (tree_type_common): Add empty_flag.
5182 (tree_decl_common): Update comments.
5183 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
5185 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5186 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
5188 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5189 * tree.c (default_is_empty_type): New function.
5190 (default_is_empty_record): New function.
5191 (arg_int_size_in_bytes): New function.
5192 (arg_size_in_bytes): New function.
5193 (get_ultimate_context): New function.
5194 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
5195 TRANSLATION_UNIT_WARN_EMPTY_P.
5196 (default_is_empty_record, arg_int_size_in_bytes,
5197 arg_size_in_bytes, get_ultimate_context): Declare.
5199 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5201 * config/arm/arm.c (cmse_clear_registers): New function.
5202 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
5203 code by call to cmse_clear_registers.
5204 (cmse_nonsecure_entry_clear_before_return): Likewise.
5206 2017-11-22 Tamar Christina <tamar.christina@arm.com>
5208 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
5209 (vdot_s32, vdotq_s32): New.
5210 (vdot_lane_u32, vdotq_lane_u32): New.
5211 (vdot_lane_s32, vdotq_lane_s32): New.
5214 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5217 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
5218 for unsigned values with fewer HWIs than the precision.
5219 (test_overflow): New function.
5220 (wide_int_cc_tests): Call it.
5222 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5224 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
5225 has the same mode class as Pmode.
5227 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5229 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
5230 padding_bits_to_clear_ptr.
5231 (cmse_nonsecure_entry_clear_before_return): Likewise.
5233 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5235 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
5236 auto_sbitap instead of integer bitfield to control register needing
5239 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5241 PR tree-optimization/83044
5242 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
5243 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
5244 that need the upper bound. Subtract offset from
5245 get_addr_base_and_unit_offset only if positive and subtract it
5246 before division by eltsize rather than after it.
5249 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
5250 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
5251 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
5255 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
5256 expand_binop with *mul_widen_optab, make sure at least one of the
5257 operands doesn't have VOIDmode.
5260 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
5262 PR rtl-optimization/82044
5263 PR tree-optimization/82042
5264 * dse.c (record_store): Check for overflow.
5265 (check_mem_read_rtx): Properly check for overflow if width == -1, call
5266 add_wild_read instead of clear_rhs_from_active_local_stores on
5267 overflow and log it into dump_file.
5269 2017-11-22 Richard Biener <rguenther@suse.de>
5271 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
5272 fake edges to exit when looking for a place to insert.
5273 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
5275 (insert_into_preds_of_block): Commit edge insertion immediately,
5276 assert that doesn't require new BBs.
5277 (fini_pre): Release expressions.
5278 (pass_pre::execute): Shuffle things around a bit, if the fn
5279 is too large do not compute AVAIL either as this is really the
5282 2017-11-22 Richard Biener <rguenther@suse.de>
5284 PR tree-optimization/83089
5285 * tree-if-conv.c (pass_if_conversion::execute): If anything
5286 changed reset SCEV and free the number of iteration estimates.
5288 2017-11-21 Martin Sebor <msebor@redhat.com>
5290 PR tree-optimization/82945
5291 * calls.h (warn_nonstring_bound): Remove unused function.
5293 2017-11-21 Martin Sebor <msebor@redhat.com>
5295 PR tree-optimization/82945
5296 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
5297 * calls.h (maybe_warn_nonstring_arg): Declare new function.
5298 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
5300 (initialize_argument_information): Call maybe_warn_nonstring_arg.
5301 * calls.h (get_attr_nonstring_decl): Declare new function.
5302 * doc/extend.texi (attribute nonstring): Update.
5303 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
5304 get_attr_nonstring_decl and handle it.
5305 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
5306 detection of nul-termination.
5307 (strlen_to_stridx): Change to a pointer.
5308 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
5309 (pass_strlen::execute): Same.
5311 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
5313 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
5314 for the new option -mprefer-vector-width=[none|128|256|512].
5315 * config/i386/i386.c (ix86_target_string): remove old style options
5316 -mprefer-avx256 and make -mprefer-avx128 as alias.
5317 (ix86_option_override_internal): Apply defaults for the
5318 -mprefer-vector-width=[128|256] option.
5319 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
5320 Implement macros to work with -mprefer-vector-width=.
5321 * config/i386/i386.opt: Implemented option
5322 -mprefer-vector-width=[none|128|256|512].
5323 * doc/invoke.texi: Documentation for
5324 -mprefer-vector-width=[none|128|256|512].
5326 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
5328 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
5330 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
5332 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
5334 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
5336 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
5338 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
5339 * expr.c (expand_expr_real_2): Likewise.
5340 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
5341 fold_binary_loc): Likewise.
5342 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
5343 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
5344 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
5345 MINUS_EXPR transformations.
5346 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
5347 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
5348 * tree-inline.c (estimate_operator_cost): Likewise.
5349 * tree-pretty-print.c (dump_generic_node, op_code_prio,
5350 op_symbol_code): Likewise.
5351 * tree-vect-stmts.c (vectorizable_operation): Likewise.
5352 * vr-values.c (extract_range_from_binary_expr): Likewise.
5353 * varasm.c (initializer_constant_valid_p_1): Likewise.
5354 * tree.def: New tree code POINTER_DIFF_EXPR.
5356 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
5358 * config/i386/i386.md (*bswap<mode>2_movbe): Add
5359 integer suffix to movbe mnemonic.
5360 (*bswaphi2_movbe): Ditto.
5361 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
5363 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5366 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
5367 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
5368 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
5370 2017-11-21 Martin Liska <mliska@suse.cz>
5372 * tree-inline.c (expand_call_inline): Remove not needed
5375 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
5376 Jakub Jelinek <jakub@redhat.com>
5379 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
5380 Remove static keyword from f variable.
5382 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5384 PR tree-optimization/83086
5385 * gimple-ssa-store-merging.c
5386 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
5387 rather than n.base_addr.
5389 2017-11-21 Martin Liska <mliska@suse.cz>
5391 PR rtl-optimization/82044
5392 PR tree-optimization/82042
5393 * dse.c (check_mem_read_rtx): Check for overflow.
5395 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5397 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
5400 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5402 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
5403 and its corresponding call arg location note.
5405 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5406 Andrew Burgess <andrew.burgess@embecosm.com>
5408 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
5410 (arc_return_slot_offset): Likewise.
5411 (arc_eh_return_address_location): New declaration.
5412 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
5413 (MUST_SAVE_REGISTER): Add exception handler case.
5414 (MUST_SAVE_RETURN_ADDR): Likewise.
5415 (arc_frame_pointer_required): Likewise.
5416 (arc_frame_pointer_needed): New function.
5417 (arc_compute_frame_size): Changed.
5418 (arc_expand_prologue): Likewise.
5419 (arc_expand_epilogue): Likewise.
5420 (arc_initial_elimination_offset): Likewise.
5421 (arc_return_slot_offset): Delete.
5422 (arc_eh_return_address_location): New function.
5423 (arc_builtin_setjmp_frame_value): Likewise.
5424 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
5425 (EH_RETURN_STACKADJ_RTX): Define.
5426 (EH_RETURN_HANDLER_RTX): Likewise.
5427 * config/arc/arc.md (eh_return): Delete.
5429 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5431 * print-tree.h (debug_vec_tree): Remove prototype.
5432 * gdbinit.in (pvt): Remove macro.
5434 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5436 PR tree-optimization/83047
5437 * gimple-ssa-store-merging.c
5438 (imm_store_chain_info::output_merged_store): If the loads with the
5439 same vuse are in different basic blocks, for load_gsi pick a load
5440 location that is dominated by the other loads.
5443 * config/i386/i386.c (ix86_memmodel_check): Start
5444 -Winvalid-memory-model diagnostics with lowercase letter.
5447 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
5448 set in_cold_section_p to the partition containing loc_list->first.
5449 When seeing loc_list->last_before_switch node, update secname and
5450 perform range_across_switch second partition handling only after that.
5453 * run-rtl-passes.c: Include debug.h.
5454 (run_rtl_passes): Call debug_hooks->assembly_start.
5455 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
5459 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
5460 OPTAB_DIRECT in calls to expand_simple_binop.
5462 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5465 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
5467 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5470 * spellcheck.h (best_match::blithely_get_best_candidate): New
5473 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5475 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
5476 with lower case letter.
5478 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
5480 * config/i386/i386.md (bswaphi2): New expander.
5481 (*bswaphi2_movbe): New insn pattern.
5482 (bswaphi -> rorhi pepehole2): New peephole pattern.
5484 2017-11-20 Jeff Law <law@redhat.com>
5486 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
5487 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5488 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
5489 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
5490 gimple-ssa-evrp-analyze.h.
5492 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
5493 set BB_VISITED here.
5494 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
5496 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5498 * tree-predcom.c: Add general comment on Store-Store chains.
5499 (split_data_refs_to_components): Postpone clearing eliminate_store_p
5501 (get_chain_last_ref_at): Rename into...
5502 (get_chain_last_write_at): ...this.
5503 (get_chain_last_write_before_load): New function.
5504 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
5505 CT_STORE_STORE when write reference is added.
5506 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
5507 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
5508 (initialize_root_vars_store_elim_1): Ditto.
5509 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
5510 definition is created.
5511 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
5512 chain by replacing it with dominant stored value.
5514 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5516 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
5518 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
5520 * vr-values.c (extract_range_from_binary_expr): Use a full range
5523 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
5525 * config/arm/arm.md (R4_REGNUM): Define constant.
5526 (nonsecure_call_internal): Remove r4 clobber.
5527 (nonsecure_call_value_internal): Likewise.
5528 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
5529 clobber and resequence match_operands.
5530 (nonsecure_call_value_reg_thumb1_v5): Likewise.
5531 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
5532 (nonsecure_call_value_reg_thumb2): Likewise.
5534 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5536 PR tree-optimization/78821
5537 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
5538 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
5539 to the address of the base rather than the base itself.
5540 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
5541 (find_bswap_or_nop_finalize): New function.
5542 (find_bswap_or_nop): Use it.
5543 (bswap_replace): Return a tree rather than bool, change first
5544 argument from gimple * to gimple_stmt_iterator, allow inserting
5545 into an empty sequence, allow ins_stmt to be NULL - then emit
5546 all stmts into gsi. Fix up MEM_REF address gimplification.
5547 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
5548 (struct store_immediate_info): Add N and INS_STMT non-static
5550 (store_immediate_info::store_immediate_info): Initialize them
5551 from newly added ctor args.
5552 (merged_store_group::apply_stores): Formatting fixes. Sort by
5554 (stmts_may_clobber_ref_p): For stores call also
5555 refs_anti_dependent_p.
5556 (gather_bswap_load_refs): New function.
5557 (imm_store_chain_info::try_coalesce_bswap): New method.
5558 (imm_store_chain_info::coalesce_immediate_stores): Use it.
5559 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
5560 (imm_store_chain_info::output_merged_store): Fail if number of
5561 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
5562 and NOP_EXPR rhs_code.
5563 (pass_store_merging::process_store): Compute n and ins_stmt, if
5564 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
5565 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
5567 (pass_store_merging::execute): Calculate dominators.
5569 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
5570 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
5571 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
5572 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5573 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
5574 class pass_optimize_bswap, bswap_replace,
5575 pass_optimize_bswap::execute): Moved to ...
5576 * gimple-ssa-store-merging.c: ... this file.
5577 Include optabs-tree.h.
5578 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
5579 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5580 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
5581 anonymous namespace, remove static keywords.
5582 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
5583 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
5585 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
5588 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
5590 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
5592 * vec.h (debug_helper): New function.
5593 (DEFINE_DEBUG_VEC): New macro.
5594 * hash-set.h (debug_helper): New function.
5595 (DEFINE_DEBUG_HASH_SET): New macro.
5596 * cfg.c (debug_slim (edge)): New function.
5597 Call DEFINE_DEBUG_VEC for edges.
5598 Call DEFINE_DEBUG_HASH_SET for edges.
5599 * cfghooks.c (debug_slim (basic_block)): New function.
5600 Call DEFINE_DEBUG_VEC for basic blocks.
5601 Call DEFINE_DEBUG_HASH_SET for basic blocks.
5602 * print-tree.c (debug_slim): New function to handle trees.
5603 Call DEFINE_DEBUG_VEC for trees.
5604 Call DEFINE_DEBUG_HASH_SET for trees.
5605 (debug (vec<tree, va_gc>) &): Remove.
5606 (debug (<vec<tree, va_gc>) *): Remove.
5607 * print-rtl.c (debug_slim): New function to handle const_rtx.
5608 Call DEFINE_DEBUG_VEC for rtx_def.
5609 Call DEFINE_DEBUG_VEC for rtx_insn.
5610 Call DEFINE_DEBUG_HASH_SET for rtx_def.
5611 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
5612 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
5613 (debug (vec<rtx_insn *> *ptr): Remove.
5614 (debug_insn_vector): Remove.
5615 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
5617 2017-11-20 Tom de Vries <tom@codesourcery.com>
5619 PR rtl-optimization/82020
5620 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
5621 IF_THEN_ELSE condition.
5623 2017-11-19 Jeff Law <law@redhat.com>
5625 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
5626 of degenerates resulting from ignoring an edge.
5628 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5631 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
5633 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5636 * profile-count.c (profile_count::to_sreal_scale): Fix return value
5637 for uninitialied counts.
5639 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5642 * tree-inline.c (estimate_num_insns): Set to 1 at least.
5644 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5647 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
5649 2017-11-19 Tom de Vries <tom@codesourcery.com>
5651 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
5654 2017-11-19 Tom de Vries <tom@codesourcery.com>
5656 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
5659 2017-11-19 Tom de Vries <tom@codesourcery.com>
5661 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5664 2017-11-19 Tom de Vries <tom@codesourcery.com>
5666 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
5668 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
5670 2017-11-19 Tom de Vries <tom@codesourcery.com>
5672 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
5674 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
5675 semicolon after MCORE_EXPORT_NAME call.
5677 2017-11-19 Tom de Vries <tom@codesourcery.com>
5680 * vmsdbgout.c (vmsdbgout_early_finish): New function.
5681 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
5683 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
5685 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
5686 (cgraph_edge::clone): Cleanup updating of profile.
5687 * ipa-cp.c (update_profiling_info): Likewise.
5688 * ipa-inline-transform.c (inline_transform): Likewise.
5689 * ipa-inline.c (inline_small_functions): Add missing space to dump.
5690 * ipa-split.c (execute_split_functions): Do not split when function
5692 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
5693 * profile-count.c (profile_count::dump): Add global0.
5694 (profile_count::to_cgraph_frequency): Do not ICE when entry is
5696 (profile_count::to_sreal_scale): Likewise.
5697 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
5698 (profile_count::combine_with_ipa_count): New function.
5699 * profile-count.h (profile_guessed_global0adjusted): New.
5700 (profile_count::adjusted_zero): New.
5701 (profile_count::global0adjusted): New.
5702 (profile_count::combine_with_ipa_count): New.
5703 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
5704 correct profile of return block of split functions.
5705 (copy_cfg_body): Remove unused profile_count.
5706 (copy_body): Likewise.
5707 (expand_call_inline): Update.
5708 (tree_function_versioning): Update.
5710 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
5712 * hash-set.h (hash_set::empty): New.
5713 * tree-ssa-threadbackward.h: Delete.
5714 * tree-ssa-threadbackward.c (class thread_jumps): New.
5715 Move max_threaded_paths into class.
5716 (fsm_find_thread_path): Remove arguments that are now in class.
5717 (profitable_jump_thread_path): Rename to...
5718 (thread_jumps::profitable_jump_thread_path): ...this.
5719 (convert_and_register_jump_thread_path): Rename to...
5720 (thread_jumps::convert_and_register_current_path): ...this.
5721 (check_subpath_and_update_thread_path): Rename to...
5722 (thread_jumps::check_subpath_and_update_thread_path): ...this.
5723 (register_jump_thread_path_if_profitable): Rename to...
5724 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
5725 (handle_phi): Rename to...
5726 (thread_jumps::handle_phi): ...this.
5727 (handle_assignment): Rename to...
5728 (thread_jumps::handle_assignment): ...this.
5729 (fsm_find_control_statement_thread_paths): Rename to...
5730 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
5731 (find_jump_threads_backwards): Rename to...
5732 (thread_jumps::find_jump_threads_backwards): ...this.
5733 Initialize path local data.
5734 (pass_thread_jumps::execute): Call find_jump_threads_backwards
5735 from within thread_jumps class.
5736 (pass_early_thread_jumps::execute): Same.
5738 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5740 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
5742 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5744 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
5745 to canonical location.
5747 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5750 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
5752 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
5754 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
5756 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
5757 rcrt1.o%s/grcrt1.o%s for -static-pie.
5759 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5761 * i386.c (ix86_multiplication_cost, ix86_division_cost,
5762 ix86_shift_rotate_cost): Break out from ...
5763 (ix86_rtx_costs): ... here.
5764 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
5767 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5769 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
5770 when entry block was promoted unlikely.
5771 (estimate_bb_frequencies): Increase frequency scale.
5772 * profile-count.h (profile_count): Export precision info.
5774 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5776 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
5777 disturbin profile of entry block.
5779 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5781 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
5784 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5786 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
5787 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
5789 2017-11-17 Jeff Law <law@redhat.com>
5791 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
5792 from evrp_dom_walker class. Various methods moved into new class.
5793 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
5794 (evrp_range_analyzer::enter): New method.
5795 (evrp_range_analyzer::leave): New method.
5796 (evrp_dom_walker): Remove delegators no longer needed by this class.
5797 Replace vr_values data member with evrp_range_analyzer
5799 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
5800 method extracted from evrp_dom_walker::before_dom_children.
5801 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
5802 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
5804 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
5805 Add private copy constructor and move assignment operators.
5806 Privatize methods and class data where trivially possible.
5807 (evrp_dom_walker::cleanup): New function, extracted from
5808 execute_early_vrp. Simplify access to class data.
5810 * vr-values.h (get_output_for_vrp): Prototype.
5811 * vr-values.c (get_output_for_vrp): New function extracted from
5812 vrp_visit_assignment_or_call and extract_range_from_stmt.
5813 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
5815 2017-11-17 Luis Machado <luis.machado@linaro.org>
5817 * config/aarch64/aarch64.c
5818 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
5819 (qdf24xx_tunings) <autoprefetcher_model>: Set to
5820 tune_params::AUTOPREFETCHER_WEAK.
5822 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5825 * config/arm/arm.c (arm_valid_target_attribute_rec):
5826 Parse "arch=" and "+<ext>".
5827 (arm_valid_target_attribute_tree): Re-init global options.
5828 (arm_option_override): Make non-static.
5829 (arm_options_perform_arch_sanity_checks): Make errors fatal.
5830 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
5831 (__ARM_FEATURE_CRC32): Support undef.
5832 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
5833 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
5835 2017-11-17 David Malcolm <dmalcolm@redhat.com>
5837 * gdbinit.in (break-on-diagnostic): New command.
5839 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5841 * config/i386/i386.c (ix86_expand_epilogue): Change simple
5842 return to indirect jump for EH return if control-flow
5843 protection is enabled. Change explicit 'false' argument in
5844 pro_epilogue_adjust_stack with a value of
5846 * config/i386/i386.md (simple_return_indirect_internal):
5847 Remove SImode restriction to support 64-bit.
5849 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5851 * combine.c (added_notes_insn): New.
5852 (try_combine): Handle added_notes_insn like added_links_insn.
5853 Rewrite return value code.
5854 (distribute_notes): Set added_notes_insn to the earliest insn we added
5857 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5859 PR rtl-optimization/82621
5860 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
5861 dest of one of those SETs is unused.
5863 2017-11-17 Richard Biener <rguenther@suse.de>
5866 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
5867 * tree-pretty-print.c (dump_generic_node): Handle
5868 annot_expr_parallel_kind.
5869 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
5870 * gimplify.c (gimple_boolify): Likewise.
5872 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5874 * config.gcc (extra_headers): Add cet.h for x86 targets.
5875 * config/i386/cet.h: New file.
5876 * doc/install.texi: Add --enable-cet/--disable-cet.
5878 2017-11-17 Richard Biener <rguenther@suse.de>
5880 PR tree-optimization/83017
5881 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
5882 (gen_parallel_loop): Properly count iterations.
5883 (parallelize_loops): Handle loop->can_be_parallel independent
5884 of flag_loop_parallelize_all. Make static profitability test match
5886 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
5887 * invoke.texi (parloops-min-per-thread): Document.
5889 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
5891 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
5892 upstreaming review comments.
5894 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5896 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
5897 with fast unaligned access.
5898 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
5900 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
5902 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
5903 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
5905 2017-11-17 Richard Biener <rguenther@suse.de>
5907 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
5908 folding of references.
5910 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
5913 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
5914 of replacing call to strncmp with corresponding call to strcmp when
5917 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
5919 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
5920 option prefer-avx256 for skylake-avx512 configuration.
5921 * config/i386/i386.c (ix86_option_override_internal): Ditto.
5922 (get_builtin_code_for_version): Ditto.
5924 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5925 Monk Chiang <sh.chiang04@gmail.com>
5927 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
5928 (FIXED_REGISTERS): Reserve more register numbers.
5929 (CALL_USED_REGISTERS): Likewise.
5930 (REG_ALLOC_ORDER): Likewise.
5931 (REG_CLASS_CONTENTS): Likewise.
5932 (REGISTER_NAMES): Likewise.
5934 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5935 Kito Cheng <kito.cheng@gmail.com>
5937 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
5939 * config/nds32/iterators.md: Add vector mode iterators and attributes.
5941 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
5943 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
5944 parameter list for vec_splats.
5946 2017-11-16 Joseph Myers <joseph@codesourcery.com>
5948 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
5950 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
5952 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
5955 * doc/install.texi: Document incompatibility of Perl >=5.6.26
5956 with the required version of automake 1.11.6.
5958 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
5960 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
5962 (DU_C2_3_power9): Correct reservation combinations.
5963 (FP_DIV_power9, VEC_DIV_power9): New.
5964 (power9-alu): Split out rotate/shift...
5965 (power9-rot): ...to here, correct dispatch resource.
5966 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
5968 (power9-fp): Correct latency.
5969 (power9-sdiv): Add div/sqrt resource.
5970 (power9-ddiv): Correct latency, add div/sqrt resource.
5971 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
5972 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
5974 (power9-qpdiv, power9-qpmul): Adjust resource usage.
5976 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
5978 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
5979 switch statement mapping KF built-ins to TF built-ins if we don't
5980 have the proper ISA 3.0 assembler support.
5982 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5984 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
5985 (lower_emutls_function_body): Do not compute it.
5987 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5989 * ipa-split.c (split_bb_info): Turn time to sreal.
5990 (split_point): Likewise.
5991 (dump_split_point): Likewise.
5992 (fine_split_points): Likewise.
5993 (execute_split_functions): Only zero split_bbs; turn time to sreals.
5995 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5997 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
6000 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6002 * predict.c (combine_predictions_for_bb): Preserve zero predicted
6004 (expensive_function_p): Remove useless assert.
6005 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
6007 2017-11-16 Martin Sebor <msebor@redhat.com>
6009 PR tree-optimization/82588
6010 PR tree-optimization/82583
6011 * tree-vrp.c (check_array_ref): Handle flexible array members,
6012 string literals, and inner indices.
6013 (search_for_addr_array): Add detail to diagnostics.
6015 2017-11-16 Nathan Sidwell <nathan@acm.org>
6019 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
6020 (SET_DECL_ASSEMBLER_NAME): Forward to
6021 overwrite_decl_assembler_name.
6022 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
6023 (overwrite_decl_assembler_name): Declare.
6024 * tree.c (overwrite_decl_assembler_name): New.
6025 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
6026 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
6027 (LANG_HOOKS_INITIALIZER): Add it.
6028 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
6029 * langhooks.c (lhd_set_decl_assembler_name): Use
6030 SET_DECL_ASSEMBLER_NAME.
6031 (lhd_overwrite_decl_assembler_name): Default implementation.
6033 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6034 Jackson Woodruff <jackson.woodruff@arm.com>
6036 PR tree-optimization/71026
6037 * match.pd: Canonicalize constant multiplies in division.
6039 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6041 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
6042 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
6043 Remove OPT_fomit_frame_pointer entry.
6044 * common/config/arc/arc-common.c: Likewise.
6045 * common/config/arm/arm-common.c: Likewise.
6046 * common/config/avr/avr-common.c: Likewise.
6047 * common/config/c6x/c6x-common.c: Likewise.
6048 * common/config/cr16/cr16-common.c: Likewise.
6049 * common/config/cris/cris-common.c: Likewise.
6050 * common/config/epiphany/epiphany-common.c: Likewise.
6051 * common/config/fr30/fr30-common.c: Likewise.
6052 * common/config/frv/frv-common.c: Likewise.
6053 * common/config/ia64/ia64-common.c: Likewise.
6054 * common/config/iq2000/iq2000-common.c: Likewise.
6055 * common/config/lm32/lm32-common.c: Likewise.
6056 * common/config/m32r/m32r-common.c: Likewise.
6057 * common/config/mcore/mcore-common.c: Likewise.
6058 * common/config/microblaze/microblaze-common.c: Likewise.
6059 * common/config/mips/mips-common.c: Likewise.
6060 * common/config/mmix/mmix-common.c: Likewise.
6061 * common/config/mn10300/mn10300-common.c: Likewise.
6062 * common/config/nios2/nios2-common.c: Likewise.
6063 * common/config/pa/pa-common.c: Likewise.
6064 * common/config/pdp11/pdp11-common.c: Likewise.
6065 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
6066 * common/config/riscv/riscv-common.c: Likewise.
6067 * common/config/rs6000/rs6000-common.c: Likewise.
6068 * common/config/rx/rx-common.c: Likewise.
6069 * common/config/s390/s390-common.c: Likewise.
6070 * common/config/sh/sh-common.c: Likewise.
6071 * common/config/sparc/sparc-common.c: Likewise.
6072 * common/config/tilegx/tilegx-common.c: Likewise.
6073 * common/config/tilepro/tilepro-common.c: Likewise.
6074 * common/config/v850/v850-common.c: Likewise.
6075 * common/config/visium/visium-common.c: Likewise.
6076 * common/config/xstormy16/xstormy16-common.c: Likewise.
6077 * common/config/xtensa/xtensa-common.c: Likewise.
6078 * invoke.texi (-fomit-frame-pointer): Update documentation.
6080 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6082 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
6084 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6086 * cfg.c (scale_bbs_frequencies_int,
6087 cale_bbs_frequencies_gcov_type): Remove.
6088 * cfg.h (scale_bbs_frequencies_int,
6089 cale_bbs_frequencies_gcov_type): Remove.
6091 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6093 * tree-ssa-loop-manip.c
6094 (scale_dominated_blocks_in_loop): Update to profile counts.
6095 (tree_transform_and_unroll_loop): Likewise.
6097 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6099 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
6100 scale_bbs_frequencies_int.
6102 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6104 * final.c (compute_alignments): Use counts rather than frequencies.
6106 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6108 * cfgloopanal.c: Include sreal.h
6109 (average_num_loop_insns): Use counts and sreal for accounting.
6111 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6113 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
6116 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6117 Kito Cheng <kito.cheng@gmail.com>
6119 * config/nds32/constraints.md: Provide more constraints.
6120 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
6121 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
6122 support constraints usage.
6124 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6126 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
6128 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6129 Kito Cheng <kito.cheng@gmail.com>
6131 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
6132 * config/nds32/nds32.opt: Refine the layout.
6133 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
6134 TARGET_EXT_STRING): Support new options.
6135 * config/nds32/nds32.h: Likewise.
6136 * config/nds32/nds32.md: Likewise.
6137 * config/nds32/nds32-predicates.c: Likewise.
6138 * config/nds32/constraints.md: Likewise.
6139 * common/config/nds32/nds32-common.c: Likewise.
6141 2017-11-16 Julia Koval <julia.koval@intel.com>
6144 * config/i386/gfniintrin.h: Add sse check.
6145 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
6147 2017-11-16 Julia Koval <julia.koval@intel.com>
6149 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
6150 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
6151 (ix86_handle_option): Handle -mavx512vbmi2.
6152 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
6153 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
6154 * config/i386/i386-c.c (__AVX512VBMI2__): New.
6155 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
6156 (ix86_valid_target_attribute_inner_p): Ditto.
6157 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
6158 * config/i386/i386.opt (mavx512vbmi2): New option.
6159 * doc/invoke.texi: Add new option.
6161 2017-11-16 Julia Koval <julia.koval@intel.com>
6163 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
6164 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
6165 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
6166 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
6167 _mm512_gf2p8mul_epi8): New intrinsics.
6168 * config/i386/i386-builtin-types.def
6169 (V64QI_FTYPE_V64QI_V64QI): New type.
6170 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
6171 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
6172 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
6173 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
6174 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
6175 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
6177 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
6179 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
6180 explicitly as a stream of bytes.
6182 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6184 * config/rs6000/altivec.h (vec_xst_be): New #define.
6185 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
6186 and externalize from *altivec_vperm_<mode>_internal.
6187 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
6189 (XL_BE_V8HI): Likewise.
6190 (XL_BE_V4SI): Likewise.
6191 (XL_BE_V4SI): Likewise.
6192 (XL_BE_V2DI): Likewise.
6193 (XL_BE_V4SF): Likewise.
6194 (XL_BE_V2DF): Likewise.
6195 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
6196 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
6197 all array entries with these keys: VSX_BUILTIN_VEC_XL,
6198 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
6199 VSX_BUILTIN_VEC_XST_BE.
6200 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
6201 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
6203 (altivec_init_builtins): Replace conditional calls to def_builtin
6204 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
6205 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
6206 with unconditional calls. Remove calls to def_builtin for
6207 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
6208 __builtin_vec_xst_be.
6209 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
6210 to define_expand, and add alternate RTL generation for P8.
6211 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
6212 vsx_ld_elemrev_v8hi.
6213 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
6214 add alternate RTL generation for P8.
6215 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
6216 vsx_ld_elemrev_v16qi.
6217 (vsx_st_elemrev_v8hi): Convert define_insn
6218 to define_expand, and add alternate RTL generation for P8.
6219 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
6220 vsx_st_elemrev_v8hi.
6221 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
6222 add alternate RTL generation for P8.
6223 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
6224 vsx_st_elemrev_v16qi.
6226 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
6229 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
6230 TARGET_AVX512ER check.
6231 (ix86_option_override_internal): Set MASK_VZEROUPPER if
6232 neither -mzeroupper nor -mno-zeroupper is used and
6233 TARGET_EMIT_VZEROUPPER is set.
6234 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
6235 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
6237 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6239 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6240 folding of vector compares.
6241 (fold_build_vec_cmp): New helper function.
6242 (fold_compare_helper): New helper function.
6243 (builtin_function_type): Add compare builtins to the list of functions
6244 having unsigned arguments. Cosmetic updates to comment indentation.
6245 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
6246 the not+eq combination.
6248 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6250 PR tree-optimization/82726
6251 PR tree-optimization/70754
6252 * tree-predcom.c (order_drefs_by_pos): New function.
6253 (combine_chains): Move code setting has_max_use_after to...
6254 (try_combine_chains): ...here. New parameter. Sort combined chains
6255 according to position information.
6256 (tree_predictive_commoning_loop): Update call to above function.
6257 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
6259 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6261 PR tree-optimization/82726
6263 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6265 PR tree-optimization/70754
6266 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6267 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6268 combined stmt before it if not NULL.
6269 (combine_chains): Process refs reversely and compute dominance point
6273 2017-02-23 Bin Cheng <bin.cheng@arm.com>
6275 PR tree-optimization/79663
6276 * tree-predcom.c (combine_chains): Process refs in reverse order
6277 only for ZERO length chains, and add explaining comment.
6279 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6281 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
6282 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
6283 * doc/invoke.texi (ARM Options): Add armv8.3-a.
6285 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6287 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
6289 2017-11-15 Martin Liska <mliska@suse.cz>
6291 * tree-cfg.c (pass_warn_function_return::execute):
6292 Compare warn_return_type for greater than zero.
6294 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
6298 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
6299 to return true on Xeon and not on Xeon Phi.
6300 (ix86_check_avx256_register): Changed to ...
6301 (ix86_check_avx_upper_register): ... this. Add extra check for
6302 VALID_AVX512F_REG_OR_XI_MODE.
6303 (ix86_avx_u128_mode_needed): Changed
6304 ix86_check_avx256_register to ix86_check_avx_upper_register.
6305 (ix86_check_avx256_stores): Changed to ...
6306 (ix86_check_avx_upper_stores): ... this. Changed
6307 ix86_check_avx256_register to ix86_check_avx_upper_register.
6308 (ix86_avx_u128_mode_after): Changed
6309 avx_reg256_found to avx_upper_reg_found. Changed
6310 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
6311 (ix86_avx_u128_mode_entry): Changed
6312 ix86_check_avx256_register to ix86_check_avx_upper_register.
6313 (ix86_avx_u128_mode_exit): Ditto.
6314 * config/i386/i386.h: (host_detect_local_cpu): New define.
6316 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6318 * config/arm/xgene1.md (xgene1): Split into automatons
6319 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
6320 (xgene1_f_load): Adjust reservations and/or types.
6321 (xgene1_f_store): Likewise.
6322 (xgene1_load_pair): Likewise.
6323 (xgene1_store_pair): Likewise.
6324 (xgene1_fp_load1): Likewise.
6325 (xgene1_load1): Likewise.
6326 (xgene1_store1): Likewise.
6327 (xgene1_move): Likewise.
6328 (xgene1_alu): Likewise.
6329 (xgene1_simd): Likewise.
6330 (xgene1_bfm): Likewise.
6331 (xgene1_neon_load1): Likewise.
6332 (xgene1_neon_store1): Likewise.
6333 (xgene1_neon_logic): Likewise.
6334 (xgene1_neon_st1): Likewise.
6335 (xgene1_neon_ld1r): Likewise.
6336 (xgene1_alu_cond): Added.
6337 (xgene1_shift_reg): Likwise.
6338 (xgene1_bfx): Likewise.
6339 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
6341 2017-11-15 Jakub Jelinek <jakub@redhat.com>
6344 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
6346 (can_widen_mult_without_libcall): New function.
6347 (expand_mul_overflow): If only checking unsigned mul overflow,
6348 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
6349 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
6350 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
6351 (expand_DIVMOD): Formatting fix.
6352 * expmed.h (expand_mult): Add NO_LIBCALL argument.
6353 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
6354 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
6356 PR tree-optimization/82977
6357 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
6358 constructed temporary to strlen_to_stridx.put.
6360 2017-11-15 Martin Liska <mliska@suse.cz>
6362 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
6363 * configure: Regenerate.
6365 2017-11-15 Martin Liska <mliska@suse.cz>
6368 * config/sh/sh-mem.cc: Use proper probability for
6371 2017-11-14 Jeff Law <law@redhat.com>
6373 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
6374 the red zone for stack_clash_protection_final_dynamic_probe targets
6375 when the total dynamic stack size is zero bytes.
6377 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
6378 blocks is post order.
6380 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
6382 * dumpfile.h (TDF_COMPARE_DEBUG): New.
6383 * final.c (rest_of_clean_state): Set it for the
6384 -fcompare-debug dump.
6385 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
6386 class when TDF_COMPARE_DEBUG is set.
6388 * dwarf2out.c (gen_producer_string): Discard
6391 2017-11-15 Joseph Myers <joseph@codesourcery.com>
6394 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
6395 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
6396 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
6397 (__TGMATH_REAL_2_3): Remove macros.
6398 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
6399 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
6401 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
6403 (remquo): Define using __TGMATH_REAL_3.
6405 2017-11-14 Jeff Law <law@redhat.com>
6407 * vr-values.c: New file with contents extracted from tree-vrp.c.
6408 * Makefile.in (OBJS): Add vr-values.o
6409 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
6410 (set_value_range, set_and_canonicalize_value_range): Likewise.
6411 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
6412 (value_range_constant_singleton, symbolic_range_p): Likewise.
6413 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
6414 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
6415 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
6416 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
6417 (find_case_label_range, find_case_label_index): Likewise.
6418 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
6419 (range_int_cst_singleton_p, value_inside_range): Likewise.
6420 (get_single_symbol): Likewise.
6421 (switch_update): Move structure definition here.
6422 (to_remove_edges, to_update_switch_stmts): Provide externs.
6423 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
6424 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
6425 (vrp_val_is_min, set_value_range): Likewise.
6426 (set_and_canonicalize_value_range, copy_value_range): Likewise.
6427 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
6428 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
6429 (range_is_nonnull, range_int_cst_p): Likewwise.
6430 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
6431 (get_single_symbol, operand_less_p): Likewise
6432 (compare_values_warnv, compare_values): Likewise.
6433 (value_inside_range, value_range_constant_singleton): Likewise.
6434 (zero_nonzero_bitgs_from_vr): Likewise.
6435 (extract_range_from_binary_expr_1): Likewise.
6436 (overflow_comparison_p): Likewise.
6437 (to_remove_edges, to_update_switch_stmts): Likewise.
6438 (find_case_label-index, find_case_label_range): Likewise.
6439 (switch_update, set_value_range_to_nonnegative): Remove.
6440 (set_value_range_to_truthvalue): Likewise.
6441 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
6442 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
6443 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
6444 (find_case_label_ranges, test_for_singularity): Likewise.
6445 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
6446 (x_vr_values): Move to its remaining use site.
6448 2017-11-10 Jeff Law <law@redhat.com>
6450 * vr-values.h (VR_INITIALIZER): Move #define here.
6451 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
6452 * Makefile.in (OBJS): Add tree-evrp.o
6453 * tree-vrp.h (assert_info): Move structure definition here.
6454 (set_value_range_to_varying): Prototype.
6455 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6456 (infer_value_range, register_edge_assert_for): Likewise.
6457 (stmt_interesting_for_vrp): Likewise.
6458 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
6459 (set_value_range_to_varying): No longer static.
6460 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6461 (infer_value_range, register_edge_assert_for): Likewise.
6463 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6465 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
6466 generate the XXBRD instruction.
6468 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
6469 (rs6000_builtin_type_compatible): Treat _Float128 and long double
6470 as being compatible if -mabi=ieeelongdouble.
6471 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
6472 to setup float128 built-ins with hardware support.
6473 (BU_FLOAT128_HW_2): Likewise.
6474 (BU_FLOAT128_HW_3): Likewise.
6475 (BU_FLOAT128_HW_VSX_1): Likewise.
6476 (BU_FLOAT128_HW_VSX_2): Likewise.
6477 (scalar_extract_expq): Change float128 built-in functions to
6478 accommodate having both KFmode and TFmode functions. Use the
6479 KFmode variant as the default.
6480 (scalar_extract_sigq): Likewise.
6481 (scalar_test_neg_qp): Likewise.
6482 (scalar_insert_exp_q): Likewise.
6483 (scalar_insert_exp_qp): Likewise.
6484 (scalar_test_data_class_qp): Likewise.
6485 (sqrtf128_round_to_odd): Delete processing the round to odd
6486 built-in functions as special built-in functions, and define them
6487 as float128 built-ins. Use the KFmode variant as the default.
6488 (truncf128_round_to_odd): Likewise.
6489 (addf128_round_to_odd): Likewise.
6490 (subf128_round_to_odd): Likewise.
6491 (mulf128_round_to_odd): Likewise.
6492 (divf128_round_to_odd): Likewise.
6493 (fmaf128_round_to_odd): Likewise.
6494 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6495 support for KFmode and TFmode xststdcqp calls.
6496 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
6497 point, switch the built-in handlers for the get/set float128
6498 exponent, get float128 mantissa, float128 test built-ins, and the
6499 float128 round to odd built-in functions. Eliminate creating the
6500 float128 round to odd built-in functions as special built-ins.
6501 (rs6000_init_builtins): Eliminate special creation of the float128
6502 round to odd built-in functions.
6503 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
6504 function insns to support both TFmode and KFmode variants.
6505 (xsxsigqp_<mode>): Likewise.
6506 (xsiexpqpf_<mode>): Likewise.
6507 (xsiexpqp_<mode>): Likewise.
6508 (xststdcqp_<mode>): Likewise.
6509 (xststdcnegqp_<mode>): Likewise.
6510 (xststdcqp_<mode>): Likewise.
6512 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6514 * tree-ssa-threadupdate.c (compute_path_counts): Remove
6515 unused path_in_freq_ptr parameter.
6516 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
6518 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6520 * ipa-inline.c (edge_badness): Dump sreal frequency.
6521 (compute_inlined_call_time): Match natural implementaiton ...
6522 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
6523 forgotten division by CGRAPH_FREQ_BASE.
6525 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6527 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
6528 Solaris 11. Update comment.
6529 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
6531 * config/sol2.h (STARTFILE_SPEC): Likewise.
6532 * configure: Regenerate.
6534 2017-11-14 Carl Love <cel@us.ibm.com>
6536 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
6537 le_ and be_ prefixes to swap* variables. Remove
6538 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
6541 2017-11-14 Jason Merrill <jason@redhat.com>
6543 Support GTY((cache)) on hash_map.
6544 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
6545 (ggc_cache_remove): Override it instead of ggc_mx.
6546 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
6547 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
6548 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
6549 (simple_cache_map_traits): Override maybe_mx.
6550 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
6551 (hash_map): Friend gt_cleare_cache.
6552 (gt_cleare_cache): New.
6553 * tree.h (tree_cache_traits): New hash_map traits class.
6554 (tree_cache_map): New typedef.
6556 2017-11-14 Richard Biener <rguenther@suse.de>
6558 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
6559 paramter and handling.
6560 (cleanup_control_flow_bb): Likewise.
6561 (cleanup_control_flow_pre): New helper performing a DFS walk
6562 to call cleanup_control_flow_bb in PRE order.
6563 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
6564 via cleanup_control_flow_pre.
6566 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
6568 * config/aarch64/aarch64-simd.md
6569 (aarch64_simd_bsl<mode>_internal): Remove DImode.
6570 (*aarch64_simd_bsl<mode>_alt): Likewise.
6571 (aarch64_simd_bsldi_internal): New.
6572 (aarch64_simd_bsldi_alt): Likewise.
6574 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6576 * tracer.c (better_p): Do not compare frequencies.
6577 * reg-stack.c (better_edge): Likewise.
6578 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
6581 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6583 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
6584 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
6585 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
6586 * lto-streamer-in.c (input_function): Use update_max_bb_count.
6587 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
6588 * predict.c (maybe_hot_frequency_p): Inline to ...
6589 (maybe_hot_count_p): ... here; rewrite to counts.
6590 (counts_to_freqs): Rename to ...
6591 (update_max_bb_count): ... this one.
6592 (expensive_function_p): Use counts.
6593 (estimate_bb_frequencies): Update.
6594 (rebuild_frequencies): Update.
6595 * predict.h (counts_to_freqs): Rename to ...
6596 (update_max_bb_count): ... this one.
6597 * profile.c (compute_branch_probabilities): Add debug info
6598 * tree-inline.c (expand_call_inline): Update debug info.
6599 (optimize_inline_calls): Use update_max_bb_count..
6600 (tree_function_versioning): Use update_max_bb_count..
6601 * value-prof.c (gimple_value_profile_transformations):
6602 Do not use update_max_bb_count.
6604 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6606 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
6607 always use frequencies.
6609 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6611 * bb-reorder.c: Remove frequencies from comments.
6612 (better_edge_p): Use profile counts.
6613 (find_traces): Dump profile counts.
6614 (rotate_loop): Use profile counts.
6615 (find_traces_1_round): Likewise.
6616 (connect_better_edge_p): Use counts instead of probabilities for
6618 (copy_bb_p): Drop early check for non-0 frequency.
6619 (sanitize_hot_paths): Update comments.
6621 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6623 * ipa-split.c (struct split_point): Add count.
6624 (consider_split): Do not compute incoming frequency; compute incoming
6625 count and store it to split_point.
6626 (split_function): Set count of the call to split part correctly.
6628 2017-11-13 Carl Love <cel@us.ibm.com>
6630 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
6632 2017-11-13 Tom Tromey <tom@tromey.com>
6634 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
6636 2017-11-13 Carl Love <cel@us.ibm.com>
6638 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
6639 Add support for builtins:
6640 unsigned int vec_first_{,miss}_match_{,or_eos}index,
6641 vector {un,}signed {char,int,short},
6642 vector {un,}signed {char,int,short}) arguments.
6643 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
6644 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
6645 Add BU_P9V_AV_2 expansions for the builtins.
6646 * config/rs6000/altivec.h (vec_first_match_index,
6647 vec_first_mismatch_index, vec_first_match_or_eos_index,
6648 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
6649 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
6650 new extern declaration.
6651 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
6652 * config/rs6000/vsx.md (first_match_index_<mode>,
6653 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
6654 first_mismatch_or_eos_index_<mode>): Add define expand.
6655 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
6656 * doc/extend.texi: Update the built-in documenation file for the new
6659 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
6661 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
6662 into the min/max operations for _Float<N> and _Float<N>X types.
6664 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6667 * dwarf2out.c (do_eh_frame): New static variable.
6668 (dwarf2out_begin_prologue): Set it.
6669 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
6671 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6673 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
6675 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
6677 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6680 * debug.h (dwarf2out_do_eh_frame): Declare.
6681 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
6682 (dwarf2out_do_frame): Use it.
6683 (dwarf2out_do_cfi_asm): Likewise.
6684 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
6685 (dwarf2out_assembly_start): Likewise.
6686 (dwarf2out_begin_prologue): Fix comment.
6687 * toplev.c (compile_file): Always call dwarf2out_frame_finish
6688 if the target needs either debug or unwind DWARF2 info.
6689 * lto-opts.c (lto_write_options): Do not save -fexceptions,
6690 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
6691 -ftrapping-math, -ftrapv and -fwrapv.
6693 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6695 * cgraph.c (cgraph_edge::sreal_frequency): New function.
6696 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
6697 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
6698 (estimate_edge_size_and_time): Likewise.
6699 (ipa_merge_fn_summary_after_inlining): Likewise.
6700 * ipa-inline.c (cgraph_freq_base_rec): Remove.
6701 (compute_uninlined_call_time): Use sreal_frequency.
6702 (compute_inlined_call_time): Likewise.
6703 (ipa_inline): Do not initialize cgraph_freq_base_rec.
6704 * profile-count.c: Include sreal.h.
6705 (profile_count::to_sreal_scale): New.
6706 * profile-count.h: Forward declare sreal.
6707 (profile_count::to_sreal_scale): Declare.
6709 2017-11-13 Nathan Sidwell <nathan@acm.org>
6711 * diagnostic.c (maybe_line_and_column): New.
6712 (diagnostic_get_location_text): Use it.
6713 (diagnostic_report_current_module): Likewise.
6714 (test_diagnostic_get_location_text): Add tests.
6716 2017-11-13 Luis Machado <luis.machado@linaro.org>
6718 * doc/md.texi (Specifying processor pipeline description): Fix
6719 incorrect latency for the div instruction example.
6721 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6723 PR tree-optimization/78821
6724 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
6725 that bit_not_p is the same.
6726 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
6727 (split_group): Count precisely bit_not_p bits in each statement.
6728 (invert_op): New function.
6729 (imm_store_chain_info::output_merged_store): Use invert_op to
6730 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
6731 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
6733 2017-11-13 Martin Liska <mliska@suse.cz>
6735 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
6736 (struct source_info): Likewise.
6737 (add_branch_counts): Likewise.
6738 (add_line_counts): Likewise.
6739 (function_summary): Likewise.
6740 (output_intermediate_line): Likewise.
6741 (generate_results): Likewise.
6743 2017-11-13 Martin Liska <mliska@suse.cz>
6745 * gcov.c (struct block_info): Remove typedef for block_t.
6746 (struct line_info): Likewise.
6747 (line_info::has_block): Likewise.
6748 (EXIT_BLOCK): Likewise.
6749 (unblock): Likewise.
6750 (circuit): Likewise.
6751 (get_cycles_count): Likewise.
6752 (process_file): Likewise.
6753 (read_graph_file): Likewise.
6754 (solve_flow_graph): Likewise.
6755 (find_exception_blocks): Likewise.
6756 (add_line_counts): Likewise.
6757 (accumulate_line_info): Likewise.
6758 (output_line_details): Likewise.
6760 2017-11-13 Martin Liska <mliska@suse.cz>
6762 * gcov.c (struct arc_info): Remove typedef for arc_t.
6763 (struct line_info): Likewise.
6764 (add_branch_counts): Likewise.
6765 (output_branch_count): Likewise.
6766 (function_info::~function_info): Likewise.
6767 (circuit): Likewise.
6768 (output_intermediate_line): Likewise.
6769 (read_graph_file): Likewise.
6770 (solve_flow_graph): Likewise.
6771 (find_exception_blocks): Likewise.
6772 (add_line_counts): Likewise.
6773 (accumulate_line_info): Likewise.
6774 (output_line_details): Likewise.
6775 (output_function_details): Likewise.
6777 2017-11-13 Martin Liska <mliska@suse.cz>
6779 * gcov.c (struct function_info): Remove typedef for function_t.
6780 (struct source_info): Likewise.
6781 (source_info::get_functions_at_location): Likewise.
6782 (solve_flow_graph): Likewise.
6783 (find_exception_blocks): Likewise.
6784 (add_line_counts): Likewise.
6785 (output_intermediate_file): Likewise.
6786 (process_file): Likewise.
6787 (generate_results): Likewise.
6788 (release_structures): Likewise.
6789 (read_graph_file): Likewise.
6790 (read_count_file): Likewise.
6791 (accumulate_line_counts): Likewise.
6792 (output_lines): Likewise.
6794 2017-11-13 Martin Liska <mliska@suse.cz>
6796 * gcov.c (function_info::function_info): Remove num_counts
6797 and add vector<gcov_type>.
6798 (function_info::~function_info): Use the vector.
6799 (process_file): Likewise.
6800 (read_graph_file): Likewise.
6801 (read_count_file): Likewise.
6802 (solve_flow_graph): Likewise.
6804 2017-11-13 Martin Liska <mliska@suse.cz>
6806 * gcov.c (function_info::is_artificial): New function.
6807 (process_file): Erase all artificial early.
6808 (generate_results): Skip as all artificial are already
6811 2017-11-13 Martin Liska <mliska@suse.cz>
6813 * gcov.c (read_graph_file): Store to global vector of functions.
6814 (read_count_file): Iterate the vector.
6815 (process_file): Likewise.
6816 (generate_results): Likewise.
6817 (release_structures): Likewise.
6819 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6821 PR tree-optimization/82954
6822 * gimple-ssa-store-merging.c
6823 (imm_store_chain_info::coalesce_immediate_stores): If
6824 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
6826 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
6828 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6829 Upddate call to ENDIAN_LANE_N.
6830 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
6831 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
6832 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
6833 and use aarch64_endian_lane_rtx.
6834 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
6836 2017-11-12 Tom de Vries <tom@codesourcery.com>
6838 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
6840 2017-11-12 Tom de Vries <tom@codesourcery.com>
6842 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
6844 2017-11-12 Tom de Vries <tom@codesourcery.com>
6846 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
6848 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6849 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6850 * defaults.h (ASM_OUTPUT_LABELREF): Same.
6852 2017-11-11 Martin Sebor <msebor@redhat.com>
6855 * doc/extend.texi (attribute nonstring): Remove spurious argument.
6858 * prefic.c (translate_name): Replace strncpy with memcpy to
6859 avoid -Wstringop-truncation.
6861 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6863 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
6865 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6867 * predict.c (maybe_hot_frequency_p): Do not use cfun.
6869 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6871 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
6874 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
6876 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
6877 enable generating XXBRH if the value is in a vector register.
6878 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
6879 value is in a vector register.
6880 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
6881 register to register bswap64's instead of doing the GPR sequence
6882 used on previous machines.
6883 (bswapdi2_xxbrd): New insn.
6884 (bswapdi2_reg): Disallow on ISA 3.0.
6885 (register to register bswap64 splitter): Do not split the insn on
6886 ISA 3.0 systems that use XXBRD.
6888 2017-11-10 Martin Sebor <msebor@redhat.com>
6891 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
6893 (find_subframework_file): Same.
6895 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6897 * auto-profile.c (afdo_indirect_call): Drop frequency.
6898 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
6899 (cgraph_node::create_edge): Drop frequency argument.
6900 (cgraph_node::create_indirect_edge): Drop frequency argument.
6901 (cgraph_edge::make_speculative): Drop frequency arguments.
6902 (cgraph_edge::resolve_speculation): Do not update frequencies
6903 (cgraph_edge::dump_edge_flags): Do not dump frequency.
6904 (cgraph_node::dump): Check consistency in IPA mode.
6905 (cgraph_edge::maybe_hot_p): Use IPA counter.
6906 (cgraph_edge::verify_count_and_frequency): Rename to ...
6907 (cgraph_edge::verify_count): ... this one; drop frequency checking.
6908 (cgraph_node::verify_node): Update.
6909 * cgraph.h (struct cgraph_edge): Drop frequency.
6910 (cgraph_edge::frequency): New function.
6911 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
6913 (cgraph_edge::rebuild_edges): Likewise.
6914 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
6915 (duplicate_thunk_for_node): Do not pass frequency.
6916 (cgraph_node::create_clone): Scale only counts.
6917 (cgraph_node::create_virtual_clone): Do not pass frequency.
6918 (cgraph_node::create_edge_including_clones): Do not pass frequency.
6919 (cgraph_node::create_version_clone): Do not pass frequency.
6920 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
6921 (cgraph_node::expand_thunk): Do not pass frequency.
6922 (cgraph_node::create_wrapper): Do not pass frequency.
6923 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
6925 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
6926 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
6927 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
6928 (gather_caller_stats): Use frequency function.
6929 (ipcp_cloning_candidate_p): Use frequency function.
6930 (ipcp_propagate_stage): Use frequency function.
6931 (get_info_about_necessary_edges): Use frequency function.
6932 (update_profiling_info): Update only IPA profile.
6933 (update_specialized_profile): Use frequency functoin.
6934 (perhaps_add_new_callers): Update only IPA profile.
6935 * ipa-devirt.c (ipa_devirt): Use IPA profile.
6936 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
6937 (dump_ipa_call_summary): Use frequency function.
6938 (estimate_edge_size_and_time): Use frequency function.
6939 (ipa_merge_fn_summary_after_inlining): Use frequency function.
6940 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
6941 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
6942 (update_noncloned_counts): ... ths one; scale counts only.
6943 (clone_inlined_nodes): Do not scale frequency.
6944 (inline_call): Do not pass frequency.
6945 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
6946 (compute_inlined_call_time): Use IPA profile.
6947 (want_inline_small_function_p): Use IPA profile.
6948 (want_inline_self_recursive_call_p): Use IPA profile.
6949 (edge_badness): Use IPA profile.
6950 (lookup_recursive_calls): Use IPA profile.
6951 (recursive_inlining): Do not pass frequency.
6952 (resolve_noninline_speculation): Do not update frequency.
6953 (inline_small_functions): Collect max of IPA profile.
6954 (dump_overall_stats): Dump IPA porfile.
6955 (dump_inline_stats): Dump IPA porfile.
6956 (ipa_inline): Collect IPA stats.
6957 * ipa-inline.h (clone_inlined_nodes): Update prototype.
6958 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
6959 (ipa_propagate_frequency): Use frequency function.
6960 (ipa_profile): Cleanup.
6961 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
6962 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
6963 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
6964 (input_node): Do not stream frequency.
6965 (input_edge): Do not stream frequency.
6966 (merge_profile_summaries): Scale only IPA profiles.
6967 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
6968 * predict.c (drop_profile): Do not recompute frequency.
6969 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
6970 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
6971 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
6972 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
6973 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
6974 * tree-inline.c (copy_bb): Do not scale frequency.
6975 (expand_call_inline): Do not scale frequency.
6976 (tree_function_versioning): Do not scale frequency.
6977 * ubsan.c (ubsan_create_edge): Do not pass frequency.
6979 2017-11-10 Julia Koval <julia.koval@intel.com>
6981 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
6982 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
6983 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
6984 (_mm256_maskz_gf2p8affine_epi64_epi8)
6985 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
6986 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
6987 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
6988 (__builtin_ia32_vgf2p8affineqb_v32qi)
6989 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
6990 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
6992 2017-11-10 Tamar Christina <tamar.christina@arm.com>
6996 (arm_option_override): Refactor.
6997 (arm_option_reconfigure_globals): New.
6998 (arm_options_perform_arch_sanity_checks): New.
6999 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
7001 (arm_options_perform_arch_sanity_checks): Likewise
7003 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
7005 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
7006 (power9-qpmul): New.
7007 * rs6000/rs6000.md ("type" attr): Add qmul.
7008 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
7009 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
7010 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
7012 2017-11-10 Martin Sebor <msebor@redhat.com>
7015 * builtins.c (compute_objsize): Handle arrays that
7016 compute_builtin_object_size likes to fail for. Make extern.
7017 * builtins.h (compute_objsize): Declare.
7018 (check_strncpy_sizes): New function.
7019 (expand_builtin_strncpy): Call check_strncpy_sizes.
7020 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
7021 -Wstringop-truncation.
7022 (gimple_fold_builtin_strncat): Same.
7023 * gimple.c (gimple_build_call_from_tree): Set call location.
7024 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
7025 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
7026 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
7027 (handle_builtin_strlen): Use strlen_to_stridx.
7028 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
7030 Use strlen_to_stridx.
7031 (pass_strlen::execute): Release strlen_to_stridx.
7032 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
7033 (-Wstringop-truncation): Document new option.
7035 2017-11-10 Martin Liska <mliska@suse.cz>
7037 PR gcov-profile/82702
7038 * gcov.c (main): Handle intermediate files in a different
7040 (get_gcov_intermediate_filename): New function.
7041 (output_gcov_file): Remove support of intermediate files.
7042 (generate_results): Allocate intermediate file.
7043 (release_structures): Clean-up properly fn_end.
7044 (output_intermediate_file): Start iterating with line 1.
7046 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7048 PR tree-optimization/82929
7049 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7050 ops_swapped_p non-static data member.
7051 (store_immediate_info::store_immediate_info): Clear it.
7052 (imm_store_chain_info::coalesce_immediate_stores): If swapping
7053 ops set ops_swapped_p.
7054 (count_multiple_uses): Handle ops_swapped_p.
7056 2017-11-10 Martin Liska <mliska@suse.cz>
7058 * coverage.c (coverage_init): Stream information about
7059 support of has_unexecuted_blocks.
7060 * doc/gcov.texi: Document that.
7061 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
7062 * gcov.c (read_graph_file): Likewise.
7063 (output_line_beginning): Fix a small issue with
7066 2017-11-10 Bin Cheng <bin.cheng@arm.com>
7068 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
7069 reference of trivial component.
7071 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7074 * gimple-ssa-store-merging.c
7075 (pass_store_merging::terminate_all_aliasing_chains): For
7076 gimple_store_p stmts also call refs_output_dependent_p.
7078 PR rtl-optimization/82913
7079 * compare-elim.c (try_merge_compare): Punt if def_insn is not
7082 2017-11-09 Jeff Law <law@redhat.com>
7084 * vr-values.h: New file with vr_values class.
7085 * tree-vrp.c: Include vr-values.h
7086 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
7087 data objects into the vr_values class.
7088 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
7089 (get_value_range): Make it a member function within vr_values class.
7090 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
7091 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
7092 (op_with_constant_singleton_value_range): Likewise.
7093 (extract_range_for_var_from_comparison_expr): Likewise.
7094 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
7095 (extract_range_from_binary_expr): Likewise.
7096 (extract_range_from_unary_expr): Likewise.
7097 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
7098 (check_for_binary_op_overflow, extract_range_basic): Likewise.
7099 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
7100 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
7101 (compare_name_with_value, compare_names): Likewise.
7102 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
7103 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
7104 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
7105 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
7106 (extract_range_from_phi_node): Likewise.
7107 (simplify_truth_ops_using_ranges): Likewise.
7108 (simplify_div_or_mod_using_ranges): Likewise.
7109 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
7110 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
7111 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
7112 (simplify_float_conversion_using_ranges): Likewise.
7113 (simplify_internal_call_using_ranges): Likewise.
7114 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
7115 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
7116 poitner via x_vr_values for calls into gimple folder.
7117 (vrp_initialize_lattice): Make this the vr_values ctor.
7118 (vrp_free_lattice): Make this the vr_values dtor.
7119 (set_vr_value): New function.
7120 (class vrp_prop): Add vr_values data member. Add various member
7121 functions as well as member functions that delegate to vr_values.
7122 (check_array_ref): Make a member function within vrp_prop class.
7123 (search_for_addr_array, vrp_initialize): Likewise.
7124 (vrp_finalize): Likewise. Revamp to avoid direct access to
7125 vr_value, values_propagated, etc.
7126 (check_array_bounds): Extract vrp_prop class instance pointer from
7127 walk info structure. Use it to call member functions.
7128 (check_all_array_refs): Make a member function within vrp_prop class.
7129 Smuggle class instance pointer via walk info structure.
7130 (x_vr_values): New local static.
7131 (vrp_valueize): Use x_vr_values to get class instance.
7132 (vr_valueize_1): Likewise.
7133 (class vrp_folder): Add vr_values data member. Add various member
7134 functions as well as member functions that delegate to vr_values.
7135 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
7136 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
7137 class instance from vr_values. Use it to call member functions.
7138 (vrp_dom_walker): Add vr_values data member.
7139 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
7140 instance via x_vr_values.
7141 (identify_jump_threads): Accept vr_values as argument. Store
7142 it into the walker structure.
7143 (evrp_dom_walker): Add vr_values class data member. Add various
7145 (evrp_dom_walker::try_find_new_range): Use vr_values data
7146 member to access the memory allocator.
7147 (evrp_dom_walker::before_dom_children): Store vr_values class
7148 instance into the vrp_folder class.
7149 (evrp_dom_walker::push_value_range): Rework to avoid direct
7150 access to num_vr_values and vr_value.
7151 (evrp_dom_walker::pop_value_range): Likewise.
7152 (execute_early_vrp): Remove call to vrp_initialize_lattice.
7153 Use vr_values to get to dump_all_value_ranges member function.
7154 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
7155 and simplify_cond_using_ranges_2 via vrp_prop class instance.
7156 Pass vr_values class instance down to identify_jump_threads.
7157 Remove call to vrp_free_lattice.
7158 (debug_all_value_ranges): Remove.
7160 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
7161 (vrp_folder): Likewise.
7163 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
7164 Get it from the existing bitmap instead.
7165 (vrp_intersect_ranges_1): Likewise.
7167 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7169 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7171 (store_immediate_info::store_immediate_info): Add bitnotp argument,
7172 set bit_not_p to it.
7173 (imm_store_chain_info::coalesce_immediate_stores): Break group
7174 if bit_not_p is different.
7175 (count_multiple_uses, split_group,
7176 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
7177 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
7178 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
7179 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
7180 to store_immediate_info ctor.
7182 2017-11-09 Jim Wilson <jimw@sifive.com>
7184 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
7185 (scan_prog_file): Likewise.
7187 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7189 * bb-reorder.c (max_entry_frequency): Remove.
7190 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
7191 connect_traces, push_to_next_round_p): Remove prototypes.
7192 (find_traces_1_round): Use counts only.
7193 (push_to_next_round_p): Likewise.
7194 (find_traces): Likewise.
7195 (rotate_loop): Likewise.
7196 (find_traces_1_round): Likewise.
7197 (connect_traces): Likewise.
7198 (edge_order): Likewise.
7200 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
7202 * config/arm/arm.c (output_return_instruction): Add comments to
7203 indicate requirement for cmse_nonsecure_entry return to account
7204 for the size of clearing instruction output here.
7205 (thumb_exit): Likewise.
7206 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
7207 return in hardfloat mode.
7209 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
7211 * config/rs6000/rs6000.c (machine_function): Add a bool,
7212 "toc_is_wrapped_separately".
7213 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
7214 if it wasn't explicitly set or unset, we are optimizing for speed, and
7215 doing separate shrink-wrapping.
7216 (rs6000_get_separate_components): Enable the TOC component if
7217 saving the TOC register in the prologue.
7218 (rs6000_components_for_bb): Handle the TOC component.
7219 (rs6000_emit_prologue_components): Store the TOC register where needed.
7220 (rs6000_set_handled_components): Mark TOC as handled, if handled.
7221 (rs6000_emit_prologue): Don't save the TOC if that is already done.
7223 2017-11-09 Martin Jambor <mjambor@suse.cz>
7225 * ipa-param-manipulation.c: New file.
7226 * ipa-param-manipulation.h: Likewise.
7227 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
7228 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
7229 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
7230 (ipa_parm_adjustment): Likewise.
7231 (ipa_parm_adjustment_vec): Likewise.
7232 (ipa_get_vector_of_formal_parms): Moved declaration to
7233 ipa-param-manipulation.h.
7234 (ipa_get_vector_of_formal_parm_types): Likewise.
7235 (ipa_modify_formal_parameters): Likewise.
7236 (ipa_modify_call_arguments): Likewise.
7237 (ipa_combine_adjustments): Likewise.
7238 (ipa_dump_param_adjustments): Likewise.
7239 (ipa_modify_expr): Likewise.
7240 (ipa_get_adjustment_candidate): Likewise.
7241 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
7242 ipa-param-manipulation.c.
7243 (ipa_get_vector_of_formal_parm_types): Likewise.
7244 (ipa_modify_formal_parameters): Likewise.
7245 (ipa_modify_call_arguments): Likewise.
7246 (ipa_modify_expr): Likewise.
7247 (get_ssa_base_param): Likewise.
7248 (ipa_get_adjustment_candidate): Likewise.
7249 (index_in_adjustments_multiple_times_p): Likewise.
7250 (ipa_combine_adjustments): Likewise.
7251 (ipa_dump_param_adjustments): Likewise.
7252 * tree-sra.c: Also include ipa-param-manipulation.h
7253 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
7256 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7257 Alan Hayward <alan.hayward@arm.com>
7258 David Sherwood <david.sherwood@arm.com>
7260 * doc/sourcebuild.texi (vect_masked_store): Document.
7262 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7263 Alan Hayward <alan.hayward@arm.com>
7264 David Sherwood <david.sherwood@arm.com>
7266 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
7268 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7269 Alan Hayward <alan.hayward@arm.com>
7270 David Sherwood <david.sherwood@arm.com>
7272 * doc/sourcebuild.texi (vect_variable_length): Document.
7274 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7275 Alan Hayward <alan.hayward@arm.com>
7276 David Sherwood <david.sherwood@arm.com>
7278 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
7280 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7281 Alan Hayward <alan.hayward@arm.com>
7282 David Sherwood <david.sherwood@arm.com>
7284 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
7286 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7287 Alan Hayward <alan.hayward@arm.com>
7288 David Sherwood <david.sherwood@arm.com>
7290 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
7291 previously undocumented selectors.
7292 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
7294 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7296 * doc/rtl.texi (const_vector): Say that elements can be
7297 const_wide_ints too.
7298 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
7299 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
7300 (gen_vec_duplicate): Use it instead of CONSTANT_P.
7301 * optabs.c (expand_vector_broadcast): Likewise.
7303 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7304 Alan Hayward <alan.hayward@arm.com>
7305 David Sherwood <david.sherwood@arm.com>
7307 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
7308 scaled index even if the unscaled address was invalid.
7309 Don't increase the complexity of using a scale in that case.
7311 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7312 Alan Hayward <alan.hayward@arm.com>
7313 David Sherwood <david.sherwood@arm.com>
7315 * doc/rtl.texi: Rewrite the subreg rules so that they partition
7316 the inner register into REGMODE_NATURAL_SIZE bytes rather than
7317 UNITS_PER_WORD bytes.
7318 * emit-rtl.c (validate_subreg): Divide subregs into blocks
7319 based on REGMODE_NATURAL_SIZE of the inner mode.
7320 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
7321 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
7322 * expmed.c (lowpart_bit_field_p): Divide the value up into
7323 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
7324 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
7325 whether something is likely to occupy more than one register.
7327 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7330 * ipa-inline-transform.c (update_noncloned_frequencies): Use
7331 profile_count::adjust_for_ipa_scaling.
7332 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
7333 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
7335 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
7337 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7339 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
7340 (split_group): Add total_orig and total_new arguments, estimate the
7341 number of statements related to the store group without store merging
7342 and with store merging.
7343 (imm_store_chain_info::output_merged_store): Adjust split_group
7344 callers, punt if estimated number of statements with store merging
7345 is not smaller than estimated number of statements without it.
7347 (handled_load): Remove has_single_use checks.
7348 (pass_store_merging::process_store): Likewise.
7350 2017-11-09 Richard Biener <rguenther@suse.de>
7352 PR tree-optimization/82902
7353 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
7355 2017-11-09 Martin Liska <mliska@suse.cz>
7358 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
7361 2017-11-09 Martin Liska <mliska@suse.cz>
7363 PR tree-optimization/82669
7364 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
7366 2017-11-09 Martin Liska <mliska@suse.cz>
7368 PR gcov-profile/48463
7369 * coverage.c (coverage_begin_function): Output also end locus
7370 of a function and information whether the function is
7372 * gcov-dump.c (tag_function): Parse and print the information.
7373 * gcov.c (INCLUDE_MAP): Add include.
7374 (INCLUDE_SET): Likewise.
7375 (struct line_info): Move earlier in the source file because
7376 of vector<line_info> in function_info structure.
7377 (line_info::line_info): Likewise.
7378 (line_info::has_block): Likewise.
7379 (struct source_info): Add new member index.
7380 (source_info::get_functions_at_location): New function.
7381 (function_info::group_line_p): New function.
7382 (output_intermediate_line): New function.
7383 (output_intermediate_file): Use the mentioned function.
7384 (struct function_start): New.
7385 (struct function_start_pair_hash): Likewise.
7386 (process_file): Add code that identifies group functions.
7387 Assign lines either to global or function scope.
7388 (generate_results): Skip artificial functions.
7389 (find_source): Assign index for each source file.
7390 (read_graph_file): Read new flag artificial and end_line.
7391 (add_line_counts): Assign it either to global of function scope.
7392 (accumulate_line_counts): Isolate core of the function to
7393 accumulate_line_info and call it for both function and global
7395 (accumulate_line_info): New function.
7396 (output_line_beginning): Fix GNU coding style.
7397 (print_source_line): New function.
7398 (output_line_details): Likewise.
7399 (output_function_details): Likewise.
7400 (output_lines): Iterate both source (global) scope and function
7402 (struct function_line_start_cmp): New class.
7403 * doc/gcov.texi: Reflect changes in documentation.
7405 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7408 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
7409 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
7410 and similarly for not instead of neg.
7412 2017-11-08 Andi Kleen <ak@linux.intel.com>
7414 * config/i386/i386.opt: Add -mforce-indirect-call.
7415 * config/i386/predicates.md: Check for flag_force_indirect_call.
7416 * doc/invoke.texi: Document -mforce-indirect-call
7418 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
7420 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
7422 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
7423 config/riscv/riscv.c (predict.h): New include.
7424 (riscv_slow_unaligned_access_p): No longer static.
7425 (riscv_block_move_straight): Add require.
7426 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7428 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7431 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
7432 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
7433 nonimmediate_operand predicate for operand 1 instead of
7436 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7438 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
7440 * config/aarch64/constraints.md (Uml): New constraint.
7441 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
7444 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7446 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
7447 of two vec_duplicates into a vec_concat.
7449 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7451 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7452 Simplify vec_merge of vec_duplicate and vec_concat.
7453 * config/aarch64/constraints.md (Utq): New constraint.
7454 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
7457 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7459 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7460 Simplify vec_merge of vec_duplicate and const_vector.
7461 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
7463 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
7464 mode iterator. Update predicate on operand 1 to
7465 handle non-const_vec constants. Delete constraints.
7466 (*aarch64_combinez_be<mode>): Likewise for operand 2.
7468 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7470 PR tree-optimization/78821
7471 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
7473 (store_operand_info::store_operand_info): Initialize it to false.
7474 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
7475 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
7476 store in the group, and if chain_info is non-NULL, to ignore altogether
7478 (compatible_load_p): Fail if bit_not_p does not match.
7479 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
7480 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
7481 (pass_store_merging::process_store): Adjust
7482 terminate_all_aliasing_chains calls to pass NULL in all current spots,
7483 call terminate_all_aliasing_chains newly when adding a store into
7484 a chain with non-NULL chain_info.
7486 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7488 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
7490 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7492 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7494 (aarch64_layout_frame): Initialise emit_frame_chain.
7495 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
7496 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
7498 2017-11-08 Martin Liska <mliska@suse.cz>
7500 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
7501 of gimple_call_internal_p.
7503 2017-11-07 Tom de Vries <tom@codesourcery.com>
7505 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
7507 2017-11-07 Tom de Vries <tom@codesourcery.com>
7509 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
7512 2017-11-08 Martin Liska <mliska@suse.cz>
7515 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
7517 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7519 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
7521 2017-11-07 Jakub Jelinek <jakub@redhat.com>
7524 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
7525 (*cmp<mode>_ccz_1): New insn with $k alternative.
7528 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
7529 RTX_COMM_COMPARE as commutative as well.
7530 (ix86_binary_operator_ok): Formatting fix.
7531 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
7532 *<code><mode>3<mask_name><round_saeonly_name>,
7533 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
7534 *<s>mul<mode>3_highpart<mask_name>,
7535 *vec_widen_umult_even_v16si<mask_name>,
7536 *vec_widen_umult_even_v8si<mask_name>,
7537 *vec_widen_umult_even_v4si<mask_name>,
7538 *vec_widen_smult_even_v16si<mask_name>,
7539 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
7540 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
7541 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
7542 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
7543 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
7544 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
7545 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
7546 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
7547 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
7548 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
7549 ix86_binary_operator_ok. Formatting fixes.
7550 (*<plusminus_insn><mode>3<mask_name><round_name>,
7551 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
7554 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
7556 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
7557 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
7558 eq and ne if TARGET_ISEL.
7559 (cmp): New code_iterator.
7560 (UNS, UNSU_, UNSIK): New code_attrs.
7561 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
7562 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
7564 ("eq<mode>3"): ... this.
7565 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
7567 ("ne<mode>3"): ... this.
7569 2017-11-07 Julia Koval <julia.koval@intel.com>
7572 * common/config/i386/i386-common.c
7573 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
7574 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
7575 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
7576 * config/i386/i386.opt: Ditto.
7577 * config/i386/i386.c (ix86_target_string): Ditto.
7578 (ix86_option_override_internal): Ditto.
7579 (ix86_init_mpx_builtins): Move MPX to args2.
7580 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
7581 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7582 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7583 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7584 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7585 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7586 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
7588 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
7591 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
7592 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
7593 (zero-extendsidi peephole2): Remove peephole.
7595 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7598 * stor-layout.c: Include attribs.h.
7599 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
7600 explicit lookup of "aligned" attribute.
7602 2017-11-07 Andrew Waterman <andrew@sifive.com>
7604 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
7605 (riscv_expand_block_move): Likewise.
7606 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
7608 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
7609 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
7610 * config/riscv/riscv.c (riscv_block_move_straight): New function.
7611 (riscv_adjust_block_mem): Likewise.
7612 (riscv_block_move_loop): Likewise.
7613 (riscv_expand_block_move): Likewise.
7614 * config/riscv/riscv.md (movmemsi): New pattern.
7616 2017-11-07 Michael Clark <michaeljclark@mac.com>
7618 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
7619 (MUSL_DYNAMIC_LINKER): Likewise.
7621 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
7623 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
7624 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
7627 2017-11-07 Richard Biener <rguenther@suse.de>
7629 * match.pd: Fix build.
7631 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
7632 Jackson Woodruff <jackson.woodruff@arm.com>
7634 PR tree-optimization/71026
7635 * match.pd: Canonicalize negate in division.
7637 2017-11-07 Sudakshina Das <sudi.das@arm.com>
7640 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
7642 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7644 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
7645 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
7647 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7649 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
7650 non-scalar integral types.
7651 * match.pd (negate_expr_p): Handle MINUS_EXPR.
7652 (-(A-B), -(~A)): New transformations.
7654 2017-11-07 Tom de Vries <tom@codesourcery.com>
7656 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
7657 semicolon after "do {} while (0)".
7658 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7659 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7660 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7661 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7662 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7664 2017-11-07 Tom de Vries <tom@codesourcery.com>
7666 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
7667 after "do {} while (0)".
7668 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7669 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7670 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7671 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7672 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7674 2017-11-07 Tom de Vries <tom@codesourcery.com>
7677 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
7679 (arm_rtx_costs_internal): Add missing semicolon after
7680 HANDLE_NARROW_SHIFT_ARITH call.
7682 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7684 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7685 disable isel if it was not set explicitly.
7687 2017-11-06 James Bowman <james.bowman@ftdichip.com>
7689 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
7690 (add_type_attribute) likewise.
7692 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
7694 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
7695 of struct ix86_frame.
7696 (ix86_initial_elimination_offset): Likewise.
7697 (ix86_expand_split_stack_prologue): Likewise.
7699 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
7701 * tree-vrp.h (enum value_range_type): Update stale comment.
7703 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7704 Alan Hayward <alan.hayward@arm.com>
7705 David Sherwood <david.sherwood@arm.com>
7707 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
7708 (aarch64_expand_vec_perm_const): Take the number of units too.
7709 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
7710 (aarch64_expand_vec_perm_const): Likewise.
7711 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
7712 (vec_perm<mode>): Update accordingly.
7714 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7715 Alan Hayward <alan.hayward@arm.com>
7716 David Sherwood <david.sherwood@arm.com>
7718 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
7719 Take the number of units too.
7720 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
7721 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
7722 but check for a vector mode before rather than after the call.
7723 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7724 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
7725 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
7726 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
7727 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
7728 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
7729 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
7730 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
7731 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
7732 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
7733 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
7734 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
7735 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
7736 (aarch64_sqdmull2_n<mode>): Update accordingly.
7738 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7739 Alan Hayward <alan.hayward@arm.com>
7740 David Sherwood <david.sherwood@arm.com>
7742 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
7743 the number of units too.
7744 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
7745 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
7746 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
7747 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
7748 (vec_store_lanesxi<mode>): Update accordingly.
7750 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7751 Alan Hayward <alan.hayward@arm.com>
7752 David Sherwood <david.sherwood@arm.com>
7754 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
7755 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
7756 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
7757 of units rather than the mode.
7758 * config/aarch64/iterators.md (nunits): New mode attribute.
7759 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
7760 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
7761 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
7762 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
7763 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
7764 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
7765 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
7766 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
7767 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
7768 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
7769 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
7770 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
7771 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
7772 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
7773 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
7774 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
7775 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
7776 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
7777 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
7778 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
7779 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
7780 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
7781 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
7782 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
7783 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
7784 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
7785 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
7786 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
7787 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
7788 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
7789 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
7790 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
7791 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
7792 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
7793 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
7794 (aarch64_simd_vec_setv2di): Likewise.
7796 2017-11-06 Carl Love <cel@us.ibm.com>
7798 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
7800 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
7801 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
7803 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
7805 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
7806 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
7807 Add power 8 macro expansions.
7808 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
7809 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
7810 power 8 instructions. (VSX_XXBR): Add iterator.
7812 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
7814 * config/arm/arm.md (predicable_short_it): Change default to "no",
7815 improve documentation, remove uses that are identical to the default.
7816 (enabled_for_depr_it): Rename to enabled_for_short_it.
7817 * gcc/config/arm/arm-fixed.md (predicable_short_it):
7818 Remove default uses.
7819 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
7820 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
7821 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
7822 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
7824 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
7827 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
7828 float128 helper macros, which are no longer used after deleting
7829 the old 'q' built-in functions, and moving the round to odd
7830 built-in functions to being special built-in functions.
7831 (BU_FLOAT128_2): Likewise.
7832 (BU_FLOAT128_1_HW): Likewise.
7833 (BU_FLOAT128_2_HW): Likewise.
7834 (BU_FLOAT128_3_HW): Likewise.
7835 (FABSQ): Delete old 'q' built-in functions.
7836 (COPYSIGNQ): Likewise.
7837 (SQRTF128_ODD): Move round to odd built-in functions to be
7838 special built-in functions, so that we can handle
7839 -mabi=ieeelongdouble.
7840 (TRUNCF128_ODD): Likewise.
7841 (ADDF128_ODD): Likewise.
7842 (SUBF128_ODD): Likewise.
7843 (MULF128_ODD): Likewise.
7844 (DIVF128_ODD): Likewise.
7845 (FMAF128_ODD): Likewise.
7846 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
7847 built-in names to 'f128'.
7848 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
7849 old 'q' built-in functions, as the machine independent code for
7850 'f128' built-in functions handles this.
7851 (rs6000_expand_builtin): Add expansion for float128 round to odd
7852 functions, keying off on -mabi=ieeelongdouble of whether to use
7853 the KFmode or TFmode variant.
7854 (rs6000_init_builtins): Initialize the _Float128 round to odd
7856 * doc/extend.texi (PowerPC Built-in Functions): Document the old
7857 _Float128 'q' built-in functions are now mapped into the new
7858 'f128' built-in functions.
7860 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
7862 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
7863 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
7865 2017-11-06 David Malcolm <dmalcolm@redhat.com>
7868 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
7869 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
7870 * toplev.c: Include "ipa-fnsummary.h".
7871 (toplev::finalize): Call ipa_fnsummary_c_finalize.
7873 2017-11-06 Jakub Jelinek <jakub@redhat.com>
7875 PR tree-optimization/82838
7876 * gimple-ssa-store-merging.c
7877 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
7878 on a separate gimple_seq which is then appended to seq.
7880 2017-11-06 Jeff Law <law@redhat.com>
7883 * config/i386/i386.c (PROBE_INTERVAL): Remove.
7884 (get_probe_interval): New functions.
7885 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
7886 (ix86_adjust_stack_and_probe): Likewise.
7887 (output_adjust_stack_and_probe): Likewise.
7888 (ix86_emit_probe_stack_range): Likewise.
7889 (ix86_expand_prologue): Likewise.
7891 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7893 PR tree-optimization/82816
7894 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
7895 if the modes of the two types are the same.
7896 (convert_plusminus_to_widen): Likewise.
7898 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7900 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
7902 (usadv16qi): New define_expand.
7903 (usadv8hi): New define_expand.
7905 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
7908 * ipa-inline-transform.c (update_noncloned_frequencies): Always
7910 (inline_transform): Likewise.
7911 * predict.c (counts_to_freqs): Remove useless conditional.
7912 * profile-count.h (profile_count::apply_scale): Move sanity check.
7913 * tree-inline.c (copy_bb): Always scale.
7914 (copy_cfg_body): Likewise.
7916 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
7919 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
7921 (*cmp_ite0): Add enabled_for_depr_it attribute.
7922 (*cmp_ite1): Likewise.
7924 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7926 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
7929 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7931 * tree-vrp.c (vrp_int_const_binop): Return true on success and
7932 return the value by pointer.
7933 (extract_range_from_multiplicative_op_1): Update accordingly.
7934 Return as soon as an operation fails.
7936 2017-11-05 Tom de Vries <tom@codesourcery.com>
7939 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
7940 (DEF_SANITIZER_BUILTIN): ... here.
7941 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
7942 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
7944 2017-11-05 Tom de Vries <tom@codesourcery.com>
7947 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
7949 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
7950 ASM_OUTPUT_BEFORE_CASE_LABEL call.
7951 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
7953 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7954 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7955 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7957 2017-11-05 Tom de Vries <tom@codesourcery.com>
7960 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
7963 2017-11-04 Michael Clark <michaeljclark@mac.com>
7965 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
7966 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
7968 (*addsi3_extended): Likewise.
7969 (*addsi3_extended2): Likewise.
7970 (<optab>si3): Likewise.
7971 (<optab>di3): Likewise.
7972 (<optab><mode>3): Likewise.
7973 (<*optabe>si3_internal): Likewise.
7974 (zero_extendqi<SUPERQI:mode>2): Likewise.
7975 (*add<mode>hi3): Likewise.
7976 (*xor<mode>hi3): Likewise.
7977 (<optab>di3): Likewise.
7978 (*<optab>si3_extend): Likewise.
7979 (*sge<u>_<X:mode><GPR:mode>): Likewise.
7980 (*slt<u>_<X:mode><GPR:mode>): Likewise.
7981 (*sle<u>_<X:mode><GPR:mode>): Likewise.
7983 2017-11-04 Andrew Waterman <andrew@sifive.com>
7985 * config/riscv/riscv.c (riscv_option_override): Conditionally set
7986 TARGET_STRICT_ALIGN based upon -mtune argument.
7988 2017-11-04 Andrew Waterman <andrew@sifive.com>
7990 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
7992 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
7994 * config/i386/i386.c (choose_basereg): Use optional scratch
7995 register and add assertion.
7996 (x86_emit_outlined_ms2sysv_save): Use scratch register when
7997 needed, and don't allocate stack.
7998 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
7999 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
8000 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
8002 2017-11-03 Jeff Law <law@redhat.com>
8004 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
8005 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
8006 to probe at the start of a noreturn function.
8008 2017-11-03 Jakub Jelinek <jakub@redhat.com>
8010 PR tree-optimization/78821
8011 * gimple-ssa-store-merging.c: Update the file comment.
8012 (MAX_STORE_ALIAS_CHECKS): Define.
8013 (struct store_operand_info): New type.
8014 (store_operand_info::store_operand_info): New constructor.
8015 (struct store_immediate_info): Add rhs_code and ops data members.
8016 (store_immediate_info::store_immediate_info): Add rhscode, op0r
8017 and op1r arguments to the ctor, initialize corresponding data members.
8018 (struct merged_store_group): Add load_align_base and load_align
8020 (merged_store_group::merged_store_group): Initialize them.
8021 (merged_store_group::do_merge): Update them.
8022 (merged_store_group::apply_stores): Pick the constant for
8023 encode_tree_to_bitpos from one of the two operands, or skip
8024 encode_tree_to_bitpos if neither operand is a constant.
8025 (class pass_store_merging): Add process_store method decl. Remove
8026 bool argument from terminate_all_aliasing_chains method decl.
8027 (pass_store_merging::terminate_all_aliasing_chains): Remove
8028 var_offset_p argument and corresponding handling.
8029 (stmts_may_clobber_ref_p): New function.
8030 (compatible_load_p): New function.
8031 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
8032 if there is overlap and rhs_code is not INTEGER_CST. For
8033 non-overlapping stores terminate group if rhs is not mergeable.
8034 (get_alias_type_for_stmts): Change first argument from
8035 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
8036 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
8037 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
8039 (get_location_for_stmts): Change first argument from
8040 auto_vec<gimple *> & to vec<gimple *> &.
8041 (struct split_store): Remove orig_stmts data member, add orig_stores.
8042 (split_store::split_store): Create orig_stores rather than orig_stmts.
8043 (find_constituent_stmts): Renamed to ...
8044 (find_constituent_stores): ... this. Change second argument from
8045 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
8046 to info structures rather than the statements.
8047 (split_group): Rename ALLOW_UNALIGNED argument to
8048 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
8049 it. Adjust find_constituent_stores caller.
8050 (imm_store_chain_info::output_merged_store): Handle rhs_code other
8051 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
8052 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
8053 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
8054 (mem_valid_for_store_merging): New function.
8055 (handled_load): New function.
8056 (pass_store_merging::process_store): New method.
8057 (pass_store_merging::execute): Use process_store method. Adjust
8058 terminate_all_aliasing_chains caller.
8060 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8062 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
8063 Return true for more constants, symbols and label references.
8064 (aarch64_valid_floating_const): Remove unused function.
8066 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8069 * config/aarch64/aarch64.c (aarch64_layout_frame):
8070 Undo forcing of LR at bottom of frame.
8072 2017-11-03 Jeff Law <law@redhat.com>
8075 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
8076 for int_registers_saved.
8078 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
8079 extracted from tree-ssa-dom.c.
8080 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
8081 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
8082 (record_equivalences_from_incoming_edge): Add additional argument
8083 to single_pred_edge_ignoring_loop_edges call.
8084 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
8085 (uncprop_dom_walker::before_dom_children): Add additional argument
8086 to single_pred_edge_ignoring_loop_edges call.
8087 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
8088 single_pred_edge_ignoring_loop_edges rather than open coding.
8089 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
8091 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
8093 * match.pd (-(-A)): Rewrite.
8095 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8097 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
8098 (rs6000_emit_int_cmove): New declaration.
8099 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
8100 (rs6000_emit_sISEL): Delete.
8101 (rs6000_emit_int_cmove): Make non-static.
8102 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
8103 instead of rs6000_emit_sISEL.
8105 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
8107 * asan.c (create_cond_insert_point): Maintain profile.
8108 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
8110 * basic-block.h (struct basic_block_def): Remove frequency.
8111 (EDGE_FREQUENCY): Use to_frequency
8112 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
8114 (find_traces): Update to use to_frequency.
8115 (find_traces_1_round): Likewise; use only IPA counts.
8116 (bb_to_key): Likewise.
8117 (connect_traces): Use IPA counts only.
8118 (copy_bb_p): Update to use to_frequency.
8119 (fix_up_crossing_landing_pad): Likewise.
8120 (sanitize_hot_paths): Likewise.
8121 * bt-load.c (basic_block_freq): Likewise.
8122 * cfg.c (init_flow): Set count_max to uninitialized.
8123 (check_bb_profile): Remove frequencies; check counts.
8124 (dump_bb_info): Do not dump frequencies.
8125 (update_bb_profile_for_threading): Update counts only.
8126 (scale_bbs_frequencies_int): Likewise.
8127 (MAX_SAFE_MULTIPLIER): Remove.
8128 (scale_bbs_frequencies_gcov_type): Update counts only.
8129 (scale_bbs_frequencies_profile_count): Update counts only.
8130 (scale_bbs_frequencies): Update counts only.
8131 * cfg.h (struct control_flow_graph): Add count-max.
8132 (update_bb_profile_for_threading): Update prototype.
8133 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
8134 (find_many_sub_basic_blocks): Likewise.
8135 * cfgcleanup.c (try_forward_edges): Likewise.
8136 (try_crossjump_to_edge): Likewise.
8137 * cfgexpand.c (expand_gimple_cond): Likewise.
8138 (expand_gimple_tailcall): Likewise.
8139 (construct_init_block): Likewise.
8140 (construct_exit_block): Likewise.
8141 * cfghooks.c (verify_flow_info): Check consistency of counts.
8142 (dump_bb_for_graph): Do not dump frequencies.
8143 (split_block_1): Do not update frequencies.
8144 (split_edge): Do not update frequencies.
8145 (make_forwarder_block): Do not update frequencies.
8146 (duplicate_block): Do not update frequencies.
8147 (account_profile_record): Do not update frequencies.
8148 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
8149 for global heuristics.
8150 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
8151 (expected_loop_iterations_unbounded): Use counts only.
8152 * cfgloopmanip.c (scale_loop_profile): Simplify.
8153 (create_empty_loop_on_edge): Simplify
8155 (duplicate_loop_to_header_edge): Simplify
8156 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
8157 (update_br_prob_note): Take care of removing note when profile
8159 (relink_block_chain): Do not dump frequency.
8160 (rtl_account_profile_record): Use to_frequency.
8161 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
8162 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
8163 (cgraph_update_edges_for_call_stmt_node): Likewise.
8164 (cgraph_edge::verify_count_and_frequency): Update.
8165 (cgraph_node::verify_node): Temporarily disable frequency verification.
8166 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
8167 to_cgraph_frequency.
8168 (cgraph_edge::rebuild_edges): Convert to ipa counts.
8169 * cgraphunit.c (init_lowered_empty_function): Do not initialize
8171 (cgraph_node::expand_thunk): Update profile.
8172 * except.c (dw2_build_landing_pads): Do not update frequency.
8173 * final.c (compute_alignments): Use to_frequency.
8174 (dump_basic_block_info): Do not dump frequency.
8175 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
8176 (dump_gimple_bb_header): Do not dump frequency.
8177 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
8179 * gimple-streamer-in.c (input_bb): Do not stream frequency.
8180 * gimple-streamer-out.c (output_bb): Do not stream frequency.
8181 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
8182 (init_before_recovery): Do not update frequency.
8183 (sched_create_recovery_edges): Do not update frequency.
8184 * hsa-gen.c (convert_switch_statements): Do not update frequency.
8185 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
8186 (ipa_cp_c_finalize): Set max_count to uninitialized.
8187 * ipa-fnsummary.c (get_minimal_bb): Use counts.
8188 (param_change_prob): Use counts.
8189 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
8191 * ipa-split.c (consider_split): Use to_frequency.
8192 (split_function): Use to_frequency.
8193 * ira-build.c (loop_compare_func): Likewise.
8194 (mark_loops_for_removal): Likewise.
8195 (mark_all_loops_for_removal): Likewise.
8196 * loop-doloop.c (doloop_modify): Do not update frequency.
8197 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
8199 * lto-streamer-in.c (input_function): Update count_max.
8200 * omp-expand.c (expand_omp_taskreg): Update count_max.
8201 * omp-simd-clone.c (simd_clone_adjust): Update profile.
8202 * predict.c (maybe_hot_frequency_p): Use to_frequency.
8203 (maybe_hot_count_p): Use ipa counts only.
8204 (maybe_hot_bb_p): Simplify.
8205 (maybe_hot_edge_p): Simplify.
8206 (probably_never_executed): Do not take frequency argument.
8207 (probably_never_executed_bb_p): Do not pass frequency.
8208 (probably_never_executed_edge_p): Likewise.
8209 (combine_predictions_for_bb): Check that profile is nonzero.
8210 (propagate_freq): Do not set frequency.
8211 (drop_profile): Simplify.
8212 (counts_to_freqs): Simplify.
8213 (expensive_function_p): Use to_frequency.
8214 (propagate_unlikely_bbs_forward): Simplify.
8215 (determine_unlikely_bbs): Simplify.
8216 (estimate_bb_frequencies): Add hack to silence graphite issues.
8217 (compute_function_frequency): Use ipa counts.
8218 (pass_profile::execute): Update.
8219 (rebuild_frequencies): Use counts only.
8220 (force_edge_cold): Use counts only.
8221 * profile-count.c (profile_count::dump): Dump new count types.
8222 (profile_count::differs_from_p): Check compatiblity.
8223 (profile_count::to_frequency): New function.
8224 (profile_count::to_cgraph_frequency): New function.
8225 * profile-count.h (struct function): Declare.
8226 (enum profile_quality): Add profile_guessed_local and
8227 profile_guessed_global0.
8228 (class profile_proability): Decrease number of bits to 29;
8229 update from_reg_br_prob_note and to_reg_br_prob_note.
8230 (class profile_count: Update comment; decrease number of bits
8231 to 61. Check compatibility.
8232 (profile_count::compatible_p): New private member function.
8233 (profile_count::ipa_p): New member function.
8234 (profile_count::operator<): Handle global zero correctly.
8235 (profile_count::operator>): Handle global zero correctly.
8236 (profile_count::operator<=): Handle global zero correctly.
8237 (profile_count::operator>=): Handle global zero correctly.
8238 (profile_count::nonzero_p): New member function.
8239 (profile_count::force_nonzero): New member function.
8240 (profile_count::max): New member function.
8241 (profile_count::apply_scale): Handle IPA scalling.
8242 (profile_count::guessed_local): New member function.
8243 (profile_count::global0): New member function.
8244 (profile_count::ipa): New member function.
8245 (profile_count::to_frequency): Declare.
8246 (profile_count::to_cgraph_frequency): Declare.
8247 * profile.c (OVERLAP_BASE): Delete.
8248 (compute_frequency_overlap): Delete.
8249 (compute_branch_probabilities): Do not use compute_frequency_overlap.
8250 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
8251 * sched-ebb.c (rank): Use counts only.
8252 * shrink-wrap.c (handle_simple_exit): Use counts only.
8253 (try_shrink_wrapping): Use counts only.
8254 (place_prologue_for_one_component): Use counts only.
8255 * tracer.c (find_best_predecessor): Use to_frequency.
8256 (find_trace): Use to_frequency.
8257 (tail_duplicate): Use to_frequency.
8258 * trans-mem.c (expand_transaction): Do not update frequency.
8259 * tree-call-cdce.c: Do not update frequency.
8260 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
8261 (gimple_merge_blocks): Likewise.
8262 (gimple_split_edge): Likewise.
8263 (gimple_duplicate_sese_region): Likewise.
8264 (gimple_duplicate_sese_tail): Likewise.
8265 (move_sese_region_to_fn): Likewise.
8266 (gimple_account_profile_record): Likewise.
8267 (insert_cond_bb): Likewise.
8268 * tree-complex.c (expand_complex_div_wide): Likewise.
8269 * tree-eh.c (lower_resx): Update profile.
8270 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
8272 (initialize_cfun): Do not initialize frequencies
8273 (freqs_to_counts): Delete.
8274 (copy_cfg_body): Ignore count parameter.
8275 (copy_body): Update.
8276 (expand_call_inline): Update count_max.
8277 (optimize_inline_calls): Update count_max.
8278 (tree_function_versioning): Update count_max.
8279 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
8280 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
8282 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
8283 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
8284 (try_peel_loop): Likewise.
8285 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
8287 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
8288 (tree_transform_and_unroll_loop): Do not use frequencies
8289 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
8290 Use reliable prediction only.
8291 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
8292 * tree-ssa-sink.c (select_best_block): Use to_frequency.
8293 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
8294 probability scaling.
8295 * tree-ssa-threadupdate.c (create_block_for_threading): Do
8296 not update frequency
8297 (any_remaining_duplicated_blocks): Likewise.
8298 (update_profile): Likewise.
8299 (estimated_freqs_path): Delete.
8300 (freqs_to_counts_path): Delete.
8301 (clear_counts_path): Delete.
8302 (ssa_fix_duplicate_block_edges): Likewise.
8303 (duplicate_thread_path): Likewise.
8304 * tree-switch-conversion.c (gen_inbound_check): Use counts.
8305 * tree-tailcall.c (decrease_profile): Do not update frequency.
8306 (eliminate_tail_call): Likewise.
8307 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
8308 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
8309 (optimize_mask_stores): Likewise.
8310 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8311 * ubsan.c (ubsan_expand_null_ifn): Update profile.
8312 (ubsan_expand_ptr_ifn): Update profile.
8313 * value-prof.c (gimple_ic): Simplify.
8314 * value-prof.h (gimple_ic): Update prototype.
8315 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
8316 * ipa-inline.c (compute_uninlined_call_time): Be sure that
8318 (want_inline_self_recursive_call_p): Likewise.
8319 (resolve_noninline_speculation): Only cummulate defined counts.
8320 (inline_small_functions): Use nonzero_p.
8321 (ipa_inline): Do not access freed node.
8323 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8325 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8326 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
8328 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
8330 * config/riscv/riscv.c (riscv_legitimize_move): Handle
8331 non-legitimate address.
8333 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8335 * config/rs6000/rs6000.md (*lt0_disi): Delete.
8336 (*lt0_<mode>di, *lt0_<mode>si): New.
8338 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8340 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
8341 TARGET_PAIRED_FLOAT.
8343 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
8344 Jim Wilson <jim.wilson@linaro.org>
8346 * config/aarch64/aarch64-cores.def (saphira): New CPU.
8347 * config/aarch64/aarch64-tune.md: Regenerated.
8348 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
8349 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
8351 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
8353 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
8354 (arc_expand_prologue): Restore blink for millicode.
8355 * config/arc/linux.h (LINK_EH_SPEC): Defined.
8357 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
8360 * config/i386/i386.c (ix86_vector_duplicate_value): Use
8361 gen_vec_duplicate after forcing the scalar into a register.
8363 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
8365 * combine (try_combine): Print the insns input to try_combine to the
8368 2017-11-02 Steve Ellcey <sellcey@cavium.com>
8371 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
8372 Remove second argument from aarch64_process_target_attr call.
8373 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
8375 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
8377 (aarch64_handle_attr_arch): Remove second argument.
8378 (aarch64_handle_attr_cpu): Ditto.
8379 (aarch64_handle_attr_tune): Ditto.
8380 (aarch64_handle_attr_isa_flags): Ditto.
8381 (aarch64_process_one_target_attr): Ditto.
8382 (aarch64_process_target_attr): Ditto.
8383 (aarch64_option_valid_attribute_p): Remove second argument.
8384 on aarch64_process_target_attr call.
8386 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8388 * diagnostic.c: Include "selftest-diagnostic.h".
8389 (selftest::assert_location_text): New function.
8390 (selftest::test_diagnostic_get_location_text): New function.
8391 (selftest::diagnostic_c_tests): Call it.
8393 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8395 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
8396 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
8397 (class selftest::test_diagnostic_context): Move to...
8398 * selftest-diagnostic.c: New file.
8399 * selftest-diagnostic.h: New file.
8401 2017-11-02 James Bowman <james.bowman@ftdichip.com>
8403 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
8404 offset range for FT32B.
8405 * config/ft32/ft32.h: option "mcompress" enables relaxation.
8406 * config/ft32/ft32.md: Add TARGET_NOPM.
8407 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
8408 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
8410 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
8412 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
8414 2017-11-02 Jeff Law <law@redhat.com>
8416 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
8417 virtual keyword on FINAL OVERRIDE members.
8419 * tree-ssa-propagate.h (ssa_propagation_engine): Group
8420 virtuals together. Add virtual destructor.
8421 (substitute_and_fold_engine): Similarly.
8423 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
8425 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
8427 2017-11-02 Richard Biener <rguenther@suse.de>
8429 PR tree-optimization/82795
8430 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
8432 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8434 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
8436 (gcc_GAS_CHECK_FEATURE): Remove.
8437 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
8438 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
8439 * configure: Regenerate.
8441 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
8443 * config/arc/arc.c (hwloop_optimize): Account for empty
8446 2017-11-02 Richard Biener <rguenther@suse.de>
8449 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
8450 Truncate ARRAY_REF index and element size.
8452 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8454 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
8456 2017-11-01 Jeff Law <law@redhat.com>
8458 * tree-ssa-ccp.c (ccp_folder): New class derived from
8459 substitute_and_fold_engine.
8460 (ccp_folder::get_value): New member function.
8461 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
8462 (ccp_fold_stmt): Remove prototype.
8463 (ccp_finalize): Call substitute_and_fold from the ccp_class.
8464 * tree-ssa-copy.c (copy_folder): New class derived from
8465 substitute_and_fold_engine.
8466 (copy_folder::get_value): Renamed from get_value.
8467 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
8468 * tree-vrp.c (vrp_folder): New class derived from
8469 substitute_and_fold_engine.
8470 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
8471 (vrp_folder::get_value): New member function.
8472 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
8473 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
8474 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
8475 provide a class interface to folder/substitute routines.
8476 (ssa_prop_fold_stmt_fn): Remove typedef.
8477 (ssa_prop_get_value_fn): Likewise.
8478 (subsitute_and_fold): Remove prototype.
8479 (replace_uses_in): Likewise.
8480 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
8481 Renamed from replace_uses_in. Call the virtual member function
8482 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
8483 (substitute_and_fold_dom_walker): Remove initialization of
8484 data member entries for calbacks. Add substitute_and_fold_engine
8485 member and initialize it.
8486 (substitute_and_fold_dom_walker::before_dom_children0: Use the
8487 member functions for get_value, replace_phi_args_in c
8488 replace_uses_in, and fold_stmt calls.
8489 (substitute_and_fold_engine::substitute_and_fold): Renamed from
8490 substitute_and_fold. Remove assert. Update ctor call.
8492 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
8493 (ssa_prop_visit_phi_fn): Likewise.
8494 (class ssa_propagation_engine): New class to provide an interface
8496 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
8498 (ssa_prop_visit_phi): Likewise.
8499 (ssa_propagation_engine::simulate_stmt): Moved into class.
8500 Call visit_phi/visit_stmt from the class rather than via
8501 file scoped static variables.
8502 (ssa_propagation_engine::simulate_block): Moved into class.
8503 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
8504 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
8505 set file scoped statics for the visit_stmt/visit_phi callbacks.
8506 * tree-complex.c (complex_propagate): New class derived from
8507 ssa_propagation_engine.
8508 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
8509 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
8510 (tree_lower_complex): Call ssa_propagate via the complex_propagate
8512 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
8513 ssa_propagation_engine.
8514 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
8515 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
8516 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
8517 * tree-ssa-copy.c (copy_prop): New class derived from
8518 ssa_propagation_engine.
8519 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
8520 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
8521 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
8522 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
8523 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
8524 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
8525 (execute_vrp): Call ssa_propagate from the vrp_prop class.
8527 2017-11-01 Jakub Jelinek <jakub@redhat.com>
8529 PR rtl-optimization/82778
8530 PR rtl-optimization/82597
8531 * compare-elim.c (struct comparison): Add in_a_setter field.
8532 (find_comparison_dom_walker::before_dom_children): Remove killed
8533 bitmap and df_simulate_find_defs call, instead walk the defs.
8534 Compute last_setter and initialize in_a_setter. Merge definitions
8535 with first initialization for a few variables.
8536 (try_validate_parallel): Use insn_invalid_p instead of
8537 recog_memoized. Return insn rather than just the pattern.
8538 (try_merge_compare): Fix up comment. Don't uselessly test if
8539 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
8541 (execute_compare_elim_after_reload): Remove df_chain_add_problem
8544 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8545 Alan Hayward <alan.hayward@arm.com>
8546 David Sherwood <david.sherwood@arm.com>
8548 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
8549 aarch64_hard_regno_nregs to get the number of registers
8552 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8553 Alan Hayward <alan.hayward@arm.com>
8554 David Sherwood <david.sherwood@arm.com>
8556 * config/aarch64/constraints.md (Upl): Rename to...
8558 * config/aarch64/aarch64.md
8559 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
8562 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8563 Alan Hayward <alan.hayward@arm.com>
8564 David Sherwood <david.sherwood@arm.com>
8566 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
8567 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
8570 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8571 Alan Hayward <alan.hayward@arm.com>
8572 David Sherwood <david.sherwood@arm.com>
8574 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
8575 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
8576 (aarch64_evpc_dup): Generate rtl direcly, rather than using
8578 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
8579 of a single element.
8580 * config/aarch64/iterators.md: Add a comment above the permute
8581 unspecs to say that they are generated directly by
8582 aarch64_expand_vec_perm_const.
8583 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
8585 2017-11-01 Nathan Sidwell <nathan@acm.org>
8587 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
8589 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8591 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
8592 and medany code models, and describe what they do.
8594 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8596 Revert accidental duplicate:
8598 * combine.c (can_change_dest_mode): Reject changes in
8599 REGMODE_NATURAL_SIZE.
8601 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
8603 PR rtl-optimization/64682
8604 PR rtl-optimization/69567
8605 PR rtl-optimization/69737
8606 PR rtl-optimization/82683
8607 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
8608 register mentioned in the note, drop the note, unless it came from I3,
8609 in which case it should go to I3 again.
8611 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8613 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
8614 and return false if not.
8615 (clear_bytes_written_by, live_bytes_read): Update accordingly.
8617 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8619 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
8620 range is known to be empty.
8622 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8623 Alan Hayward <alan.hayward@arm.com>
8624 David Sherwood <david.sherwood@arm.com>
8626 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
8627 and CONST_VECTOR_NUNITS instead of computing the number of units from
8628 the byte sizes of the vector and element.
8629 (simplify_binary_operation_1): Likewise.
8630 (simplify_const_binary_operation): Likewise.
8631 (simplify_ternary_operation): Likewise.
8633 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8634 Alan Hayward <alan.hayward@arm.com>
8635 David Sherwood <david.sherwood@arm.com>
8637 * var-tracking.c (INT_MEM_OFFSET): Replace with...
8638 (int_mem_offset): ...this new function.
8639 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
8640 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
8641 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
8644 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8645 Alan Hayward <alan.hayward@arm.com>
8646 David Sherwood <david.sherwood@arm.com>
8648 * lower-subreg.c (interesting_mode_p): New function.
8649 (compute_costs, find_decomposable_subregs, decompose_register)
8650 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
8651 (resolve_clobber, dump_choices): Use it.
8653 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8654 Alan Hayward <alan.hayward@arm.com>
8655 David Sherwood <david.sherwood@arm.com>
8657 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
8659 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8660 Alan Hayward <alan.hayward@arm.com>
8661 David Sherwood <david.sherwood@arm.com>
8663 * alias.c (find_base_value, find_base_term): Only process integer
8664 truncations. Check the precision rather than the size.
8666 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8667 Alan Hayward <alan.hayward@arm.com>
8668 David Sherwood <david.sherwood@arm.com>
8670 * machmode.h (is_narrower_int_mode): New function
8671 * optabs.c (expand_float, expand_fix): Use it.
8672 * dwarf2out.c (rotate_loc_descriptor): Likewise.
8674 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8675 Alan Hayward <alan.hayward@arm.com>
8676 David Sherwood <david.sherwood@arm.com>
8678 * rtl.h (narrower_subreg_mode): New function.
8679 * ira-color.c (update_costs_from_allocno): Use it.
8681 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8682 Alan Hayward <alan.hayward@arm.com>
8683 David Sherwood <david.sherwood@arm.com>
8685 * optabs-query.h (convert_optab_p): New function, split out from...
8686 (convert_optab_handler): ...here.
8687 (widening_optab_handler): Delete.
8688 (find_widening_optab_handler): Remove permit_non_widening parameter.
8689 (find_widening_optab_handler_and_mode): Likewise. Provide an
8690 override that operates on mode class wrappers.
8691 * optabs-query.c (widening_optab_handler): Delete.
8692 (find_widening_optab_handler_and_mode): Remove permit_non_widening
8693 parameter. Assert that the two modes are the same class and that
8694 the "from" mode is narrower than the "to" mode. Use
8695 convert_optab_handler instead of widening_optab_handler.
8696 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
8697 instead of widening_optab_handler.
8698 * expr.c (expand_expr_real_2): Update calls to
8699 find_widening_optab_handler.
8700 * optabs.c (expand_widen_pattern_expr): Likewise.
8701 (expand_binop_directly): Take the insn_code as a parameter.
8702 (expand_binop): Only call find_widening_optab_handler for
8703 conversion optabs; use optab_handler otherwise. Update calls
8704 to find_widening_optab_handler and expand_binop_directly.
8705 Use convert_optab_handler instead of widening_optab_handler.
8706 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
8707 find_widening_optab_handler and use scalar_mode rather than
8709 (convert_plusminus_to_widen): Likewise.
8711 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8712 Alan Hayward <alan.hayward@arm.com>
8713 David Sherwood <david.sherwood@arm.com>
8715 * machmode.h (fixed_size_mode): New class.
8716 * rtl.h (get_pool_mode): Return fixed_size_mode.
8717 * gengtype.c (main): Add fixed_size_mode.
8718 * target.def (get_raw_result_mode): Return a fixed_size_mode.
8719 (get_raw_arg_mode): Likewise.
8720 * doc/tm.texi: Regenerate.
8721 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
8722 * targhooks.c (default_get_reg_raw_mode): Likewise.
8723 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
8724 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
8725 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
8726 (msp430_get_raw_result_mode): Likewise.
8727 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
8728 * dbxout.c (dbxout_parms): Require fixed-size modes.
8729 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
8730 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8731 * omp-low.c (lower_oacc_reductions): Likewise.
8732 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
8733 (simplify_subreg): Update accordingly.
8734 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
8735 (force_const_mem): Update accordingly. Return NULL_RTX for modes
8736 that aren't fixed-size.
8737 (get_pool_mode): Return a fixed_size_mode.
8738 (output_constant_pool_2): Take a fixed_size_mode.
8740 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8741 Alan Hayward <alan.hayward@arm.com>
8742 David Sherwood <david.sherwood@arm.com>
8744 * doc/rtl.texi (vec_series): Document.
8745 (const): Say that the operand can be a vec_series.
8746 * rtl.def (VEC_SERIES): New rtx code.
8747 * rtl.h (const_vec_series_p_1): Declare.
8748 (const_vec_series_p): New function.
8749 * emit-rtl.h (gen_const_vec_series): Declare.
8750 (gen_vec_series): Likewise.
8751 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
8752 (gen_vec_series): Likewise.
8753 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
8754 * simplify-rtx.c (simplify_unary_operation): Handle negations
8756 (simplify_binary_operation_series): New function.
8757 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
8758 (test_vector_ops_series): New function.
8759 (test_vector_ops): Call it.
8760 * config/powerpcspe/altivec.md (altivec_lvsl): Use
8761 gen_const_vec_series.
8762 (altivec_lvsr): Likewise.
8763 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
8765 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8766 Alan Hayward <alan.hayward@arm.com>
8767 David Sherwood <david.sherwood@arm.com>
8769 * doc/rtl.texi (const): Update description of address constants.
8770 Say that vector constants are allowed too.
8771 * common.md (E, F): Use CONSTANT_P instead of checking for
8773 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
8774 checking for CONST_VECTOR.
8775 * expmed.c (make_tree): Use build_vector_from_val for a CONST
8777 * expr.c (expand_expr_real_2): Check for vector modes instead
8778 of checking for CONST_VECTOR.
8779 * rtl.h (const_vec_p): New function.
8780 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
8781 (unwrap_const_vec_duplicate): Handle them here too.
8783 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8784 David Malcolm <dmalcolm@redhat.com>
8785 Alan Hayward <alan.hayward@arm.com>
8786 David Sherwood <david.sherwood@arm.com>
8788 * rtl.h (vec_duplicate_p): New function.
8789 * selftest-rtl.c (assert_rtx_eq_at): New function.
8790 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
8791 (assert_rtx_eq_at): Declare.
8792 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
8793 * selftest-run-tests.c (selftest::run_tests): Call it.
8794 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
8795 (simplify_unary_operation_1): Recursively handle vector duplicates.
8796 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
8798 (simplify_subreg): Handle subregs of vector duplicates.
8799 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
8800 (selftest::simplify_rtx_c_tests): New functions.
8802 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8803 Alan Hayward <alan.hayward@arm.com>
8804 David Sherwood <david.sherwood@arm.com>
8806 * emit-rtl.h (gen_const_vec_duplicate): Declare.
8807 (gen_vec_duplicate): Likewise.
8808 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
8810 (gen_const_vector): ...here.
8811 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
8812 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
8813 whose elements are all equal.
8814 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
8815 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
8816 (simplify_relational_operation): Likewise.
8817 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
8819 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
8820 (aarch64_expand_vector_init): Likewise.
8821 * config/arm/arm.c (neon_vdup_constant): Likewise.
8822 (neon_expand_vector_init): Likewise.
8823 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
8824 (arm_block_set_unaligned_vect): Likewise.
8825 (arm_block_set_aligned_vect): Likewise.
8826 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
8827 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
8828 (expand_vec_perm_even_odd_pack): Likewise.
8829 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
8830 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
8831 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
8832 gen_const_vec_duplicate.
8833 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
8834 * config/mips/mips.c (mips_gen_const_int_vector): Use
8835 gen_const_vec_duplicate.
8836 (mips_expand_vector_init): Use CONST0_RTX.
8837 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
8838 (define_split): Use gen_const_vec_duplicate.
8839 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
8840 (define_split): Use gen_const_vec_duplicate.
8841 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
8842 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
8843 * config/spu/spu.c (spu_const): Likewise.
8845 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8846 Alan Hayward <alan.hayward@arm.com>
8847 David Sherwood <david.sherwood@arm.com>
8849 * combine.c (can_change_dest_mode): Reject changes in
8850 REGMODE_NATURAL_SIZE.
8852 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
8854 * configure.ac (--enable-libssp): New.
8855 (gcc_cv_libc_provides_ssp): Check for explicit setting before
8856 trying to determine target-specific default. Adjust indentation.
8857 * configure: Regenerated.
8858 * doc/install.texi (Configuration): Expand --disable-libssp
8861 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
8863 config/i386/i386.c (ix86_expand_epilogue): Correct stack
8866 2017-10-31 Martin Jambor <mjambor@suse.cz>
8869 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
8871 2017-10-31 David Malcolm <dmalcolm@redhat.com>
8873 * auto-profile.c (autofdo_source_profile::read): Use
8874 UNKNOWN_LOCATION rather than 0.
8875 * diagnostic-core.h (warning_at_rich_loc): Rename to...
8876 (warning_at): ...this overload.
8877 (warning_at_rich_loc_n): Rename to...
8878 (warning_n): ...this overload.
8879 (error_at_rich_loc): Rename to...
8880 (error_at): ...this overload.
8881 (pedwarn_at_rich_loc): Rename to...
8882 (pedwarn): ...this overload.
8883 (permerror_at_rich_loc): Rename to...
8884 (permerror): ...this overload.
8885 (inform_at_rich_loc): Rename to...
8886 (inform): ...this overload.
8887 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
8888 (diagnostic_n_impl_richloc): Rename to...
8889 (diagnostic_n_impl): ...this rich_location *-based decl.
8890 (inform_at_rich_loc): Rename to...
8891 (inform): ...this, and add an assertion.
8892 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
8893 (warning_at_rich_loc): Rename to...
8894 (warning_at): ...this, and add an assertion.
8895 (warning_at_rich_loc_n): Rename to...
8896 (warning_n): ...this, and add an assertion.
8897 (warning_n): Update location_t-based implementation for removal of
8898 location_t-based diagnostic_n_impl.
8899 (pedwarn_at_rich_loc): Rename to...
8900 (pedwarn): ...this, and add an assertion.
8901 (permerror_at_rich_loc): Rename to...
8902 (permerror): ...this, and add an assertion.
8903 (error_n): Update for removal of location_t-based diagnostic_n_impl.
8904 (error_at_rich_loc): Rename to...
8905 (error_at): ...this, and add an assertion.
8906 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
8907 (driver::do_spec_on_infiles): Likewise.
8908 * substring-locations.c (format_warning_va): Update for renaming
8909 of inform_at_rich_loc.
8911 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
8913 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
8914 _Float<N>X built-in functions so that the variant without the
8915 "__builtin_" prefix is only enabled for the GNU C and Objective C
8916 languages when they are in non-strict ANSI/ISO mode.
8917 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
8918 * target.def (floatn_builtin_p): Add a target hook to control
8919 whether _Float<N> and _Float<N>X built-in functions without the
8920 "__builtin_" prefix are enabled, and return true for C and
8921 Objective C in the default hook. Include langhooks.h in
8923 * targhooks.h (default_floatn_builtin_p): Likewise.
8924 * targhooks.c (default_floatn_builtin_p): Likewise.
8925 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
8926 floatn_builtin_p target hook.
8927 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
8929 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
8930 Eric Botcazou <ebotcazou@adacore.com>
8932 PR rtl-optimization/81803
8933 * lra-constraints.c (curr_insn_transform): Also reload the whole
8934 register for a strict subreg no wider than a word if this is for
8935 a WORD_REGISTER_OPERATIONS target.
8937 2017-10-31 Jason Merrill <jason@redhat.com>
8939 * gdbinit.in: Skip over inlines from timevar.h.
8941 2017-10-31 Martin Liska <mliska@suse.cz>
8943 * doc/gcov.texi: Document new option.
8944 * gcov.c (print_usage): Likewise print it.
8945 (process_args): Support the argument.
8946 (format_count): New function.
8947 (format_gcov): Use the function.
8949 2017-10-31 Martin Liska <mliska@suse.cz>
8951 * gcov.c (struct name_map): do not use typedef.
8952 Define operator== and operator<.
8953 (name_search): Remove.
8954 (name_sort): Remove.
8955 (main): Do not allocate names.
8956 (process_file): Add vertical space.
8957 (generate_results): Use std::find.
8958 (release_structures): Do not release memory.
8959 (find_source): Use std::find.
8961 2017-10-31 Martin Liska <mliska@suse.cz>
8963 * gcov.c (struct line_info): Remove it's typedef.
8964 (line_info::line_info): Add proper ctor.
8965 (line_info::has_block): Do not use a typedef.
8966 (struct source_info): Do not use typedef.
8967 (circuit): Likewise.
8968 (get_cycles_count): Likewise.
8969 (output_intermediate_file): Iterate via vector iterator.
8970 (add_line_counts): Use std::vector methods.
8971 (accumulate_line_counts): Likewise.
8972 (output_lines): Likewise.
8974 2017-10-31 Martin Liska <mliska@suse.cz>
8976 * gcov.c (struct source_info): Remove typedef.
8977 (source_info::source_info): Add proper ctor.
8978 (accumulate_line_counts): Use struct, not it's typedef.
8979 (output_gcov_file): Likewise.
8980 (output_lines): Likewise.
8981 (main): Do not allocate an array.
8982 (output_intermediate_file): Use size of vector container.
8983 (process_file): Resize the vector.
8984 (generate_results): Do not preallocate, use newly added vector
8986 (release_structures): Do not release sources.
8987 (find_source): Use vector methods.
8988 (add_line_counts): Do not use typedef.
8990 2017-10-31 Martin Liska <mliska@suse.cz>
8992 * doc/gcov.texi: Document that.
8993 * gcov.c (add_line_counts): Mark lines with a non-executed
8995 (output_line_beginning): Handle such lines.
8996 (output_lines): Pass new argument.
8997 (output_intermediate_file): Print it in intermediate format.
8999 2017-10-31 Martin Liska <mliska@suse.cz>
9001 * color-macros.h: New file.
9002 * diagnostic-color.c: Factor out color related to macros to
9004 * doc/gcov.texi: Document -k option.
9005 * gcov.c (INCLUDE_STRING): Include string.h.
9006 (print_usage): Add -k option.
9007 (process_args): Parse it.
9008 (pad_count_string): New function.
9009 (output_line_beginning): Likewise.
9010 (DEFAULT_LINE_START): New macro.
9011 (output_lines): Support color output.
9013 2017-10-31 Martin Liska <mliska@suse.cz>
9015 PR gcov-profile/82633
9016 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
9017 their interaction with GCOV infrastructure.
9018 * configure.ac: Add -fkeep-{inline,static}-functions to
9020 * configure: Regenerate.
9022 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
9025 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
9027 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
9030 * config/rs6000/rs6000.md (allocate_stack): Force update interval
9031 into a register if it does not fit into an immediate offset field.
9033 2017-10-31 Olivier Hainque <hainque@adacore.com>
9035 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
9037 2017-10-31 Julia Koval <julia.koval@intel.com>
9039 * config.gcc: Add gfniintrin.h.
9040 * config/i386/gfniintrin.h: New.
9041 * config/i386/i386-builtin-types.def
9042 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
9043 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
9044 __builtin_ia32_vgf2p8affineinvqb_v32qi,
9045 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
9046 __builtin_ia32_vgf2p8affineinvqb_v16qi,
9047 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
9048 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
9049 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
9050 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
9051 V64QI_FTYPE_V64QI_V64QI_INT): New types.
9052 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
9053 * config/i386/immintrin.h: Include gfniintrin.h.
9054 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
9056 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
9058 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
9060 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9062 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
9063 (arm_ashldi3_1bit): Remove pattern.
9064 (ashrdi3): Remove shift by 1 expansion.
9065 (arm_ashrdi3_1bit): Remove pattern.
9066 (lshrdi3): Remove shift by 1 expansion.
9067 (arm_lshrdi3_1bit): Remove pattern.
9068 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
9069 cost of ashldi3 by 1.
9070 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
9071 (<shift>di3_neon): Likewise.
9073 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9075 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
9076 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
9077 and (*aarch64_simd_mov<VQ:mode>).
9078 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
9079 pattern alternative.
9080 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
9081 attributes to match pattern alternative.
9083 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
9085 * config.gcc (powerpc*-*-*): Add emmintrin.h.
9086 * config/rs6000/emmintrin.h: New file.
9087 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
9089 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9091 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
9092 * (movdi_vfp_cortexa8): Remove pattern.
9094 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9096 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
9098 (Specific, alpha*-dec-osf5.1): Remove.
9099 (Specific, mips-sgi-irix5): Remove.
9100 (Specific, mips-sgi-irix6): Remove.
9102 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9105 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
9106 arguments to clear_bit_region_be.
9108 2017-10-30 Jim Wilson <wilson@tuliptree.org>
9110 * gimplify.c: Include memmodel.h.
9112 2017-10-30 Martin Jambor <mjambor@suse.cz>
9114 * omp-grid.c (grid_attempt_target_gridification): Also insert a
9115 condition whether loop should be executed at all.
9117 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
9119 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
9120 gimple folding of vec_madd() intrinsics.
9121 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
9122 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
9123 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
9125 2017-10-30 Richard Biener <rguenther@suse.de>
9127 PR tree-optimization/82762
9129 2017-10-23 Richard Biener <rguenther@suse.de>
9131 PR tree-optimization/82129
9133 2017-08-01 Richard Biener <rguenther@suse.de>
9135 PR tree-optimization/81181
9136 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9137 (compute_antic): ... end of iteration here.
9139 2017-10-30 Joseph Myers <joseph@codesourcery.com>
9141 * doc/invoke.texi (C Dialect Options): Document -std=c17,
9142 -std=iso9899:2017 and -std=gnu17.
9143 * doc/standards.texi (C Language): Document C17 support.
9144 * doc/cpp.texi (Overview): Mention -std=c17.
9145 (Standard Predefined Macros): Document C11 and C17 values of
9146 __STDC_VERSION__. Do not refer to C99 support as incomplete.
9147 * doc/extend.texi (Inline): Do not list individual options for
9148 standards newer than C99.
9149 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
9151 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
9154 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
9156 * asan.c (asan_finish_file): Align asan globals array by shadow
9159 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9162 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
9163 (struct store_immediate_info): Add bitregion_start and bitregion_end
9165 (store_immediate_info::store_immediate_info): Add brs and bre
9166 arguments and initialize bitregion_{start,end} from those.
9167 (struct merged_store_group): Add bitregion_start, bitregion_end,
9168 align_base and mask fields. Drop unnecessary struct keyword from
9169 struct store_immediate_info. Add do_merge method.
9170 (clear_bit_region_be): Use memset instead of loop storing zeros.
9171 (merged_store_group::do_merge): New method.
9172 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
9173 stores as long as the surrounding bitregions have no gaps.
9174 (merged_store_group::merge_overlapping): Use do_merge.
9175 (merged_store_group::apply_stores): Test that bitregion_{start,end}
9176 is byte aligned, rather than requiring that start and width are
9177 byte aligned. Drop unnecessary struct keyword from
9178 struct store_immediate_info. Allocate and populate also mask array.
9179 Make start of the arrays relative to bitregion_start rather than
9180 start and size them according to bitregion_{end,start} difference.
9181 (struct imm_store_chain_info): Drop unnecessary struct keyword from
9182 struct store_immediate_info.
9183 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
9184 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
9185 struct keyword from struct store_immediate_info.
9186 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
9187 between stores as long as the surrounding bitregions have no gaps.
9189 (struct split_store): Add orig non-static data member.
9190 (split_store::split_store): Initialize orig to false.
9191 (find_constituent_stmts): Return store_immediate_info *, non-NULL
9192 if there is exactly a single original stmt. Change stmts argument
9193 to pointer from reference, if NULL, don't push anything to it. Add
9194 first argument, use it to optimize skipping over orig stmts that
9195 are known to be before bitpos already. Simplify.
9196 (split_group): Return unsigned int count how many stores are or
9197 would be needed rather than a bool. Add allow_unaligned argument.
9198 Change split_stores argument from reference to pointer, if NULL,
9199 only do a dry run computing how many stores would be produced.
9200 Rewritten algorithm to use both alignment and misalign if
9201 !allow_unaligned and handle bitfield stores with gaps.
9202 (imm_store_chain_info::output_merged_store): Set start_byte_pos
9203 from bitregion_start instead of start. Compute allow_unaligned
9204 here, if true, do 2 split_group dry runs to compute which one
9205 produces fewer stores and prefer aligned if equal. Punt if
9206 new count is bigger or equal than original before emitting any
9207 statements, rather than during that. Remove no longer needed
9208 new_ssa_names tracking. Replace num_stmts with
9209 split_stores.length (). Use 32-bit stack allocated entries
9210 in split_stores auto_vec. Try to reuse original store lhs/rhs1
9211 if possible. Handle bitfields with gaps.
9212 (pass_store_merging::execute): Ignore bitsize == 0 stores.
9213 Compute bitregion_{start,end} for the stores and construct
9214 store_immediate_info with that. Formatting fixes.
9216 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
9219 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
9220 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
9222 2017-10-29 Jim Wilson <wilson@tuliptree.org>
9224 * gimplify.c: Include tm_p.h.
9226 * common.opt (gcoff): Re-add as ignored option.
9227 (gcoff1, gcoff2, gcoff3): Likewise.
9229 * Makefile.in (OBJS): Delete sdbout.o.
9230 (GTFILES): Delete $(srcdir)/sdbout.c.
9231 * debug.h: Delete sdb_debug_hooks.
9232 * final.c: Delete sdbout.h include.
9233 (final_scan_insn): Delete SDB_DEBUG check.
9234 (rest_of_clean_state): Likewise.
9235 * output.h: Delete sdb_begin_function_line.
9238 * toplev.c: Delete sdbout.h include.
9239 (process_options): Delete SDB_DEBUG check.
9240 * tree-core.h (tree_type_common): Delete pointer field of
9242 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
9243 TYPE_SYMTAB_POINTER.
9244 * tree.h (TYPE_SYMTAB_POINTER): Delete.
9245 (TYPE_SYMTAB_IS_POINTER): Delete.
9246 (TYPE_SYMTAB_IS_DIE): Renumber.
9247 * xcoffout.c: Refer to former sdbout.c file.
9248 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
9250 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
9251 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
9252 (Debugging Options): Delete -gcoff.
9253 (-gstabs): Delete SDB reference.
9255 (-gcoff@var{level}): Delete.
9256 * doc/passes.texi (Debugging information output): Delete SDB and
9257 sdbout.c references.
9258 * doc/tm.texi: Regenerate.
9259 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
9260 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
9262 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
9263 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
9264 (SDB_DEBUGGING_INFO): Delete.
9265 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
9266 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
9267 * target.def (output_source_filename): Delete COFF reference.
9269 * common.opt (gcoff): Delete.
9270 (gxcoff+): Update Negative chain.
9271 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
9273 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
9274 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
9275 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
9277 (expand_function_start): Change sdb reference to past tense.
9278 (expand_function_end): Change sdb reference to past tense.
9279 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
9280 * opts.c (debug_type_names): Delete coff entry.
9281 (common_handle_option): Delete OPT_gcoff case.
9282 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
9284 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
9285 * config/cris/cris.h: Delete SDB reference in comment.
9286 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
9287 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
9288 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
9289 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
9291 (ix86_expand_prologue): Likewise.
9292 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
9293 * config/ia64/ia64.h: Likewise.
9294 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
9295 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
9297 * config/mmix/mmix.h: Likewise.
9298 * config/nds32/nds32.c: Likewise.
9299 * config/stormy/storym16.h: Likewise.
9300 * config/visium/visium.h: Likewise.
9301 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
9303 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
9305 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
9306 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
9307 FRAME_POINTER_REGNUM point at high end of local var area.
9309 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
9311 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
9312 Move comment around. Do not reset best_edge for a copiable
9313 destination if the copy would cause a partition change.
9314 (better_edge_p): Remove redundant check.
9316 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9318 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
9320 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
9322 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
9323 for math functions that have _Float<N> and _Float<N>X variants.
9324 (mathfn_built_in_2): Add support for math functions that have
9325 _Float<N> and _Float<N>X variants.
9326 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
9327 (expand_builtin_mathfn_ternary): Add support for fma with
9328 _Float<N> and _Float<N>X variants.
9329 (expand_builtin): Likewise.
9330 (fold_builtin_3): Likewise.
9331 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
9332 create math function _Float<N> and _Float<N>X variants as external
9334 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
9335 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
9336 the __builtin_ prefix and if not strict ansi, without the prefix.
9337 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
9338 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
9339 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
9340 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
9341 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
9342 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
9343 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
9344 function signatures for fma _Float<N> and _Float<N>X variants.
9345 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
9346 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
9347 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
9348 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
9349 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
9350 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
9351 * gencfn-macros.c (print_case_cfn): Add support for math functions
9352 that have _Float<N> and _Float<N>X variants.
9353 (print_define_operator_list): Likewise.
9354 (fltfn_suffixes): Likewise.
9356 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
9357 for math functions that have _Float<N> and _Float<N>X variants.
9358 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
9359 and _Float<N>X variants.
9360 (COPYSIGN): Likewise.
9363 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
9364 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
9366 (integer_valued_read_call_p): Likewise.
9367 * fold-const-call.c (fold_const_call_ss): Likewise.
9368 (fold_const_call_sss): Add support for copysign, fmin, and fmax
9369 _Float<N> and _Float<N>X variants.
9370 (fold_const_call_ssss): Add support for fma _Float<N> and
9371 _Float<N>X variants.
9372 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
9373 support for copysign and fma _Float<N> and _Float<N>X variants.
9374 (backprop::process_builtin_call_use): Likewise.
9375 * tree-call-cdce.c (can_test_argument_range); Add support for
9376 sqrt _Float<N> and _Float<N>X variants.
9377 (edom_only_function): Likewise.
9378 (get_no_error_domain): Likewise.
9379 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
9380 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
9381 copysign _Float<N> and _Float<N>X variants.
9382 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
9383 handled by machine independent code.
9384 (FMAF128): Likewise.
9385 * doc/cpp.texi (Common Predefined Macros): Document defining
9386 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
9387 fma _Float<N> and _Float<N>X variants.
9389 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9392 * config/i386/i386-modes.def (CCFPU): Remove definition.
9393 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
9394 (ix86_cc_modes_compatible): Ditto.
9395 (ix86_expand_carry_flag_compare): Ditto.
9396 (ix86_expand_int_movcc): Ditto.
9397 (ix86_expand_int_addcc): Ditto.
9398 (ix86_reverse_condition): Ditto.
9399 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
9400 Return true/false for unordered/ordered fp comparisons.
9401 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
9402 (ix86_prepare_fp_compare_args): Update for rename.
9403 (ix86_expand_fp_compare): Update for rename. Generate unordered
9404 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
9405 (ix86_expand_sse_compare_and_jump): Ditto.
9406 * config/i386/predicates.md (fcmov_comparison_operator):
9407 Remove CCFPU mode handling.
9408 (ix86_comparison_operator): Ditto.
9409 (ix86_carry_flag_operator): Ditto.
9410 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
9411 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
9412 (*cmpu<mode>_cc_i387): Ditto.
9413 (FPCMP): Remove mode iterator.
9414 (unord): Remove mode attribute.
9415 (unord_subst): New define_subst transformation
9416 (unord): New define_subst attribute.
9418 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
9419 (*cmpi<unord>xf_i387): Ditto.
9420 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
9421 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
9422 using unord_subst transformation.
9423 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
9424 (round_saeonly): Also handle CCFP mode.
9425 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
9426 Remove UNSPEC_SAHF unspec handling.
9428 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9430 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
9432 2017-10-27 Jeff Law <law@redhat.com>
9434 * gimple-ssa-sprintf.c: Include domwalk.h.
9435 (class sprintf_dom_walker): New class, derived from dom_walker.
9436 (sprintf_dom_walker::before_dom_children): New function.
9437 (struct call_info): Moved into sprintf_dom_walker class
9438 (compute_formath_length, handle_gimple_call): Likewise.
9439 (sprintf_length::execute): Call the dominator walker rather
9440 than walking the statements.
9442 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
9443 gimple statement locations.
9444 (check_array_bounds): Corresponding changes. Get the statement's
9445 location directly from wi->stmt.
9447 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
9450 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
9452 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9454 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
9455 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
9457 2017-10-27 Jakub Jelinek <jakub@redhat.com>
9460 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
9461 * config/i386/i386.c (maybe_get_pool_constant): Removed.
9462 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
9463 maybe_get_pool_constant.
9464 * config/i386/predicates.md (zero_extended_scalar_load_operand):
9467 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9469 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
9470 2.26 caveat. Update gas and gld versions.
9471 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
9474 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9476 * cgraph.h (set_malloc_flag): Declare.
9477 * cgraph.c (set_malloc_flag_1): New function.
9478 (set_malloc_flag): Likewise.
9479 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
9480 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
9482 (read_ipa_call_summary): Add support for reading is_return_callee.
9483 (write_ipa_call_summary): Stream is_return_callee.
9484 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
9485 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
9486 ipa-prop.h, ipa-fnsummary.h.
9487 (pure_const_names): Change to static.
9488 (malloc_state_e): Define.
9489 (malloc_state_names): Define.
9490 (funct_state_d): Add field malloc_state.
9491 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
9492 (check_retval_uses): New function.
9493 (malloc_candidate_p): Likewise.
9494 (analyze_function): Add support for malloc attribute.
9495 (pure_const_write_summary): Stream malloc_state.
9496 (pure_const_read_summary): Add support for reading malloc_state.
9497 (dump_malloc_lattice): New function.
9498 (propagate_malloc): New function.
9499 (warn_function_malloc): New function.
9500 (ipa_pure_const::execute): Call propagate_malloc and
9501 ipa_free_fn_summary.
9502 (pass_local_pure_const::execute): Add support for malloc attribute.
9503 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
9504 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
9506 2017-10-27 Martin Liska <mliska@suse.cz>
9508 PR gcov-profile/82457
9509 * doc/invoke.texi: Document that one needs a non-strict ISO mode
9510 for fork-like functions to be properly instrumented.
9512 2017-10-27 Richard Biener <rguenther@suse.de>
9515 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
9516 info when we redirected EH.
9518 2017-10-26 Michael Collison <michael.collison@arm.com>
9520 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
9522 (<optab>_trunchf<GPI:mode>2: New pattern.
9523 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
9524 * config/aarch64/iterators.md (wv): New mode attribute.
9525 (vf, VF): New mode attributes.
9526 (vgp, VGP): New mode attributes.
9527 (s): Update attribute with SImode and DImode prefixes.
9529 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9531 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
9532 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
9533 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
9534 (nios2_option_overide): Initialize it. Don't allow R0-relative
9535 addressing with PIC.
9536 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
9537 (nios2_symbolic_constant_p): Likewise.
9538 (nios2_legitimate_address_p): Likewise.
9539 (nios2_r0rel_section_name_p): New.
9540 (nios2_symbol_ref_in_r0rel_data_p): New.
9541 (nios2_emit_move_sequence): Handle r0rel_constant_p.
9542 (r0rel_constant_p): New.
9543 (nios2_print_operand_address): Handle r0rel_constant_p.
9544 (nios2_cdx_narrow_form_p): Likewise.
9545 * config/nios2/nios2.opt (mr0rel-sec=): New option.
9546 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
9547 (Nios II Options): Document -mr0rel-sec.
9549 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9551 * config/nios2/nios2.c: Include xregex.h.
9552 (nios2_gprel_sec_regex): New.
9553 (nios2_option_overide): Initialize it. Don't allow GP-relative
9554 addressing with PIC.
9555 (nios2_small_section_name_p): Check for regex match.
9556 * config/nios2/nios2.opt (mgprel-sec=): New option.
9557 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
9558 (Nios II Options): Document -mgprel-sec.
9560 2017-10-26 Jim Wilson <wilson@tuliptree.org>
9562 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
9564 2017-10-26 Tom de Vries <tom@codesourcery.com>
9566 PR tree-optimization/82707
9567 * gimple.c (gimple_copy): Fix unsharing of
9568 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
9570 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
9572 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
9573 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
9574 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
9575 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
9576 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
9577 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
9578 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
9579 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
9580 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
9581 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
9582 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
9583 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
9584 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
9585 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
9586 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
9587 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
9588 _mm512_mask_cmpunord_ps_mask): New intrinsics.
9590 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9592 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
9594 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
9595 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
9596 warning to rs6000.c. Remove the Undocumented flag, since it has
9598 (-mabi=ibmlongdouble): Likewise.
9599 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
9600 already set, set the default format for long double.
9601 (rs6000_debug_reg_global): Print whether long double is IBM or
9603 (rs6000_option_override_internal): Rework setting long double
9604 format. Only warn if the user is changing the long double default
9605 and they did not use -Wno-psabi.
9606 * doc/invoke.texi (PowerPC options): Update the documentation for
9607 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
9609 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9610 Alan Hayward <alan.hayward@arm.com>
9611 David Sherwood <david.sherwood@arm.com>
9613 * rtl.h (wider_subreg_mode): New function.
9614 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
9615 rather than an unsigned int *.
9616 * ira-color.c (regno_max_ref_width): Replace with...
9617 (regno_max_ref_mode): ...this new variable.
9618 (coalesced_pseudo_reg_slot_compare): Update accordingly.
9619 Use wider_subreg_mode.
9620 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
9621 rather than an unsigned int *.
9622 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
9623 (process_alt_operands): Likewise.
9624 (invariant_p): Likewise.
9625 * lra-spills.c (assign_mem_slot): Likewise.
9626 (add_pseudo_to_slot): Likewise.
9627 * lra.c (collect_non_operand_hard_regs): Likewise.
9628 (add_regs_to_insn_regno_info): Likewise.
9629 * reload1.c (regno_max_ref_width): Replace with...
9630 (regno_max_ref_mode): ...this new variable.
9631 (reload): Update accordingly. Update call to
9632 ira_sort_regnos_for_alter_reg.
9633 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
9634 (init_eliminable_invariants): Update to use regno_max_ref_mode.
9635 (scan_paradoxical_subregs): Likewise.
9637 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9639 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
9640 (aarch64_frame): Add emit_frame_chain boolean.
9641 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9642 Move eh_return case to aarch64_layout_frame.
9643 (aarch64_layout_frame): Initialize emit_frame_chain.
9644 (aarch64_expand_prologue): Use emit_frame_chain.
9646 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9648 * config/aarch64/aarch64.c (aarch64_layout_frame):
9649 Ensure LR is always stored at the bottom of the callee-saves.
9650 Remove rarely used frame layout which saves callee-saves at top of
9651 frame, so the store of LR can be used as a valid probe in all cases.
9653 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9655 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
9656 Improve unaligned TImode/TFmode base/offset split.
9658 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9659 Alan Hayward <alan.hayward@arm.com>
9660 David Sherwood <david.sherwood@arm.com>
9662 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
9663 * combine.c (find_single_use_1): Likewise.
9664 (expand_field_assignment): Likewise.
9665 (move_deaths): Likewise.
9666 * lra-constraints.c (simplify_operand_subreg): Likewise.
9667 (curr_insn_transform): Likewise.
9668 * lra.c (collect_non_operand_hard_regs): Likewise.
9669 (add_regs_to_insn_regno_info): Likewise.
9670 * rtlanal.c (reg_referenced_p): Likewise.
9671 (covers_regno_no_parallel_p): Likewise.
9673 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9675 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
9676 Don't print any bits outside the precision of the value.
9677 * wide-int.cc (test_printing): Add some new tests.
9679 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9681 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
9682 supports -xbrace_comment option.
9683 * configure: Regenerate.
9684 * config.in: Regenerate.
9685 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
9686 (ASM_CPU_SPEC): Use it.
9688 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9690 * target.def (static_rtx_alignment): New hook.
9691 * targhooks.h (default_static_rtx_alignment): Declare.
9692 * targhooks.c (default_static_rtx_alignment): New function.
9693 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
9694 * doc/tm.texi: Regenerate.
9695 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
9696 instead of targetm.constant_alignment. Remove call to
9698 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9699 (cris_preferred_mininum_alignment): New function, split out from...
9700 (cris_constant_alignment): ...here.
9701 (cris_static_rtx_alignment): New function.
9702 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
9704 (ix86_constant_alignment): ...here.
9705 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9706 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9707 (mmix_static_rtx_alignment): New function.
9708 * config/spu/spu.c (spu_static_rtx_alignment): New function.
9709 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9711 2017-10-26 Tamar Christina <tamar.christina@arm.com>
9714 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
9715 Add flag_trapping_math and flag_fp_int_builtin_inexact.
9717 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
9719 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
9720 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
9721 mark as a sign-extending load.
9722 (local_pic_load_u): Define.
9724 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
9727 * fold-const.c (operand_equal_for_comparison_p): Also return true
9728 if ARG0 is a simple variant of ARG1 with narrower precision.
9729 (fold_ternary_loc): Always pass unstripped operands to the predicate.
9731 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
9733 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
9735 * i386.h (processor_costs): Add gather_static, gather_per_elt,
9736 scatter_static, scatter_per_elt.
9737 * x86-tune-costs.h: Add new cost entries.
9739 2017-10-25 Richard Biener <rguenther@suse.de>
9741 * tree-ssa-sccvn.h (vn_eliminate): Declare.
9742 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
9743 class pass_fre): Move to ...
9744 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
9745 class pass_fre): ... here and adjust for statistics.
9747 2017-10-25 Jakub Jelinek <jakub@redhat.com>
9750 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
9751 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
9752 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
9753 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
9756 2017-10-25 Richard Biener <rguenther@suse.de>
9758 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
9759 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
9760 eliminate_push_avail, eliminate_insert): Move inside...
9761 (class eliminate_dom_walker): ... this class in preparation
9763 (fini_eliminate): Remove by merging with ...
9764 (eliminate): ... this function. Adjust for class changes.
9765 (pass_pre::execute): Remove fini_eliminate call.
9766 (pass_fre::execute): Likewise.
9768 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9771 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
9772 (VPERMI2, VPERMI2I): New mode iterators.
9773 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
9774 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
9776 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
9777 mode iterator. Remove 3 old define_insn patterns.
9778 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
9779 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
9780 VPERMI2 mode iterator, remove the other two expanders.
9781 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
9782 to use VPERMI2 mode iterator, add another alternative for vpermi2*
9783 instructions, remove the other two patterns.
9784 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
9785 mode iterator, remove the other two patterns.
9786 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
9787 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
9788 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
9789 and adjust argument order accordingly.
9790 (ix86_expand_vec_perm): Adjust caller.
9791 (expand_vec_perm_1): Likewise.
9792 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
9793 (expand_vec_perm_vpermt2_vpshub2): ... this.
9794 (ix86_expand_vec_perm_const_1): Adjust caller.
9795 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
9798 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
9799 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
9800 (vec_shl_<mode>): Remove unused expander.
9801 (avx512bw_<shift_insn><mode>3): New define_insn.
9802 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
9803 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
9805 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
9808 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
9811 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9813 PR rtl-optimization/82396
9814 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
9815 (autopref_multipass_init): Simplify initialization.
9816 (autopref_rank_data): Simplify sort order.
9817 * gcc/sched-int.h (autopref_multipass_data_): Remove
9818 multi_mem_insn_p, min_offset and max_offset.
9820 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9823 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9824 Check special value of flag_omit_frame_pointer.
9825 (aarch64_can_eliminate): Likewise.
9826 (aarch64_override_options_after_change_1): Simplify handling of
9827 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
9829 2017-10-24 Richard Biener <rguenther@suse.de>
9831 PR tree-optimization/82697
9832 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
9833 zero for conditional load and unconditional store.
9835 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9837 * doc/install.texi: Document bootstrap-cet.
9839 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9842 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
9843 ENDBR instruction at function entrance if function is only
9846 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9849 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
9850 patterns to better describe from which operation the CF is computed.
9851 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
9852 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
9853 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
9854 is 0, use _0 suffixed expanders instead of emitting a comparison
9857 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
9859 * config/i386/i386.md(*movsf_internal, *movdf_internal):
9860 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
9862 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
9865 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
9866 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
9867 * loop-iv.c (iv_get_reaching_def): Likewise.
9868 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
9869 variable is promoted and the partition contains undefined values.
9871 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9873 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
9875 (nios2_address_cost): Define.
9876 (nios2_legitimize_address): Recognize (exp + constant) directly.
9877 (TARGET_ADDRESS_COST): Define.
9879 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9881 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
9882 (nios2_symbolic_memory_operand_p): Declare.
9883 (nios2_split_large_constant): Declare.
9884 (nios2_split_symbolic_memory_operand): Declare.
9885 * config/nios2/nios2.c: Adjust includes.
9886 (nios2_symbolic_constant_allowed): New.
9887 (nios2_symbolic_constant_p): New.
9888 (nios2_plus_symbolic_constant_p): New.
9889 (nios2_valid_addr_expr_p): Recognize addresses involving
9891 (nios2_legitimate_address_p): Likewise, also LO_SUM.
9892 (nios2_symbolic_memory_operand_p): New.
9893 (nios2_large_constant_p): New.
9894 (nios2_split_large_constant): New.
9895 (nios2_split_plus_large_constant): New.
9896 (nios2_split_symbolic_memory_operand): New.
9897 (nios2_legitimize_address): Code refactoring. Handle addresses
9898 involving symbolic constants.
9899 (nios2_emit_move_sequence): Likewise.
9900 (nios2_print_operand): Improve error output.
9901 (nios2_print_operand_address): Handle LO_SUM.
9902 (nios2_cdx_narrow_form_p): Likewise.
9903 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
9904 operands involving symbolic constants.
9905 (movhi_internal, movsi_internal): Likewise.
9906 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
9907 (extendhisi2, extendqi<mode>2): Likewise.
9909 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9911 * tree-pass.h (PROP_rtl_split_insns): Define.
9912 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
9914 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9916 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
9918 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9921 * target.def (const_not_ok_for_debug_p): Default to
9922 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
9923 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
9924 * targhooks.c (default_const_not_ok_for_debug_p): New function.
9925 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
9926 which targetm.const_not_ok_for_debug_p returned true.
9927 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
9929 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
9931 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
9932 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
9933 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
9935 (ix86_const_not_ok_for_debug_p): New function.
9936 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
9937 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
9939 2017-10-23 David Malcolm <dmalcolm@redhat.com>
9942 * system.h: Conditionally include "unique-ptr.h" if
9943 INCLUDE_UNIQUE_PTR is defined.
9944 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
9945 of defining INCLUDE_UNIQUE_PTR before including "system.h".
9947 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
9949 * config/rl78/rl78.md: New define_expand "subdi3".
9951 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
9954 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
9955 DF_REF_INSN if DF_REF_INSN_INFO is false.
9957 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
9959 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
9960 xmm_move instead of sse_move.
9961 (sse_store_index): New function.
9962 (ix86_register_move_cost): Be more sensible about mismatch stall;
9963 model AVX moves correctly; make difference between sse->integer and
9965 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
9966 moves; make difference between SSE and AVX.
9967 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
9968 and zmm_move. Increase size of sse load and store tables;
9969 add unaligned load and store tables; add ssemmx_to_integer.
9970 * x86-tune-costs.h: Update all entries according to real
9971 move latencies from Agner Fog's manual and chip documentation.
9973 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9976 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
9977 * config/i386/constraints.md (Wf): New constraint.
9978 * config/i386/i386.md (UNSPEC_SBB): New unspec.
9979 (cmp<dwi>_doubleword): Removed.
9980 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
9981 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
9982 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
9983 expand with cmp<dwi>_doubleword. For LTU and GEU use
9984 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
9986 * common.opt (gcolumn-info): Enable by default.
9987 * doc/invoke.texi (gcolumn-info): Document new default.
9989 2017-10-23 Richard Biener <rguenther@suse.de>
9991 PR tree-optimization/82672
9992 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
9993 Fold the stmt if we propagated into it.
9995 2017-10-23 Richard Biener <rguenther@suse.de>
9997 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
9998 (bitmap_remove_expr_from_set): ... this. All callers call this
9999 for non-constant values.
10000 (bitmap_set_subtract): Rename to...
10001 (bitmap_set_subtract_expressions): ... this. Adjust and
10003 (bitmap_set_contains_value): Remove superfluous check.
10004 (bitmap_set_replace_value): Inline into single caller ...
10005 (bitmap_value_replace_in_set): ... here and simplify.
10006 (dependent_clean): Merge into ...
10007 (clean): ... this using an overload. Adjust.
10008 (prune_clobbered_mems): Adjust.
10009 (compute_antic_aux): Likewise.
10010 (compute_partial_antic_aux): Likewise.
10012 2017-10-23 Richard Biener <rguenther@suse.de>
10014 PR tree-optimization/82129
10016 2017-08-01 Richard Biener <rguenther@suse.de>
10018 PR tree-optimization/81181
10019 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10020 (compute_antic): ... end of iteration here.
10022 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10024 * target.def (starting_frame_offset): New hook.
10025 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
10026 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
10027 * doc/tm.texi.in: Regenerate.
10028 * hooks.h (hook_hwi_void_0): Declare.
10029 * hooks.c (hook_hwi_void_0): New function.
10030 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
10031 STARTING_FRAME_OFFSET.
10032 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
10033 * reload1.c (reload): Likewise.
10034 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
10035 instead of STARTING_FRAME_OFFSET.
10036 * function.c (try_fit_stack_local): Likewise.
10037 (assign_stack_local_1): Likewise
10038 (instantiate_virtual_regs): Likewise.
10039 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
10040 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
10041 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
10042 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
10043 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
10044 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
10045 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
10046 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
10047 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
10048 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
10049 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
10050 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
10051 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
10052 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
10053 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
10054 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
10055 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
10056 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
10057 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
10058 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
10059 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
10060 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
10061 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
10062 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
10063 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
10064 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
10065 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
10066 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
10067 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
10068 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
10069 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
10070 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
10071 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
10072 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
10073 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
10074 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
10075 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
10076 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
10077 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
10078 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
10079 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
10080 * config/avr/avr.c (avr_starting_frame_offset): Make static and
10081 return a HOST_WIDE_INT.
10082 (avr_builtin_setjmp_frame_value): Use it instead of
10083 STARTING_FRAME_OFFSET.
10084 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10085 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
10086 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
10088 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10089 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
10090 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
10091 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10092 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
10093 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
10094 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10095 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
10096 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
10097 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10098 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
10099 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
10101 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10102 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
10103 * config/mips/mips.c (mips_compute_frame_info): Refer to
10104 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10105 (mips_starting_frame_offset): New function.
10106 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10107 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
10108 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
10109 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
10110 and return a HOST_WIDE_INT.
10111 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10112 (mmix_initial_elimination_offset): Refer to
10113 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10114 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
10115 * config/pa/pa.c (pa_starting_frame_offset): New function.
10116 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
10117 (pa_expand_prologue): Likewise.
10118 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10119 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
10120 !FRAME_GROWS_DOWNWARD handling to...
10121 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10122 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
10123 !FRAME_GROWS_DOWNWARD handling to...
10124 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10125 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
10126 !FRAME_GROWS_DOWNWARD handling to...
10127 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10128 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
10130 (rs6000_starting_frame_offset): New function.
10131 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
10132 !FRAME_GROWS_DOWNWARD handling to...
10133 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10134 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
10135 !FRAME_GROWS_DOWNWARD handling to...
10136 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10137 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
10138 !FRAME_GROWS_DOWNWARD handling to...
10139 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10140 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
10141 (rs6000_starting_frame_offset): New function.
10142 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
10143 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
10144 * config/vax/vax.c (vax_starting_frame_offset): New function.
10145 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
10146 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10147 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
10148 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
10149 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10150 * system.h (STARTING_FRAME_OFFSET): Poison.
10152 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10154 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10155 SCALAR_TYPE_MODE instead of TYPE_MODE.
10157 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10158 Alan Hayward <alan.hayward@arm.com>
10159 David Sherwood <david.sherwood@arm.com>
10161 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
10163 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10164 Alan Hayward <alan.hayward@arm.com>
10165 David Sherwood <david.sherwood@arm.com>
10167 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
10169 2017-10-23 Richard Biener <rguenther@suse.de>
10171 PR tree-optimization/82129
10172 * tree-ssa-pre.c (bitmap_set_and): Remove.
10173 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
10174 canonicalizing expressions in the set to those with lowest
10175 ID rather than taking that from the first edge.
10177 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10179 * combine.c (rtx_equal_for_field_assignment_p): Use
10180 byte_lowpart_offset.
10182 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10183 Alan Hayward <alan.hayward@arm.com>
10184 David Sherwood <david.sherwood@arm.com>
10186 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
10187 to a SUBREG_PROMOTED_VAR.
10189 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10190 Alan Hayward <alan.hayward@arm.com>
10191 David Sherwood <david.sherwood@arm.com>
10193 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
10194 (expand_debug_source_expr): Likewise.
10195 * combine.c (combine_simplify_rtx): Likewise.
10196 * cse.c (fold_rtx): Likewise.
10197 * optabs.c (expand_float): Likewise.
10198 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10199 (simplify_binary_operation_1): Likewise.
10201 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10202 Alan Hayward <alan.hayward@arm.com>
10203 David Sherwood <david.sherwood@arm.com>
10205 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
10206 (record_promoted_value): Likewise.
10207 * expr.c (expand_expr_real_2): Likewise.
10208 * ree.c (update_reg_equal_equiv_notes): Likewise.
10209 (combine_set_extension): Likewise.
10210 * rtlanal.c (low_bitmask_len): Likewise.
10211 * simplify-rtx.c (neg_const_int): Likewise.
10212 (simplify_binary_operation_1): Likewise.
10214 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10215 Alan Hayward <alan.hayward@arm.com>
10216 David Sherwood <david.sherwood@arm.com>
10218 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
10219 * regcprop.c (maybe_mode_change): Likewise.
10220 * reload1.c (alter_reg): Likewise.
10222 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10224 * inchash.h (inchash::hash::add_wide_int): New function.
10225 * lto-streamer-out.c (hash_tree): Use it.
10227 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10229 * inchash.h (inchash::hash::add_wide_int): Rename to...
10230 (inchash::hash::add_hwi): ...this.
10231 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
10232 (polymorphic_call_target_hasher::hash): Likewise.
10233 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
10234 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
10235 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
10236 * lto-streamer-out.c (hash_tree): Likewise.
10237 * optc-save-gen.awk: Likewise.
10238 * tree.c (add_expr): Likewise.
10240 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10243 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
10244 for ordered inequality comparisons even with TARGET_IEEE_FP.
10246 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10249 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
10250 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10251 Expand with cmp<dwi>_doubleword.
10253 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10255 * extend.texi: Add x86 specific to 'nocf_check' attribute.
10256 List CET intrinsics.
10257 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
10258 specific to -fcf-protection option.
10260 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10262 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
10263 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
10264 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
10265 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
10266 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
10267 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
10268 (extra_objs): Add cet.o for Linux/x86 targets.
10269 (tmake_file): Add i386/t-cet for Linux/x86 targets.
10270 * config/i386/cet.c: New file.
10271 * config/i386/cetintrin.h: Likewise.
10272 * config/i386/t-cet: Likewise.
10273 * config/i386/cpuid.h (bit_SHSTK): New.
10274 (bit_IBT): Likewise.
10275 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
10276 pass IBT and SHSTK bits.
10277 * config/i386/i386-builtin-types.def
10278 (VOID_FTYPE_UNSIGNED_PVOID): New.
10279 (VOID_FTYPE_UINT64_PVOID): Likewise.
10280 * config/i386/i386-builtin.def: Add CET intrinsics.
10281 * config/i386/i386-c.c (ix86_target_macros_internal): Add
10282 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
10283 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
10284 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
10286 * config/i386/i386.c (rest_of_insert_endbranch): New.
10287 (pass_data_insert_endbranch): Likewise.
10288 (pass_insert_endbranch): Likewise.
10289 (make_pass_insert_endbranch): Likewise.
10290 (ix86_notrack_prefixed_insn_p): Likewise.
10291 (ix86_target_string): Add -mibt, -mshstk flags.
10292 (ix86_option_override_internal): Add flag_cf_protection
10294 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
10295 (ix86_print_operand): Add 'notrack' prefix output.
10296 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
10297 (ix86_expand_builtin): Expand CET intrinsics.
10298 (x86_output_mi_thunk): Add 'endbranch' instruction.
10299 * config/i386/i386.h (TARGET_IBT): New.
10300 (TARGET_IBT_P): Likewise.
10301 (TARGET_SHSTK): Likewise.
10302 (TARGET_SHSTK_P): Likewise.
10303 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
10304 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
10305 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
10306 (builtin_setjmp_setup): New pattern.
10307 (builtin_longjmp): Likewise.
10308 (rdssp<mode>): Likewise.
10309 (incssp<mode>): Likewise.
10310 (saveprevssp): Likewise.
10311 (rstorssp): Likewise.
10312 (wrss<mode>): Likewise.
10313 (wruss<mode>): Likewise.
10314 (setssbsy): Likewise.
10315 (clrssbsy): Likewise.
10316 (nop_endbr): Likewise.
10317 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
10319 * config/i386/immintrin.h: Include <cetintrin.h>.
10320 * config/i386/linux-common.h
10321 (file_end_indicate_exec_stack_and_cet): New prototype.
10322 (TARGET_ASM_FILE_END): New.
10324 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10326 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
10327 latencies instead of having separate table; make difference between
10328 integer and float costs.
10329 * i386.h (processor_costs): Remove scalar_stmt_cost,
10330 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
10331 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
10333 * x86-tune-costs.h: Remove entries which has been removed in
10334 procesor_costs from all tables; make cond_taken_branch_cost
10335 and cond_not_taken_branch_cost COST_N_INSNS based.
10337 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10339 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
10341 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10343 * config/i386/i386.md (isa): Remove fma_avx512f.
10344 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
10345 <avx512>_fmadd_<mode>_mask3<round_name>,
10346 <avx512>_fmsub_<mode>_mask<round_name>,
10347 <avx512>_fmsub_<mode>_mask3<round_name>,
10348 <avx512>_fnmadd_<mode>_mask<round_name>,
10349 <avx512>_fnmadd_<mode>_mask3<round_name>,
10350 <avx512>_fnmsub_<mode>_mask<round_name>,
10351 <avx512>_fnmsub_<mode>_mask3<round_name>,
10352 <avx512>_fmaddsub_<mode>_mask<round_name>,
10353 <avx512>_fmaddsub_<mode>_mask3<round_name>,
10354 <avx512>_fmsubadd_<mode>_mask<round_name>,
10355 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
10356 (*vec_widen_umult_even_v16si<mask_name>,
10357 *vec_widen_smult_even_v16si<mask_name>): Likewise.
10358 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
10360 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10362 * extend.texi: Add 'nocf_check' documentation.
10363 * gimple.texi: Add second parameter to
10364 gimple_build_call_from_tree.
10365 * invoke.texi: Add -fcf-protection documentation.
10366 * rtl.texi: Add REG_CALL_NOTRACK documenation.
10368 2017-10-20 Richard Biener <rguenther@suse.de>
10370 PR tree-optimization/82473
10371 * tree-vect-loop.c (vectorizable_reduction): Properly get at
10372 the largest input type.
10374 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10376 * c-attribs.c (handle_nocf_check_attribute): New function.
10377 (c_common_attribute_table): Add 'nocf_check' handling.
10378 * gimple-parser.c: Add second argument NULL to
10379 gimple_build_call_from_tree.
10380 * attrib.c (comp_type_attributes): Check nocf_check attribute.
10381 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
10383 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
10384 * common.opt: Add fcf-protection flag.
10385 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
10386 * flag-types.h: Add enum cf_protection_level.
10387 * gimple.c (gimple_build_call_from_tree): Add second parameter.
10388 Add 'nocf_check' attribute propagation to gimple call.
10389 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
10390 (gimple_build_call_from_tree): Update prototype.
10391 (gimple_call_nocf_check_p): New function.
10392 (gimple_call_set_nocf_check): Likewise.
10393 * gimplify.c: Add second argument to gimple_build_call_from_tree.
10394 * ipa-icf.c: Add nocf_check attribute in statement hash.
10395 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
10396 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
10397 * toplev.c (process_options): Add flag_cf_protection handling.
10399 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10401 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
10403 2017-10-20 Richard Biener <rguenther@suse.de>
10405 PR tree-optimization/82603
10406 * tree-if-conv.c (predicate_mem_writes): Make sure to only
10407 remove false predicated stores.
10409 2017-10-20 Richard Biener <rguenther@suse.de>
10411 * graphite-isl-ast-to-gimple.c
10412 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
10413 Remove return value and simplify, dump copied stmt after lhs
10415 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
10416 Reduce dump verbosity.
10417 (gsi_insert_earliest): Likewise.
10418 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
10419 * graphite.c (print_global_statistics): Adjust dumping.
10420 (print_graphite_scop_statistics): Likewise.
10421 (print_graphite_statistics): Do not dump loops here.
10422 (graphite_transform_loops): But here.
10424 2017-10-20 Nicolas Roche <roche@adacore.com>
10426 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
10427 * configure: Regenerate.
10429 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10432 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
10433 functions when optimizing replace GIMPLE_RETURN stmts with
10434 calls to __builtin_unreachable ().
10437 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
10438 for -fsanitize=thread link of executables.
10439 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
10440 link of executables.
10443 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
10444 New mode iterators.
10445 (<shift_insn><mode>3<mask_name>): Change the last of the 3
10446 define_insns for logical vector shifts to use VI248_AVX512BW
10447 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
10448 condition, useless isa and prefix attributes. Change the first
10449 2 of these define_insns to ...
10450 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
10451 define_insn for avx512vl.
10452 (<shift_insn><mode>3): ... and this, new define_insn without
10453 masking for non-avx512vl.
10456 * config/i386/sse.md (*andnot<mode>3,
10457 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
10458 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
10459 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
10460 not applied use empty suffix even for TARGET_AVX512VL.
10461 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
10462 is applied, supply evex,evex or evex,evex,evex instead of just
10465 2017-10-20 Julia Koval <julia.koval@intel.com>
10467 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
10468 (OPTION_MASK_ISA_GFNI_UNSET): New.
10469 (ix86_handle_option): Handle OPT_mgfni.
10470 * config/i386/cpuid.h (bit_GFNI): New.
10471 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
10472 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
10473 * config/i386/i386.c (ix86_target_string): Add -mgfni.
10474 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
10475 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
10476 * config/i386/i386.opt: Add mgfni.
10478 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
10480 * config/msp430/msp430.c (msp430_option_override): Disable
10481 -fdelete-null-pointer-checks.
10482 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
10484 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10486 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
10487 of x87 and SSE instructions.
10489 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10491 * asan.c (create_cond_insert_point): Do not update edge count.
10492 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
10493 (afdo_propagate_circuit): Likewise.
10494 (afdo_calculate_branch_prob): Likewise.
10495 (afdo_annotate_cfg): Likewise.
10496 * basic-block.h (struct edge_def): Remove count.
10497 (edge_def::count): New accessor.
10498 * bb-reorder.c (rotate_loop): Update.
10499 (find_traces_1_round): Update.
10500 (connect_traces): Update.
10501 (sanitize_hot_paths): Update.
10502 * cfg.c (unchecked_make_edge): Update.
10503 (make_single_succ_edge): Update.
10504 (check_bb_profile): Update.
10505 (dump_edge_info): Update.
10506 (update_bb_profile_for_threading): Update.
10507 (scale_bbs_frequencies_int): Update.
10508 (scale_bbs_frequencies_gcov_type): Update.
10509 (scale_bbs_frequencies_profile_count): Update.
10510 (scale_bbs_frequencies): Update.
10511 * cfganal.c (connect_infinite_loops_to_exit): Update.
10512 * cfgbuild.c (compute_outgoing_frequencies): Update.
10513 (find_many_sub_basic_blocks): Update.
10514 * cfgcleanup.c (try_forward_edges): Update.
10515 (try_crossjump_to_edge): Update
10516 * cfgexpand.c (expand_gimple_cond): Update
10517 (expand_gimple_tailcall): Update
10518 (construct_exit_block): Update
10519 * cfghooks.c (verify_flow_info): Update
10520 (redirect_edge_succ_nodup): Update
10521 (split_edge): Update
10522 (make_forwarder_block): Update
10523 (duplicate_block): Update
10524 (account_profile_record): Update
10525 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10526 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10527 * cfgloopmanip.c (scale_loop_profile): Update.
10529 (lv_adjust_loop_entry_edge): Update.
10530 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10531 (force_nonfallthru_and_redirect): Update.
10532 (purge_dead_edges): Update.
10533 (rtl_flow_call_edges_add): Update.
10534 * cgraphunit.c (init_lowered_empty_function): Update.
10535 (cgraph_node::expand_thunk): Update.
10536 * gimple-pretty-print.c (dump_probability): Update.
10537 (dump_edge_probability): Update.
10538 * gimple-ssa-isolate-paths.c (isolate_path): Update.
10539 * haifa-sched.c (sched_create_recovery_edges): Update.
10540 * hsa-gen.c (convert_switch_statements): Update.
10541 * ifcvt.c (dead_or_predicable): Update.
10542 * ipa-inline-transform.c (inline_transform): Update.
10543 * ipa-split.c (split_function): Update.
10544 * ipa-utils.c (ipa_merge_profiles): Update.
10545 * loop-doloop.c (add_test): Update.
10546 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
10547 * lto-streamer-in.c (input_cfg): Update.
10548 (input_function): Update.
10549 * lto-streamer-out.c (output_cfg): Update.
10550 * modulo-sched.c (sms_schedule): Update.
10551 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10552 * predict.c (maybe_hot_edge_p): Update.
10553 (unlikely_executed_edge_p): Update.
10554 (probably_never_executed_edge_p): Update.
10555 (dump_prediction): Update.
10556 (drop_profile): Update.
10557 (propagate_unlikely_bbs_forward): Update.
10558 (determine_unlikely_bbs): Update.
10559 (force_edge_cold): Update.
10560 * profile.c (compute_branch_probabilities): Update.
10561 * reg-stack.c (better_edge): Update.
10562 * shrink-wrap.c (handle_simple_exit): Update.
10563 * tracer.c (better_p): Update.
10564 * trans-mem.c (expand_transaction): Update.
10565 (split_bb_make_tm_edge): Update.
10566 * tree-call-cdce.c: Update.
10567 * tree-cfg.c (gimple_find_sub_bbs): Update.
10568 (gimple_split_edge): Update.
10569 (gimple_duplicate_sese_region): Update.
10570 (gimple_duplicate_sese_tail): Update.
10571 (gimple_flow_call_edges_add): Update.
10572 (insert_cond_bb): Update.
10573 (execute_fixup_cfg): Update.
10574 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
10575 * tree-complex.c (expand_complex_div_wide): Update.
10576 * tree-eh.c (lower_resx): Update.
10577 (unsplit_eh): Update.
10578 (cleanup_empty_eh_move_lp): Update.
10579 * tree-inline.c (copy_edges_for_bb): Update.
10580 (freqs_to_counts): Update.
10581 (copy_cfg_body): Update.
10582 * tree-ssa-dce.c (remove_dead_stmt): Update.
10583 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10584 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10585 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10586 (unloop_loops): Update.
10587 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10588 * tree-ssa-loop-split.c (connect_loops): Update.
10589 (split_loop): Update.
10590 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10591 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10592 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10593 * tree-ssa-reassoc.c (branch_fixup): Update.
10594 * tree-ssa-tail-merge.c (replace_block_by): Update.
10595 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10596 (compute_path_counts): Update.
10597 (update_profile): Update.
10598 (recompute_probabilities): Update.
10599 (update_joiner_offpath_counts): Update.
10600 (estimated_freqs_path): Update.
10601 (freqs_to_counts_path): Update.
10602 (clear_counts_path): Update.
10603 (ssa_fix_duplicate_block_edges): Update.
10604 (duplicate_thread_path): Update.
10605 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10606 (case_bit_test_cmp): Update.
10607 (collect_switch_conv_info): Update.
10608 (gen_inbound_check): Update.
10609 (do_jump_if_equal): Update.
10610 (emit_cmp_and_jump_insns): Update.
10611 * tree-tailcall.c (decrease_profile): Update.
10612 (eliminate_tail_call): Update.
10613 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10614 (vect_do_peeling): Update.
10615 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10616 * ubsan.c (ubsan_expand_null_ifn): Update.
10617 (ubsan_expand_ptr_ifn): Update.
10618 * value-prof.c (gimple_divmod_fixed_value): Update.
10619 (gimple_mod_pow2): Update.
10620 (gimple_mod_subtract): Update.
10621 (gimple_ic): Update.
10622 (gimple_stringop_fixed_value): Update.
10624 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10627 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
10629 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
10631 PR rtl-optimization/82395
10632 * ira-color.c (allocno_priority_compare_func): Fix comparison step
10633 based on non_spilled_static_chain_regno_p.
10635 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10637 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
10638 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
10639 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
10641 2017-10-19 Martin Sebor <msebor@redhat.com>
10643 PR tree-optimization/82596
10644 * tree.c (array_at_struct_end_p): Handle STRING_CST.
10646 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10648 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
10649 (get_mem_refs_of_builtin_call): Likewise.
10650 * builtins.c (expand_builtin_apply): Adjust call to
10651 allocate_dynamic_stack_space.
10652 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
10653 the third argument to allocate_dynamic_stack_space, otherwise -1.
10654 (expand_builtin): Deal with all alloca variants.
10655 (is_inexpensive_builtin): Likewise.
10656 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
10657 * calls.c (special_function_p): Deal with all alloca variants.
10658 (initialize_argument_information): Adjust call to
10659 allocate_dynamic_stack_space.
10660 (expand_call): Likewise.
10661 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
10662 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
10663 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
10664 use it for the stack usage computation.
10665 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
10666 * function.c (gimplify_parameters): Call build_alloca_call_expr.
10667 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
10668 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
10669 (in_loop_p): Remove first argument and useless check.
10670 (pass_walloca::execute): Remove useless test and adjust call to above.
10671 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
10672 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
10673 (gimplify_call_expr): Deal with all alloca variants.
10674 * hsa-gen.c (gen_hsa_alloca): Likewise.
10675 (gen_hsa_insns_for_call): Likewise.
10676 * ipa-pure-const.c (special_builtin_state): Likewise.
10677 * tree-chkp.c (chkp_build_returned_bound): Likewise.
10678 * tree-object-size.c (alloc_object_size): Likewise.
10679 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
10680 (call_may_clobber_ref_p_1): Likewise.
10681 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
10682 (ccp_fold_stmt): Likewise.
10683 (optimize_stack_restore): Likewise.
10684 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
10685 (mark_all_reaching_defs_necessary_1): Likewise.
10686 (propagate_necessity): Likewise.
10687 (eliminate_unnecessary_stmts): Likewise.
10688 * tree.c (build_common_builtin_nodes): Build
10689 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
10690 (build_alloca_call_expr): New function.
10691 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
10692 (CASE_BUILT_IN_ALLOCA): Likewise.
10693 (build_alloca_call_expr): Declare.
10694 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
10696 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10699 * dwarf2out.c (new_die_raw): New static inline function.
10700 (new_die): Use it to create the DIE.
10701 (add_AT_external_die_ref): Likewise.
10702 (clone_die): Likewise.
10703 (clone_as_declaration): Likewise.
10704 (dwarf2out_vms_debug_main_pointer): Likewise.
10705 (base_type_die): Likewise. Remove early return for corner cases.
10706 Do not call add_pubtype on the DIE here.
10707 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
10708 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
10709 typedefs for base types with DW_AT_endianity. Make sure a DIE with
10710 native order exists for base types, attach the DIE manually and call
10711 add_pubtype on it. Do not equate a reverse order DIE to the type.
10713 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
10715 * config/arm/arm.c (align_ok_ldrd_strd): New function.
10716 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
10718 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
10720 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10722 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
10723 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
10724 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
10725 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
10726 * opts.c (sanitizer_opts): Add builtin.
10727 * ubsan.c (instrument_builtin): New function.
10728 (pass_ubsan::execute): Call it.
10729 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
10730 * doc/invoke.texi: Document -fsanitize=builtin.
10732 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
10733 builtins, store max (log2 (align), 0) into uchar field instead of
10734 align into uptr field.
10735 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
10736 store uchar 0 field instead of uptr 0 field.
10737 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
10738 instead of passing one address of struct with 2 locations pass
10739 two addresses of structs with 1 location each.
10740 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
10741 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
10742 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
10743 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
10744 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
10745 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
10746 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
10747 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
10749 2017-10-19 Martin Liska <mliska@suse.cz>
10752 * file-find.c (remove_prefix): Remove.
10753 * file-find.h (remove_prefix): Likewise.
10754 * gcc-ar.c: Remove smartness of lookup.
10756 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
10758 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
10759 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
10760 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
10762 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10765 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
10766 (setcc + and to xor + setcc): New peephole2.
10768 2017-10-19 Tom de Vries <tom@codesourcery.com>
10770 * doc/sourcebuild.texi (Test Directives, Variants of
10771 dg-require-support): Add dg-require-stack-size.
10773 2017-10-19 Martin Liska <mliska@suse.cz>
10776 * gimplify.c (gimplify_decl_expr): Do not instrument variables
10777 that have a large alignment.
10778 (gimplify_target_expr): Likewise.
10780 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
10782 PR rtl-optimization/82602
10783 * ira.c (rtx_moveable_p): Return false for volatile asm.
10785 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
10788 * config/i386/i386-modes.def (CCGZ): New CC mode.
10789 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
10790 * config/i386/predicates.md (ix86_comparison_operator):
10792 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10793 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
10794 with double-word subtraction.
10795 (put_condition_code): Handle CCGZmode.
10797 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
10799 * wide-int.cc (debug (const wide_int &)): New.
10800 (debug (const wide_int *)): New.
10801 (debug (const widest_int &)): New.
10802 (debug (const widest_int *)): New.
10804 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
10806 PR middle-end/82556
10807 * lra-constraints.c (curr_insn_transform): Use non-input operand
10808 instead of output one for matched reload.
10810 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10812 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
10813 (tree-ssa-loop-ivopts.h): New header file.
10814 (struct builtin_info): New fields.
10815 (classify_builtin_1): Compute and record base and offset parts for
10816 memset builtin partition by calling strip_offset.
10817 (offset_cmp, fuse_memset_builtins): New functions.
10818 (finalize_partitions): Fuse adjacent memset partitions by calling
10820 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
10821 Expose the interface.
10822 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
10824 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10826 PR tree-optimization/82574
10827 * tree-loop-distribution.c (find_single_drs): New parameter. Check
10828 that data reference must be executed exactly once per iteration
10829 against the outermost loop in nest.
10830 (classify_partition): Update call to above function.
10832 2017-10-18 Richard Biener <rguenther@suse.de>
10834 PR tree-optimization/82591
10835 * graphite.c (graphite_transform_loops): Move code gen message
10837 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10838 Here. Handle scop_to_isl_ast failing.
10839 (scop_to_isl_ast): Limit the number of ISL operations.
10841 2017-10-18 Richard Biener <rguenther@suse.de>
10843 * graphite-isl-ast-to-gimple.c
10844 (translate_isl_ast_to_gimple::set_rename): Simplify.
10845 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
10846 (graphite_copy_stmts_from_block): ... here.
10847 (copy_bb_and_scalar_dependences): Simplify.
10848 (add_parameters_to_ivs_params): Canonicalize.
10849 (generate_entry_out_of_ssa_copies): Simplify.
10850 * graphite-sese-to-poly.c (extract_affine_name): Simplify
10851 by passing in ISL dimension.
10852 (parameter_index_in_region_1): Rename to ...
10853 (parameter_index_in_region): ... this.
10854 (extract_affine): Adjust assert, pass down parameter index.
10855 (add_param_constraints): Use range-info when available.
10856 (build_scop_context): Adjust.
10857 * sese.c (new_sese_info): Adjust.
10858 (free_sese_info): Likewise.
10859 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
10860 Remove unused typedefs.
10861 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
10863 2017-10-18 Martin Liska <mliska@suse.cz>
10865 * combine.c (simplify_compare_const): Add gcc_fallthrough.
10867 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10869 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
10870 (s390_sched_init): Do not reset s390_sched_state if we entered the
10871 current basic block via a fallthru edge and all others are unlikely.
10873 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10875 * config/s390/s390.c (NUM_SIDES): New variable.
10876 (LONGRUNNING_THRESHOLD): New variable.
10877 (LATENCY_FACTOR): New variable.
10878 (s390_sched_score): Decrease score for long-running instructions on
10880 (s390_sched_variable_issue): Perform bookkeeping for long-running
10883 2017-10-18 Richard Biener <rguenther@suse.de>
10885 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
10886 Simplify with removal of the parameter rename map.
10887 (set_rename): Likewise.
10888 (should_copy_to_new_region): Likewise.
10889 (graphite_copy_stmts_from_block): Likewise.
10890 (copy_bb_and_scalar_dependences): Remove initialization of
10891 unused copied_bb_map.
10892 (copy_def): Remove.
10893 (copy_internal_parameters): Likewise.
10894 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
10895 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
10896 Use INTEGRAL_TYPE_P.
10897 (parameter_index_in_region_1): Rename to ...
10898 (assign_parameter_index_in_region): ... this. Assert we have
10899 a parameter we handle.
10900 (scan_tree_for_params): Adjust.
10901 * sese.h (parameter_rename_map_t): Remove.
10902 (struct sese_info_t): Remove unused parameter_rename_map and
10903 copied_bb_map members.
10904 * sese.c (new_sese_info): Adjust.
10905 (free_sese_info): Likewise.
10907 2017-10-18 Martin Liska <mliska@suse.cz>
10910 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
10911 on an abnormal edge.
10913 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
10915 * doc/invoke.texi (ffunction-sections and fdata-sections):
10918 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10920 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
10921 the use statement can throw internally.
10923 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10925 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
10926 any RTX present on the RHS of a SET.
10927 * compare-elim.c (try_eliminate_compare): Restore comment.
10929 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10931 * langhooks.h (struct lang_hooks): Document that tree_size langhook
10932 may be also called on tcc_type nodes.
10933 * langhooks.c (lhd_tree_size): Likewise.
10935 2017-10-17 David Malcolm <dmalcolm@redhat.com>
10937 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
10938 format_warning_at_substring.
10939 (maybe_warn): Convert source_range * param to a location_t. Pass
10940 UNKNOWN_LOCATION rather than NULL to fmtwarn.
10941 (format_directive): Remove code to extract source_ranges and
10942 source_range * in favor of just a location_t.
10943 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
10945 * substring-locations.c (format_warning_va): Convert
10946 source_range * param to a location_t.
10947 (format_warning_at_substring): Likewise.
10948 * substring-locations.h (format_warning_va): Likewise.
10949 (format_warning_at_substring): Likewise.
10951 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
10953 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
10955 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
10956 and scatter/gather ops.
10958 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
10959 vec_gather_load and vec_scatter_store.
10960 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
10961 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
10962 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
10963 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
10964 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
10965 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
10967 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
10969 * reg-stack.c (compare_for_stack_reg): Add bool argument.
10970 Detect FTST instruction and handle its register pops. Only pop
10971 second operand if can_pop_second_op is true.
10972 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
10973 set can_pop_second_op to false in the compare_for_stack_reg call.
10975 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
10976 output_fp_compare for stack register operands.
10977 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
10978 instructions here. Do not emit stack register pops here. Assert
10979 that FCOMPP pops next to top stack register. Rewrite function.
10981 2017-10-17 Nathan Sidwell <nathan@acm.org>
10983 PR middle-end/82577
10984 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
10985 use DECL_ASSEMBLER_NAME_RAW.
10987 PR middle-end/82546
10988 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
10991 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
10992 Wilco Dijkstra <wilco.dijkstra@arm.com>
10994 * builtins.c (expand_builtin_update_setjmp_buf): Add a
10995 converstion to Pmode from the buf_addr.
10997 2017-10-17 Richard Biener <rguenther@suse.de>
10999 * graphite-dependences.c (scop_get_reads_and_writes): Change
11000 output parameters to references.
11002 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
11004 PR 71026/tree-optimization
11005 * fold-const.c (distribute_real_division): Removed.
11006 (fold_binary_loc): Remove calls to distribute_real_divison.
11008 2017-10-17 Richard Biener <rguenther@suse.de>
11010 * graphite-scop-detection.c
11011 (scop_detection::stmt_has_simple_data_refs_p): Always use
11012 the full nest as region.
11013 (try_generate_gimple_bb): Likewise.
11014 * sese.c (scalar_evolution_in_region): Simplify now that
11015 SCEV can handle instantiation in regions.
11016 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
11017 in the non-loop part of a function if requested.
11019 2017-10-17 Richard Biener <rguenther@suse.de>
11021 PR tree-optimization/82563
11022 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
11024 (graphite_regenerate_ast_isl): Call it.
11025 * graphite-scop-detection.c (build_scops): Remove entry edge split.
11027 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11029 PR tree-optimization/82549
11030 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
11031 Formatting fixes. Instead of calling make_bit_field_ref with negative
11034 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
11036 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
11037 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
11038 _mm_maskz_reduce_ss): New.
11039 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
11040 __builtin_ia32_reducess_mask): Ditto..
11041 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
11042 * config/i386/sse.md (reduces<mode>): Renamed to ...
11043 (reduces<mode><mask_scalar_name>): ... this.
11044 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
11046 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
11047 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
11049 2017-10-16 David Malcolm <dmalcolm@redhat.com>
11051 * Makefile.in (OBJS): Add unique-ptr-tests.o.
11052 * selftest-run-tests.c (selftest::run_tests): Call
11053 selftest::unique_ptr_tests_cc_tests.
11054 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
11055 * unique-ptr-tests.cc: New file.
11057 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
11060 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11062 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11063 (make_hard_regno_born, make_hard_regno_dead): Update
11064 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
11066 2017-10-16 Jeff Law <law@redhat.com>
11068 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11070 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
11072 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
11074 2017-10-16 Olivier Hainque <hainque@adacore.com>
11076 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
11077 with_cpu if we were configured for an e500v2 target cpu name.
11079 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
11081 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
11082 * doc/invoke.texi: Document +nodsp as a valid extension for
11085 2017-10-16 Martin Liska <mliska@suse.cz>
11087 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
11088 (test_set_range): Likewise.
11089 (test_range_functions): Rename to ...
11090 (test_bit_in_range): ... this.
11091 (sbitmap_c_tests): Add new test.
11093 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11095 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
11097 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
11098 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
11100 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11102 * config/aarch64/aarch64-builtins.c
11103 (aarch64_types_quadopu_lane_qualifiers): New.
11104 (TYPES_QUADOPU_LANE): New.
11105 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
11106 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
11107 (aarch64_<sur>dot_laneq<vsi2qi>): New.
11108 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
11109 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
11110 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
11111 (Vdottype, DOTPROD): New.
11112 (sur): Add SDOT and UDOT.
11114 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11116 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
11117 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
11118 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
11119 Add TARGET_DOTPROD.
11120 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
11121 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
11122 Enable TARGET_DOTPROD.
11123 (cortex-a75.cortex-a55): Likewise.
11124 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
11126 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11128 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
11129 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
11131 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
11133 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
11134 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
11135 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
11136 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
11137 * config/arm/types.md (neon_dot, neon_dot_q): New.
11138 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
11140 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11142 * config/arm/arm.h (TARGET_DOTPROD): New.
11143 * config/arm/arm.c (arm_arch_dotprod): New.
11144 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
11145 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
11146 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
11147 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
11148 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
11149 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
11150 * doc/invoke.texi (armv8.2-a): Document dotprod
11152 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11154 * i386.c (ix86_vec_cost): New function.
11155 (ix86_rtx_costs): Handle vector operations better.
11156 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
11157 * x86-tune-costs.h: Add new costs to all tables.
11159 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11161 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
11163 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
11164 divsd, sqrtss and sqrtsd
11165 * x86-tune-costs.h: Add new entries to all costs.
11166 (znver1_cost): Fix to match real instruction latencies.
11168 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11169 Michael Collison <michael.collison@arm.com>
11171 * compare-elim.c: Include emit-rtl.h.
11172 (can_merge_compare_into_arith): New function.
11173 (try_validate_parallel): Likewise.
11174 (try_merge_compare): Likewise.
11175 (try_eliminate_compare): Call the above when no previous clobber
11177 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
11180 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11182 PR middle-end/62263
11183 PR middle-end/82498
11184 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
11185 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
11187 PR middle-end/62263
11188 PR middle-end/82498
11189 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
11190 to be any operand_equal_p operands. For & (B - 1) require
11191 B to be power of 2. Recognize
11192 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
11194 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11197 * optabs.c (expand_memory_blockage): Fix call of
11198 targetm.have_memory_blockage.
11200 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11203 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
11204 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
11205 objects to extra_objs instead of overwriting it.
11207 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11209 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
11210 Use any_fp_register_operand as operand[3] predicate. Simplify
11211 equality test for operands[2] and operands[4] memory location.
11212 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
11213 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
11214 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
11215 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
11216 any_fp_register_operand as operand[1] predicate. Simplify
11217 equality test for operands[0] and operands[3] memory location.
11218 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
11219 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
11220 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
11222 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11224 * target-insns.def: Add memory_blockage.
11225 * optabs.c (expand_memory_blockage): New function.
11226 (expand_asm_memory_barrier): Rename ...
11227 (expand_asm_memory_blockage): ... to this.
11228 (expand_mem_thread_fence): Call expand_memory_blockage
11229 instead of expand_asm_memory_barrier.
11230 (expand_mem_singnal_fence): Ditto.
11231 (expand_atomic_load): Ditto.
11232 (expand_atomic_store): Ditto.
11233 * doc/md.texi (Standard Pattern Names For Generation):
11234 Document memory_blockage instruction pattern.
11236 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
11238 * config/rl78/rl78.c (rl78_emit_libcall): New function.
11239 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
11240 * config/rl78/rl78.md: New define_expand "adddi3".
11242 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11244 * cfghooks.c (verify_flow_info): Disable check that all probabilities
11247 2017-10-13 Jeff Law <law@redhat.com>
11249 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
11251 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11254 * internal-fn.c (expand_mul_overflow): If both operands have
11255 the same highpart of -1 or 0 and the topmost bit of lowpart
11256 is different, overflow is if res <= 0 rather than res < 0.
11258 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
11260 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
11261 TARGET_P9_VECTOR code for unaligned_load case.
11263 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11265 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
11267 2017-10-13 Nathan Sidwell <nathan@acm.org>
11269 * tree-core.h (tree_contains_struct): Make bool.
11270 * tree.c (tree_contains_struct): Likewise.
11271 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
11272 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
11273 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
11274 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
11275 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
11277 2017-10-13 Richard Biener <rguenther@suse.de>
11279 * graphite-isl-ast-to-gimple.c
11280 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
11281 parameters and dominance check.
11282 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
11283 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
11284 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
11285 Do not update SSA form here or do intermediate IL verification.
11286 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
11287 (graphite_initialize): Remove check on the number of loops in
11288 the function and inline into graphite_transform_loops.
11289 (graphite_finalize): Inline into graphite_transform_loops.
11290 (graphite_transform_loops): Perform SSA update and IL verification
11292 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
11294 2017-10-13 Richard Biener <rguenther@suse.de>
11296 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
11297 graphite_expression_type_precision): Avoid global constructor
11299 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
11300 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
11301 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
11302 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
11303 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
11304 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
11306 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
11309 * config/i386/i386.h (ix86_red_zone_size): New.
11310 * config/i386/i386.md (push peephole2s): Replace
11311 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
11313 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11314 Alan Hayward <alan.hayward@arm.com>
11315 David Sherwood <david.sherwood@arm.com>
11317 * combine.c (can_change_dest_mode): Reject changes in
11318 REGMODE_NATURAL_SIZE.
11320 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11321 Alan Hayward <alan.hayward@arm.com>
11322 David Sherwood <david.sherwood@arm.com>
11324 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
11325 (expand_debug_source_expr): Likewise.
11326 * combine.c (combine_simplify_rtx): Likewise.
11327 * cse.c (fold_rtx): Likewise.
11328 * fwprop.c (canonicalize_address): Likewise.
11329 * targhooks.c (default_shift_truncation_mask): Likewise.
11331 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11332 Alan Hayward <alan.hayward@arm.com>
11333 David Sherwood <david.sherwood@arm.com>
11335 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
11336 (widened_mode): Likewise.
11337 (expand_unop): Likewise.
11338 * ree.c (transform_ifelse): Likewise.
11339 (merge_def_and_ext): Likewise.
11340 (combine_reaching_defs): Likewise.
11341 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11343 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11344 Alan Hayward <alan.hayward@arm.com>
11345 David Sherwood <david.sherwood@arm.com>
11347 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
11348 * combine.c (gen_lowpart_for_combine): Likewise.
11349 * dwarf2out.c (rtl_for_decl_location): Likewise.
11350 * final.c (alter_subreg): Likewise.
11351 * rtlhooks.c (gen_lowpart_general): Likewise.
11352 (gen_lowpart_if_possible): Likewise.
11354 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11355 Alan Hayward <alan.hayward@arm.com>
11356 David Sherwood <david.sherwood@arm.com>
11358 * calls.c (expand_call): Use subreg_lowpart_offset.
11359 * cse.c (cse_insn): Likewise.
11360 * regcprop.c (copy_value): Likewise.
11361 (copyprop_hardreg_forward_1): Likewise.
11363 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11366 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
11367 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
11368 =Q constraints to +Q and into insn condition add check
11369 that operands[0] and operands[1] are equal.
11370 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
11371 =Q constraints to +Q and into insn condition add check
11372 that operands[0] is equal to either operands[1] or operands[2].
11375 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
11376 instead of handling MINUS_EXPR twice (once for each argument),
11377 canonicalize operand order and handle just once, use rtype where
11378 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
11381 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
11382 any values of __C while still being pattern recognizable as a simple
11383 rotate instruction.
11385 2017-10-13 Richard Biener <rguenther@suse.de>
11387 PR tree-optimization/82451
11389 2017-10-02 Richard Biener <rguenther@suse.de>
11391 PR tree-optimization/82355
11392 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11393 a mapping for the enclosing loop but avoid generating one for
11394 the loop tree root.
11395 (copy_bb_and_scalar_dependences): Remove premature codegen
11396 error on PHIs in blocks duplicated into multiple places.
11397 * graphite-scop-detection.c
11398 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11399 in the region use it as loop and nest to analyze the DR in.
11400 (try_generate_gimple_bb): Likewise.
11401 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11402 (add_loop_constraints): For blocks in a loop not in the region
11403 create a dimension with a single iteration.
11404 * sese.h (gbb_loop_at_index): Remove assert.
11406 * cfgloop.c (loop_preheader_edge): For the loop tree root
11407 return the single successor of the entry block.
11408 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11409 Reset the SCEV hashtable and niters.
11410 * graphite-scop-detection.c
11411 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
11412 assert that we only have POLYNOMIAL_CHREC that vary in loops
11413 contained in the region.
11414 (scop_detection::graphite_can_represent_expr): Adjust.
11415 (scop_detection::stmt_has_simple_data_refs_p): For loops
11416 not in the region set loop to NULL. The nest is now the
11417 entry edge to the region.
11418 (try_generate_gimple_bb): Likewise.
11419 * sese.c (scalar_evolution_in_region): Adjust for
11420 instantiate_scev change.
11421 * tree-data-ref.h (graphite_find_data_references_in_stmt):
11422 Make nest parameter the edge into the region.
11423 (create_data_ref): Likewise.
11424 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
11425 entry edge into a region and adjust instantiate_scev calls.
11426 (create_data_ref): Likewise.
11427 (graphite_find_data_references_in_stmt): Likewise.
11428 (find_data_references_in_stmt): Pass the loop preheader edge
11429 from the nest argument.
11430 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
11431 parameter the edge into the region.
11432 (instantiate_parameters): Use the loop preheader edge as entry.
11433 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
11435 (get_instantiated_value_entry): Make instantiate_below parameter
11436 the edge into the region.
11437 (instantiate_scev_name): Likewise. Adjust dominance checks,
11438 when we cannot use loop-based instantiation instantiate by
11439 walking use-def chains.
11440 (instantiate_scev_poly): Adjust.
11441 (instantiate_scev_binary): Likewise.
11442 (instantiate_scev_convert): Likewise.
11443 (instantiate_scev_not): Likewise.
11444 (instantiate_array_ref): Remove.
11445 (instantiate_scev_3): Likewise.
11446 (instantiate_scev_2): Likewise.
11447 (instantiate_scev_1): Likewise.
11448 (instantiate_scev_r): Do not blindly handle N-operand trees.
11449 Do not instantiate array-refs. Handle all constants and invariants.
11450 (instantiate_scev): Make instantiate_below parameter
11451 the edge into the region.
11452 (resolve_mixers): Use the loop preheader edge for the region
11453 parameter to instantiate_scev_r.
11454 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
11456 2017-10-13 Richard Biener <rguenther@suse.de>
11458 PR tree-optimization/82525
11459 * graphite-isl-ast-to-gimple.c
11460 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
11462 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
11463 Fail code generation when we cannot represent the isl integer.
11464 (binary_op_to_tree): Elide modulo operations that are no-ops
11465 in the type we code generate. Remove now superfluous code
11468 2017-10-13 Richard Biener <rguenther@suse.de>
11470 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
11471 (scop_detection::harmful_loop_in_region): Remove premature
11472 IV type restriction.
11473 (scop_detection::graphite_can_represent_scev): We can handle
11474 pointer IVs just fine.
11476 2017-10-13 Alan Modra <amodra@gmail.com>
11478 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
11479 "Clobbers and Scratch Registers". Add paragraph on
11480 alternative to clobbers for scratch registers and OpenBLAS
11483 2017-10-13 Alan Modra <amodra@gmail.com>
11485 * doc/extend.texi (Clobbers): Correct vax example. Delete old
11486 example of a memory input for a string of known length. Move
11487 commentary out of table. Add a number of new examples
11488 covering array memory inputs.
11490 2017-10-12 Martin Liska <mliska@suse.cz>
11492 PR tree-optimization/82493
11493 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
11494 (test_range_functions): New function.
11495 (sbitmap_c_tests): Likewise.
11496 * selftest-run-tests.c (selftest::run_tests): Run new tests.
11497 * selftest.h (sbitmap_c_tests): New function.
11499 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11501 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11503 * config/rs6000/amo.h: Fix spacing issue.
11505 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11508 * config/i386/i386.md (*ashl<mode>3_mask_1,
11509 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
11510 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
11513 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11515 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
11516 (profile_probability): Set max_probability
11517 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
11519 * profile-count.c (profile_probability::differs_from_p): Do not
11520 rely on max_probaiblity == 10000
11522 2017-10-12 Jeff Law <law@redhat.com>
11524 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
11527 2017-10-12 Martin Sebor <msebor@redhat.com>
11531 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
11532 (handle_alias_pairs): Call it.
11533 * common.opt (-Wattribute-alias): New option.
11534 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
11535 * doc/invoke.texi (-Wattribute-alias): Document.
11537 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
11540 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11542 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11544 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11545 (make_hard_regno_born, make_hard_regno_dead): Update
11546 bb_killed_pseudos and bb_gen_pseudos.
11548 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11550 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
11552 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
11554 * config/alpha/alpha.c (alpha_split_conditional_move):
11555 Use std::swap instead of manually swapping.
11556 (alpha_stdarg_optimize_hook): Ditto.
11557 (alpha_canonicalize_comparison): Ditto.
11559 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11561 * tree-loop-distribution.c (struct builtin_info): New struct.
11562 (struct partition): Refactor fields into struct builtin_info.
11563 (partition_free): Free struct builtin_info.
11564 (build_size_arg_loc, build_addr_arg_loc): Delete.
11565 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
11566 information from struct builtin_info.
11567 (find_single_drs): New function refactored from classify_partition.
11568 Also moved builtin validity checks to this function.
11569 (compute_access_range, alloc_builtin): New functions.
11570 (classify_builtin_st, classify_builtin_ldst): New functions.
11571 (classify_partition): Refactor code into functions find_single_drs,
11572 classify_builtin_st and classify_builtin_ldst.
11573 (distribute_loop): Don't do runtime alias check when distributing
11575 (find_seed_stmts_for_distribution): New function.
11576 (pass_loop_distribution::execute): Refactor code finding seed
11577 stmts into above function. Support distribution for the innermost
11578 two-level loop nest. Adjust dump information.
11580 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11582 * tree-loop-distribution.c: Adjust the general comment.
11583 (NUM_PARTITION_THRESHOLD): New macro.
11584 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
11585 (classify_partition): Skip builtin pattern of loop nest's inner loop.
11586 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
11587 in call to build_partition_graph.
11588 (finalize_partitions): New parameter. Make loop distribution more
11589 conservative by fusing more partitions.
11590 (distribute_loop): Don't do runtime alias check in case of loop nest
11592 (find_seed_stmts_for_distribution): New function.
11593 (prepare_perfect_loop_nest): New function.
11594 (pass_loop_distribution::execute): Refactor code finding seed stmts
11595 and loop nest into above functions. Support loop nest distribution.
11596 Adjust dump information accordingly.
11598 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11600 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
11601 and set PTYPE_SEQUENTIAL for merged partition.
11603 2017-10-12 Richard Biener <rguenther@suse.de>
11605 PR tree-optimization/69728
11607 2017-09-19 Richard Biener <rguenther@suse.de>
11609 PR tree-optimization/69728
11610 * graphite-sese-to-poly.c (schedule_error): New global.
11611 (add_loop_schedule): Handle empty domain by failing the
11613 (build_original_schedule): Handle schedule_error.
11615 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
11616 domain by returning an unchanged schedule.
11618 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11620 * genrecog.c (validate_pattern): For VEC_SELECT verify that
11621 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
11623 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
11625 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
11628 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11631 * expr.c (store_field): Don't optimize away bitsize == 0 store
11632 from CALL_EXPR with addressable return type.
11634 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11636 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
11637 * config/rs6000/rs6000.md (sel): Delete mode attribute.
11638 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
11639 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
11640 TARGET_ISEL instead of TARGET_ISEL<sel>.
11642 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
11644 * config/rs6000/rs6000.c
11645 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
11647 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11649 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
11651 (reg_or_cint_operand): Fix comment.
11652 (reg_or_zero_operand): New predicate.
11653 * config/rs6000/rs6000-protos.h (output_isel): Delete.
11654 * config/rs6000/rs6000.c (output_isel): Delete.
11655 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
11656 instead of reg_or_cint_operand. Output instruction directly (not via
11658 (isel_unsigned_<mode>): Ditto.
11659 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
11660 gpc_reg_operand. Add an instruction alternative for this. Output
11661 instruction directly.
11662 (*isel_reversed_unsigned_<mode>): Ditto.
11664 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11666 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
11667 (TARGET_CANONICALIZE_COMPARISON): Define.
11669 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
11672 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
11673 Check whether the dest is REG before adding REG_EQUIV note.
11675 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11678 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11680 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11681 (make_hard_regno_born, make_hard_regno_dead): Update
11682 bb_killed_pseudos and bb_gen_pseudos.
11684 2017-10-11 Nathan Sidwell <nathan@acm.org>
11686 * incpath.h (enum incpath_kind): Name enum, prefix values.
11687 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
11688 * incpath.c (heads, tails): Use INC_MAX.
11689 (add_env_var_paths, add_standard_paths): Use incpath_kind.
11690 (merge_include_chains, split_quote_chain,
11691 register_include_chains): Update incpath_kind names.
11692 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
11693 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
11695 (add_framework_path, darwin_register_objc_includes): Likewise.
11696 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
11698 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11700 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
11701 Do not use float_operator operator predicate.
11702 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
11703 * config/i386/predicates.md (float_operator): Remove predicate.
11705 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11707 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
11708 (*jccxf_i387): Ditto.
11709 (*jcc<mode>_i387): Ditto.
11710 (*jccu<mode>_i387): Ditto.
11711 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
11712 (*jcc_*_i387 splitters): Remove.
11713 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
11714 * config/i386/i386.c (ix86_split_fp_branch): Remove.
11715 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
11718 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11720 * profile-count.h (slow_safe_scale_64bit): New function.
11721 (safe_scale_64bit): New inline.
11722 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
11723 * profile-count.c: Include wide-int.h
11724 (slow_safe_scale_64bit): New.
11726 2017-10-11 Nathan Sidwell <nathan@acm.org>
11728 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
11729 HAS_DECL_ASSEMBLER_NAME_P.
11730 * gimple-expr.c (gimple_decl_printable_name: Check
11731 HAS_DECL_ASSEMBLER_NAME_P too.
11732 * ipa-utils.h (type_in_anonymous_namespace_p): Check
11733 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
11734 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
11735 * passes.c (rest_of_decl_compilation): Check
11736 HAS_DECL_ASSEMBLER_NAME_P too.
11737 * recog.c (verify_changes): Likewise.
11738 * tree-pretty-print.c (dump_decl_name): Likewise.
11739 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
11741 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
11742 (SET_DECL_ASSEMBLER_NAME): Use it.
11743 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
11744 (COPY_DECL_ASSEMBLER_NAME): Likewise.
11745 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
11747 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11749 * config.gcc (i386, x86_64): Add extra objects.
11750 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
11751 (ix86_min_insn_size): Declare.
11752 (ix86_issue_rate): Declare.
11753 (ix86_adjust_cost): Declare.
11754 (ia32_multipass_dfa_lookahead): Declare.
11755 (ix86_macro_fusion_p): Declare.
11756 (ix86_macro_fusion_pair_p): Declare.
11757 (ix86_bd_has_dispatch): Declare.
11758 (ix86_bd_do_dispatch): Declare.
11759 (ix86_core2i7_init_hooks): Declare.
11760 (ix86_atom_sched_reorder): Declare.
11761 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
11762 (COSTS_N_BYTES): Move to x86-tune-costs.h.
11763 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
11764 (rip_relative_addr_p): Rename to ...
11765 (ix86_rip_relative_addr_p): ... this one; export.
11766 (memory_address_length): Update.
11767 (ix86_issue_rate): Move to x86-tune-sched.c.
11768 (ix86_flags_dependent): Move to x86-tune-sched.c.
11769 (ix86_agi_dependent): Move to x86-tune-sched.c.
11770 (exact_dependency_1): Move to x86-tune-sched.c.
11771 (exact_store_load_dependency): Move to x86-tune-sched.c.
11772 (ix86_adjust_cost): Move to x86-tune-sched.c.
11773 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
11774 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
11775 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
11776 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
11777 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
11778 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
11779 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
11780 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
11781 (min_insn_size): Rename to ...
11782 (ix86_min_insn_size): ... this one; export.
11783 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
11784 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
11785 (core2i7_first_cycle_multipass_backtrack): Move to
11786 x86-tune-sched-core.c.
11787 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
11788 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
11789 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
11790 (ix86_avoid_jump_mispredicts): Update.
11791 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
11792 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
11793 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
11794 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
11795 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
11796 (MAX_INSN): Move to ix86-tune-sched-bd.c.
11797 (MAX_IMM): Move to ix86-tune-sched-bd.c.
11798 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
11799 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
11800 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
11801 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
11802 (MAX_STORE): Move to ix86-tune-sched-bd.c.
11803 (BIG): Move to ix86-tune-sched-bd.c.
11804 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
11805 (enum insn_path): Move to ix86-tune-sched-bd.c.
11806 (get_mem_group): Move to ix86-tune-sched-bd.c.
11807 (is_cmp): Move to ix86-tune-sched-bd.c.
11808 (dispatch_violation): Move to ix86-tune-sched-bd.c.
11809 (is_branch): Move to ix86-tune-sched-bd.c.
11810 (is_prefetch): Move to ix86-tune-sched-bd.c.
11811 (init_window): Move to ix86-tune-sched-bd.c.
11812 (allocate_window): Move to ix86-tune-sched-bd.c.
11813 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
11814 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
11815 (process_end_window): Move to ix86-tune-sched-bd.c.
11816 (allocate_next_window): Move to ix86-tune-sched-bd.c.
11817 (find_constant): Move to ix86-tune-sched-bd.c.
11818 (get_num_immediates): Move to ix86-tune-sched-bd.c.
11819 (has_immediate): Move to ix86-tune-sched-bd.c.
11820 (get_insn_path): Move to ix86-tune-sched-bd.c.
11821 (get_insn_group): Move to ix86-tune-sched-bd.c.
11822 (count_num_restricted): Move to ix86-tune-sched-bd.c.
11823 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
11824 (add_insn_window): Move to ix86-tune-sched-bd.c.
11825 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
11826 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
11827 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
11828 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
11829 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
11830 (do_dispatch): Move to ix86-tune-sched-bd.c.
11831 (has_dispatch): Move to ix86-tune-sched-bd.c.
11832 * config/i386/t-i386: Add new object files.
11833 * config/i386/x86-tune-costs.h: New file.
11834 * config/i386/x86-tune-sched-atom.c: New file.
11835 * config/i386/x86-tune-sched-bd.c: New file.
11836 * config/i386/x86-tune-sched-core.c: New file.
11837 * config/i386/x86-tune-sched.c: New file.
11839 2017-10-11 Liu Hao <lh_mouse@126.com>
11841 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
11842 the generic version below instead.
11843 (should_colorize): Recognize Windows consoles as terminals
11845 * pretty-print.c [__MINGW32__] (write_all): New function.
11846 [__MINGW32__] (find_esc_head): Likewise.
11847 [__MINGW32__] (find_esc_terminator): Likewise.
11848 [__MINGW32__] (eat_esc_sequence): Likewise.
11849 [__MINGW32__] (mingw_ansi_fputs): New function that handles
11851 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
11854 2017-10-11 Richard Biener <rguenther@suse.de>
11856 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
11857 Properly call analyze_scalar_evolution with the loop of the stmt.
11859 2017-10-11 Richard Biener <rguenther@suse.de>
11861 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
11862 * tree-core.h (tree_base): Add chrec_var union member.
11863 * tree.h (CHREC_VAR): Remove.
11864 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
11865 * tree-chrec.h (build_polynomial_chrec): Adjust.
11866 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
11867 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
11869 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
11871 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
11872 * match.pd: ... here.
11873 ((T) X == (T) Y): Relax condition.
11875 2017-10-11 Bin Cheng <bin.cheng@arm.com>
11877 PR tree-optimization/82472
11878 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
11880 (break_alias_scc_partitions): Update postorder number.
11882 2017-10-11 Martin Liska <mliska@suse.cz>
11885 * opts.c (parse_no_sanitize_attribute): Do not use error_value
11887 * opts.h (parse_no_sanitize_attribute): Remove last argument.
11889 2017-10-11 Martin Liska <mliska@suse.cz>
11891 * print-rtl.c (print_insn): Move declaration of idbuf
11892 to same scope as name.
11894 2017-10-11 Martin Liska <mliska@suse.cz>
11899 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11900 volatile arguments.
11902 2017-10-11 Martin Liska <mliska@suse.cz>
11905 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11906 volatile arguments.
11908 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
11910 * config.gcc (default_gnu_indirect_function): Default to yes for
11911 arm*-*-linux* with glibc.
11913 2017-10-11 Richard Biener <rguenther@suse.de>
11915 * tree-scalar-evolution.c (get_scalar_evolution): Handle
11916 default-defs and types we do not want to analyze.
11917 (interpret_loop_phi): Replace unreachable code with an assert.
11918 (compute_scalar_evolution_in_loop): Remove and inline ...
11919 (analyze_scalar_evolution_1): ... here, replacing condition with
11920 what makes the intent clearer. Remove handling of cases
11921 get_scalar_evolution now handles.
11923 2017-10-10 Jim Wilson <wilson@tuliptree.org>
11925 PR rtl-optimization/81434
11926 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
11927 comment for main loop. In sched_group_found if, also add checks for
11928 pass and min_cost_group.
11930 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
11932 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
11933 (rs6000_insn_cost): New function.
11934 * config/rs6000/rs6000.md (cost): New attribute.
11936 2017-10-10 Jakub Jelinek <jakub@redhat.com>
11937 H.J. Lu <hongjiu.lu@intel.com>
11941 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
11942 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
11943 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
11944 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
11945 to OPTION_MASK_ISA_AVX512VL - builtins that have both
11946 OPTION_MASK_ISA_MMX and some other bit set require both
11947 mmx and the ISAs without the mmx bit.
11948 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
11949 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
11950 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
11951 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
11952 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
11953 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
11954 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
11955 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
11956 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
11957 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
11958 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
11959 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
11960 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
11961 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
11962 Add OPTION_MASK_ISA_MMX.
11964 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
11966 * config.gcc (armv7*-*-freebsd*): New target.
11967 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
11969 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
11971 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
11972 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
11975 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
11977 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
11978 a template parameter.
11979 (WIDE_INT_REF_FOR): Update accordingly.
11980 * tree.h (wi::int_traits <const_tree>): Delete.
11981 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
11982 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
11983 (wi::tree_to_wide_ref): New typedef.
11984 (wi::to_wide): New function.
11985 * calls.c (get_size_range): Use wi::to_wide when operating on
11986 trees as wide_ints.
11987 * cgraph.c (cgraph_node::create_thunk): Likewise.
11988 * config/i386/i386.c (ix86_data_alignment): Likewise.
11989 (ix86_local_alignment): Likewise.
11990 * dbxout.c (stabstr_O): Likewise.
11991 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
11992 * expr.c (const_vector_from_tree): Likewise.
11993 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
11994 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
11995 (fold_negate_expr_1, int_const_binop_1, const_binop)
11996 (fold_convert_const_int_from_real, optimize_bit_field_compare)
11997 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
11998 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
11999 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
12000 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
12001 (fold_not_const, round_up_loc): Likewise.
12002 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
12003 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
12004 (alloca_call_type): Likewise.
12005 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
12006 * godump.c (go_output_typedef): Likewise.
12007 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
12008 * internal-fn.c (get_min_precision): Likewise.
12009 * ipa-cp.c (ipcp_store_vr_results): Likewise.
12010 * ipa-polymorphic-call.c
12011 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
12012 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
12013 (ipa_modify_call_arguments): Likewise.
12014 * match.pd: Likewise.
12015 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
12016 * print-tree.c (print_node_brief, print_node): Likewise.
12017 * stmt.c (expand_case): Likewise.
12018 * stor-layout.c (layout_type): Likewise.
12019 * tree-affine.c (tree_to_aff_combination): Likewise.
12020 * tree-cfg.c (group_case_labels_stmt): Likewise.
12021 * tree-data-ref.c (dr_analyze_indices): Likewise.
12022 (prune_runtime_alias_test_list): Likewise.
12023 * tree-dump.c (dequeue_and_dump): Likewise.
12024 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
12025 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
12026 * tree-pretty-print.c (dump_generic_node): Likewise.
12027 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
12028 (simple_iv_with_niters): Likewise.
12029 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
12030 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
12031 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
12032 * tree-ssa-loop-niter.c (split_to_var_and_offset)
12033 (refine_value_range_using_guard, number_of_iterations_ne_max)
12034 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
12035 (get_cst_init_from_scev, record_nonwrapping_iv)
12036 (scev_var_range_cant_overflow): Likewise.
12037 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
12038 * tree-ssa-pre.c (compute_avail): Likewise.
12039 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
12040 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
12041 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
12042 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
12043 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
12044 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
12045 (dump_case_nodes, try_switch_expansion): Likewise.
12046 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
12047 (vect_do_peeling): Likewise.
12048 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
12049 * tree-vect-stmts.c (vectorizable_load): Likewise.
12050 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
12051 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
12052 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
12053 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
12054 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
12055 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
12056 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
12057 (evrp_dom_walker::before_dom_children): Likewise.
12058 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
12059 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
12060 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
12061 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
12062 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
12063 (get_range_pos_neg): Likewise.
12064 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
12065 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
12066 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
12067 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
12068 * config/avr/avr.c (avr_fold_builtin): Likewise.
12069 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
12070 * config/msp430/msp430.c (msp430_attr): Likewise.
12071 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
12072 * config/powerpcspe/powerpcspe-c.c
12073 (altivec_resolve_overloaded_builtin): Likewise.
12074 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
12075 (rs6000_expand_ternop_builtin): Likewise.
12076 * config/rs6000/rs6000-c.c
12077 (altivec_resolve_overloaded_builtin): Likewise.
12078 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
12079 (rs6000_expand_ternop_builtin): Likewise.
12080 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
12082 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12084 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
12085 when copying loop nest with only one inner loop.
12087 2017-10-10 Richard Biener <rguenther@suse.de>
12089 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
12090 blocks if SCEV is active.
12091 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
12093 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
12094 (scev_initialize): Assert we are not yet initialized.
12096 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12098 * tree-loop-distribution.c (generate_loops_for_partition): Remove
12099 inner loop's exit stmt by making it always exit the loop, otherwise
12100 we would generate an infinite empty loop.
12102 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12104 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
12105 renaming variables in new preheader if it's deleted.
12107 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12109 * tree-loop-distribution.c (struct partition): Remove unused field
12110 loops of the structure.
12111 (partition_alloc, partition_free): Ditto.
12112 (build_rdg_partition_for_vertex): Ditto.
12114 2017-10-09 Jeff Law <law@redhat.com>
12116 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
12117 return type to match prototype and documentation.
12119 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12121 * config/rs6000/rs6000.c (processor_costs): Move to ...
12122 * config/rs6000/rs6000.h: ... here.
12123 (rs6000_cost): Declare.
12125 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
12127 * except.c (setjmp_fn): New global variable.
12128 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
12129 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
12130 if DONT_USE_BUILTIN_SETJMP is defined.
12132 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12134 * target.def (insn_cost): New hook.
12135 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
12136 * doc/tm.texi: Regenerate.
12137 * rtlanal.c (insn_cost): Use the new hook.
12139 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12141 * combine.c (combine_validate_cost): Compute the new insn_cost,
12142 not just pattern_cost.
12143 (try_combine): Adjust comment.
12145 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12147 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
12149 * combine.c (uid_insn_cost): Adjust comment.
12150 (combine_validate_cost): Adjust comment. Use pattern_cost instead
12152 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
12153 * dse.c (find_shift_sequence): Ditto.
12154 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
12155 (bb_valid_for_noce_process_p): Use pattern_cost.
12156 * rtl.h (insn_rtx_cost): Delete.
12157 (pattern_cost): New prototype.
12158 (insn_cost): New prototype.
12159 * rtlanal.c (insn_rtx_cost): Rename to...
12160 (pattern_cost): ... this.
12163 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
12165 * config/i386/i386.md (*jcc_2): Remove insn pattern.
12166 (*jcc<mode>_0_r_i387): Ditto.
12167 (*jccxf_r_i387): Ditto.
12168 (*jcc<mode>_r_i387): Ditto.
12169 (*jccu<mode>_r_i387): Ditto.
12170 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
12171 (*jcc): Rename from *jcc_1.
12173 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12175 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
12176 deferred rescans after the lvx/stvx recombination pre-pass.
12178 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
12180 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
12181 memory operation instruction support.
12182 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
12183 (rs6000-ibm-aix[789]*): Likewise.
12184 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
12185 Document new functions.
12187 2017-10-09 Richard Biener <rguenther@suse.de>
12189 PR tree-optimization/82397
12190 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
12191 equality only for semantically equal trees.
12193 2017-10-09 Richard Biener <rguenther@suse.de>
12195 PR tree-optimization/82449
12196 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
12197 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
12198 allow constant addresses.
12199 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
12202 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12204 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
12207 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12210 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
12213 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12216 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
12218 2017-10-09 Jakub Jelinek <jakub@redhat.com>
12221 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
12222 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
12224 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
12226 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
12227 (WI_BINARY_PREDICATE_RESULT): Likewise.
12228 (wi::binary_traits::operator_result): New type.
12229 (wi::binary_traits::predicate_result): Likewise.
12230 (generic_wide_int::operator~, unary generic_wide_int::operator-)
12231 (generic_wide_int::operator==, generic_wide_int::operator!=)
12232 (generic_wide_int::operator&, generic_wide_int::and_not)
12233 (generic_wide_int::operator|, generic_wide_int::or_not)
12234 (generic_wide_int::operator^, generic_wide_int::operator+
12235 (binary generic_wide_int::operator-, generic_wide_int::operator*):
12237 (operator~, unary operator-, operator==, operator!=, operator&)
12238 (operator|, operator^, operator+, binary operator-, operator*): New
12240 * expr.c (get_inner_reference): Use wi::bit_and_not.
12241 * fold-const.c (fold_binary_loc): Likewise.
12242 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
12243 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
12244 (bit_value_binop): Likewise.
12245 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
12246 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
12247 (extract_range_from_binary_expr_1): Likewise.
12248 (masked_increment): Likewise.
12249 (simplify_bit_ops_using_ranges): Likewise.
12251 2017-10-09 Martin Jambor <mjambor@suse.cz>
12254 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
12255 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
12256 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
12258 (hsa_fixup_mov_insn_type): New function.
12259 (hsa_op_with_type::get_in_type): Use it.
12260 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
12261 immediates in an assert.
12262 (hsa_op_with_type::extend_int_to_32bit): New method.
12263 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
12264 types. Convert to dest type if necessary.
12265 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
12266 (reg_for_gimple_ssa): Pass false as min32int to
12267 hsa_type_for_scalar_tree_type.
12268 (gen_hsa_addr): Fixup type when creating addresable temporary.
12269 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
12270 (gen_hsa_unary_operation): Extend operands and convert to dest type if
12271 necessary. Call hsa_fixup_mov_insn_type.
12272 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
12273 extend operands and convert to dest type if necessary.
12274 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
12275 to dest type if necessary.
12276 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
12277 if conversion nt necessary and size matches.
12278 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
12279 to dest type if necessary.
12280 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
12281 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
12283 (gen_hsa_clrsb): Likewise.
12284 (gen_hsa_ffs): Likewise.
12285 (gen_hsa_divmod): Extend operands and convert to dest type if
12287 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
12289 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12291 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
12292 Remove empty default arguments. Use a brace block as output
12294 (conditional return): Ditto.
12296 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
12297 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
12299 (group_ending_nop): Ditto.
12300 (doloop_end): Ditto.
12301 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
12302 (splitters for those): Ditto.
12304 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12306 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
12307 a conditional jump (and the compare for it) so that pc_rtx is the
12309 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
12310 for the deleted and renamed ctr<mode>_internal[234] patterns.
12311 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
12312 Delete second conditional return pattern.
12313 (ctr<mode>_internal2): Delete this second bdnz pattern.
12314 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
12315 (ctr<mode>_internal4): Delete this second bdz pattern.
12317 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12319 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
12320 (always_initialized_rtx_for_ssa_name_p): New predicate.
12321 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
12322 (finish_out_of_ssa): Free new field of SA.
12323 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
12324 * tree-ssa-coalesce.c: Include tree-ssa.h.
12325 (get_parm_default_def_partitions): Remove extern keyword.
12326 (get_undefined_value_partitions): New function.
12327 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
12328 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
12329 uninitialized bits.
12330 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
12332 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12334 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
12336 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12338 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
12339 for targets that preffer 128bit.
12341 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12343 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
12345 2017-10-08 Olivier Hainque <hainque@adacore.com>
12347 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
12348 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
12350 (thumb_set_return_address): Likewise.
12352 2017-10-08 Olivier Hainque <hainque@adacore.com>
12354 * common/config/arm/arm-common.c (arm_except_unwind_info):
12355 Handle DWARF2_UNWIND_INFO.
12357 2017-10-07 Michael Collison <michael.collison@arm.com>
12359 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
12362 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
12364 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
12365 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
12366 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
12367 defined, force the creation of a new block for a dispatch label.
12369 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
12371 * invoke.texi (Wsuggest-attribute=cold): Document.
12372 * common.opt (Wsuggest-attribute=cold): New
12373 * ipa-pure-const.c (warn_function_cold): New function.
12374 * predict.c (compute_function_frequency): Use it.
12375 * predict.h (warn_function_cold): Declare.
12377 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
12379 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
12382 2017-10-06 Martin Liska <mliska@suse.cz>
12384 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
12385 keyword for member functions.
12386 (struct sanopt_tree_couple): New struct.
12387 (struct sanopt_tree_couple_hash): New function.
12388 (struct sanopt_ctx): Add new hash_map.
12389 (has_dominating_ubsan_ptr_check): New function.
12390 (record_ubsan_ptr_check_stmt): Likewise.
12391 (maybe_optimize_ubsan_ptr_ifn): Likewise.
12392 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
12393 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
12395 2017-10-06 Sudakshina Das <sudi.das@arm.com>
12398 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
12399 aarch64_simd_valid_immediate on CONST_VECTORs.
12400 (aarch64_reg_or_bic_imm): Likewise.
12402 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
12404 PR rtl-optimization/82396
12405 * haifa-sched.c (ready_sort_real): Disable qsort checking.
12407 2017-10-06 Sebastian Pop <sebpop@gmail.com>
12409 * graphite-dependences.c (scop_get_reads): Move code to...
12410 (scop_get_must_writes): Move code to...
12411 (scop_get_may_writes): Move code to...
12412 (scop_get_reads_and_writes): ... here.
12413 (scop_get_dependences): Call scop_get_reads_and_writes.
12415 2017-10-06 Jakub Jelinek <jakub@redhat.com>
12417 PR tree-optimization/82434
12418 * fold-const.h (can_native_encode_type_p,
12419 can_native_encode_string_p): Remove.
12420 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
12421 don't encode anything, just return what would be otherwise returned.
12422 (native_encode_fixed, native_encode_complex, native_encode_vector):
12424 (native_encode_string): Likewise. Inline by hand
12425 can_native_encode_string_p.
12426 (can_native_encode_type_p): Remove.
12427 (can_native_encode_string_p): Remove.
12428 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
12429 STRING_CSTs using can_native_encode_string_p, test all
12430 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
12431 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
12432 argument from native_encode_expr.
12433 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
12434 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
12437 2017-10-06 Richard Biener <rguenther@suse.de>
12439 PR tree-optimization/82397
12440 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
12441 operand_equal_p but rely on data_ref_compare_tree for detecting
12443 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
12444 to match up with dr_group_sort_cmp.
12446 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12449 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
12451 * config/s390/s390-builtin-types.def: Regenerate.
12453 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12456 * config/s390/s390-builtin-types.def: Regenerate.
12457 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
12458 Change flag from B_VXE to B_VX.
12459 (s390_vec_min_dbl): Remove B_VXE flag.
12461 2017-10-06 Richard Biener <rguenther@suse.de>
12463 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
12464 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
12465 translate_isl_ast_to_gimple::is_valid_rename,
12466 translate_isl_ast_to_gimple::get_rename,
12467 translate_isl_ast_to_gimple::get_def_bb_for_const,
12468 translate_isl_ast_to_gimple::get_new_name,
12469 translate_isl_ast_to_gimple::collect_all_ssa_names,
12470 translate_isl_ast_to_gimple::copy_loop_phi_args,
12471 translate_isl_ast_to_gimple::collect_all_ssa_names,
12472 translate_isl_ast_to_gimple::copy_loop_phi_args,
12473 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
12474 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
12475 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
12476 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
12477 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
12478 translate_isl_ast_to_gimple::copy_cond_phi_args,
12479 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
12480 translate_isl_ast_to_gimple::edge_for_new_close_phis,
12481 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
12482 translate_isl_ast_to_gimple::rename_uses,
12483 translate_isl_ast_to_gimple::rename_all_uses): Remove.
12484 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
12485 (set_rename_for_each_def): Likewise.
12486 (graphite_copy_stmts_from_block): Handle debug stmt resetting
12487 here. Handle rewriting SCEV analyzable uses here.
12488 (copy_bb_and_scalar_dependences): Generate code for PHI
12490 (graphite_regenerate_ast_isl): Adjust.
12491 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
12492 (add_write, add_read): New functions.
12493 (build_cross_bb_scalars_def): Use it and simplify.
12494 (build_cross_bb_scalars_use): Likewise.
12495 (graphite_find_cross_bb_scalar_vars): Inline into...
12496 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
12497 simulating out-of-SSA. Compute liveout and add dependencies.
12498 (build_scops): Force an empty entry block.
12499 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
12501 (sese_build_liveouts): Declare.
12502 (sese_trivially_empty_bb_p): Likewise.
12503 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
12504 compute liveout and debug_liveout.
12505 (sese_bad_liveouts_use): Remove.
12506 (sese_reset_debug_liveouts_bb): Likewise.
12507 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
12508 (sese_build_liveouts): Build liveout and debug_liveout and store
12510 (new_sese_info): Adjust.
12511 (free_sese_info): Likewise.
12512 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
12513 do not build liveout here.
12514 (move_sese_in_condition): Adjust region entry.
12515 (scev_analyzable_p): Match up with chrec_apply requirements.
12516 (sese_trivially_empty_bb_p): New.
12517 * tree-into-ssa.c (get_reaching_def): Properly support generating
12518 default-defs for incremental rewrite of anonymous names.
12520 2017-10-06 Richard Biener <rguenther@suse.de>
12522 * graphite-sese-to-poly.c (extract_affine): For casts increasing
12523 precision do not perform modulo reduction.
12525 2017-10-06 Richard Biener <rguenther@suse.de>
12527 PR tree-optimization/82436
12528 * tree-vect-slp.c (vect_supported_load_permutation_p): More
12529 conservatively choose the vectorization factor when checking
12530 whether we can perform the required load permutation.
12531 (vect_transform_slp_perm_load): Assert when we may not fail.
12533 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
12535 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
12536 message for incompatible -msdata=* and -mcall-* options.
12538 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12540 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
12541 rate for post-reload scheduling.
12543 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12545 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
12547 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12549 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
12550 to improve monte carlo in scimark.
12552 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12554 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
12555 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
12556 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
12557 pentium4_cost, nocona_cost): Set reassociation width to 1.
12558 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
12559 width to 2 for fp operations and 1 otherwise.
12560 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
12562 (atom_cost): Set reassociation width to 2.
12563 (slm_cost, generic_cost): Set fp reassociation width
12564 to 2 and 1 otherwise.
12565 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
12566 (core_cost): Set fp reassociation width to 4 and vector to 2.
12567 (ix86_reassociation_width): Rewrite using cost table; special case
12568 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
12569 and TARGET_AVX128_OPTIMAL.
12570 * config/i386/i386.h (processor_costs): Add
12571 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
12572 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
12573 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
12574 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
12575 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
12576 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
12578 2017-10-05 Nathan Sidwell <nathan@acm.org>
12580 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
12582 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12584 * config/arm/arm.c (arm_test_fpu_data): New.
12585 (arm_run_selftests): Call arm_test_fpu_data.
12587 2017-10-04 Nathan Sidwell <nathan@acm.org>
12589 * toplev.c (toplev::main): Remove excess parens on pretty_printer
12591 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
12593 2017-10-04 Sudakshina Das <sudi.das@arm.com>
12595 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
12596 check type for aarch64_simd_valid_immediate.
12597 (aarch64_output_simd_mov_immediate): Update prototype.
12598 (aarch64_simd_valid_immediate): Update prototype.
12599 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
12600 support for ORR-immediate.
12601 (and<mode>3): modified pattern to add support for BIC-immediate.
12602 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
12603 now checks for valid immediate for BIC and ORR based on new enum
12605 (aarch64_output_simd_mov_immediate): Function now used to output
12606 BIC/ORR imm as well based on new enum argument.
12607 * config/aarch64/constraints.md (Do): New vector immediate constraint.
12609 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
12610 (aarch64_reg_or_bic_imm): Likewise.
12612 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12614 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
12615 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
12617 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12621 PR rtl-optimization/82396
12622 * haifa-sched.c (autopref_multipass_init): Simplify
12624 (autopref_rank_data): Simplify sort order.
12625 * sched-int.h (autopref_multipass_data_): Remove
12626 multi_mem_insn_p, min_offset and max_offset.
12628 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12630 * doc/sourcebuild.texi: Document vect_peeling_profitable.
12632 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12634 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
12635 vect_doubleint_cvt.
12637 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12639 * doc/sourcebuild.texi: Document vect_long_mult.
12641 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
12643 PR tree-optimization/82413
12644 * fold-const.c (build_range_check): Use widest_int when comparing
12645 the maximum ETYPE value with HIGH.
12647 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12649 PR rtl-optimization/82396
12650 * haifa-sched.c (autopref_multipass_init): Simplify
12652 (autopref_rank_data): Simplify sort order.
12653 * sched-int.h (autopref_multipass_data_): Remove
12654 multi_mem_insn_p, min_offset and max_offset.
12656 2017-10-04 Jakub Jelinek <jakub@redhat.com>
12658 PR tree-optimization/82381
12659 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
12660 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
12663 PR tree-optimization/82374
12664 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
12665 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
12666 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
12667 current_function_decl to the new decl.
12669 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
12671 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
12672 helper macro for IEEE float128 hardware built-in functions.
12673 (SQRTF128_ODD): Add built-in functions with the round-to-odd
12675 (TRUNCF128_ODD): Likewise.
12676 (ADDF128_ODD): Likewise.
12677 (SUBF128_ODD): Likewise.
12678 (MULF128_ODD): Likewise.
12679 (DIVF128_ODD): Likewise.
12680 (FMAF128_ODD): Likewise.
12681 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
12682 UNSPEC_TRUNC_ROUND_TO_ODD.
12683 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
12684 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
12685 floating point round to odd instructions.
12686 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
12687 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
12688 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
12689 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
12690 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
12691 (trunc<mode>sf2_hw): Change the truncate with round to odd
12692 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
12693 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
12694 to odd hardware instructions.
12695 (sub<mode>3_odd): Likewise.
12696 (mul<mode>3_odd): Likewise.
12697 (div<mode>3_odd): Likewise.
12698 (sqrt<mode>2_odd): Likewise.
12699 (fma<mode>4_odd): Likewise.
12700 (fms<mode>4_odd): Likewise.
12701 (nfma<mode>4_odd): Likewise.
12702 (nfms<mode>4_odd): Likewise.
12703 (trunc<mode>df2_odd): Change the truncate with round to odd
12704 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
12706 * doc/extend.texi (PowerPC built-in functions): Update documentation
12707 for existing IEEE float128-bit built-in functions. Add built-in
12708 functions that generate the IEEE 128-bit floating point round to
12711 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
12713 PR rtl-optimization/77729
12714 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
12715 to (X&(C1&~C2))|C2 transformations.
12717 2017-10-03 Martin Jambor <mjambor@suse.cz>
12719 PR tree-optimization/82363
12720 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
12721 mismatch, mark lacc written regardless of racc.
12723 2017-10-03 Jakub Jelinek <jakub@redhat.com>
12725 PR tree-optimization/82381
12726 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
12727 stmt_to_insert nor wheather SSA_NAMEs are default defs.
12728 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
12729 fallthrough into reassoc_stmt_dominates_stmt_p.
12732 * combine.c (combine_instructions): Don't combine in unreachable
12735 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
12738 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
12739 function to not use the have_cpu variable. Do not set cpu_index,
12740 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
12741 or the default cpu.
12742 (rs6000_valid_attribute_p): Remove duplicate initializations of
12743 old_optimize and func_optimize.
12744 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
12745 (rs6000_activate_target_options): Make global.
12746 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
12749 2017-10-02 Jakub Jelinek <jakub@redhat.com>
12751 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
12752 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
12753 Set *poffset to 0 and *psize and *pmax_size to -1 if
12754 *poffset + *psize overflows in HOST_WIDE_INT.
12756 PR tree-optimization/82387
12757 PR tree-optimization/82388
12758 PR tree-optimization/82389
12759 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
12760 instead of live_bytes non-NULL.
12762 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
12765 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
12768 2017-10-02 Richard Biener <rguenther@suse.de>
12770 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
12771 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
12772 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
12774 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12776 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
12777 requested precision matches the type's.
12778 * calls.c (alloc_max_size): Calculate the new candidate size as
12779 a widest_int and use wi::to_widest when comparing it with the
12780 current candidate size.
12781 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
12782 zero rather than integer_zero_node.
12783 * match.pd: Check for a no-op conversion before using wi::add
12784 rather than after. Use tree_to_uhwi when summing small shift
12785 counts into an unsigned int.
12787 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12788 Alan Hayward <alan.hayward@arm.com>
12789 David Sherwood <david.sherwood@arm.com>
12792 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
12793 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
12794 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
12795 POINTER_AND_FP_REGS.
12797 2017-10-02 Richard Biener <rguenther@suse.de>
12799 PR tree-optimization/82355
12800 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12801 a mapping for the enclosing loop but avoid generating one for
12802 the loop tree root.
12803 (copy_bb_and_scalar_dependences): Remove premature codegen
12804 error on PHIs in blocks duplicated into multiple places.
12805 * graphite-scop-detection.c
12806 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12807 in the region use it as loop and nest to analyze the DR in.
12808 (try_generate_gimple_bb): Likewise.
12809 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12810 (add_loop_constraints): For blocks in a loop not in the region
12811 create a dimension with a single iteration.
12812 * sese.h (gbb_loop_at_index): Remove assert.
12814 2017-10-01 Kevin Buettner <kevinb@redhat.com>
12816 * omp-expand.c (adjust_context_scope): New function.
12817 (expand_parallel_call): Call adjust_context_scope.
12819 2017-10-01 Jeff Law <law@redhat.com>
12821 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
12822 dom_opt_dom_walker class with direct access to private members.
12823 Add comments. Call test_for_singularity.
12824 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
12825 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
12826 m_dummy_cond anymore.
12827 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
12829 (pass_dominator:execute): Build the dummy_cond here and pass it
12830 to the dom_opt_dom_walker ctor.
12831 (test_for_singularity): New function.
12833 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
12834 Maya Rashish <coypu@sdf.org>
12836 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
12837 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
12838 (alpha*-*-netbsd*) Use nbsd_tm_file.
12839 (arm*-*-netbsdelf*) Likewise.
12840 (i[34567]86-*-netbsdelf*) Likewise.
12841 (x86_64-*-netbsd*) Likewise.
12842 (mips*-*-netbsd*) Likewise.
12843 (powerpc-*-netbsd*) Likewise.
12844 (sh*-*-netbsd*) Likewise.
12845 (sparc-*-netbsdelf*) Likewise.
12846 (sparc64-*-netbsd*) Likewise.
12847 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
12849 (vax-*-netbsdelf*) Likewise.
12850 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
12851 (UINT_FAST8_TYPE) Likewise.
12852 (INT_FAST16_TYPE) Check CHAR_FAST16.
12853 (UINT_FAST16_TYPE) Likewise.
12855 2017-09-30 Jakub Jelinek <jakub@redhat.com>
12858 * config/i386/i386.md
12859 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
12860 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
12861 *divmodsi4_zext_2): New define_insn_and_split.
12862 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
12863 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
12864 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
12865 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
12866 New define_insn_and_split.
12867 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
12868 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
12869 operands[1] having DImode when mode is SImode.
12871 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
12872 always SImode for DIV and MOD in REG_EQUAL notes.
12874 2017-09-29 Yury Gribov <tetra2005@gmail.com>
12876 PR middle-end/82319
12877 * match.pd: Fix handling of NaNs in pattern.
12879 2017-09-29 Jeff Law <law@redhat.com>
12881 * sbitmap.c (bitmap_bit_in_range_p): New function.
12882 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
12883 * tree-ssa-dse.c (live_bytes_read): New function.
12884 (dse_classify_store): Ignore reads of dead bytes.
12886 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
12887 typos and whitespace errors.
12888 * config/i386/predicates.md (address_no_seg_operand): Likewise.
12889 * config/s390/s390.c (s390_emit_prologue): Likewise.
12891 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12894 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
12895 with a symbol for LRA.
12897 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12899 PR rtl-optimization/82338
12900 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
12902 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
12904 * genmodes.c (calc_wider_mode): Suppress qsort macro.
12905 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
12906 (qsort_chk): Declare.
12907 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
12908 (qsort_chk): New function.
12910 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12912 PR tree-optimization/82337
12913 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12914 phi definition if the PHI result appears in an abnormal PHI.
12915 (find_basis_for_base_expr): Don't record a basis if the LHS of the
12916 basis appears in an abnormal PHI.
12918 2017-09-29 Richard Biener <rguenther@suse.de>
12920 * graphite-isl-ast-to-gimple.c
12921 (translate_isl_ast_to_gimple::set_codegen_error): New function.
12922 (binary_op_to_tree): Use it.
12923 (get_rename_from_scev): Likewise.
12924 (copy_loop_phi_nodes): Likewise.
12925 (copy_bb_and_scalar_dependences): Likewise.
12926 (translate_pending_phi_nodes): Likewise.
12928 2017-09-29 Jakub Jelinek <jakub@redhat.com>
12931 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
12932 for movabsq $(i32 << shift), r64.
12934 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
12936 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
12937 index when encoding %esp as %rsp to avoid 0x67 prefix.
12939 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
12941 * config/i386/i386.md (*movsf_internal, *movdf_internal):
12942 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
12944 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
12946 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
12947 Extensions with more than 16 double VFP registers.
12948 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
12949 to_clear_mask and all code related to it. Replace the remaining
12950 entry by a sbitmap and adapt code accordingly.
12952 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
12954 * brig-builtins.def: Change pure attributes to const.
12956 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12958 * config.gcc (default_gnu_indirect_function): Default to yes for
12959 sparc*-*-linux* with glibc.
12961 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12963 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
12964 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
12965 when creating .init_array and .fini_array sections with priority
12968 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
12971 * config/aarch64/aarch64.c
12972 (aarch64_builtin_support_vector_misalignment): Always return false
12973 when misalignment is unknown.
12975 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12977 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
12978 this function to return false if the definition used by the swap
12979 instruction is artificial, or if the memory address from which the
12980 constant value is loaded is not represented by a base address held
12981 in a register or if the base address register is a frame or stack
12982 pointer. Additionally, return false if the base address of the
12983 loaded constant is a SYMBOL_REF but is not considered to be a
12985 (replace_swapped_load_constant): New function.
12986 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
12987 loaded constant vector with a load of a swapped constant vector.
12989 2017-09-27 Carl Love <cel@us.ibm.com>
12991 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
12992 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
12993 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
12996 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
12998 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
12999 first, always call autopref_rank_data otherwise.
13001 2017-09-27 Richard Biener <rguenther@suse.de>
13003 * graphite-scop-detection.c (find_scop_parameters): Move
13004 loop bound handling ...
13005 (gather_bbs::before_dom_children): ... here, avoiding the need
13006 to build scop_info->loop_nest.
13007 (record_loop_in_sese): Remove.
13008 * sese.h (sese_info_t::loop_nest): Remove.
13009 * sese.c (new_sese_info): Do not allocate loop_nest.
13010 (free_sese_info): Do not free loop_nest.
13012 2017-09-27 Jakub Jelinek <jakub@redhat.com>
13015 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
13016 lhs from calls if the lhs has addressable type.
13018 2017-09-27 Richard Biener <rguenther@suse.de>
13020 * graphite.h (scop::max_alias_set): New member.
13021 * graphite-scop-detection.c: Remove references to non-existing
13022 --param in comments.
13023 (build_alias_sets): Record the maximum alias set used for drs.
13024 (build_scops): Support zero as unlimited for
13025 --param graphite-max-arrays-per-scop.
13026 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
13027 and inline into ...
13028 (build_poly_sr_1): ... here. Compute alias set based on the
13029 maximum alias set used for drs rather than
13030 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
13032 2017-09-27 Richard Biener <rguenther@suse.de>
13034 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
13035 --param loop-block-tile-size=0 to disable tiling.
13037 2017-09-27 Richard Biener <rguenther@suse.de>
13039 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
13040 (graphite-max-nb-scop-params): Document special value zero.
13041 * domwalk.h (dom_walker::STOP): New symbolical constant.
13042 (dom_walker::dom_walker): Add optional parameter for bb to
13044 (dom_walker::~dom_walker): Declare.
13045 (dom_walker::before_dom_children): Document STOP return value.
13046 (dom_walker::m_user_bb_to_rpo): New member.
13047 (dom_walker::m_bb_to_rpo): Likewise.
13048 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
13049 mapping here if not provided by the user.
13050 (dom_walker::~dom_walker): Free bb to RPO mapping if not
13051 provided by the user.
13052 (dom_walker::STOP): Define.
13053 (dom_walker::walk): Do not compute bb to RPO mapping here.
13054 Support STOP return value from before_dom_children to stop
13056 * graphite-optimize-isl.c (optimize_isl): If the schedule
13057 is the same still generate code if -fgraphite-identity
13058 or -floop-parallelize-all are given.
13059 * graphite-scop-detection.c: Include cfganal.h.
13060 (gather_bbs::gather_bbs): Get and pass through bb to RPO
13062 (gather_bbs::before_dom_children): Return STOP for BBs
13064 (build_scops): Compute bb to RPO mapping and pass it to
13065 the domwalk. Treat --param graphite-max-nb-scop-params=0
13066 as not limiting the number of params.
13067 * graphite.c (graphite_initialize): Remove limit on the
13068 number of basic-blocks in a function.
13069 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
13070 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
13071 default value of 10.
13073 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13075 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
13076 Adjust code to eliminate needing to do the shift right 32-bits
13077 operation after XSCVDPSPN.
13079 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13081 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
13082 ((X / Y) != 0 -> X >= Y): Likewise.
13084 2017-09-26 Carl Love <cel@us.ibm.com>
13086 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
13087 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
13088 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
13089 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
13090 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
13091 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
13092 definitions and overloading.
13093 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
13094 statement for P9V_BUILTIN_XST_LEN_R.
13095 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
13096 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
13097 define_expand and define_insn for the instructions and builtins.
13098 * doc/extend.texi: Update the built-in documentation file for the new
13099 built-in functions.
13100 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
13101 define_insn for the instructions
13103 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
13106 * gcc/config/netbsd-protos.h: New file.
13107 * gcc/config/netbsd.c: New file.
13108 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
13109 * gcc/config/t-netbsd: New file.
13110 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
13111 (tmake_file) Add t-netbsd.
13112 (extra_objs) Add netbsd.o.
13114 2017-09-26 Janus Weil <janus@gcc.gnu.org>
13118 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
13120 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13122 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
13123 sign extension from a vector register to a GPR by doing a 32-bit
13124 direct move and then an EXTSW.
13125 (extendsi<mode>2 splitter): Likewise.
13126 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
13127 right or vector extract after doing XSCVDPSPN. Use
13128 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
13130 (movdi_from_sf_zero_ext): Likewise.
13131 (reload_gpr_from_vsxsf): Likewise.
13132 (p8_mfvsrd_4_disf): Delete, no longer used.
13133 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
13134 then needing to move the SFmode to a GPR to use the XSCVDPSP
13135 instruction instead of FRSP and XSCVDPSPN.
13136 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
13137 it is adjacent to the other XSCVSPDP insns.
13138 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
13139 SFmode to be in traditional Altivec registers.
13140 (vsx_xscvdpspn): Eliminate useless alternative constraint.
13141 (vsx_xscvspdpn): Likewise.
13142 (vsx_xscvspdpn_scalar): Likewise.
13144 2017-09-26 Martin Jambor <mjambor@suse.cz>
13146 * tree-sra.c (compare_access_positions): Put integral types first,
13147 stabilize sorting of integral types, remove conditions putting
13148 non-full-precision integers last.
13149 (sort_and_splice_var_accesses): Disable scalarization if a
13150 non-integert would be represented by a non-full-precision integer.
13152 2017-09-26 Joseph Myers <joseph@codesourcery.com>
13154 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
13155 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
13156 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
13157 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
13158 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
13159 conditionals inside the function instead of around it. Call
13160 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
13161 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
13163 2017-09-26 Richard Biener <rguenther@suse.de>
13165 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
13167 (scop_detection::build_scop_breadth): ... this. Removed.
13168 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
13169 (scop_detection::harmful_stmt_in_bb): Likewise.
13170 (scop_detection::graphite_can_represent_stmt): Likewise.
13171 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
13172 (scop_detection::can_represent_loop): Remove recursion, fold in ...
13173 (scop_detection::can_represent_loop_1): ... this. Removed.
13174 (scop_detection::harmful_loop_in_region): Simplify after inlining
13175 the above and remove more quadraticness.
13176 (build_scops): Adjust.
13177 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
13180 2017-09-26 Jakub Jelinek <jakub@redhat.com>
13183 * config/i386/i386.c (ix86_print_operand_address_as): Only test
13184 REGNO (base) == SP_REG if base is a REG.
13186 PR middle-end/35691
13187 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
13188 if it is different SSA_NAME.
13189 (optimize_range_tests_cmp_bitwise): New function.
13190 (optimize_range_tests): Call it.
13192 2017-09-26 Richard Biener <rguenther@suse.de>
13194 PR tree-optimization/82321
13195 * graphite.c (canonicalize_loop_closed_ssa): Properly check
13196 for the def being inside the loop.
13198 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13200 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
13202 * config/s390/s390-builtins.def: Fix constraint on op4.
13204 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13206 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
13207 independent expanders.
13208 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
13209 ("vec_ordered", "vec_unordered"): New expanders.
13211 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13213 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
13216 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13218 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
13219 vec_unpacks_lo_v16qi.
13220 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
13222 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13224 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
13225 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
13226 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
13228 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13230 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
13232 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
13234 ("*vec_slb<mode>"): New insn pattern.
13235 ("vec_shr_<mode>"): New expander.
13236 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
13237 and force the shift count operand to V16QImode.
13238 ("vec_srb<mode>"): Set shift count mode to V16QI.
13240 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13242 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
13243 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
13244 ("vec_widen_smult_hi_<mode>"): New expander definitions.
13246 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
13249 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
13251 2017-09-26 Richard Biener <rguenther@suse.de>
13253 PR tree-optimization/82320
13254 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
13257 2017-09-25 Jeff Law <law@redhat.com>
13259 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
13260 prototype for new argument.
13261 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
13262 mostly extracted from rs6000_emit_allocate_stack.
13263 (rs6000_emit_probe_stack_range_stack_clash): New function.
13264 (rs6000_emit_allocate_stack): Call
13265 rs6000_emit_probe_stack_range_stack_clash as needed.
13266 (rs6000_emit_probe_stack_range): Add additional argument
13267 to call to gen_probe_stack_range{si,di}.
13268 (output_probe_stack_range): New.
13269 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
13270 (output_probe_stack_range_stack_clash): New.
13271 (rs6000_emit_prologue): Emit notes into dump file as requested.
13272 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
13273 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
13274 Add additional operand and pass it to output_probe_stack_range.
13276 2017-09-25 Bin Cheng <bin.cheng@arm.com>
13278 PR tree-optimization/82163
13279 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
13280 (checking_verify_loop_closed_ssa): New parameter.
13281 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
13282 (check_loop_closed_ssa_stmt): Delete.
13283 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
13284 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
13285 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
13288 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
13290 * brig-builtins.def: Treat HSAIL barrier builtins as
13291 setjmp/longjump style functions.
13293 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13295 * target.def (constant_alignment): New hook.
13296 * defaults.h (CONSTANT_ALIGNMENT): Delete.
13297 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
13298 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
13299 * doc/tm.texi: Regenerate.
13300 * targhooks.h (default_constant_alignment): Declare.
13301 (constant_alignment_word_strings): Likewise.
13302 * targhooks.c (default_constant_alignment): New function.
13303 (constant_alignment_word_strings): Likewise.
13304 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
13305 instead of CONSTANT_ALIGNMENT.
13306 * varasm.c (align_variable, get_variable_align, build_constant_desc)
13307 (force_const_mem): Likewise.
13308 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
13309 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
13310 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
13311 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13312 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
13314 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
13315 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13316 constant_alignment_word_strings.
13317 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
13318 (CONSTANT_ALIGNMENT): Likewise.
13319 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13320 (arm_constant_alignment): New function.
13321 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
13322 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13323 constant_alignment_word_strings.
13324 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
13325 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13326 constant_alignment_word_strings.
13327 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
13328 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13329 (cris_constant_alignment): New function.
13330 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
13331 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13332 (epiphany_constant_alignment): New function.
13333 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
13334 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13335 constant_alignment_word_strings.
13336 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
13337 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13338 constant_alignment_word_strings.
13339 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
13340 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13341 constant_alignment_word_strings.
13342 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
13343 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
13344 * config/i386/i386.c (ix86_constant_alignment): Make static.
13345 Use the same interface as the target hook.
13346 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13347 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
13348 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13349 constant_alignment_word_strings.
13350 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
13351 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
13352 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13353 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
13354 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13355 constant_alignment_word_strings.
13356 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
13357 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13358 constant_alignment_word_strings.
13359 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
13360 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13361 constant_alignment_word_strings.
13362 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
13363 * config/microblaze/microblaze.c (microblaze_constant_alignment):
13365 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13366 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
13367 * config/mips/mips.c (mips_constant_alignment): New function.
13368 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13369 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
13370 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
13371 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13372 (mmix_constant_alignment): Make static. Use the same interface
13373 as the target hook.
13374 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
13375 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13376 constant_alignment_word_strings.
13377 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
13378 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13379 constant_alignment_word_strings.
13380 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
13381 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13382 constant_alignment_word_strings.
13383 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
13384 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13385 (rs6000_constant_alignment): New function.
13386 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
13387 * config/riscv/riscv.c (riscv_constant_alignment): New function.
13388 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13389 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
13390 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13391 (rs6000_constant_alignment): New function.
13392 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
13393 * config/s390/s390.c (s390_constant_alignment): New function.
13394 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13395 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
13396 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13397 constant_alignment_word_strings.
13398 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
13399 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13400 (sparc_constant_alignment): New function.
13401 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
13402 * config/spu/spu.c (spu_constant_alignment): New function.
13403 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13404 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
13405 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13406 constant_alignment_word_strings.
13407 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
13408 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13409 constant_alignment_word_strings.
13410 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
13411 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13412 constant_alignment_word_strings.
13413 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
13414 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13415 (visium_constant_alignment): New function.
13416 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
13417 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13418 (xtensa_constant_alignment): New function.
13419 * system.h (CONSTANT_ALIGNMENT): Poison.
13421 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
13423 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13424 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
13425 (rs6000_builtin_valid_without_lhs): New helper function.
13426 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13427 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
13429 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13431 * target.h (vec_perm_indices): Use unsigned short rather than
13433 (auto_vec_perm_indices): Likewise.
13434 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
13435 Use unsigned int rather than unsigned char.
13436 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
13438 2017-09-25 Richard Biener <rguenther@suse.de>
13440 * cfgloop.h (sort_sibling_loops): Declare.
13441 * cfgloop.c (sort_sibling_loops_cmp): New helper.
13442 (sort_sibling_loops): New function sorting the sibling loop list
13444 * graphite.c (graphite_transform_loops): Sort sibling loops.
13446 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
13448 * target.def (vec_perm_const_ok): Change sel parameter to
13450 * optabs-query.c (can_vec_perm_p): Update accordingly.
13451 * doc/tm.texi: Regenerate.
13452 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
13453 auto_vec_perm_indices and remove separate nelt field.
13454 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
13455 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
13456 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
13457 (aarch64_expand_vec_perm_const): Update accordingly.
13458 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
13459 to vec_perm_indices.
13460 * config/arm/arm.c (expand_vec_perm_d): Change perm to
13461 auto_vec_perm_indices and remove separate nelt field.
13462 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
13463 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
13464 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
13466 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
13467 to vec_perm_indices.
13468 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
13469 sel to vec_perm_indices.
13470 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
13471 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
13472 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
13474 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
13477 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
13480 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
13481 on the FUNCTION_DECL function context if it has a DIE that is a
13484 2017-09-25 Richard Biener <rguenther@suse.de>
13486 PR tree-optimization/82285
13487 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
13490 2017-09-25 Tom de Vries <tom@codesourcery.com>
13494 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
13497 2017-09-25 Richard Biener <rguenther@suse.de>
13499 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
13500 ISL errors other than isl_error_quota happen. Dump if the
13501 schedule is the same.
13502 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
13503 errors instead of aborting inside ISL.
13505 2017-09-25 Iain Sandoe <iain@codesourcery.com>
13508 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
13509 of libgcc_eh for m64.
13510 * config/i386/darwin64.h: Likewise.
13512 2017-09-25 Richard Biener <rguenther@suse.de>
13514 PR middle-end/82144
13515 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
13516 attribute for incomplete types nor twice for complete ones.
13518 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
13521 * config/i386/i386.c (ix86_print_operand_address_as): Encode
13522 %esp as %rsp to avoid 0x67 prefix if there is no index or base
13525 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
13528 * config/i386/i386.opt (mprefer-avx256): Use
13529 ix86_target_flags variable.
13530 * config/i386/i386.c (ix86_target_string): Move
13531 -mprefer-avx256 to flag2_opts.
13533 2017-09-22 Jakub Jelinek <jakub@redhat.com>
13535 PR middle-end/35691
13536 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
13537 and x != -1 | y != -1 into (x & y) != -1.
13539 2017-09-22 Steve Ellcey <sellcey@cavium.com>
13541 * config.gcc: Add new case statement to set
13542 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
13543 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
13544 s390x-*-linux* case statements. Added aarch64 to the list of
13545 supported architectures.
13547 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13549 PR tree-optimization/82289
13550 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
13551 STMT_VINFO_RELEVANT_P.
13553 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13554 Alan Hayward <alan.hayward@arm.com>
13555 David Sherwood <david.sherwood@arm.com>
13557 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
13558 for VR_RANGE only; don't allow VR_ANTI_RANGE.
13559 (extract_range_from_binary_expr_1): Don't call
13560 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
13562 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13563 Alan Hayward <alan.hayward@arm.com>
13564 David Sherwood <david.sherwood@arm.com>
13566 * target.def (preferred_vector_alignment): New hook.
13567 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
13569 * doc/tm.texi: Regenerate.
13570 * targhooks.h (default_preferred_vector_alignment): Declare.
13571 * targhooks.c (default_preferred_vector_alignment): New function.
13572 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
13574 (DR_TARGET_ALIGNMENT): New macro.
13575 (aligned_access_p): Update commentary.
13576 (vect_known_alignment_in_bytes): New function.
13577 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
13579 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
13580 Calculate the misalignment based on the target alignment rather than
13582 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
13583 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
13584 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
13585 the target alignment, rather than masking the element misalignment
13586 with the number of elements in a vector. Also use the target
13587 alignment when calculating the maximum number of peels.
13588 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
13589 instead of TYPE_ALIGN_UNIT.
13590 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
13591 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
13592 (vect_create_addr_base_for_vector_ref): Update call accordingly.
13593 (vect_create_data_ref_ptr): Likewise.
13594 (vect_setup_realignment): Realign by ANDing with
13595 -DR_TARGET_MISALIGNMENT.
13596 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
13597 the number of peels based on DR_TARGET_ALIGNMENT.
13598 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
13599 with the guaranteed alignment boundary when deciding whether
13601 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
13602 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
13603 (ensure_base_align): Remove stmt_info parameter. Get the
13604 target base alignment from DR_TARGET_ALIGNMENT.
13605 (vectorizable_store): Update call accordingly. Interpret
13606 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
13608 (vectorizable_load): Likewise.
13610 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13611 Alan Hayward <alan.hayward@arm.com>
13612 David Sherwood <david.sherwood@arm.com>
13614 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
13615 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
13616 (vect_enhance_data_refs_alignment): Likewise.
13618 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13620 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13621 error. Only quit immediately if parsing is complete.
13622 (BEGIN): Initialize fatal_err and parse_done.
13623 (begin fpu, end fpu): Check number of arguments.
13624 (begin arch, end arch): Likewise.
13625 (begin cpu, end cpu): Likewise.
13626 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13627 (optalias): Likewise.
13629 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13631 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13632 * config/arm/arm-isa.h: Delete. Move definitions to ...
13633 * arm-cpus.in: ... here. Use new feature and fgroup values.
13634 * config/arm/arm.c (arm_option_override): Use lower case for feature
13636 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13637 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13638 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13640 (print_isa_bits_for): New function.
13641 (gen_isa): New function.
13642 (gen_comm_data): Use print_isa_bits_for.
13643 (define feature): New keyword.
13644 (define fgroup): New keyword.
13645 * config/arm/t-arm (TM_H): Remove.
13646 (GTM_H): Add arm-isa.h.
13647 (arm-isa.h): Add rule to generate file.
13648 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13649 case for feature bit names.
13651 2017-09-22 Richard Biener <rguenther@suse.de>
13653 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
13655 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
13656 print of no dependency loops ...
13657 * graphite.c (graphite_transform_loops): ... here.
13658 (canonicalize_loop_closed_ssa_form): Work from inner to outer
13660 (same_close_phi_node, remove_duplicate_close_phi,
13661 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
13662 (canonicalize_loop_closed_ssa): ... here and simplify.
13663 * graphite-optimize-isl.c: Include tree-vectorizer.h.
13664 (optimize_isl): Use dump_printf_loc to tell when we stopped
13665 optimizing because of an ISL timeout.
13667 2017-09-22 Richard Biener <rguenther@suse.de>
13669 PR tree-optimization/82291
13670 * tree-if-conv.c (predicate_mem_writes): Make sure to
13671 remove writes in blocks predicated with false.
13673 2017-09-22 Richard Biener <rguenther@suse.de>
13675 * sese.c: Include cfganal.h.
13676 (if_region_set_false_region): Remove.
13677 (create_if_region_on_edge): Likewise.
13678 (move_sese_in_condition): Re-implement without destroying
13681 2017-09-22 Richard Biener <rguenther@suse.de>
13683 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
13684 Verify both BBs contain loop PHI nodes before dispatching to
13685 copy_loop_phi_args.
13686 (graphite_regenerate_ast_isl): Do not recompute dominators,
13687 do not verify three times. Restructure for clarity.
13688 * graphite-scop-detection.c (same_close_phi_node,
13689 remove_duplicate_close_phi, make_close_phi_nodes_unique,
13690 defined_in_loop_p, canonicalize_loop_closed_ssa,
13691 canonicalize_loop_closed_ssa_form): Simplify, remove excess
13692 checking and SSA rewrite, move to ...
13693 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
13694 (graphite_initialize): Do not pass in ctx, do not reset the
13695 SCEV cache, compute only dominators.
13696 (graphite_transform_loops): Allocate ISL ctx after
13697 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
13698 Maintain post-dominators only around build_scops.
13699 * sese.c (if_region_set_false_region): Make static. Free
13700 and recompute dominators.
13701 (move_sese_in_condition): Assert we don't get called with
13702 post-dominators computed.
13703 * sese.h (if_region_set_false_region): Remove.
13705 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
13707 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
13708 mode attribute for TARGET_AVX512VL.
13710 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
13712 * config/i386/i386.opt (mprefer-avx256): New option.
13713 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
13715 (ix86_preferred_simd_mode): Return 256-bit AVX modes
13716 for TARGET_PREFER_AVX256.
13717 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
13719 2017-09-21 Jeff Law <law@redhat.com>
13721 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
13722 Fix dump output if the only stack space is for pushed registers.
13724 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13726 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
13729 2017-09-21 Martin Sebor <msebor@redhat.com>
13732 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
13733 code (in C++) or code that triggers warnings.
13735 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
13737 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
13739 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
13741 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
13742 * sched-rgn.c: Ditto.
13743 * sel-sched-ir.c: Ditto.
13745 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
13747 * toplev.h (set_random_seed): Adjust return type.
13748 * toplev.c (init_local_tick): Move eager initialization of random_seed
13749 to get_random_seed. Adjust comment.
13750 (init_random_seed): Inline to get_random_seed, delete.
13751 (get_random_seed): Initialize random_seed lazily.
13752 (set_random_seed): Do not return previous value.
13753 (print_switch_value): Do not call get_random_seed.
13755 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
13757 * cgraph.c (delete_function_version): New, broken out from...
13758 (cgraph_node::delete_function_version): ...here. Rename to
13759 cgraph_node::delete_function_version_by_decl. Update all uses.
13760 (cgraph_node::remove): Call delete_function_version.
13762 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13765 * tree-inline.c (expand_call_inline): Emit clobber stmts for
13766 VAR_DECLs to which addressable non-volatile parameters are mapped
13767 and for id->retvar after the return value assignment. Clear
13768 id->retval and id->retbnd after inlining.
13770 2017-09-21 Richard Biener <rguenther@suse.de>
13772 PR tree-optimization/82276
13773 PR tree-optimization/82244
13774 * tree-vrp.c (build_assert_expr_for): Set
13775 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
13777 (remove_range_assertions): Revert earlier change.
13779 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
13782 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
13784 2017-09-21 Richard Biener <rguenther@suse.de>
13786 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
13787 Restore valid IL after code generation errors.
13788 * graphite.c (graphite_transform_loops): Diagnose code
13789 generation issues as MSG_MISSED_OPTIMIZATION and continue
13790 with processing SCOPs.
13792 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13793 Alan Hayward <alan.hayward@arm.com>
13794 David Sherwood <david.sherwood@arm.com>
13796 * calls.c (compute_argument_addresses): Use simplify_gen_binary
13797 rather than choosing between plus_constant and gen_rtx_<CODE>.
13798 * expr.c (emit_push_insn): Likewise.
13799 (expand_expr_real_2): Likewise.
13801 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13802 Alan Hayward <alan.hayward@arm.com>
13803 David Sherwood <david.sherwood@arm.com>
13805 * loop-unroll.c (split_iv): Call copy_rtx on the step.
13807 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13808 Alan Hayward <alan.hayward@arm.com>
13809 David Sherwood <david.sherwood@arm.com>
13811 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
13812 calling tree_to_uhwi.
13814 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13815 Alan Hayward <alan.hayward@arm.com>
13816 David Sherwood <david.sherwood@arm.com>
13818 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
13819 INTEGER_CST rather than a negative test for ADDR_EXPR.
13821 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13822 Alan Hayward <alan.hayward@arm.com>
13823 David Sherwood <david.sherwood@arm.com>
13825 * tree-vrp.c (extract_range_from_binary_expr_1): Check
13826 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
13828 2017-09-21 Richard Biener <rguenther@suse.de>
13830 PR tree-optimization/71351
13831 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
13832 graphite_create_new_loop_guard): Remove, fold remaining parts
13834 (translate_isl_ast_node_for): ... here and simplify.
13836 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13839 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
13840 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
13841 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
13842 alternative always use QI mode, for -Os imov (=R,R) alternative
13843 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
13846 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13847 Jeff Law <law@redhat.com>
13849 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
13850 (allocate_stack_space): New function, partially extracted from
13851 s390_emit_prologue.
13852 (s390_emit_prologue): Track offset to most recent stack probe.
13853 Code to allocate space moved into allocate_stack_space.
13854 Dump actions when no stack is allocated.
13855 (s390_prologue_plus_offset): New function.
13856 (s390_emit_stack_probe): Likewise.
13858 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
13860 * common.opt (Wa, Wl, Wp, g, gz=): Add
13862 (gno-column-info): Remove.
13863 (gcolumn-info): Drop RejectNegative.
13864 (gno-): New prefix.
13865 (gno-record-gcc-switches): Remove.
13866 (grecord-gcc-switches): Drop RejectNegative.
13867 (gno-split-dwarf): Remove.
13868 (gsplit-dwarf): Drop RejectNegative.
13869 (gno-strict-dwarf): Remove.
13870 (gstrict-dwarf): Drop RejectNegative.
13871 * config/darwin.opt (gfull, gused): Add RejectNegative.
13872 * dwarf2out.c (gen_producer_string): Drop
13873 gno-record-gcc-switches handler.
13874 * optc-gen.awk: Add g to prefixes with negative forms.
13875 * opts-common.c (remapping_prefix_p): New.
13876 (find_opt): Check it.
13877 (generate_canonical_option): Test g prefix.
13878 (option_map): Add -gno- mapping.
13879 (add_misspelling_candidates): Check remapping_prefix_p.
13881 2017-09-20 Jeff Law <law@redhat.com>
13883 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
13884 thinko in stack clash protection support.
13886 * explow.c (compute_stack_clash_protection_loop_data): Use
13887 CONST_INT_P instead of explicit test. Verify object is a
13888 CONST_INT_P before looking at INTVAL.
13889 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
13890 instead of explicit test.
13892 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
13895 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
13896 address instead of to r1 and r11.
13898 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
13900 * config.gcc: Support "knm".
13901 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
13902 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13904 * config/i386/i386.c (m_KNM): Define.
13905 (processor_target_table): Add "knm".
13907 (ix86_option_override_internal): Add "knm".
13908 (ix86_issue_rate): Add PROCESSOR_KNM.
13909 (ix86_adjust_cost): Ditto.
13910 (ia32_multipass_dfa_lookahead): Ditto.
13911 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
13912 (fold_builtin_cpu): Add M_INTEL_KNM.
13913 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
13914 (processor_type): Add PROCESSOR_KNM.
13915 * config/i386/x86-tune.def: Add m_KNM.
13916 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
13918 2017-09-20 Richard Biener <rguenther@suse.de>
13920 PR tree-optimization/80213
13921 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
13922 are allowed in empty BBs as well.
13923 (canonicalize_loop_closed_ssa): Also look for other complex
13925 (scop_detection::get_sese): Include the loop-closed PHI block
13927 (scop_detection::merge_sese): Remove code adding extra blocks.
13928 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
13929 (build_scops): Assert the final returned scop is invalid.
13931 2017-09-20 Richard Biener <rguenther@suse.de>
13933 PR tree-optimization/82264
13934 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
13936 (vn_phi_lookup): Likewise.
13937 (vn_phi_insert): Likewise.
13939 2017-09-20 Jakub Jelinek <jakub@redhat.com>
13941 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
13942 that fits into uhwi or shwi, add DW_AT_const_value regardless
13943 of early_dwarf without going through RTL, using add_AT_unsigned
13946 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
13947 (DEBUG_LTO_ABBREV_SECTION): Likewise.
13948 (DEBUG_LTO_MACINFO_SECTION): Likewise.
13949 (DEBUG_MACRO_SECTION): Likewise.
13950 (DEBUG_LTO_MACRO_SECTION): Likewise.
13951 (DEBUG_STR_DWO_SECTION): Likewise.
13952 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
13953 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
13954 (DEBUG_LTO_DWO_LINE_SECTION): Define.
13955 (DEBUG_LTO_LINE_STR_SECTION): Define.
13956 (init_sections_and_labels): Initialize debug_line_str_section
13957 variable. Initialize debug_loc_section for -gdwarf-5 to
13958 DEBUG_LOCLISTS_SECTION. Formatting fixes.
13960 2017-09-20 Richard Biener <rguenther@suse.de>
13962 * graphite-sese-to-poly.c (extract_affine): Properly handle
13963 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
13965 2017-09-20 Richard Biener <rguenther@suse.de>
13967 PR tree-optimization/81373
13968 * graphite-scop-detection.c (build_cross_bb_scalars_def):
13969 Force SESE live-out defs to be handled even if they are
13972 2017-09-19 Jeff Law <law@redhat.com>
13974 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
13975 nothing for stack adjustments with REG_STACK_CHECK.
13976 * sched-deps.c (parse_add_or_inc): Reject insns with
13977 REG_STACK_CHECK from dependency breaking.
13978 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
13979 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
13980 * reg-notes.def (STACK_CHECK): New note.
13982 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
13983 (ix86_expand_prologue): Dump stack clash info as needed.
13984 Call ix86_adjust_stack_and_probe_stack_clash as needed.
13986 * function.c (dump_stack_clash_frame_info): New function.
13987 * function.h (dump_stack_clash_frame_info): Prototype.
13988 (enum stack_clash_probes): New enum.
13990 * config/alpha/alpha.c (alpha_expand_prologue): Also check
13991 flag_stack_clash_protection.
13992 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
13993 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
13994 (arm_frame_pointer_required): Likewise.
13995 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
13996 (ia64_expand_prologue): Likewise.
13997 * config/mips/mips.c (mips_expand_prologue): Likewise.
13998 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
13999 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14000 (sparc_flat_expand_prologue): Likewise.
14001 * config/spu/spu.c (spu_expand_prologue): Likewise.
14003 * explow.c: Include "params.h".
14004 (anti_adjust_stack_and_probe_stack_clash): New function.
14005 (get_stack_check_protect): Likewise.
14006 (compute_stack_clash_protection_loop_data): Likewise.
14007 (emit_stack_clash_protection_loop_start): Likewise.
14008 (emit_stack_clash_protection_loop_end): Likewise.
14009 (allocate_dynamic_stack_space): Use get_stack_check_protect.
14010 Use anti_adjust_stack_and_probe_stack_clash.
14011 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
14012 (emit_stack_clash_protection_loop_start): Likewise.
14013 (emit_stack_clash_protection_loop_end): Likewise.
14014 * rtl.h (get_stack_check_protect): Prototype.
14015 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
14016 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
14017 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
14019 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
14021 * doc/tm.texi: Rebuilt.
14022 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
14023 get_stack_check_protect.
14024 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
14025 * config/arm/arm.c (arm_expand_prologue): Likewise.
14026 (arm_frame_pointer_required): Likewise.
14027 * config/i386/i386.c (ix86_expand_prologue): Likewise.
14028 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
14029 * config/mips/mips.c (mips_expand_prologue): Likewise.
14030 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
14031 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
14032 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14033 (sparc_flat_expand_prologue): Likewise.
14035 * common.opt (-fstack-clash-protection): New option.
14036 * flag-types.h (enum stack_check_type): Note difference between
14037 -fstack-check= and -fstack-clash-protection.
14038 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
14039 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
14040 * toplev.c (process_options): Issue warnings/errors for cases
14041 not handled with -fstack-clash-protection.
14042 * doc/invoke.texi (-fstack-clash-protection): Document new option.
14043 (-fstack-check): Note additional problem with -fstack-check=generic.
14044 Note that -fstack-check is primarily for Ada and refer users
14045 to -fstack-clash-protection for stack-clash-protection.
14046 Document new params for stack clash protection.
14048 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
14050 * config/i386/i386.c (ix86_split_long_move): Do not handle
14051 address used for LEA in a special way.
14053 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
14055 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
14057 2017-09-19 Martin Sebor <msebor@redhat.com>
14060 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
14061 of incompatible types.
14063 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
14065 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14066 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
14067 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14068 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
14070 2017-09-19 Richard Biener <rguenther@suse.de>
14072 PR tree-optimization/82244
14073 * tree-vrp.c (remove_range_assertions): Do not propagate
14074 a constant to abnormals but replace the assert with a copy.
14076 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
14078 PR rtl-optimization/57878
14079 PR rtl-optimization/68988
14080 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
14081 avoidance test involving non_reload_pseudos. Move frequency test
14082 below the general fragmentation avoidance test.
14084 2017-09-19 Richard Biener <rguenther@suse.de>
14086 PR tree-optimization/69728
14087 * graphite-sese-to-poly.c (schedule_error): New global.
14088 (add_loop_schedule): Handle empty domain by failing the
14090 (build_original_schedule): Handle schedule_error.
14092 2017-09-19 Richard Biener <rguenther@suse.de>
14094 * graphite-scop-detection.c (scop_detection::can_represent_loop):
14095 Do not iterate to sibling loops but only to siblings of inner
14098 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
14101 * config/m68k/m68k.md (moveq feeding equality comparison): Check
14102 that the registers are different.
14104 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
14106 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
14107 to processor_model and "amdfam17h" to arch_names_table.
14108 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
14110 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14113 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
14115 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14116 Alan Hayward <alan.hayward@arm.com>
14117 David Sherwood <david.sherwood@arm.com>
14119 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
14121 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
14123 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
14124 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
14125 vect_schedule_slp_instance.
14126 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
14127 Update call to vect_slp_analyze_node_operations. Simplify return
14129 (vect_slp_analyze_bb_1): Update call accordingly.
14130 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
14131 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
14132 (vect_schedule_slp): Update call accordingly.
14134 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14135 Alan Hayward <alan.hayward@arm.com>
14136 David Sherwood <david.sherwood@arm.com>
14138 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
14139 with types that aren't in fact scalar.
14141 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14143 * tree-vect-slp.c (vect_record_max_nunits): New function,
14145 (vect_build_slp_tree_1): ...here.
14146 (vect_build_slp_tree_2): Call it for phis too.
14148 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14150 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
14151 to vect_get_vec_def_for_operand when getting the mask operand.
14153 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14154 Alan Hayward <alan.hayward@arm.com>
14155 David Sherwood <david.sherwood@arm.com>
14157 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
14160 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14161 Alan Hayward <alan.hayward@arm.com>
14162 David Sherwood <david.sherwood@arm.com>
14164 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
14165 calculation for vector booleans.
14167 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14168 Alan Hayward <alan.hayward@arm.com>
14169 David Sherwood <david.sherwood@arm.com>
14171 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
14173 (vect_transform_stmt): ...here.
14174 (vect_analyze_stmt): Use it instead of calling
14175 vectorizable_live_operation directly.
14177 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
14179 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
14180 non-SIMT targets in acc vector loops.
14182 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
14184 * configure.ac: Add arc and check if assembler supports gdwarf2.
14185 * configure: Regenerate.
14187 2017-09-18 Richard Biener <rguenther@suse.de>
14189 PR tree-optimization/82220
14190 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
14191 epilogue niters from the min_profitable_iters compute.
14193 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14196 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
14197 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
14198 (ix86_init_pic_reg): Revert 2017-09-01 changes.
14200 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
14203 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
14204 switching to a new text section.
14206 2017-09-18 Richard Biener <rguenther@suse.de>
14208 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
14210 (build_alias_set): Reject aliases with no access function.
14212 2017-09-18 Richard Biener <rguenther@suse.de>
14214 PR tree-optimization/79622
14215 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
14217 (build_cross_bb_scalars_use): Likewise.
14219 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
14221 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
14223 2017-09-18 Alan Modra <amodra@gmail.com>
14226 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
14227 stack_pointer_rtx for count 0. Update comments. Break up
14228 large rtl expression.
14230 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
14232 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
14233 Increase to 20 bytes.
14234 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
14235 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
14236 or avx version of the stub.
14238 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
14241 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
14242 compute the minimum stack alignment. Also update preferred stack
14243 boundary for leaf functions.
14245 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
14247 PR tree-optimization/82228
14248 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
14251 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14253 * common/config/nds32/nds32-common.c
14254 (nds32_option_optimization_table): Refine formatting.
14255 (nds32_option_optimization_table): Use -fsched-pressure and
14256 -fomit-frame-pointer for specific optimization level.
14258 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14260 * config/nds32/nds32.c: Refine formatting and comments.
14261 * config/nds32/nds32.h: Likewise.
14262 * config/nds32/nds32.md: Likewise.
14263 * config/nds32/nds32-cost.c: Likewise.
14264 * config/nds32/nds32-isr.c: Likewise.
14265 * config/nds32/nds32-md-auxiliary.c: Likewise.
14266 * config/nds32/nds32-multiple.md: Likewise.
14267 * config/nds32/nds32-predicates.c: Likewise.
14269 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
14270 Jakub Jelinek <jakub@redhat.com>
14272 Add support for -std=c++2a.
14273 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
14275 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
14277 2017-09-15 Steve Ellcey <sellcey@cavium.com>
14280 * doc/extend.texi (Common Function Attributes): Add
14281 references to ARM, AArch64, and S/390 specific attributes.
14282 (Function Specific Option Pragmas): Add AArch64 and S/390
14283 to list of back ends that support the target pragma.
14285 2017-09-15 Nathan Sidwell <nathan@acm.org>
14287 * doc/standards.texi: Fix C++17 description. Update URLs for
14290 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
14292 * common.opt (Wcast-align=strict): New warning option.
14293 * doc/invoke.texi: Document -Wcast-align=strict.
14295 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
14297 * cgraph.h (cgraph_thunk_info): Add comments.
14298 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
14299 assert for VIRTUAL_* arguments stricter.
14301 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
14303 PR tree-optimization/71026
14304 * match.pd: Move RDIV patterns from fold-const.c
14305 * fold-const.c (distribute_real_division): Removed.
14306 (fold_binary_loc): Remove calls to distribute_real_divison.
14308 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14310 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
14311 c++1z and gnu++1z as deprecated. Change other references to
14312 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
14313 Change -Wc++1z-compat to -Wc++17-compat.
14314 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
14315 * dwarf2out.c (highest_c_language): Handle C++17.
14316 (gen_compile_unit_die): Likewise.
14318 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14320 PR rtl-optimization/82192
14321 * combine.c (make_extraction): Don't look through non-paradoxical
14322 SUBREGs or TRUNCATE if pos + len is or might be bigger than
14325 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14326 Alan Hayward <alan.hayward@arm.com>
14327 David Sherwood <david.sherwood@arm.com>
14329 * target.def (function_arg_offset): New hook.
14330 * targhooks.h (default_function_arg_offset): Declare.
14331 * targhooks.c (default_function_arg_offset): New function.
14332 * function.c (locate_and_pad_parm): Use
14333 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
14334 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
14335 (TARGET_FUNCTION_ARG_OFFSET): ...this.
14336 * doc/tm.texi: Regenerate.
14337 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
14338 * config/spu/spu.c (spu_function_arg_offset): New function.
14339 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
14340 * system.h (FUNCTION_ARG_OFFSET): Poison.
14342 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14343 Alan Hayard <alan.hayward@arm.com>
14344 David Sherwood <david.sherwood@arm.com>
14346 * target.def (truly_noop_truncation): New hook.
14347 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
14348 than TRULY_NOOP_TRUNCATION.
14349 * hooks.h (hook_bool_uint_uint_true): Declare.
14350 * hooks.c (hook_bool_uint_uint_true): New function.
14351 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
14352 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
14353 * doc/tm.texi: Regenerate.
14354 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
14355 rather than TRULY_NOOP_TRUNCATION in comments.
14356 (simplify_comparison): Likewise.
14357 (record_truncated_value): Likewise.
14358 * expmed.c (extract_bit_field_1): Likewise.
14359 (extract_split_bit_field): Likewise.
14360 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
14361 instead of TRULY_NOOP_TRUNCATION.
14362 * function.c (assign_parm_setup_block): Likewise.
14363 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
14364 * rtlhooks.c: Include target.h.
14365 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
14366 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
14367 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
14368 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
14369 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
14370 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
14371 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
14372 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
14373 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
14374 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
14375 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
14376 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
14377 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
14378 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
14379 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
14380 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
14381 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
14382 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
14383 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
14384 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
14385 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
14386 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
14387 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
14388 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
14389 * config/mips/mips.c (mips_truly_noop_truncation): New function.
14390 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14391 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14392 rather than TRULY_NOOP_TRUNCATION in comments.
14393 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
14394 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
14395 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
14396 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
14397 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
14398 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
14399 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
14400 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
14401 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
14402 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
14403 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
14404 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14405 rather than TRULY_NOOP_TRUNCATION in comments.
14406 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
14407 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
14408 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
14409 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
14410 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
14411 TRULY_NOOP_TRUNCATION condition.
14412 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
14413 (TRULY_NOOP_TRUNCATION): Delete.
14414 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
14415 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
14416 * config/spu/spu.c (spu_truly_noop_truncation): New function.
14417 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14418 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
14419 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
14420 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
14421 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14422 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14423 rather than TRULY_NOOP_TRUNCATION in comments.
14424 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
14425 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
14426 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
14427 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
14428 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
14429 * system.h (TRULY_NOOP_TRUNCATION): Poison.
14431 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
14434 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
14435 (*cmp_ior): Likewise.
14436 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
14437 (*ior_scc_scc_cmp): Likewise.
14438 (*and_scc_scc): Likewise.
14439 (*and_scc_scc_cmp): Likewise.
14441 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14442 Alan Hayard <alan.hayward@arm.com>
14443 David Sherwood <david.sherwood@arm.com>
14445 * target.def (can_change_mode_class): New hook.
14446 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
14447 (hard_regno_nregs): Likewise.
14448 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
14449 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
14450 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
14451 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
14452 (LOAD_EXTEND_OP): Update accordingly.
14453 * doc/tm.texi: Regenerate.
14454 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
14455 CANNOT_CHANGE_MODE_CLASS.
14456 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
14457 (REG_CAN_CHANGE_MODE_P): ...this new macro.
14458 * combine.c (simplify_set): Update accordingly.
14459 * emit-rtl.c (validate_subreg): Likewise.
14460 * recog.c (general_operand): Likewise.
14461 * regcprop.c (mode_change_ok): Likewise.
14462 * reload1.c (choose_reload_regs): Likewise.
14463 (inherit_piecemeal_p): Likewise.
14464 * rtlanal.c (simplify_subreg_regno): Likewise.
14465 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
14466 instead of CANNOT_CHANGE_MODE_CLASS.
14467 (reload_cse_simplify_operands): Likewise.
14468 * reload.c (push_reload): Use targetm.can_change_mode_class
14469 instead of CANNOT_CHANGE_MODE_CLASS.
14470 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
14471 REG_CANNOT_CHANGE_MODE_P.
14472 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14473 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
14474 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14475 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14476 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14477 (arm_can_change_mode_class): New function.
14478 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14479 than CANNOT_CHANGE_MODE_CLASS in comments.
14480 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14481 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
14482 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
14483 (ix86_can_change_mode_class): ...this new function, inverting the
14484 sense of the return value.
14485 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14486 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14487 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14488 (ia64_can_change_mode_class): New function.
14489 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14490 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
14491 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
14492 (m32c_can_change_mode_class): ...this new function, inverting the
14493 sense of the return value.
14494 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14495 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14496 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
14497 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
14498 (mips_can_change_mode_class): ...this new function, inverting the
14499 sense of the return value.
14500 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14501 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14502 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14503 (msp430_can_change_mode_class): New function.
14504 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14505 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
14506 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14507 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14508 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14509 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
14510 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14511 (pa_cannot_change_mode_class): Replace with...
14512 (pa_can_change_mode_class): ...this new function, inverting the
14513 sense of the return value.
14514 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14515 than CANNOT_CHANGE_MODE_CLASS in comments.
14516 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14517 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
14518 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14519 (pdp11_cannot_change_mode_class): Replace with...
14520 (pdp11_can_change_mode_class): ...this new function, inverting the
14521 sense of the return value.
14522 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14523 * config/powerpcspe/powerpcspe-protos.h
14524 (rs6000_cannot_change_mode_class_ptr): Delete.
14525 * config/powerpcspe/powerpcspe.c
14526 (rs6000_cannot_change_mode_class_ptr): Delete.
14527 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14528 (rs6000_option_override_internal): Assign to
14529 targetm.can_change_mode_class instead of
14530 rs6000_cannot_change_mode_class_ptr.
14531 (rs6000_cannot_change_mode_class): Replace with...
14532 (rs6000_can_change_mode_class): ...this new function, inverting the
14533 sense of the return value.
14534 (rs6000_debug_cannot_change_mode_class): Replace with...
14535 (rs6000_debug_can_change_mode_class): ...this new function.
14536 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14537 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
14538 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14539 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14540 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
14542 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
14543 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14544 (rs6000_option_override_internal): Assign to
14545 targetm.can_change_mode_class instead of
14546 rs6000_cannot_change_mode_class_ptr.
14547 (rs6000_cannot_change_mode_class): Replace with...
14548 (rs6000_can_change_mode_class): ...this new function, inverting the
14549 sense of the return value.
14550 (rs6000_debug_cannot_change_mode_class): Replace with...
14551 (rs6000_debug_can_change_mode_class): ...this new function.
14552 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14553 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
14554 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
14555 (s390_can_change_mode_class): ...this new function, inverting the
14556 sense of the return value.
14557 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14558 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14559 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
14560 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14561 (sh_cannot_change_mode_class): Replace with...
14562 (sh_can_change_mode_class): ...this new function, inverting the
14563 sense of the return value.
14564 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14565 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14566 (sparc_can_change_mode_class): New function.
14567 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14568 * config/spu/spu.c (spu_can_change_mode_class): New function.
14569 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14570 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14571 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14572 (visium_can_change_mode_class): New function.
14573 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
14575 2017-09-15 Richard Biener <rguenther@suse.de>
14577 PR tree-optimization/82217
14578 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
14579 but not undefined case.
14581 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14584 * postreload.c (reload_cse_simplify_operands): Skip
14585 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
14587 2017-09-15 Richard Biener <rguenther@suse.de>
14589 PR tree-optimization/68823
14590 * graphite-scop-detection.c (build_alias_set): If we have a
14591 possible dependence check whether we can handle them by just
14592 looking at the DRs DR_ACCESS_FNs.
14593 (build_scops): If build_alias_set fails, fail the SCOP.
14595 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
14597 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
14598 to support float128 built-in functions that require the ISA 3.0
14600 (BU_FLOAT128_3_HW): Likewise.
14601 (SQRTF128): Add support for the IEEE 128-bit square root and fma
14602 built-in functions.
14603 (FMAF128): Likewise.
14605 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14606 support for built-in functions that need the ISA 3.0 IEEE 128-bit
14607 floating point instructions.
14608 (rs6000_invalid_builtin): Likewise.
14609 (rs6000_builtin_mask_names): Likewise.
14610 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
14611 (RS6000_BTM_FLOAT128_HW): Likewise.
14612 (RS6000_BTM_COMMON): Likewise.
14613 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
14615 * doc/extend.texi (RS/6000 built-in functions): Document the
14616 IEEE 128-bit floating point square root and fused multiply-add
14617 built-in functions.
14619 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
14621 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
14622 reg (r2) isn't in the set of registers defined in the prologue.
14624 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14625 Alan Hayward <alan.hayward@arm.com>
14626 David Sherwood <david.sherwood@arm.com>
14628 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
14629 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
14630 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
14631 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
14632 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
14634 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
14635 max_vectorization_factor.
14636 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
14638 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14639 Alan Hayward <alan.hayward@arm.com>
14640 David Sherwood <david.sherwood@arm.com>
14642 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
14643 (vect_worthwhile_without_simd_p): Declare.
14644 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
14645 (vectorizable_reduction): Use it.
14646 * tree-vect-stmts.c (vectorizable_shift): Likewise.
14647 (vectorizable_operation): Likewise.
14649 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14650 Alan Hayward <alan.hayward@arm.com>
14651 David Sherwood <david.sherwood@arm.com>
14653 * tree-vectorizer.h (vect_get_num_copies): New function.
14654 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
14655 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14656 (vectorizable_induction): Likewise.
14657 (vectorizable_live_operation): Likewise.
14658 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14659 (vectorizable_bswap): Likewise.
14660 (vectorizable_call): Likewise.
14661 (vectorizable_conversion): Likewise.
14662 (vectorizable_assignment): Likewise.
14663 (vectorizable_shift): Likewise.
14664 (vectorizable_operation): Likewise.
14665 (vectorizable_store): Likewise.
14666 (vectorizable_load): Likewise.
14667 (vectorizable_condition): Likewise.
14668 (vectorizable_comparison): Likewise.
14669 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
14671 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14672 Alan Hayward <alan.hayward@arm.com>
14673 David Sherwood <david.sherwood@arm.com>
14675 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
14676 of vect_init_vector.
14678 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14679 Alan Hayward <alan.hayward@arm.com>
14680 David Sherwood <david.sherwood@arm.com>
14682 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
14683 an inline wrapper that provides a location.
14684 (gimple_build_vector): Likewise.
14685 * gimple-fold.c (gimple_build_vector_from_val): New function.
14686 (gimple_build_vector): Likewise.
14687 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
14688 functions to build the initial value. Always return a gimple value.
14689 (get_initial_defs_for_reduction): Likewise. Only compute
14691 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
14692 vect_init_vector on the results from get_initial_def(s)_for_reduction.
14693 (vectorizable_induction): Use gimple_build_vector rather than
14696 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14697 Alan Hayward <alan.hayward@arm.com>
14698 David Sherwood <david.sherwood@arm.com>
14700 * target.h (vec_perm_indices): New typedef.
14701 (auto_vec_perm_indices): Likewise.
14702 * optabs-query.h: Include target.h
14703 (can_vec_perm_p): Take a vec_perm_indices *.
14704 * optabs-query.c (can_vec_perm_p): Likewise.
14705 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
14706 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14707 * tree-vect-generic.c (lower_vec_perm): Likewise.
14708 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
14709 (vect_grouped_load_supported): Likewise.
14710 (vect_shift_permute_load_chain): Likewise.
14711 (vect_permute_store_chain): Use auto_vec_perm_indices.
14712 (vect_permute_load_chain): Likewise.
14713 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
14714 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
14715 Update uses of can_vec_perm_p.
14716 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
14717 mode with a number of elements. Take a vec_perm_indices *.
14718 (vect_create_epilog_for_reduction): Update accordingly.
14719 Use auto_vec_perm_indices.
14720 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
14721 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
14722 (vect_transform_slp_perm_load): Likewise.
14723 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
14724 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
14725 (vect_gen_perm_mask_checked): Likewise.
14726 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
14727 (vect_gen_perm_mask_checked): Likewise.
14728 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
14729 (vectorizable_store): Likewise.
14730 (vectorizable_load): Likewise.
14731 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
14732 (vectorizable_bswap): Likewise.
14734 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14735 Alan Hayward <alan.hayward@arm.com>
14736 David Sherwood <david.sherwood@arm.com>
14738 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
14739 * tree.c (build_vector): Likewise.
14740 (build_vector_from_ctor): Update accordingly.
14741 (build_vector_from_val): Likewise.
14742 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14743 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14744 * tree-vect-generic.c (add_rshift): Likewise.
14745 (expand_vector_divmod): Likewise.
14746 (optimize_vector_constructor): Likewise.
14747 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
14748 (vect_transform_slp_perm_load): Likewise.
14749 (vect_schedule_slp_instance): Likewise.
14750 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
14751 (vectorizable_call): Likewise.
14752 (vect_gen_perm_mask_any): Likewise. Add elements in order.
14753 * expmed.c (make_tree): Likewise.
14754 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
14755 a vector passed to build_vector.
14756 (fold_convert_const): Likewise.
14757 (exact_inverse): Likewise.
14758 (fold_ternary_loc): Likewise.
14759 (fold_relational_const): Likewise.
14760 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
14761 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
14762 (const_unop): Likewise. Store the reduction accumulator in a
14763 variable rather than an array.
14764 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
14765 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
14766 the new vector, rather than constructing it after the input arrays.
14767 (native_interpret_vector): Use auto_vec<tree> when building
14768 a vector passed to build_vector. Add elements in order.
14769 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
14770 auto_vec<tree> when building a vector passed to build_vector.
14771 (vect_create_epilog_for_reduction): Likewise.
14772 (vectorizable_induction): Likewise.
14773 (get_initial_def_for_reduction): Likewise. Fix indentation of
14775 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
14777 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
14778 passed to build_vector.
14780 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14781 Alan Hayward <alan.hayward@arm.com>
14782 David Sherwood <david.sherwood@arm.com>
14784 * tree-core.h (tree_base::u): Add an "nelts" field.
14785 (tree_vector): Use VECTOR_CST_NELTS as the length.
14786 * tree.c (tree_size): Likewise.
14787 (make_vector): Initialize VECTOR_CST_NELTS.
14788 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
14789 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
14790 TYPE_VECTOR_SUBPARTS.
14791 * expr.c (const_vector_mask_from_tree): Consistently use "units"
14792 as the number of units, setting it from VECTOR_CST_NELTS.
14793 (const_vector_from_tree): Likewise.
14794 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
14795 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
14796 (fold_negate_expr_1): Likewise.
14797 (fold_convert_const): Likewise.
14798 (const_binop): Likewise. Differentiate the number of output and
14800 (const_unop): Likewise.
14801 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
14802 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
14803 in cases that did the opposite.
14805 2017-09-14 Richard Biener <rguenther@suse.de>
14807 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
14810 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
14812 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
14814 2017-09-14 Jakub Jelinek <jakub@redhat.com>
14817 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
14818 if and where to split a bb, except for splitting before debug insn
14819 sequences followed by non-label real insn. Delete debug insns
14820 in between basic blocks.
14822 * combine.c (make_compound_operation_int): Formatting fixes.
14824 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
14825 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14826 * config/netbsd.h (LINK_EH_SPEC): Likewise.
14827 * config/sol2.h (LINK_EH_SPEC): Likewise.
14828 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14829 * config/s390/linux.h (LINK_SPEC): Likewise.
14830 * config/freebsd.h (LINK_EH_SPEC): Likewise.
14831 * config/openbsd.h (LINK_EH_SPEC): Likewise.
14832 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14833 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
14834 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
14835 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14836 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
14837 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
14839 2017-09-13 Jakub Jelinek <jakub@redhat.com>
14841 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
14843 (ENDFILE_LINUX_SPEC): Likewise.
14844 (LINK_EH_SPEC): Likewise.
14845 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
14846 (LINK_OS_LINUX_SPEC32): Likewise.
14847 (LINK_OS_LINUX_SPEC64): Likewise.
14848 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
14849 (LINK_OS_LINUX_SPEC): Likewise.
14851 2017-09-13 Martin Liska <mliska@suse.cz>
14853 PR middle-end/82154
14854 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
14855 CASE_HIGH is NULL_TREE.
14857 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14858 Alan Hayward <alan.hayward@arm.com>
14859 David Sherwood <david.sherwood@arm.com>
14861 * target.def (secondary_memory_needed): New hook.
14862 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
14863 instead of SECONDARY_MEMORY_NEEDED.
14864 (secondary_memory_needed_mode): Likewise.
14865 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
14866 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
14867 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
14868 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
14869 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
14870 * doc/tm.texi: Regenerate.
14871 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
14872 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
14873 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14874 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
14875 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
14876 * config/i386/i386.c (inline_secondary_memory_needed): Put the
14877 mode argument first and change the reg_class arguments to reg_class_t.
14878 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
14879 Make static. Update the call to inline_secondary_memory_needed.
14880 (ix86_register_move_cost): Update the call to
14881 inline_secondary_memory_needed.
14882 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14883 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
14885 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
14886 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14888 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
14889 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
14890 * config/mips/mips.c (mips_secondary_memory_needed): Make static
14891 and match hook interface. Add comment from mips.h.
14892 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14893 * config/mmix/mmix.md (truncdfsf2): Refer to
14894 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14896 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
14897 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
14898 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14899 (pa_secondary_memory_needed): New function.
14900 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
14901 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
14902 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14903 (pdp11_secondary_memory_needed): Make static and match hook interface.
14904 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
14905 * config/powerpcspe/powerpcspe-protos.h
14906 (rs6000_secondary_memory_needed_ptr): Delete.
14907 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
14909 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14910 (rs6000_option_override_internal): Assign to
14911 targetm.secondary_memory_needed rather than
14912 rs6000_secondary_memory_needed_ptr.
14913 (rs6000_secondary_memory_needed): Match hook interface.
14914 (rs6000_debug_secondary_memory_needed): Likewise.
14915 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
14916 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
14917 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
14918 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14919 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
14920 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
14922 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
14923 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14924 (rs6000_option_override_internal): Assign to
14925 targetm.secondary_memory_needed rather than
14926 rs6000_secondary_memory_needed_ptr.
14927 (rs6000_secondary_memory_needed): Match hook interface.
14928 (rs6000_debug_secondary_memory_needed): Likewise.
14929 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
14930 * config/s390/s390.c (s390_secondary_memory_needed): New function.
14931 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14932 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
14933 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14934 (sparc_secondary_memory_needed): New function.
14935 * lra-constraints.c (check_and_process_move): Refer to
14936 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14938 (curr_insn_transform): Likewise.
14939 (process_alt_operands): Use targetm.secondary_memory_needed
14940 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14941 (check_secondary_memory_needed_p): Likewise.
14942 (choose_split_class): Likewise.
14943 * reload.c: Unconditionally include code that was previously
14944 conditional on SECONDARY_MEMORY_NEEDED.
14945 (push_secondary_reload): Use targetm.secondary_memory_needed
14946 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14947 (push_reload): Likewise.
14948 * reload1.c: Unconditionally include code that was previously
14949 conditional on SECONDARY_MEMORY_NEEDED.
14950 (choose_reload_regs): Use targetm.secondary_memory_needed
14951 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14952 (gen_reload): Likewise.
14953 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
14955 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14956 Alan Hayward <alan.hayward@arm.com>
14957 David Sherwood <david.sherwood@arm.com>
14959 * target.def (secondary_memory_needed_mode): New hook:
14960 * targhooks.c (default_secondary_memory_needed_mode): Declare.
14961 * targhooks.h (default_secondary_memory_needed_mode): New function.
14962 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
14963 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
14964 * doc/tm.texi: Regenerate.
14965 * lra-constraints.c (check_and_process_move): Use
14966 targetm.secondary_memory_needed_mode instead of
14967 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
14968 (curr_insn_transform): Likewise.
14969 * reload.c (get_secondary_mem): Likewise.
14970 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14971 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
14973 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14974 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14975 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
14976 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14977 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
14979 * config/powerpcspe/powerpcspe-protos.h
14980 (rs6000_secondary_memory_needed_mode): Delete.
14981 * config/powerpcspe/powerpcspe.c
14982 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14983 (rs6000_secondary_memory_needed_mode): Make static.
14984 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14985 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
14987 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14989 (rs6000_secondary_memory_needed_mode): Make static.
14990 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14991 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
14992 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14993 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14994 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14996 (sparc_secondary_memory_needed_mode): New function.
14997 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
14999 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
15001 * config/aarch64/constraints.md (Umq): New constraint.
15002 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15004 (mov<mode>): Update condition.
15006 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15008 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
15009 when bitposition is the same.
15011 2017-09-13 Richard Biener <rguenther@suse.de>
15013 * dwarf2out.c (output_die_symbol): Remove.
15014 (output_die): Do not output a DIEs symbol.
15016 2017-09-13 Richard Biener <rguenther@suse.de>
15018 PR middle-end/82128
15019 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
15020 default-def to avoid breaking iterator update with the weird
15021 interaction with cgraph_update_edges_for_call_stmt_node.
15023 2017-09-13 Richard Biener <rguenther@suse.de>
15025 * tree-cfg.c (verify_gimple_assign_binary): Add verification
15026 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
15027 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
15028 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
15030 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
15032 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
15033 Disable pc relative literal load irrespective of
15034 TARGET_FIX_ERR_A53_84341 for default.
15036 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
15038 * config/sparc/sparc.c (output_return): Output the source location of
15039 the insn in the delay slot, if any.
15040 (output_sibcall): Likewise.
15042 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
15045 * common.opt (-static-pie): New alias.
15046 (shared): Negate static-pie.
15047 (-no-pie): Update help text.
15049 (static-pie): New option.
15050 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
15051 -static-pie support.
15052 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
15053 (LINK_EH_SPEC): Likewise.
15054 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15055 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15056 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15057 * gcc.c (LINK_COMMAND_SPEC): Likewise.
15058 (init_gcc_specs): Likewise.
15059 (init_spec): Likewise.
15060 (display_help): Update help message for -pie.
15061 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
15064 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
15066 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
15067 (movdi_aarch64): Likewise.
15068 (movti_aarch64): Likewise.
15070 2017-09-12 Simon Wright <simon@pushface.org>
15073 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
15074 calculation of the minor version, always output as 0.
15076 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15079 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15080 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
15081 on it early, rather than manual conversion late. For
15082 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
15083 instead of performing manual conversion.
15085 2017-09-12 Carl Love <cel@us.ibm.com>
15087 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
15088 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
15090 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15091 VMULOSW): Add definitions.
15092 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15093 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15094 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15095 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15096 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15098 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15100 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
15102 (movti_aarch64): Likewise.
15103 (movdf_aarch64): Likewise.
15104 (movtf_aarch64): Likewise.
15105 (load_pairdi): Likewise.
15106 (store_pairdi): Likewise.
15107 (load_pairdf): Likewise.
15108 (store_pairdf): Likewise.
15109 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
15110 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
15111 (ldr_got_small_<mode>): Likewise.
15112 (ldr_got_small_28k_<mode>): Likewise.
15113 (ldr_got_tiny): Likewise.
15114 * config/aarch64/iterators.md (ldst_sz): New.
15115 (ldpstp_sz): Likewise.
15116 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
15118 (thunderx_load): Split load_8 to load_16.
15119 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
15121 (thunderx2t99_storepair_basic): Split store_8 to store_16.
15122 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
15123 (xgene1_store_pair): Split store_8 to store_16.
15124 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
15125 (falkor_st_0_st_sd): Split store_8 to store_16.
15127 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15129 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
15130 and store1/2/3/4 to store_4/8/12/16.
15131 * config/aarch64/aarch64.md: Update for rename.
15132 * config/arm/arm.md: Likewise.: Likewise.
15133 * config/arm/arm.c: Likewise.
15134 * config/arm/thumb1.md: Likewise.
15135 * config/arm/thumb2.md: Likewise.
15136 * config/arm/vfp.md: Likewise.
15137 * config/arm/arm-generic.md: Likewise.
15138 * config/arm/arm1020e.md: Likewise.
15139 * config/arm/arm1026ejs.md: Likewise.
15140 * config/arm/arm1136jfs.md: Likewise.
15141 * config/arm/arm926ejs.md: Likewise.
15142 * config/arm/cortex-a15.md: Likewise.
15143 * config/arm/cortex-a17.md: Likewise.
15144 * config/arm/cortex-a5.md: Likewise.
15145 * config/arm/cortex-a53.md: Likewise.
15146 * config/arm/cortex-a57.md: Likewise.
15147 * config/arm/cortex-a7.md: Likewise.
15148 * config/arm/cortex-a8.md: Likewise.
15149 * config/arm/cortex-a9.md: Likewise.
15150 * config/arm/cortex-m4.md: Likewise.
15151 * config/arm/cortex-m7.md: Likewise.
15152 * config/arm/cortex-r4.md: Likewise.
15153 * config/arm/exynos-m1.md: Likewise.
15154 * config/arm/fa526.md: Likewise.
15155 * config/arm/fa606te.md: Likewise.
15156 * config/arm/fa626te.md: Likewise.
15157 * config/arm/fa726te.md: Likewise.
15158 * config/arm/fmp626.md: Likewise.
15159 * config/arm/iwmmxt.md: Likewise.
15160 * config/arm/ldmstm.md: Likewise.
15161 * config/arm/marvell-pj4.md: Likewise.
15162 * config/arm/xgene1.md: Likewise.
15163 * config/aarch64/thunderx.md: Likewise.
15164 * config/aarch64/thunderx2t99.md: Likewise.
15165 * config/aarch64/falkor.md: Likewise.
15167 2017-09-12 Martin Liska <mliska@suse.cz>
15169 * attribs.c (private_lookup_attribute): New function.
15170 * attribs.h (private_lookup_attribute): Declared here.
15171 (lookup_attribute): Called from this place.
15173 2017-09-12 Richard Biener <rguenther@suse.de>
15175 PR tree-optimization/82157
15176 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
15177 stmts with side-effects.
15179 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15180 Alan Hayward <alan.hayward@arm.com>
15181 David Sherwood <david.sherwood@arm.com>
15183 * target.def (hard_regno_nregs): New hook.
15184 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
15185 * targhooks.h (default_hard_regno_nregs): Declare.
15186 * targhooks.c (default_hard_regno_nregs): New function.
15187 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
15188 (TARGET_HARD_REGNO_NREGS): ...this hook.
15189 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
15190 (CLASS_MAX_NREGS): Likewise.
15191 * doc/tm.texi: Regenerate.
15192 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
15193 instead of HARD_REGNO_NREGS.
15194 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
15195 HARD_REGNO_NREGS in the comment.
15196 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
15197 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
15198 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
15199 Return an unsigned int.
15200 (TARGET_HARD_REGNO_NREGS): Redefine.
15201 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
15202 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
15203 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15204 (arc_hard_regno_nregs): New function.
15205 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
15206 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
15207 (arm_hard_regno_nregs): New function.
15208 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
15209 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
15210 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
15211 (TARGET_HARD_REGNO_NREGS): Redefine.
15212 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
15213 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
15214 (HARD_REGNO_NREGS): Delete.
15215 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
15216 (cr16_hard_regno_nregs): New function.
15217 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
15218 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
15219 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
15220 (cris_hard_regno_nregs): New function.
15221 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
15222 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
15223 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15224 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
15225 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
15226 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
15227 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
15228 (frv_hard_regno_nregs): Make static. Take and return an
15230 (frv_class_max_nregs): Remove outdated copy of documentation.
15231 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
15232 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
15233 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
15234 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
15235 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
15236 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
15237 (TARGET_HARD_REGNO_NREGS): Redefine.
15238 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
15239 (CLASS_MAX_NREGS): Update comment.
15240 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
15241 (ia64_hard_regno_nregs): New function.
15242 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
15243 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
15244 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
15245 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
15246 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
15248 (m32c_hard_regno_nregs): Likewise. Make static.
15249 (TARGET_HARD_REGNO_NREGS): Redefine.
15250 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
15251 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
15252 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
15253 (m68k_hard_regno_nregs): New function.
15254 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
15255 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
15256 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
15257 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
15258 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
15259 Take and return an unsigned int.
15260 (TARGET_HARD_REGNO_NREGS): Redefine.
15261 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
15262 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15263 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
15264 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
15265 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
15266 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
15267 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
15268 (msp430_hard_regno_nregs): Make static. Take and return an
15270 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
15271 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
15272 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
15273 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
15274 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
15275 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
15276 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
15277 (TARGET_HARD_REGNO_NREGS): Redefine.
15278 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
15279 (PA_HARD_REGNO_NREGS): ...this.
15280 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
15281 (PA_HARD_REGNO_NREGS): ...this.
15282 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15283 (pa_hard_regno_nregs): New function.
15284 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
15285 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
15286 (pdp11_hard_regno_nregs): New function.
15287 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
15288 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
15289 (rs6000_hard_regno_nregs_hook): New function.
15290 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
15291 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
15292 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
15293 Take and return an unsigned int. Move earlier in file.
15294 (TARGET_HARD_REGNO_NREGS): Redefine.
15295 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
15296 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
15297 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
15298 (rl78_hard_regno_nregs): Make static. Take and return an
15300 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
15301 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
15302 (rs6000_hard_regno_nregs_hook): New function.
15303 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
15304 * config/rx/rx.c (rx_hard_regno_nregs): New function.
15305 (TARGET_HARD_REGNO_NREGS): Redefine.
15306 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
15307 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
15308 instead of HARD_REGNO_NREGS.
15309 (s390_hard_regno_nregs): New function.
15310 (s390_hard_regno_mode_ok): Add comment from s390.h.
15311 (TARGET_HARD_REGNO_NREGS): Redefine.
15312 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
15313 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
15314 (sh_hard_regno_nregs): New function.
15315 (sh_pass_in_reg_p): Use it.
15316 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
15317 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15318 (sparc_hard_regno_nregs): New function.
15319 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
15320 * config/spu/spu.c (spu_hard_regno_nregs): New function.
15321 (spu_function_arg_advance): Use it, supplying a valid register number.
15322 (TARGET_HARD_REGNO_NREGS): Redefine.
15323 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
15324 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
15325 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
15326 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
15327 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
15328 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
15329 (CLASS_MAX_NREGS): Remove copy of old documentation.
15330 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
15331 (visium_hard_regno_nregs): New function.
15332 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
15333 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
15334 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15335 xtensa_hard_regno_nregs): New function.
15336 * system.h (HARD_REGNO_NREGS): Poison.
15338 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15340 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
15341 hard_regno_nregs instead of HARD_REGNO_NREGS.
15342 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
15343 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
15344 (c6x_expand_epilogue): Likewise.
15345 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
15346 (frv_read_iacc_argument): Likewise.
15347 * config/sh/sh.c: Include regs.h.
15348 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15349 (regs_used): Likewise.
15350 (output_stack_adjust): Likewise.
15351 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
15352 * expmed.c: Include regs.h.
15353 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15354 * ree.c: Include regs.h.
15355 (combine_reaching_defs): Use hard_regno_nregs instead of
15357 (add_removable_extension): Likewise.
15359 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15361 * regs.h (hard_regno_nregs): Turn into a function.
15362 (end_hard_regno): Update accordingly.
15363 * caller-save.c (setup_save_areas): Likewise.
15364 (save_call_clobbered_regs): Likewise.
15365 (replace_reg_with_saved_mem): Likewise.
15366 (insert_restore): Likewise.
15367 (insert_save): Likewise.
15368 * combine.c (can_change_dest_mode): Likewise.
15369 (move_deaths): Likewise.
15370 (distribute_notes): Likewise.
15371 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
15372 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
15373 (rs6000_split_multireg_move): Likewise.
15374 (rs6000_register_move_cost): Likewise.
15375 (rs6000_memory_move_cost): Likewise.
15376 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
15377 (rs6000_split_multireg_move): Likewise.
15378 (rs6000_register_move_cost): Likewise.
15379 (rs6000_memory_move_cost): Likewise.
15380 * cselib.c (cselib_reset_table): Likewise.
15381 (cselib_lookup_1): Likewise.
15382 * emit-rtl.c (set_mode_and_regno): Likewise.
15383 * function.c (aggregate_value_p): Likewise.
15384 * ira-color.c (setup_profitable_hard_regs): Likewise.
15385 (check_hard_reg_p): Likewise.
15386 (calculate_saved_nregs): Likewise.
15387 (assign_hard_reg): Likewise.
15388 (improve_allocation): Likewise.
15389 (calculate_spill_cost): Likewise.
15390 * ira-emit.c (modify_move_list): Likewise.
15391 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
15392 (ira_hard_reg_in_set_p): Likewise.
15393 * ira.c (setup_reg_mode_hard_regset): Likewise.
15394 (clarify_prohibited_class_mode_regs): Likewise.
15395 (check_allocation): Likewise.
15396 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15397 (lra_setup_reg_renumber): Likewise.
15398 (setup_try_hard_regno_pseudos): Likewise.
15399 (spill_for): Likewise.
15400 (assign_hard_regno): Likewise.
15401 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
15402 * lra-constraints.c (in_class_p): Likewise.
15403 (lra_constraint_offset): Likewise.
15404 (simplify_operand_subreg): Likewise.
15405 (lra_constraints): Likewise.
15406 (split_reg): Likewise.
15407 (split_if_necessary): Likewise.
15408 (invariant_p): Likewise.
15409 (inherit_in_ebb): Likewise.
15410 * lra-lives.c (process_bb_lives): Likewise.
15411 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
15412 (get_hard_regs): Likewise.
15413 (do_remat): Likewise.
15414 * lra-spills.c (assign_spill_hard_regs): Likewise.
15415 * mode-switching.c (create_pre_exit): Likewise.
15416 * postreload.c (reload_combine_recognize_pattern): Likewise.
15417 * recog.c (peep2_find_free_register): Likewise.
15418 * regcprop.c (kill_value_regno): Likewise.
15419 (set_value_regno): Likewise.
15420 (copy_value): Likewise.
15421 (maybe_mode_change): Likewise.
15422 (find_oldest_value_reg): Likewise.
15423 (copyprop_hardreg_forward_1): Likewise.
15424 * regrename.c (check_new_reg_p): Likewise.
15425 (regrename_do_replace): Likewise.
15426 * reload.c (push_reload): Likewise.
15427 (combine_reloads): Likewise.
15428 (find_dummy_reload): Likewise.
15429 (operands_match_p): Likewise.
15430 (find_reloads): Likewise.
15431 (find_equiv_reg): Likewise.
15432 (reload_adjust_reg_for_mode): Likewise.
15433 * reload1.c (count_pseudo): Likewise.
15434 (count_spilled_pseudo): Likewise.
15435 (find_reg): Likewise.
15436 (clear_reload_reg_in_use): Likewise.
15437 (free_for_value_p): Likewise.
15438 (allocate_reload_reg): Likewise.
15439 (choose_reload_regs): Likewise.
15440 (reload_adjust_reg_for_temp): Likewise.
15441 (emit_reload_insns): Likewise.
15442 (delete_output_reload): Likewise.
15443 * rtlanal.c (subreg_get_info): Likewise.
15444 * sched-deps.c (sched_analyze_reg): Likewise.
15445 * sel-sched.c (init_regs_for_mode): Likewise.
15446 (mark_unavailable_hard_regs): Likewise.
15447 (choose_best_reg_1): Likewise.
15448 (verify_target_availability): Likewise.
15449 * valtrack.c (dead_debug_insert_temp): Likewise.
15450 * var-tracking.c (track_loc_p): Likewise.
15451 (emit_note_insn_var_location): Likewise.
15452 * varasm.c (make_decl_rtl): Likewise.
15453 * reginfo.c (choose_hard_reg_mode): Likewise.
15454 (init_reg_modes_target): Refer directly to
15455 this_target_regs->x_hard_regno_nregs.
15457 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15459 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
15460 instead of hard_regno_nregs.
15462 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15464 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
15465 end_hard_regno instead of hard_regno_nregs.
15466 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
15467 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
15468 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
15469 * ira-color.c (improve_allocation): Likewise.
15470 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15471 * lra-lives.c (mark_regno_live): Likewise.
15472 (mark_regno_dead): Likewise.
15473 * lra-remat.c (operand_to_remat): Likewise.
15474 * lra.c (collect_non_operand_hard_regs): Likewise.
15475 * postreload.c (reload_combine_note_store): Likewise.
15476 (move2add_valid_value_p): Likewise.
15477 * reload.c (regno_clobbered_p): Likewise.
15479 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15481 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
15483 * config/v850/v850.c (v850_reorg): Likewise.
15484 * reload.c (refers_to_regno_for_reload_p): Likewise.
15485 (find_equiv_reg): Likewise.
15486 * reload1.c (reload_reg_reaches_end_p): Likewise.
15488 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15490 * caller-save.c (add_used_regs): Use REG_NREGS instead of
15492 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
15493 * config/arm/arm.c (output_move_neon): Likewise.
15494 (arm_attr_length_move_neon): Likewise.
15495 (neon_split_vcombine): Likewise.
15496 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
15497 (c6x_mark_reg_written): Likewise.
15498 (c6x_dwarf_register_span): Likewise.
15499 * config/i386/i386.c (ix86_save_reg): Likewise.
15500 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
15501 (rws_access_reg): Likewise.
15502 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
15503 * mode-switching.c (create_pre_exit): Likewise.
15504 * ree.c (combine_reaching_defs): Likewise.
15505 (add_removable_extension): Likewise.
15506 * regcprop.c (find_oldest_value_reg): Likewise.
15507 (copyprop_hardreg_forward_1): Likewise.
15508 * reload.c (reload_inner_reg_of_subreg): Likewise.
15509 (push_reload): Likewise.
15510 (combine_reloads): Likewise.
15511 (find_dummy_reload): Likewise.
15512 (reload_adjust_reg_for_mode): Likewise.
15513 * reload1.c (find_reload_regs): Likewise.
15514 (forget_old_reloads_1): Likewise.
15515 (reload_reg_free_for_value_p): Likewise.
15516 (reload_adjust_reg_for_temp): Likewise.
15517 (emit_reload_insns): Likewise.
15518 (delete_output_reload): Likewise.
15519 * sel-sched.c (choose_best_reg_1): Likewise.
15520 (choose_best_pseudo_reg): Likewise.
15522 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15523 Alan Hayward <alan.hayward@arm.com>
15524 David Sherwood <david.sherwood@arm.com>
15526 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
15527 * target.def (slow_unaligned_access): New hook.
15528 * targhooks.h (default_slow_unaligned_access): Declare.
15529 * targhooks.c (default_slow_unaligned_access): New function.
15530 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
15531 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
15532 * doc/tm.texi: Regenerate.
15533 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
15534 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
15535 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
15537 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
15538 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
15540 (rs6000_slow_unaligned_access): New function.
15541 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15542 (expand_block_compare): Likewise.
15543 (expand_strn_compare): Likewise.
15544 (rs6000_rtx_costs): Likewise.
15545 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
15546 (riscv_slow_unaligned_access): Likewise.
15547 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
15548 (riscv_slow_unaligned_access_p): ...this and make static.
15549 (riscv_option_override): Update accordingly.
15550 (riscv_slow_unaligned_access): New function.
15551 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15552 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
15553 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15554 (rs6000_slow_unaligned_access): New function.
15555 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15556 (rs6000_rtx_costs): Likewise.
15557 * config/rs6000/rs6000-string.c (expand_block_compare)
15558 (expand_strn_compare): Use targetm.slow_unaligned_access instead
15559 of SLOW_UNALIGNED_ACCESS.
15560 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
15561 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
15562 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
15563 of SLOW_UNALIGNED_ACCESS.
15564 * expmed.c (simple_mem_bitfield_p): Likewise.
15565 * expr.c (alignment_for_piecewise_move): Likewise.
15566 (emit_group_load_1): Likewise.
15567 (emit_group_store): Likewise.
15568 (copy_blkmode_from_reg): Likewise.
15569 (emit_push_insn): Likewise.
15570 (expand_assignment): Likewise.
15571 (store_field): Likewise.
15572 (expand_expr_real_1): Likewise.
15573 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
15574 * lra-constraints.c (simplify_operand_subreg): Likewise.
15575 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15576 * gimple-ssa-store-merging.c: Likewise in block comment at start
15578 * tree-ssa-strlen.c: Include target.h.
15579 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
15580 of SLOW_UNALIGNED_ACCESS.
15581 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
15583 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15585 PR rtl-optimization/82185
15586 * expmed.c (emit_store_flag_int): Only test tem if it has been
15589 2017-09-12 Richard Biener <rguenther@suse.de>
15591 PR middle-end/82149
15592 * match.pd ((FTYPE) N CMP CST): Fix typo.
15594 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
15596 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
15598 (mips_near_type_p): Add 'short_call' attribute as a synonym
15600 * doc/extend.texi (short_call): Document new function attribute.
15602 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15605 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
15606 assertion check that in the condition.
15607 (get_atomic_generic_size): Likewise. Before testing if parameter
15608 has pointer type, if it has array type, call for C++
15609 default_conversion to perform array-to-pointer conversion.
15611 2017-09-12 Richard Biener <rguenther@suse.de>
15613 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
15614 for operations we cannot scalarize.
15616 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
15618 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
15619 vectors heap vectors. Clean up comments.
15620 Make visited_bbs a reference.
15621 (profitable_jump_thread_path): Make GC
15622 vectors heap vectors. Clean up comments.
15624 (convert_and_register_jump_thread_path): Make GC vectors heap
15626 (check_subpath_and_update_thread_path): Same. Clean up comments.
15627 Make visited_bbs a reference.
15628 (handle_phi): Abstract common code to to
15629 register_jump_thread_path_if_profitable.
15630 Rename VAR_BB to DEF_BB.
15632 Make GC vectors heap vectors.
15633 Make visited_bbs a reference.
15634 (handle_assignment): Same.
15635 (register_jump_thread_path_if_profitable): New.
15636 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
15638 Make GC vectors heap vectors. Clean up comments.
15639 Make visited_bbs a reference.
15640 (find_jump_threads_backwards): Make visited_bbs live in the stack.
15641 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
15644 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
15647 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
15648 words of E_DImode object are reachable by xtensa_uimm8x4 access.
15650 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
15652 Revert r251800 and r251799.
15654 2017-09-11 Martin Jambor <mjambor@suse.cz>
15657 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
15658 arguments in advance.
15659 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
15660 use it to find predecessor edges.
15661 (naive_outof_ssa): Collect vector of predecessors.
15663 2017-09-08 Jason Merrill <jason@redhat.com>
15665 PR c++/70029 - ICE with ref-qualifier and -flto
15666 * langhooks.h (struct lang_hooks_for_types): Add
15667 copy_lang_qualifiers.
15668 * attribs.c (build_type_attribute_qual_variant): Use it.
15669 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
15671 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
15672 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
15674 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
15677 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
15678 (*mulsi3_sp64): New instruction.
15679 (mulsi3): New expander.
15681 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
15683 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
15685 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15687 * sancov.c: Include memmodel.h.
15689 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
15692 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
15695 2017-09-07 Carl Love <cel@us.ibm.com>
15697 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
15698 the sldi instruction.
15700 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
15702 * sancov.c: Include tm_p.h.
15704 2017-09-07 Jakub Jelinek <jakub@redhat.com>
15707 * output.h (switch_to_other_text_partition): New declaration.
15708 * varasm.c (switch_to_other_text_partition): New function.
15709 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
15710 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
15711 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
15712 to the other text partition before emitting LCL label and switch back
15713 after emitting the word after it.
15715 2017-09-07 Richard Biener <rguenther@suse.de>
15717 * passes.def (pass_split_crit_edges): Remove instance before PRE.
15718 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
15719 critical edges here, after loop init.
15720 (pass_data_pre): Remove PROP_no_crit_edges flags.
15721 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
15722 for valueization of call args to avoid leaking VN_TOP.
15723 (visit_use): Assert we do not visit default defs.
15724 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
15725 Use error_mark_node to more easily detect leaking VN_TOP.
15726 All default-defs are varying, not VN_TOP. Mark them visited.
15727 (run_scc_vn): Make code match comment.
15729 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15731 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
15732 OPTION_MASK_FLOAT128_KEYWORD.
15733 (POWERPC_MASKS): Likewise.
15734 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
15735 support for the -mfloat128-type option, and make -mfloat128
15736 default on PowerPC Linux systems. Define or undefine
15737 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
15738 Define __float128 to be __ieee128 if IEEE 128-bit support is
15739 enabled, or undefine it.
15740 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
15741 Delete defining __FLOAT128_TYPE__.
15742 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
15743 -mfloat128-type option and make -mfloat128 default on PowerPC
15745 (TARGET_FLOAT128_TYPE): Likewise.
15746 (-mfloat128-type): Likewise.
15747 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15748 Delete the -mfloat128-type option and make -mfloat128 default on
15749 PowerPC Linux systems. Always use __ieee128 to be the keyword for
15750 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
15751 128-bit floating point is enabled. Change tests from using
15752 -mfloat128-type to -mfloat128.
15753 (rs6000_mangle_type): Use the correct mangling for the __float128
15754 type even if normal long double is restricted to 64-bits.
15755 (floatn_mode): Enable the _Float128 type by default on VSX Linux
15757 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
15758 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
15759 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
15761 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
15762 documentation for -mfloat128.
15764 2017-09-06 Olivier Hainque <hainque@adacore.com>
15766 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
15768 2017-09-06 Wish Wu <wishwu007@gmail.com>
15769 Jakub Jelinek <jakub@redhat.com>
15771 * asan.c (initialize_sanitizer_builtins): Add
15772 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
15773 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
15774 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
15775 BT_FN_VOID_UINT64_PTR variables.
15776 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
15777 (BT_FN_VOID_UINT16_UINT16): Likewise.
15778 (BT_FN_VOID_UINT32_UINT32): Likewise.
15779 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
15780 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
15781 (BT_FN_VOID_UINT64_PTR): Likewise.
15782 * common.opt (flag_sanitize_coverage): New variable.
15783 (fsanitize-coverage=trace-pc): Remove.
15784 (fsanitize-coverage=): Add.
15785 * flag-types.h (enum sanitize_coverage_code): New enum.
15786 * fold-const.c (fold_range_test): Disable non-short-circuit
15787 optimization if flag_sanitize_coverage.
15788 (fold_truth_andor): Likewise.
15789 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15790 * opts.c (COVERAGE_SANITIZER_OPT): Define.
15791 (coverage_sanitizer_opts): New array.
15792 (get_closest_sanitizer_option): Add OPTS argument, handle also
15793 OPT_fsanitize_coverage_.
15794 (parse_sanitizer_options): Adjusted to also handle
15795 OPT_fsanitize_coverage_.
15796 (common_handle_option): Add OPT_fsanitize_coverage_.
15797 * sancov.c (instrument_comparison, instrument_switch): New function.
15798 (sancov_pass): Add trace-cmp support.
15799 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
15800 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
15801 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
15802 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
15803 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
15804 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
15805 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
15806 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
15807 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
15808 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
15810 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15812 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
15813 error. Only quit immediately if parsing is complete.
15814 (BEGIN): Initialize fatal_err and parse_done.
15815 (begin fpu, end fpu): Check number of arguments.
15816 (begin arch, end arch): Likewise.
15817 (begin cpu, end cpu): Likewise.
15818 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
15819 (optalias): Likewise.
15821 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15823 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
15824 * config/arm/arm-isa.h: Delete. Move definitions to ...
15825 * arm-cpus.in: ... here. Use new feature and fgroup values.
15826 * config/arm/arm.c (arm_option_override): Use lower case for feature
15828 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
15829 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
15830 * config/arm/parsecpu.awk (END): Add new command 'isa'.
15832 (print_isa_bits_for): New function.
15833 (gen_isa): New function.
15834 (gen_comm_data): Use print_isa_bits_for.
15835 (define feature): New keyword.
15836 (define fgroup): New keyword.
15837 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
15838 (arm-isa.h): Add rule to generate file.
15839 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
15840 case for feature bit names.
15842 2017-09-06 Richard Biener <rguenther@suse.de>
15844 * tree-ssa-pre.c (NECESSARY): Remove.
15845 (create_expression_by_pieces): Do not touch pass-local flags.
15846 (insert_into_preds_of_block): Likewise.
15847 (do_pre_regular_insertion): Likewise.
15848 (eliminate_insert): Likewise.
15849 (eliminate_dom_walker::before_dom_children): Likewise.
15850 (fini_eliminate): Do not look at inserted_exprs.
15851 (mark_operand_necessary): Remove.
15852 (remove_dead_inserted_code): Replace with simple work-list
15853 algorithm based on inserted_exprs and SSA uses.
15854 (pass_pre::execute): Re-order fini_eliminate and
15855 remove_dead_inserted_code.
15857 2017-09-06 Olivier Hainque <hainque@adacore.com>
15859 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
15860 for VxWorks 7. Adjust surrounding comments.
15862 2017-09-06 Richard Biener <rguenther@suse.de>
15864 * gimple-ssa-strength-reduction.c
15865 (find_candidates_dom_walker::before_dom_children): Also allow
15868 2017-09-06 Richard Biener <rguenther@suse.de>
15870 PR tree-optimization/82108
15871 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
15872 for gap in the non-permutation SLP case.
15874 2017-09-06 Martin Jambor <mjambor@suse.cz>
15876 PR tree-optimization/82078
15877 * tree-sra.c (sort_and_splice_var_accesses): Move call to
15878 add_access_to_work_queue...
15879 (build_accesses_from_assign): ...here.
15880 (propagate_all_subaccesses): Make sure racc is the group
15881 representative, if there is one.
15883 2017-09-06 Jakub Jelinek <jakub@redhat.com>
15885 PR middle-end/82095
15886 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
15889 2017-09-06 Richard Biener <rguenther@suse.de>
15891 * gimple-ssa-strength-reduction.c
15892 (find_candidates_dom_walker::before_doom_children): Use a
15893 type and not a mode check.
15895 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
15898 * config/arm/predicates.md (arm_general_adddi_operand): Create new
15900 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
15902 2017-09-05 Jeff Law <law@redhat.com>
15904 PR tree-optimization/64910
15905 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
15906 cases where we have 3 or more operands.
15908 2017-09-05 Jakub Jelinek <jakub@redhat.com>
15910 PR middle-end/81768
15911 * omp-low.c (lower_omp_for): Recompute tree invariant if
15912 gimple_omp_for_initial/final is ADDR_EXPR.
15914 PR middle-end/81768
15915 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
15916 into gimple val before gimplification fo the COND_EXPR.
15918 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
15920 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
15921 REGION_COPY argument.
15922 (thread_through_all_blocks): Remove unused argument to
15923 duplicate_thread_path.
15925 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15926 Alan Hayward <alan.hayward@arm.com>
15927 David Sherwood <david.sherwood@arm.com>
15929 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
15930 Take a scalar_mode rather than a machine_mode.
15931 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15932 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
15933 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15934 (aarch64_gen_adjusted_ldpstp): Likewise.
15935 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
15937 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15938 Alan Hayward <alan.hayward@arm.com>
15939 David Sherwood <david.sherwood@arm.com>
15941 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
15942 Take a scalar_int_mode instead of a machine_mode.
15943 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15944 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15945 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15946 (aarch64_simd_attr_length_rglist): Delete.
15947 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
15948 a scalar_int_mode instead of a machine_mode.
15949 (aarch64_add_offset): Likewise.
15950 (aarch64_internal_mov_immediate): Likewise
15951 (aarch64_add_constant_internal): Likewise.
15952 (aarch64_add_constant): Likewise.
15953 (aarch64_movw_imm): Likewise.
15954 (aarch64_rtx_arith_op_extract_p): Likewise.
15955 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15956 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15957 Remove assert that the mode isn't a vector.
15958 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15959 (aarch64_expand_mov_immediate): Update calls after above changes.
15960 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
15961 (aarch64_and_bitmask_imm): Check for scalar integer modes.
15962 (aarch64_move_imm): Likewise.
15963 (aarch64_can_const_movi_rtx_p): Likewise.
15964 (aarch64_strip_extend): Likewise.
15965 (aarch64_extr_rtx_p): Likewise.
15966 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
15967 a CONST_INT when the mode parameter is VOIDmode.
15968 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
15970 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15972 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
15973 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15974 (bitwise_type_for_mode): Update accordingly.
15976 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15978 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
15979 * stor-layout.c (mode_for_size_tree): Likewise.
15980 (mode_for_array): Update accordingly.
15981 (layout_decl): Likewise.
15982 (compute_record_mode): Likewise. Only set the mode once.
15984 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15986 * target.def (get_mask_mode): Change return type to opt_mode.
15988 * doc/tm.texi: Regenerate.
15989 * targhooks.h (default_get_mask_mode): Return an opt_mode.
15990 * targhooks.c (default_get_mask_mode): Likewise.
15991 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15992 * optabs-query.c (can_vec_mask_load_store_p): Update use of
15993 targetm.get_mask_mode.
15994 * tree.c (build_truth_vector_type): Likewise.
15996 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15998 * machmode.h (mode_for_vector): Return an opt_mode.
15999 * stor-layout.c (mode_for_vector): Likewise.
16000 (mode_for_int_vector): Update accordingly.
16001 (layout_type): Likewise.
16002 * config/i386/i386.c (emit_memmov): Likewise.
16003 (ix86_expand_set_or_movmem): Likewise.
16004 (ix86_expand_vector_init): Likewise.
16005 (ix86_get_mask_mode): Likewise.
16006 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
16008 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
16009 * expmed.c (extract_bit_field_1): Likewise.
16010 * expr.c (expand_expr_real_2): Likewise.
16011 * optabs-query.c (can_vec_perm_p): Likewise.
16012 (can_vec_mask_load_store_p): Likewise.
16013 * optabs.c (expand_vec_perm): Likewise.
16014 * targhooks.c (default_get_mask_mode): Likewise.
16015 * tree-vect-stmts.c (vectorizable_store): Likewise.
16016 (vectorizable_load): Likewise.
16017 (get_vectype_for_scalar_type_and_size): Likewise.
16019 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16021 * machmode.h (mode_for_int_vector): New function.
16022 * stor-layout.c (mode_for_int_vector): Likewise.
16023 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
16024 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
16025 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
16026 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
16027 (s390_expand_vcond): Likewise.
16029 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16031 * machmode.h (opt_machine_mode): New type.
16032 (opt_mode<T>): Allow construction from anything that can be
16034 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
16035 (mode_for_size): Return an opt_machine_mode.
16036 * stor-layout.c (mode_for_size): Likewise.
16037 (mode_for_size_tree): Update call accordingly.
16038 (bitwise_mode_for_mode): Likewise.
16039 (make_fract_type): Likewise.
16040 (make_accum_type): Likewise.
16041 * caller-save.c (replace_reg_with_saved_mem): Update call
16043 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16044 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16045 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16046 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16047 * expmed.c (extract_bit_field_1): Likewise.
16048 * reload.c (get_secondary_mem): Likewise.
16049 * varasm.c (assemble_integer): Likewise.
16050 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
16053 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16055 * machmode.h (decimal_float_mode_for_size): New function.
16056 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
16057 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
16058 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
16059 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
16060 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
16061 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
16063 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16065 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
16066 (get_builtin_sync_mode): Likewise.
16067 (expand_ifn_atomic_compare_exchange): Likewise.
16068 (expand_builtin_atomic_clear): Likewise.
16069 (expand_builtin_atomic_test_and_set): Likewise.
16070 (fold_builtin_atomic_always_lock_free): Likewise.
16071 * calls.c (compute_argument_addresses): Likewise.
16072 (emit_library_call_value_1): Likewise.
16073 (store_one_arg): Likewise.
16074 * combine.c (combine_instructions): Likewise.
16075 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
16076 * config/arm/arm.c (arm_function_value): Likewise.
16077 (aapcs_allocate_return_reg): Likewise.
16078 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
16079 * config/i386/i386.c (construct_container): Likewise.
16080 (ix86_gimplify_va_arg): Likewise.
16081 (ix86_expand_sse_cmp): Likewise.
16082 (emit_memmov): Likewise.
16083 (emit_memset): Likewise.
16084 (expand_small_movmem_or_setmem): Likewise.
16085 (ix86_expand_pextr): Likewise.
16086 (ix86_expand_pinsr): Likewise.
16087 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
16088 * config/microblaze/microblaze.c (microblaze_block_move_straight):
16090 * config/mips/mips.c (mips_function_value_1) Likewise.
16091 (mips_block_move_straight): Likewise.
16092 (mips_expand_ins_as_unaligned_store): Likewise.
16093 * config/powerpcspe/powerpcspe.c
16094 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16095 (rs6000_darwin64_record_arg_flush): Likewise.
16096 * config/rs6000/rs6000.c
16097 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16098 (rs6000_darwin64_record_arg_flush): Likewise.
16099 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
16100 (sparc_function_value_1): Likewise.
16101 * config/spu/spu.c (adjust_operand): Likewise.
16102 (spu_emit_branch_or_set): Likewise.
16103 (arith_immediate_p): Likewise.
16104 * emit-rtl.c (gen_lowpart_common): Likewise.
16105 * expr.c (expand_expr_real_1): Likewise.
16106 * function.c (assign_parm_setup_block): Likewise.
16107 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
16108 * reload1.c (alter_reg): Likewise.
16109 * stor-layout.c (mode_for_vector): Likewise.
16110 (layout_type): Likewise.
16112 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16114 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
16115 (spu_convert_move): Likewise.
16116 * lower-subreg.c (resolve_simple_move): Likewise.
16118 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16121 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
16122 define_insn to a define_expand.
16123 (altivec_vsum2sws_direct): New define_insn.
16124 (altivec_vsumsws): Convert from a define_insn to a define_expand.
16126 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16128 * config/arm/arm.c (arm_option_params_internal): Improve setting of
16131 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
16136 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
16137 realign stack if stack alignment needed is less than incoming
16140 2017-09-05 Marek Polacek <polacek@redhat.com>
16143 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
16146 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16148 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
16150 2017-09-05 Richard Biener <rguenther@suse.de>
16152 PR tree-optimization/82084
16153 * fold-const.c (can_native_encode_string_p): Handle wide characters.
16155 2017-09-05 Richard Biener <rguenther@suse.de>
16157 PR tree-optimization/82102
16158 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
16160 2017-09-05 Martin Liska <mliska@suse.cz>
16162 PR tree-optimization/82032
16163 * tree-cfg.c (generate_range_test): New function.
16164 * tree-cfg.h (generate_range_test): Declared here.
16165 * tree-cfgcleanup.c (convert_single_case_switch): New function.
16166 (cleanup_control_expr_graph): Use it.
16167 * tree-switch-conversion.c (try_switch_expansion): Remove
16169 (emit_case_nodes): Use generate_range_test.
16171 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16174 * config/i386/i386.md (*<btsc><mode>_mask): Add
16175 TARGET_USE_BT to insn constraint.
16176 (*btr<mode>_mask): Ditto.
16178 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
16180 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
16181 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
16183 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
16186 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
16187 TARGET_NEON and TARGET_IWMMXT.
16188 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
16189 TARGET_NEON and TARGET_IWMMXT.
16190 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
16192 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16194 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
16195 (ix86_rewrite_tls_address): Ditto.
16196 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
16197 (ix86_rewrite_tls_address_1): Ditto.
16198 (ix86_rewrite_tls_address): Ditto.
16199 * config/i386/predicates.md (tls_address_pattern): New predicate.
16200 * config/i386/i386.md (TLS address splitter): New splitter.
16202 2017-09-04 Richard Biener <rguenther@suse.de>
16204 PR tree-optimization/82084
16205 * fold-const.h (can_native_encode_string_p): Declare.
16206 * fold-const.c (can_native_encode_string_p): Factor out from ...
16207 (native_encode_string): ... here.
16208 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
16209 vectorizing stores from constants we later cannot handle.
16211 2017-09-04 Marek Polacek <polacek@redhat.com>
16214 * doc/invoke.texi: Update -Wtautological-compare documentation.
16216 2017-09-04 Jeff Law <law@redhat.com>
16218 PR tree-optimization/64910
16219 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
16220 swap the first and last operand if the last is a constant.
16222 2017-09-04 Marek Polacek <polacek@redhat.com>
16225 * convert.c (do_narrow): When sanitizing signed integer overflows,
16226 bail out for signed types.
16227 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
16229 2017-09-04 Richard Biener <rguenther@suse.de>
16231 PR tree-optimization/82060
16232 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16233 Move devirtualization after stmt folding and before EH/AB/noreturn
16234 cleanup to get the stmt refs canonicalized. Use a bool instead
16235 of gimple_modified_p since that doesn't work for NOPs. Schedule
16236 NOPs generated by folding for removal.
16238 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16239 Alan Hayward <alan.hayward@arm.com>
16240 David Sherwood <david.sherwood@arm.com>
16242 * coretypes.h (pad_direction): New enum.
16243 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
16244 (FUNCTION_ARG_PADDING): Likewise.
16245 * target.def (function_arg_padding): New hook.
16246 * targhooks.h (default_function_arg_padding): Declare.
16247 * targhooks.c (default_function_arg_padding): New function.
16248 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
16249 (TARGET_FUNCTION_ARG_PADDING): ...this.
16250 * doc/tm.texi: Regenerate.
16251 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
16252 instead of direction.
16253 (compute_argument_addresses): Likewise.
16254 (load_register_parameters): Likewise.
16255 (emit_library_call_value_1): Likewise.
16256 (store_one_arg): Use targetm.calls.function_arg_padding instead
16257 of FUNCTION_ARG_PADDING.
16258 (must_pass_in_stack_var_size_or_pad): Likewise.
16259 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
16260 (emit_group_store): Likewise.
16261 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
16262 instead of FUNCTION_ARG_PADDING.
16263 (emit_push_insn): Likewise, and propagate enum change throughout
16265 * function.h (direction): Delete.
16266 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
16268 * function.c (assign_parm_find_stack_rtl): Likewise.
16269 (assign_parm_setup_block_p): Likewise.
16270 (assign_parm_setup_block): Likewise.
16271 (gimplify_parameters): Likewise.
16272 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
16273 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
16275 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
16276 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16277 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
16278 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
16279 (aarch64_function_arg_padding): ...this new function.
16280 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
16281 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16282 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
16283 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16284 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
16285 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16286 (arm_pad_arg_upward): Replace with...
16287 (arm_function_arg_padding): ...this new function.
16288 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
16290 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
16291 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
16292 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16293 (ia64_hpux_function_arg_padding): Replace with...
16294 (ia64_function_arg_padding): ...this new function. Use pad_direction
16295 instead of direction. Check for TARGET_HPUX.
16296 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
16297 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16298 (iq2000_function_arg_padding): New function.
16299 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
16300 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
16301 (mips_function_arg_padding): ...this new function.
16302 (mips_pad_reg_upward): Update accordingly.
16303 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16304 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
16305 targetm.calls.function_arg_padding.
16306 (FUNCTION_ARG_PADDING): Delete.
16307 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16308 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
16309 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16310 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
16311 (nios2_block_reg_padding): Return pad_direction instead of direction.
16312 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
16313 instead of direction.
16314 (nios2_function_arg_padding): Likewise. Make static.
16315 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16316 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
16317 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
16318 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
16319 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16320 (pa_function_arg_padding): Make static. Return pad_direction instead
16322 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
16323 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16324 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
16325 instead of direction. Use targetm.calls.function_arg_padding.
16326 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
16327 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16328 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
16329 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
16330 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
16332 (function_arg_padding): Rename to...
16333 (rs6000_function_arg_padding): ...this. Make static. Return
16334 pad_direction instead of direction.
16335 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16336 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
16337 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16338 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
16339 instead of direction. Use targetm.calls.function_arg_padding.
16340 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
16341 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16342 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
16343 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
16344 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16345 (function_arg_padding): Rename to...
16346 (rs6000_function_arg_padding): ...this. Make static. Return
16347 pad_direction instead of direction.
16348 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16349 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
16350 * config/s390/s390.c (s390_function_arg_padding): New function.
16351 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16352 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
16353 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
16354 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16355 (function_arg_padding): Rename to...
16356 (sparc_function_arg_padding): ...this. Make static. Return
16357 pad_direction instead of direction.
16358 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
16359 * config/spu/spu.c (spu_function_arg_padding): New function.
16360 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16361 * system.h (FUNCTION_ARG_PADDING): Poison.
16363 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16364 Alan Hayward <alan.hayward@arm.com>
16365 David Sherwood <david.sherwood@arm.com>
16367 * target.def (modes_tieable_p): New hook.
16368 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
16369 (TARGET_MODES_TIEABLE_P): ...this.
16370 * doc/tm.texi.in: Regenerate.
16371 * hooks.h (hook_bool_mode_mode_true): Declare.
16372 * hooks.c (hook_bool_mode_mode_true): New function.
16373 * combine.c (subst): Use targetm.modes_tieable_p instead of
16375 * dse.c (find_shift_sequence): Likewise.
16376 * expmed.c (extract_low_bits): Likewise.
16377 * lower-subreg.c: Include target.h.
16378 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
16380 * rtlanal.c (rtx_cost): Likewise.
16381 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
16382 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
16383 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
16384 (TARGET_MODES_TIEABLE_P): Redefine.
16385 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
16386 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
16387 (TARGET_MODES_TIEABLE_P): Redefine.
16388 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
16389 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
16390 (arc_modes_tieable_p): New function.
16391 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
16392 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
16393 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
16394 (arm_modes_tieable_p): Make static.
16395 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
16396 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
16397 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
16398 (TARGET_MODES_TIEABLE_P): Redefine.
16399 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
16400 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
16401 (TARGET_MODES_TIEABLE_P): Redefine.
16402 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
16403 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
16404 (cr16_modes_tieable_p): New function.
16405 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
16406 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
16407 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
16408 (TRULY_NOOP_TRUNCATION): Update comment.
16409 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
16410 (TRULY_NOOP_TRUNCATION): Update comment.
16411 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
16412 (frv_modes_tieable_p): New function.
16413 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
16414 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
16415 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
16416 (TARGET_MODES_TIEABLE_P): Redefine.
16417 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
16418 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
16419 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
16420 (TARGET_MODES_TIEABLE_P): Redefine.
16421 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
16422 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
16423 (ia64_modes_tieable_p): New function.
16424 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
16425 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
16426 (iq2000_modes_tieable_p): New function.
16427 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
16428 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
16429 (lm32_modes_tieable_p): New function.
16430 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
16431 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
16432 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
16433 (TARGET_MODES_TIEABLE_P): Redefine.
16434 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
16435 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
16436 (m32r_modes_tieable_p): New function.
16437 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
16438 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
16439 (m68k_modes_tieable_p): New function.
16440 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
16441 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
16442 (mcore_modes_tieable_p): New function.
16443 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
16444 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
16446 (TARGET_MODES_TIEABLE_P): Redefine.
16447 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
16448 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
16449 * config/mips/mips.c (mips_modes_tieable_p): Make static.
16450 (TARGET_MODES_TIEABLE_P): Redefine.
16451 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
16452 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
16453 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
16454 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
16455 (mn10300_modes_tieable_p): ...this and make static.
16456 (TARGET_MODES_TIEABLE_P): Redefine.
16457 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
16458 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
16459 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
16460 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
16461 (msp430_modes_tieable_p): Make static.
16462 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
16463 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
16464 (TARGET_MODES_TIEABLE_P): Redefine.
16465 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
16466 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
16467 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
16468 (TARGET_MODES_TIEABLE_P): Redefine.
16469 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
16470 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
16471 * config/pa/pa.c (pa_modes_tieable_p): Make static.
16472 (TARGET_MODES_TIEABLE_P): Redefine.
16473 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
16474 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
16475 (pdp11_modes_tieable_p): New function.
16476 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
16477 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
16478 (rs6000_modes_tieable_p): New function.
16479 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16480 * config/powerpcspe/powerpcspe.md: Update comment.
16481 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
16482 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
16483 (TARGET_MODES_TIEABLE_P): Redefine.
16484 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
16485 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
16486 (rl78_modes_tieable_p): New function.
16487 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
16488 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
16489 (rs6000_modes_tieable_p): New function.
16490 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16491 * config/rs6000/rs6000.md: Update comment.
16492 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
16493 * config/rx/rx.c (rx_modes_tieable_p): New function.
16494 (TARGET_MODES_TIEABLE_P): Redefine.
16495 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
16496 * config/s390/s390.c (s390_modes_tieable_p): New function.
16497 (TARGET_MODES_TIEABLE_P): Redefine.
16498 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
16499 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
16500 (sh_modes_tieable_p): New function.
16501 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
16502 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
16503 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
16504 (sparc_modes_tieable_p): Make static.
16505 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
16506 * config/spu/spu.c (spu_modes_tieable_p): New function.
16507 (TARGET_MODES_TIEABLE_P): Redefine.
16508 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
16509 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
16510 (TARGET_MODES_TIEABLE_P): Redefine.
16511 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
16512 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
16513 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
16514 * config/v850/v850.c (v850_modes_tieable_p): New function.
16515 (TARGET_MODES_TIEABLE_P): Redefine.
16516 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
16517 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
16518 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
16519 (visium_modes_tieable_p): New function.
16520 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
16521 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
16522 (xtensa_modes_tieable_p): New function.
16523 * system.h (MODES_TIEABLE_P): Poison.
16525 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16526 Alan Hayward <alan.hayward@arm.com>
16527 David Sherwood <david.sherwood@arm.com>
16529 * target.def (hard_regno_mode_ok): New hook.
16530 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
16531 (TARGET_HARD_REGNO_MODE_OK): ...this.
16532 * doc/tm.texi.in: Regenerate.
16533 * hooks.h (hook_bool_uint_mode_true): Declare.
16534 * hooks.c (hook_bool_uint_mode_true): New function.
16535 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
16536 HARD_REGNO_MODE_OK.
16537 * genpreds.c (write_insn_preds_c): Add an include of target.h.
16538 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
16539 instead of HARD_REGNO_MODE_OK.
16540 * caller-save.c: Include target.h.
16541 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
16542 HARD_REGNO_MODE_OK.
16543 * combine.c (can_combine_p): Likewise.
16544 (combinable_i3pat): Likewise.
16545 (can_change_dest_mode): Likewise.
16546 * expr.c (init_expr_target): Likewise.
16547 (convert_move): Likewise.
16548 (convert_modes): Likewise.
16549 * ira.c (setup_prohibited_class_mode_regs): Likewise.
16550 (setup_prohibited_mode_move_regs): Likewise.
16551 * ira.h (target_ira): Likewise.
16552 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16553 * lra-constraints.c (process_alt_operands): Likewise.
16554 (split_reg): Likewise.
16555 * recog.c (peep2_find_free_register): Likewise.
16556 * ree.c (combine_reaching_defs): Likewise.
16557 * regcprop.c (maybe_mode_change): Likewise.
16558 * reginfo.c (init_reg_sets_1): Likewise.
16559 (choose_hard_reg_mode): Likewise.
16560 (simplifiable_subregs): Likewise.
16561 * regrename.c (check_new_reg_p): Likewise.
16562 * reload.c (find_valid_class): Likewise.
16563 (find_valid_class_1): Likewise.
16564 (reload_inner_reg_of_subreg): Likewise.
16565 (push_reload): Likewise.
16566 (combine_reloads): Likewise.
16567 (find_dummy_reload): Likewise.
16568 (find_reloads): Likewise.
16569 * reload1.c (find_reg): Likewise.
16570 (set_reload_reg): Likewise.
16571 (allocate_reload_reg): Likewise.
16572 (choose_reload_regs): Likewise.
16573 (reload_adjust_reg_for_temp): Likewise.
16574 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
16575 (simplify_subreg_regno): Likewise.
16576 * sel-sched.c (init_regs_for_mode): Likewise.
16577 * varasm.c (make_decl_rtl): Likewise.
16578 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
16579 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
16580 HARD_REGNO_MODE_OK.
16581 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
16582 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
16583 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16584 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
16585 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
16586 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16587 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
16588 (arc_mode_class): Delete.
16589 (HARD_REGNO_MODE_OK): Delete.
16590 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16591 (arc_hard_regno_mode_ok): Rename old array to...
16592 (arc_hard_regno_mode_ok_modes): ...this.
16593 (arc_conditional_register_usage): Update accordingly.
16594 (arc_mode_class): Make static.
16595 (arc_hard_regno_mode_ok): New function.
16596 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
16597 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
16598 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16599 (arm_hard_regno_mode_ok): Make static.
16600 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
16601 HARD_REGNO_MODE_OK.
16602 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
16603 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
16604 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
16606 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16607 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
16608 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
16609 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
16610 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
16611 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16612 * config/bfin/predicates.md (valid_reg_operand): Use
16613 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
16614 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
16615 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
16616 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16617 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
16618 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
16619 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16620 (cr16_hard_regno_mode_ok): Make static and return a bool.
16621 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
16622 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16623 (cris_hard_regno_mode_ok): New function.
16624 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
16625 (epiphany_mode_class): Delete.
16626 (HARD_REGNO_MODE_OK): Delete.
16627 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
16628 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16629 (hard_regno_mode_ok): Rename to...
16630 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
16631 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
16632 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
16633 HARD_REGNO_MODE_OK.
16634 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
16635 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
16636 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16637 (frv_hard_regno_mode_ok): Make static and return a bool.
16638 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
16639 HARD_REGNO_MODE_OK.
16640 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
16641 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
16642 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
16643 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
16645 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16646 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
16647 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
16648 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
16650 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16651 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
16652 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16653 (ia64_hard_regno_mode_ok): New function.
16654 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
16655 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16656 (iq2000_hard_regno_mode_ok): New function.
16657 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
16658 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16659 (lm32_hard_regno_mode_ok): New function.
16660 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
16661 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
16662 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
16663 instead of HARD_REGNO_MODE_OK.
16664 (m32c_hard_regno_ok): Rename to...
16665 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
16666 (m32c_cannot_change_mode_class): Update accordingly.
16667 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16668 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
16669 (m32r_mode_class): Delete.
16670 (HARD_REGNO_MODE_OK): Delete.
16671 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16672 (m32r_hard_regno_mode_ok): Rename to...
16673 (m32r_hard_regno_modes): ...this.
16674 (m32r_mode_class): Make static.
16675 (m32r_hard_regno_mode_ok): New function.
16676 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
16677 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
16678 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16679 (m68k_hard_regno_mode_ok): Make static.
16680 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
16681 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16682 (mcore_hard_regno_mode_ok): New function.
16683 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
16684 (HARD_REGNO_MODE_OK): Delete.
16685 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
16687 (microblaze_hard_regno_mode_ok_p): ...this and make static.
16688 (microblaze_hard_regno_mode_ok): New function.
16689 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16690 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
16691 (mips_hard_regno_mode_ok): Delete.
16692 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
16693 (mips_hard_regno_mode_ok_p): ...this and make static.
16694 (mips_hard_regno_mode_ok_p): Rename to...
16695 (mips_hard_regno_mode_ok_uncached): ...this.
16696 (mips_hard_regno_mode_ok): New function.
16697 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
16698 of HARD_REGNO_MODE_OK.
16699 (mips_option_override): Update after above name changes.
16700 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16701 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
16702 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
16703 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
16704 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
16705 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16706 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
16707 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
16708 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
16709 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16710 (msp430_hard_regno_mode_ok): Make static and return a bool.
16711 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
16712 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
16713 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
16715 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16716 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
16717 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
16718 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
16719 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
16720 (PA_HARD_REGNO_MODE_OK): ...this
16721 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
16722 (PA_HARD_REGNO_MODE_OK): ...this.
16723 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16724 (pa_hard_regno_mode_ok): New function.
16725 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
16726 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16727 (pdp11_hard_regno_mode_ok): New function.
16728 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
16729 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
16731 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
16733 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16734 (rs6000_hard_regno_mode_ok): Rename to...
16735 (rs6000_hard_regno_mode_ok_uncached): ...this.
16736 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16737 (rs6000_hard_regno_mode_ok): New function.
16738 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
16739 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
16740 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
16741 (riscv_hard_regno_mode_ok): ...this and make static.
16742 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16743 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
16744 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
16745 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16746 (rl78_hard_regno_mode_ok): Make static and return bool.
16747 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
16748 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
16750 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
16751 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16752 (rs6000_hard_regno_mode_ok): Rename to...
16753 (rs6000_hard_regno_mode_ok_uncached): ...this.
16754 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16755 (rs6000_hard_regno_mode_ok): New function.
16756 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
16757 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
16758 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16759 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
16760 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
16761 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
16762 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16763 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
16764 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
16765 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16766 (sh_hard_regno_mode_ok): Make static.
16767 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
16768 instead of HARD_REGNO_MODE_OK.
16769 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
16770 (sparc_mode_class): Delete.
16771 (HARD_REGNO_MODE_OK): Delete.
16772 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16773 (hard_regno_mode_classes): Make static.
16774 (sparc_mode_class): Likewise.
16775 (sparc_hard_regno_mode_ok): New function.
16776 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
16777 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
16778 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
16780 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16781 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
16782 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
16783 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
16784 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
16785 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16786 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
16787 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
16788 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16789 (visium_hard_regno_mode_ok): New function.
16790 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
16791 instead of HARD_REGNO_MODE_OK.
16792 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
16793 (HARD_REGNO_MODE_OK): Delete.
16794 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
16795 (xtensa_hard_regno_mode_ok_p): ...this and make static.
16796 (xtensa_option_override): Update accordingly.
16797 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16798 (xtensa_hard_regno_mode_ok): New function.
16799 * system.h (HARD_REGNO_MODE_OK): Poison.
16801 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16802 Alan Hayward <alan.hayward@arm.com>
16803 David Sherwood <david.sherwood@arm.com>
16805 * target.def (hard_regno_call_part_clobbered): New hook.
16806 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
16807 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
16808 * doc/tm.texi: Regenerate.
16809 * hooks.h (hook_bool_uint_mode_false): Declare.
16810 * hooks.c (hook_bool_uint_mode_false): New function.
16811 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16812 * cselib.c (cselib_process_insn): Use
16813 targetm.hard_regno_call_part_clobbered instead of
16814 HARD_REGNO_CALL_PART_CLOBBERED.
16815 * ira-conflicts.c (ira_build_conflicts): Likewise.
16816 * ira-costs.c (ira_tune_allocno_costs): Likewise.
16817 * lra-constraints.c (need_for_call_save_p): Likewise.
16818 * lra-lives.c: Include target.h.
16819 (check_pseudos_live_through_calls): Use
16820 targetm.hard_regno_call_part_clobbered instead of
16821 HARD_REGNO_CALL_PART_CLOBBERED.
16822 * regcprop.c: Include target.h.
16823 (copyprop_hardreg_forward_1): Use
16824 targetm.hard_regno_call_part_clobbered instead of
16825 HARD_REGNO_CALL_PART_CLOBBERED.
16826 * reginfo.c (choose_hard_reg_mode): Likewise.
16827 * regrename.c (check_new_reg_p): Likewise.
16828 * reload.c (find_equiv_reg): Likewise.
16829 * reload1.c (emit_reload_insns): Likewise.
16830 * sched-deps.c (deps_analyze_insn): Likewise.
16831 * sel-sched.c (init_regs_for_mode): Likewise.
16832 (mark_unavailable_hard_regs): Likewise.
16833 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
16834 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16835 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
16837 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16838 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16839 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
16841 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
16843 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16844 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16845 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
16847 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16848 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16849 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
16851 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16852 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
16854 * config/powerpcspe/powerpcspe.c
16855 (rs6000_hard_regno_call_part_clobbered): New function.
16856 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16857 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16858 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
16860 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16861 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16862 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
16864 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16865 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16866 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
16868 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16869 Alan Hayward <alan.hayward@arm.com>
16870 David Sherwood <david.sherwood@arm.com>
16872 * rtl.h (subreg_memory_offset): Declare.
16873 * emit-rtl.c (subreg_memory_offset): New function.
16874 * expmed.c (store_bit_field_1): Use it.
16875 * expr.c (undefined_operand_subword_p): Likewise.
16876 * simplify-rtx.c (simplify_subreg): Likewise.
16878 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
16880 PR rtl-optimization/57448
16883 * optabs.c (expand_atomic_load): Place compiler memory barriers if
16884 using atomic_load pattern.
16885 (expand_atomic_store): Likewise.
16887 2017-09-04 Jakub Jelinek <jakub@redhat.com>
16890 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
16891 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
16892 handling. Use replace_call_with_value with NULL instead of
16893 gsi_replace, unlink_stmt_vdef and release_defs.
16895 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
16898 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
16900 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16903 * rtl.h (emit_library_call_value_1): Declare.
16904 (emit_library_call): Replace declaration with a series of overloads.
16905 Remove the parameter count argument.
16906 (emit_library_call_value): Likewise.
16907 * calls.c (emit_library_call_value_1): Make global. Replace varargs
16908 with an "rtx_mode_t *".
16909 (emit_library_call_value): Delete.
16910 (emit_library_call): Likewise.
16911 * asan.c (asan_emit_stack_protection): Update calls accordingly.
16912 (asan_emit_allocas_unpoison): Likewise.
16913 * builtins.c (expand_builtin_powi): Likewise.
16914 (expand_asan_emit_allocas_unpoison): Likewise.
16915 * cfgexpand.c (expand_main_function): Likewise.
16916 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
16917 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
16918 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
16919 * config/arm/arm.c (arm_trampoline_init): Likewise.
16920 (arm_call_tls_get_addr): Likewise.
16921 (arm_expand_divmod_libfunc): Likewise.
16922 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
16923 (smulsi3_highpart): Likewise.
16924 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
16925 (c6x_expand_compare): Likewise.
16926 (c6x_expand_movmem): Likewise.
16927 * config/frv/frv.c (frv_trampoline_init): Likewise.
16928 * config/i386/i386.c (ix86_trampoline_init): Likewise.
16929 (ix86_expand_divmod_libfunc): Likewise.
16930 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
16931 (ia64_expand_compare): Likewise.
16932 (ia64_profile_hook): Likewise.
16933 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
16934 (nonlocal_goto): Likewise.
16935 (restore_stack_nonlocal): Likewise.
16936 * config/m32r/m32r.c (block_move_call): Likewise.
16937 (m32r_trampoline_init): Likewise.
16938 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
16939 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
16940 (m68k_call_m68k_read_tp): Likewise.
16941 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
16942 (microblaze_expand_divide): Likewise.
16943 * config/mips/mips.h (mips_args): Likewise.
16944 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
16945 (MIPS_ICACHE_SYNC): Likewise.
16946 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
16947 (nios2_trampoline_init): Likewise.
16948 * config/pa/pa.c (hppa_tls_call): Likewise.
16949 (pa_trampoline_init): Likewise.
16950 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
16951 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
16952 (expand_strn_compare): Likewise.
16953 (rs6000_generate_compare): Likewise.
16954 (rs6000_expand_float128_convert): Likewise.
16955 (output_profile_hook): Likewise.
16956 (rs6000_trampoline_init): Likewise.
16957 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
16958 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
16959 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
16960 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
16961 (rs6000_generate_compare): Likewise.
16962 (rs6000_expand_float128_convert): Likewise.
16963 (output_profile_hook): Likewise.
16964 (rs6000_trampoline_init): Likewise.
16965 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
16966 * config/sh/sh.c (sh_trampoline_init): Likewise.
16967 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
16968 (sparc_emit_float_lib_cmp): Likewise.
16969 (sparc32_initialize_trampoline): Likewise.
16970 (sparc64_initialize_trampoline): Likewise.
16971 (sparc_profile_hook): Likewise.
16972 * config/spu/spu.c (ea_load_store): Likewise.
16973 * config/spu/spu.md (floatunssidf2): Likewise.
16974 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
16975 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
16976 * config/visium/visium.c (expand_block_move_4): Likewise.
16977 (expand_block_move_2): Likewise.
16978 (expand_block_move_1): Likewise.
16979 (expand_block_set_4): Likewise.
16980 (expand_block_set_2): Likewise.
16981 (expand_block_set_1): Likewise.
16982 (visium_trampoline_init): Likewise.
16983 (visium_profile_hook): Likewise.
16984 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
16985 (xtensa_setup_frame_addresses): Likewise.
16986 (xtensa_trampoline_init): Likewise.
16987 * except.c (sjlj_emit_function_enter): Likewise.
16988 (sjlj_emit_function_exit): Likewise.
16989 * explow.c (allocate_dynamic_stack_space): Likewise.
16990 (probe_stack_range): Likewise.
16991 * expr.c (convert_mode_scalar): Likewise.
16992 * optabs.c (expand_binop): Likewise.
16993 (expand_twoval_binop_libfunc): Likewise.
16994 (expand_unop): Likewise.
16995 (prepare_cmp_insn): Likewise.
16996 (prepare_float_lib_cmp): Likewise.
16997 (expand_float): Likewise.
16998 (expand_fix): Likewise.
16999 (expand_fixed_convert): Likewise.
17000 (maybe_emit_sync_lock_test_and_set): Likewise.
17001 (expand_atomic_compare_and_swap): Likewise.
17002 (expand_mem_thread_fence): Likewise.
17003 (expand_atomic_fetch_op): Likewise.
17005 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
17007 * doc/generic.texi (OpenACC): Adjust URL.
17008 * doc/invoke.texi (C Dialect Options): Ditto.
17010 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
17012 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
17013 predicate for operand 1. Add (m,<S>) constraint.
17014 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
17015 Prevent memory operand 1 with register operand 2.
17017 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
17019 PR rtl-optimization/82024
17020 * combine.c (try_combine): If the combination result is a PARALLEL,
17021 and we only need to retain the SET in there that would be placed
17022 at I2, check that we can place that at I3 instead, before doing so.
17024 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17027 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
17029 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
17030 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
17033 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
17034 Jeff Law <law@redhat.com>
17036 * varasm.c (bss_initializer_p): Do not put constants into .bss
17037 (categorize_decl_for_section): Handle bss_initializer_p returning
17038 false when DECL_INITIAL is NULL.
17040 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17043 * config/s390/s390.c (s390_can_inline_p): New function.
17045 2017-09-01 Jeff Law <law@redhat.com>
17047 PR tree-optimization/82052
17048 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
17049 Always initialize the returned slot after a hash table miss
17050 when INSERT is true.
17052 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
17054 * config/s390/s390.md (mem_signal_fence): Remove.
17055 * doc/md.texi (mem_signal_fence): Remove.
17056 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
17058 * target-insns.def (mem_signal_fence): Remove.
17060 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17063 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
17066 * asan.c (create_odr_indicator): Strip name encoding from assembler
17067 name before appending it after __odr_asan_.
17069 2017-09-01 Martin Liska <mliska@suse.cz>
17071 PR tree-optimization/82059
17072 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
17073 frequency only when an edge is redirected.
17075 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17077 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
17078 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
17079 (arc_conditional_register_usage): Remove ARC600 lp_count
17081 (arc_file_start): Emit Tag_ARC_CPU_variation.
17082 (arc_can_use_doloop_p): New conditions to use ZOLs.
17083 (hwloop_fail): New function.
17084 (hwloop_optimize): Likewise.
17085 (hwloop_pattern_reg): Likewise.
17086 (arc_doloop_hooks): New struct, to be used with reorg_loops.
17087 (arc_reorg_loops): New function, calls reorg_loops.
17088 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
17089 (arc600_corereg_hazard): Remove ZOL checking, case handled by
17091 (arc_loop_hazard): Remove function, functionality moved into
17093 (arc_hazard): Remove arc_loop_hazard call.
17094 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
17095 into hwloop_optimize.
17096 (arc_label_align): Remove ZOL handling.
17097 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
17098 * config/arc/arc.md (doloop_begin): Remove pattern.
17099 (doloop_begin_i): Likewise.
17100 (doloop_end_i): Likewise.
17101 (doloop_fallback): Likewise.
17102 (doloop_fallback_m): Likewise.
17103 (doloop_end): Reimplement expand.
17104 (arc_lp): New pattern for LP instruction.
17105 (loop_end): New pattern.
17106 (loop_fail): Likewise.
17107 (decrement_and_branch_until_zero): Likewise.
17108 * config/arc/arc.opt (mlpc-width): New option.
17109 * doc/invoke.texi (mlpc-width): Document option.
17111 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17113 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
17114 (arc_ccfsm_advance): Fix checking for delay slots.
17115 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
17117 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17119 * config/arc/arc.md (movqi_insn): Add stores to save constant long
17121 (movhi_insn): Update store instruction constraint which are saving
17122 6-bit short immediates.
17123 (movsi_insn): Consider also short scaled load operations.
17124 (zero_extendhisi2_i): Use Usd constraint instead of T.
17125 (extendhisi2_i): Add q constraint.
17126 (arc_clzsi2): Add type and length attributes.
17127 (arc_ctzsi2): Likewise.
17128 * config/arc/constraints.md (Usc): Update constraint, the
17129 assembler can parse two relocations for a single instruction.
17131 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17133 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
17134 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
17136 2017-08-31 Olivier Hainque <hainque@adacore.com>
17138 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
17139 match as powerpc-wrs-vxworks*.
17141 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
17143 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
17144 register constraint for by-element operand.
17145 (aarch64_mls_elt_merge<mode>): Likewise.
17147 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17149 * config/arc/arc.c (arc_can_follow_jump): Check for short
17152 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17154 * config.gcc: Use g.opt for arc.
17155 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
17156 functionality moved to ...
17157 (legitimate_scaled_address_p): New function, ...here.
17158 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
17159 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
17160 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
17162 (arc_override_options): Handle G option.
17163 (arc_output_pic_addr_const): Correct function definition.
17164 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
17165 (arc_decl_anon_ns_mem_p): Delete.
17166 (arc_in_small_data_p): Overhaul this function to take into
17167 consideration the value given via G option.
17168 (arc_rewrite_small_data_1): Renamed and corrected old
17169 arc_rewrite_small_data function.
17170 (arc_rewrite_small_data): New function.
17171 (small_data_pattern): Don't use pic_offset_table_rtx.
17172 * config/arc/arc.h (CC1_SPEC): Recognize G option.
17173 * config/arc/simdext.md (movmisalignv2hi): Use
17174 prepare_move_operands function.
17176 (movmisalign*): Likewise.
17177 * doc/invoke.texi (ARC options): Document -G option.
17179 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17181 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
17183 * config/arc/arc.c (arc_print_operand): Output scalled address for
17184 sdata whenever is possible.
17185 (arc_in_small_data_p): Allow sdata for 64bit datum when double
17186 load/stores are available.
17187 (compact_sda_memory_operand): Check for the alignment required by
17188 code density instructions.
17189 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
17191 * config/arc/constraints.md (Usd): Update constraint.
17192 (Us0): New constraint.
17193 (Usc): Update constraint.
17195 2017-08-31 Richard Biener <rguenther@suse.de>
17197 PR middle-end/82054
17198 * dwarf2out.c (dwarf2out_early_global_decl): Process each
17199 function only once.
17201 2017-08-31 Tamar Christina <tamar.christina@arm.com>
17203 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
17204 Resize type_signature.
17206 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17207 Alan Hayward <alan.hayward@arm.com>
17208 David Sherwood <david.sherwood@arm.com>
17210 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
17211 subregs whose inner modes can be stored in GPRs.
17212 (aarch64_classify_index): Likewise.
17214 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17215 Alan Hayward <alan.hayward@arm.com>
17216 David Sherwood <david.sherwood@arm.com>
17218 * config/aarch64/iterators.md (V_cmp_result): Rename to...
17219 (V_INT_EQUIV): ...this.
17220 (v_cmp_result): Rename to...
17221 (v_int_equiv): ...this.
17222 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
17223 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
17224 (copysign<mode>3): Likewise.
17225 (aarch64_simd_bsl<mode>_internal): Likewise.
17226 (aarch64_simd_bsl<mode>): Likewise.
17227 (vec_cmp<mode><mode>): Likewise.
17228 (vcond<mode><mode>): Likewise.
17229 (vcond<v_cmp_mixed><mode>): Likewise.
17230 (vcondu<mode><v_cmp_mixed>): Likewise.
17231 (aarch64_cm<optab><mode>): Likewise.
17232 (aarch64_cmtst<mode>): Likewise.
17233 (aarch64_fac<optab><mode>): Likewise.
17234 (vec_perm_const<mode>): Likewise.
17235 (vcond_mask_<mode><v_cmp_result>): Rename to...
17236 (vcond_mask_<mode><v_int_equiv>): ...this.
17237 (vec_cmp<mode><v_cmp_result>): Rename to...
17238 (vec_cmp<mode><v_int_equiv>): ...this.
17240 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17241 Alan Hayward <alan.hayward@arm.com>
17242 David Sherwood <david.sherwood@arm.com>
17244 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
17246 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
17247 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
17248 (UNSPEC_LD4_DREG): New unspecs.
17249 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
17250 (aarch64_ld2<mode>_dreg_be): Replace with...
17251 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
17253 (aarch64_ld3<mode>_dreg_le)
17254 (aarch64_ld3<mode>_dreg_be): Replace with...
17255 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
17257 (aarch64_ld4<mode>_dreg_le)
17258 (aarch64_ld4<mode>_dreg_be): Replace with...
17259 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
17262 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17264 PR tree-optimization/81987
17265 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
17266 insert an initializer in a location not dominated by the stride
17269 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17271 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
17272 on the entire header of the finally block in the fallthru case.
17274 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17276 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
17278 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
17280 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
17281 rs6000_emit_move_from_cr and call renamed function.
17282 (rs6000_emit_prologue): Call renamed functions.
17283 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
17284 movesi_from_cr, remove volatile CRs.
17286 2017-08-30 Jon Beniston <jon@beniston.com>
17287 Richard Biener <rguenther@suse.de>
17289 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
17290 of VECTOR_MODE_P check.
17291 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
17292 element vector types.
17294 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17296 * df.h (df_read_modify_subreg_p): Remove in favor of...
17297 * rtl.h (read_modify_subreg_p): ...this new function. Take a
17298 const_rtx instead of an rtx.
17299 * cprop.c (local_cprop_find_used_regs): Update accordingly.
17300 * df-problems.c (df_word_lr_mark_ref): Likewise.
17301 * ira-lives.c (mark_pseudo_reg_live): Likewise.
17302 (mark_pseudo_reg_dead): Likewise.
17303 (mark_ref_dead): Likewise.
17304 * reginfo.c (init_subregs_of_mode): Likewise.
17305 * sched-deps.c (sched_analyze_1): Likewise.
17306 * df-scan.c (df_def_record_1): Likewise.
17307 (df_uses_record): Likewise.
17308 (df_read_modify_subreg_p): Remove in favor of...
17309 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
17310 const_rtx instead of an rtx.
17312 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17313 Alan Hayward <alan.hayward@arm.com>
17314 David Sherwood <david.sherwood@arm.com>
17316 * rtl.h (partial_subreg_p): New function.
17317 * caller-save.c (save_call_clobbered_regs): Use it.
17318 * calls.c (expand_call): Likewise.
17319 * combine.c (combinable_i3pat): Likewise.
17320 (simplify_set): Likewise.
17321 (make_extraction): Likewise.
17322 (make_compound_operation_int): Likewise.
17323 (gen_lowpart_or_truncate): Likewise.
17324 (force_to_mode): Likewise.
17325 (make_field_assignment): Likewise.
17326 (reg_truncated_to_mode): Likewise.
17327 (record_truncated_value): Likewise.
17328 (move_deaths): Likewise.
17329 * cse.c (record_jump_cond): Likewise.
17330 (cse_insn): Likewise.
17331 * cselib.c (cselib_lookup_1): Likewise.
17332 * expmed.c (extract_bit_field_using_extv): Likewise.
17333 * function.c (assign_parm_setup_reg): Likewise.
17334 * ifcvt.c (noce_convert_multiple_sets): Likewise.
17335 * ira-build.c (create_insn_allocnos): Likewise.
17336 * lra-coalesce.c (merge_pseudos): Likewise.
17337 * lra-constraints.c (match_reload): Likewise.
17338 (simplify_operand_subreg): Likewise.
17339 (curr_insn_transform): Likewise.
17340 * lra-lives.c (process_bb_lives): Likewise.
17341 * lra.c (new_insn_reg): Likewise.
17342 (lra_substitute_pseudo): Likewise.
17343 * regcprop.c (mode_change_ok): Likewise.
17344 (maybe_mode_change): Likewise.
17345 (copyprop_hardreg_forward_1): Likewise.
17346 * reload.c (push_reload): Likewise.
17347 (find_reloads): Likewise.
17348 (find_reloads_subreg_address): Likewise.
17349 * reload1.c (alter_reg): Likewise.
17350 (eliminate_regs_1): Likewise.
17351 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17353 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
17355 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
17356 back to if statements, including unpack.
17358 2017-08-30 Martin Liska <mliska@suse.cz>
17360 PR inline-asm/82001
17361 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
17363 (func_checker::compare_asm_inputs_outputs): ... this function.
17364 (func_checker::compare_gimple_asm): Use the function to compare
17365 also ASM constrains.
17366 * ipa-icf-gimple.h: Rename the function.
17368 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17369 Alan Hayward <alan.hayward@arm.com>
17370 David Sherwood <david.sherwood@arm.com>
17372 * coretypes.h (complex_mode): New type.
17373 * gdbhooks.py (build_pretty_printer): Handle it.
17374 * machmode.h (complex_mode): New class.
17375 (complex_mode::includes_p): New function.
17376 (is_complex_int_mode): Likewise.
17377 (is_complex_float_mode): Likewise.
17378 * genmodes.c (get_mode_class): Handle complex mode classes.
17379 * function.c (expand_function_end): Use is_complex_int_mode.
17381 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17382 Alan Hayward <alan.hayward@arm.com>
17383 David Sherwood <david.sherwood@arm.com>
17385 * coretypes.h (scalar_mode_pod): New typedef.
17386 * gdbhooks.py (build_pretty_printer): Handle it.
17387 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
17388 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
17389 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
17390 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
17391 as_a <scalar_mode>.
17393 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17394 Alan Hayward <alan.hayward@arm.com>
17395 David Sherwood <david.sherwood@arm.com>
17397 * machmode.h (mode_for_vector): Take a scalar_mode instead
17399 * stor-layout.c (mode_for_vector): Likewise.
17400 * explow.c (promote_mode): Use as_a <scalar_mode>.
17401 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
17403 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17404 Alan Hayward <alan.hayward@arm.com>
17405 David Sherwood <david.sherwood@arm.com>
17407 * target.def (preferred_simd_mode): Take a scalar_mode
17408 instead of a machine_mode.
17409 * targhooks.h (default_preferred_simd_mode): Likewise.
17410 * targhooks.c (default_preferred_simd_mode): Likewise.
17411 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
17412 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
17413 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
17414 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
17415 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
17416 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
17417 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
17418 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
17420 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
17421 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
17422 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
17423 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
17424 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
17425 * doc/tm.texi: Regenerate.
17426 * optabs-query.c (can_vec_mask_load_store_p): Return false for
17429 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17430 Alan Hayward <alan.hayward@arm.com>
17431 David Sherwood <david.sherwood@arm.com>
17433 * target.def (scalar_mode_supported_p): Take a scalar_mode
17434 instead of a machine_mode.
17435 * targhooks.h (default_scalar_mode_supported_p): Likewise.
17436 * targhooks.c (default_scalar_mode_supported_p): Likewise.
17437 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
17438 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17439 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
17440 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
17441 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
17442 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
17443 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
17444 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
17445 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
17446 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
17447 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
17448 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
17450 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
17451 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
17452 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17453 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17454 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
17456 * doc/tm.texi: Regenerate.
17458 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17459 Alan Hayward <alan.hayward@arm.com>
17460 David Sherwood <david.sherwood@arm.com>
17462 * coretypes.h (opt_scalar_mode): New typedef.
17463 * gdbhooks.py (build_pretty_printers): Handle it.
17464 * machmode.h (mode_iterator::get_2xwider): Add overload for
17466 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
17468 * expr.c (convert_mode_scalar): Likewise.
17469 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17470 * optabs.c (expand_float): Likewise.
17471 (expand_fix): Likewise.
17472 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17474 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17475 Alan Hayward <alan.hayward@arm.com>
17476 David Sherwood <david.sherwood@arm.com>
17478 * optabs.c (expand_float): Explicitly check for scalars before
17479 using a branching expansion.
17480 (expand_fix): Likewise.
17482 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17483 Alan Hayward <alan.hayward@arm.com>
17484 David Sherwood <david.sherwood@arm.com>
17486 * expr.c (convert_mode): Split scalar handling out into...
17487 (convert_mode_scalar): ...this new function. Treat the modes
17490 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17491 Alan Hayward <alan.hayward@arm.com>
17492 David Sherwood <david.sherwood@arm.com>
17494 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
17496 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
17498 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17499 Alan Hayward <alan.hayward@arm.com>
17500 David Sherwood <david.sherwood@arm.com>
17502 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
17503 rather than a machine_mode.
17504 (fixed_from_string): Likewise.
17505 (fixed_convert): Likewise.
17506 (fixed_convert_from_int): Likewise.
17507 (fixed_convert_from_real): Likewise.
17508 (real_convert_from_fixed): Likewise.
17509 * fixed-value.c (fixed_from_double_int): Likewise.
17510 (fixed_from_string): Likewise.
17511 (fixed_convert): Likewise.
17512 (fixed_convert_from_int): Likewise.
17513 (fixed_convert_from_real): Likewise.
17514 (real_convert_from_fixed): Likewise.
17515 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
17517 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17518 Alan Hayward <alan.hayward@arm.com>
17519 David Sherwood <david.sherwood@arm.com>
17521 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
17522 of separate mode class checks. Do not allow vector modes here.
17523 (immed_wide_int_const): Use as_a <scalar_mode>.
17524 * explow.c (trunc_int_for_mode): Likewise.
17525 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
17526 (wi::shwi): Likewise.
17527 (wi::min_value): Likewise.
17528 (wi::max_value): Likewise.
17529 * dwarf2out.c (loc_descriptor): Likewise.
17530 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
17531 for CONST_WIDE_INT.
17533 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17534 Alan Hayward <alan.hayward@arm.com>
17535 David Sherwood <david.sherwood@arm.com>
17537 * tree.h (SCALAR_TYPE_MODE): New macro.
17538 * expr.c (expand_expr_addr_expr_1): Use it.
17539 (expand_expr_real_2): Likewise.
17540 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
17541 (fold_convert_const_fixed_from_int): Likewise.
17542 (fold_convert_const_fixed_from_real): Likewise.
17543 (native_encode_fixed): Likewise
17544 (native_encode_complex): Likewise
17545 (native_encode_vector): Likewise.
17546 (native_interpret_fixed): Likewise.
17547 (native_interpret_real): Likewise.
17548 (native_interpret_complex): Likewise.
17549 (native_interpret_vector): Likewise.
17550 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
17551 (simd_clone_adjust_argument_types): Likewise.
17552 (simd_clone_init_simd_arrays): Likewise.
17553 (simd_clone_adjust): Likewise.
17554 * stor-layout.c (layout_type): Likewise.
17555 * tree.c (build_minus_one_cst): Likewise.
17556 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17557 * tree-inline.c (estimate_move_cost): Likewise.
17558 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
17559 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
17560 (vectorizable_reduction): Likewise.
17561 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
17562 (vect_recog_mixed_size_cond_pattern): Likewise.
17563 (check_bool_pattern): Likewise.
17564 (adjust_bool_pattern): Likewise.
17565 (search_type_for_mask_1): Likewise.
17566 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
17567 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17568 (vectorizable_load): Likewise.
17569 (vectorizable_store): Likewise.
17570 * ubsan.c (ubsan_encode_value): Likewise.
17571 * varasm.c (output_constant): Likewise.
17573 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17574 Alan Hayward <alan.hayward@arm.com>
17575 David Sherwood <david.sherwood@arm.com>
17577 * coretypes.h (scalar_mode): New class.
17578 * machmode.h (scalar_mode): Likewise.
17579 (scalar_mode::includes_p): New function.
17580 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
17581 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
17582 * genmodes.c (get_mode_class): Handle remaining scalar modes.
17583 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
17584 * expmed.c (store_bit_field_1): Likewise.
17585 (extract_bit_field_1): Likewise.
17586 * expr.c (write_complex_part): Likewise.
17587 (read_complex_part): Likewise.
17588 (emit_move_complex_push): Likewise.
17589 (expand_expr_real_2): Likewise.
17590 * function.c (assign_parm_setup_reg): Likewise.
17591 (assign_parms_unsplit_complex): Likewise.
17592 * optabs.c (expand_binop): Likewise.
17593 * rtlanal.c (subreg_get_info): Likewise.
17594 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17595 * varasm.c (output_constant_pool_2): Likewise.
17597 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17598 Alan Hayward <alan.hayward@arm.com>
17599 David Sherwood <david.sherwood@arm.com>
17601 * expmed.c (extract_high_half): Use scalar_int_mode and remove
17603 (expmed_mult_highpart_optab): Likewise.
17604 (expmed_mult_highpart): Likewise.
17606 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17607 Alan Hayward <alan.hayward@arm.com>
17608 David Sherwood <david.sherwood@arm.com>
17610 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
17611 instead of a machine_mode.
17612 (builtin_memset_read_str): Likewise.
17613 * builtins.c (c_readstr): Likewise.
17614 (builtin_memcpy_read_str): Likewise.
17615 (builtin_strncpy_read_str): Likewise.
17616 (builtin_memset_read_str): Likewise.
17617 (builtin_memset_gen_str): Likewise.
17618 (expand_builtin_signbit): Use scalar_int_mode for local variables.
17619 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
17620 instead of a machine_mode.
17621 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
17623 (make_extraction): Likewise.
17624 (try_widen_shift_mode): Take and return scalar_int_modes instead
17626 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
17627 a scalar_int_mode instead of a machine_mode.
17628 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
17629 (avr_addr_space_pointer_mode): Likewise.
17630 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
17631 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
17632 (msp430_unwind_word_mode): Likewise.
17633 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
17634 (spu_addr_space_pointer_mode): Likewise.
17635 (spu_addr_space_address_mode): Likewise.
17636 (spu_libgcc_cmp_return_mode): Likewise.
17637 (spu_libgcc_shift_count_mode): Likewise.
17638 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
17639 (rl78_addr_space_pointer_mode): Likewise.
17640 (fl78_unwind_word_mode): Likewise.
17641 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
17643 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
17644 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
17645 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
17646 (mips_valid_pointer_mode): Likewise.
17647 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
17648 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
17649 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
17651 (ft32_addr_space_address_mode): Likewise.
17652 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
17653 scalar_int_mode instead of a machine_mode.
17654 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
17656 (m32c_addr_space_address_mode): Likewise.
17657 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
17658 (rs6000_eh_return_filter_mode): Likewise.
17659 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
17660 (rs6000_eh_return_filter_mode): Likewise.
17661 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
17662 (s390_libgcc_shift_count_mode): Likewise.
17663 (s390_unwind_word_mode): Likewise.
17664 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
17666 * target.def (mode_rep_extended): Likewise.
17667 (valid_pointer_mode): Likewise.
17668 (addr_space.valid_pointer_mode): Likewise.
17669 (eh_return_filter_mode): Return a scalar_int_mode rather than
17671 (libgcc_cmp_return_mode): Likewise.
17672 (libgcc_shift_count_mode): Likewise.
17673 (unwind_word_mode): Likewise.
17674 (addr_space.pointer_mode): Likewise.
17675 (addr_space.address_mode): Likewise.
17676 * doc/tm.texi: Regenerate.
17677 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
17679 (do_jump): Use scalar_int_mode for local variables.
17680 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
17681 rather than a machine_mode.
17682 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
17683 (scompare_loc_descriptor_wide): Likewise.
17684 (scompare_loc_descriptor_narrow): Likewise.
17685 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
17687 * except.c (sjlj_emit_dispatch_table): Likewise.
17688 (expand_builtin_eh_copy_values): Likewise.
17689 * explow.c (convert_memory_address_addr_space_1): Likewise.
17690 Take a scalar_int_mode rather than a machine_mode.
17691 (convert_memory_address_addr_space): Take a scalar_int_mode rather
17692 than a machine_mode.
17693 (memory_address_addr_space): Use scalar_int_mode for local variables.
17694 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
17695 rather than a machine_mode.
17696 * expmed.c (mask_rtx): Likewise.
17697 (init_expmed_one_conv): Likewise.
17698 (expand_mult_highpart_adjust): Likewise.
17699 (extract_high_half): Likewise.
17700 (expmed_mult_highpart_optab): Likewise.
17701 (expmed_mult_highpart): Likewise.
17702 (expand_smod_pow2): Likewise.
17703 (expand_sdiv_pow2): Likewise.
17704 (emit_store_flag_int): Likewise.
17705 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
17707 (extract_low_bits): Likewise.
17708 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
17710 * expr.c (pieces_addr::adjust): Likewise.
17711 (can_store_by_pieces): Likewise.
17712 (store_by_pieces): Likewise.
17713 (clear_by_pieces_1): Likewise.
17714 (expand_expr_addr_expr_1): Likewise.
17715 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
17716 (expand_expr_real_1): Likewise.
17717 (try_casesi): Likewise.
17718 * final.c (shorten_branches): Likewise.
17719 * fold-const.c (fold_convert_const_int_from_fixed): Change the
17720 type of "mode" to machine_mode.
17721 * internal-fn.c (expand_arith_overflow_result_store): Take a
17722 scalar_int_mode rather than a machine_mode.
17723 (expand_mul_overflow): Use scalar_int_mode for local variables.
17724 * loop-doloop.c (doloop_modify): Likewise.
17725 (doloop_optimize): Likewise.
17726 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
17727 than a machine_mode.
17728 (expand_doubleword_shift_condmove): Likewise.
17729 (expand_doubleword_shift): Likewise.
17730 (expand_doubleword_clz): Likewise.
17731 (expand_doubleword_popcount): Likewise.
17732 (expand_doubleword_parity): Likewise.
17733 (expand_absneg_bit): Use scalar_int_mode for local variables.
17734 (prepare_float_lib_cmp): Likewise.
17735 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
17736 rather than a machine_mode.
17737 (convert_memory_address_addr_space): Likewise.
17738 (get_mode_bounds): Likewise.
17739 (get_address_mode): Return a scalar_int_mode rather than a
17741 * rtlanal.c (get_address_mode): Likewise.
17742 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
17743 than a machine_mode.
17744 * targhooks.c (default_mode_rep_extended): Likewise.
17745 (default_valid_pointer_mode): Likewise.
17746 (default_addr_space_valid_pointer_mode): Likewise.
17747 (default_eh_return_filter_mode): Return a scalar_int_mode rather
17748 than a machine_mode.
17749 (default_libgcc_cmp_return_mode): Likewise.
17750 (default_libgcc_shift_count_mode): Likewise.
17751 (default_unwind_word_mode): Likewise.
17752 (default_addr_space_pointer_mode): Likewise.
17753 (default_addr_space_address_mode): Likewise.
17754 * targhooks.h (default_eh_return_filter_mode): Likewise.
17755 (default_libgcc_cmp_return_mode): Likewise.
17756 (default_libgcc_shift_count_mode): Likewise.
17757 (default_unwind_word_mode): Likewise.
17758 (default_addr_space_pointer_mode): Likewise.
17759 (default_addr_space_address_mode): Likewise.
17760 (default_mode_rep_extended): Take a scalar_int_mode rather than
17762 (default_valid_pointer_mode): Likewise.
17763 (default_addr_space_valid_pointer_mode): Likewise.
17764 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
17766 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
17767 rather than a machine_mode.
17768 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
17769 for local variables.
17770 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17771 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
17772 than a machine_mode.
17774 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17775 Alan Hayward <alan.hayward@arm.com>
17776 David Sherwood <david.sherwood@arm.com>
17778 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
17779 the mode argument to scalar_int_mode.
17780 (do_jump_by_parts_zero_rtx): Likewise.
17781 (do_jump_by_parts_equality_rtx): Likewise.
17782 (do_jump_by_parts_greater): Take a mode argument.
17783 (do_jump_by_parts_equality): Likewise.
17784 (do_jump_1): Update calls accordingly.
17786 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17787 Alan Hayward <alan.hayward@arm.com>
17788 David Sherwood <david.sherwood@arm.com>
17790 * is-a.h (safe_dyn_cast): New function.
17791 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
17792 (jump_table_for_label): Likewise.
17793 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
17794 instead of an rtx_insn *.
17795 (shorten_branches): Use dyn_cast instead of LABEL_P and
17796 JUMP_TABLE_DATA_P. Use jump_table_for_label and
17797 rtx_jump_table_data::get_data_mode.
17798 (final_scan_insn): Likewise.
17800 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17801 Alan Hayward <alan.hayward@arm.com>
17802 David Sherwood <david.sherwood@arm.com>
17804 * combine.c (try_combine): Use is_a <scalar_int_mode> when
17805 trying to combine a full-register integer set with a subreg
17808 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17809 Alan Hayward <alan.hayward@arm.com>
17810 David Sherwood <david.sherwood@arm.com>
17812 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
17813 that is always either address_mode or pointer_mode.
17815 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17816 Alan Hayward <alan.hayward@arm.com>
17817 David Sherwood <david.sherwood@arm.com>
17819 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
17820 when the two are known to be equal.
17822 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17823 Alan Hayward <alan.hayward@arm.com>
17824 David Sherwood <david.sherwood@arm.com>
17826 * simplify-rtx.c (simplify_const_unary_operation): Use
17827 is_a <scalar_int_mode> instead of checking for a nonzero
17828 precision. Forcibly convert op_mode to a scalar_int_mode
17829 in that case. More clearly differentiate the operand and
17830 result modes and use the former when deciding what the value
17831 of a count-bits operation should be. Use is_int_mode instead
17832 of checking for a MODE_INT. Remove redundant check for whether
17833 this mode has a zero precision.
17835 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17836 Alan Hayward <alan.hayward@arm.com>
17837 David Sherwood <david.sherwood@arm.com>
17839 * optabs.c (widen_leading): Change the type of the mode argument
17840 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
17841 (widen_bswap): Likewise.
17842 (expand_parity): Likewise.
17843 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
17844 (expand_ffs): Likewise.
17845 (epand_unop): Check for scalar integer modes before calling the
17848 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17849 Alan Hayward <alan.hayward@arm.com>
17850 David Sherwood <david.sherwood@arm.com>
17852 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
17854 (expand_expr_real_1): Update call accordingly.
17856 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17857 Alan Hayward <alan.hayward@arm.com>
17858 David Sherwood <david.sherwood@arm.com>
17860 * expmed.c (store_bit_field_using_insv): Add op0_mode and
17861 value_mode arguments. Use scalar_int_mode internally.
17862 (store_bit_field_1): Rename the new integer mode from imode
17863 to op0_mode and use it instead of GET_MODE (op0). Update calls
17864 to store_split_bit_field, store_bit_field_using_insv and
17865 store_fixed_bit_field.
17866 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
17867 Use scalar_int_mode internally. Use a bit count rather than a mode
17868 when calculating the largest bit size for get_best_mode.
17869 Update calls to store_split_bit_field and store_fixed_bit_field_1.
17870 (store_fixed_bit_field_1): Add mode and value_mode arguments.
17871 Remove assertion that OP0 has a scalar integer mode.
17872 (store_split_bit_field): Add op0_mode and value_mode arguments.
17873 Update calls to extract_fixed_bit_field.
17874 (extract_bit_field_using_extv): Add an op0_mode argument.
17875 Use scalar_int_mode internally.
17876 (extract_bit_field_1): Rename the new integer mode from imode to
17877 op0_mode and use it instead of GET_MODE (op0). Update calls to
17878 extract_split_bit_field, extract_bit_field_using_extv and
17879 extract_fixed_bit_field.
17880 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
17881 to extract_split_bit_field and extract_fixed_bit_field_1.
17882 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
17883 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
17884 on the target mode.
17885 (extract_split_bit_field): Add an op0_mode argument. Update call
17886 to extract_fixed_bit_field.
17888 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17889 Alan Hayward <alan.hayward@arm.com>
17890 David Sherwood <david.sherwood@arm.com>
17892 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
17893 * explow.c (hard_function_value): Likewise.
17894 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
17895 convert_to_mode call outside the loop.
17896 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
17897 for the mode iterator. Require the mode specified by max_pieces
17899 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
17901 (copy_blkmode_to_reg): Likewise.
17902 (set_storage_via_setmem): Likewise.
17903 * optabs.c (prepare_cmp_insn): Likewise.
17904 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17905 * stor-layout.c (finish_bitfield_representative): 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 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
17912 * expr.c (convert_move): Use them.
17913 (convert_modes): Likewise.
17914 (store_expr_with_bounds): Likewise.
17916 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17917 Alan Hayward <alan.hayward@arm.com>
17918 David Sherwood <david.sherwood@arm.com>
17920 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
17921 parameter for the mode of "x". Remove the "known_x", "known_mode"
17922 and "known_ret" arguments. Change the type of the mode argument
17923 to scalar_int_mode.
17924 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
17925 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
17926 (reg_num_sign_bit_copies_for_combine): Likewise.
17927 * rtlanal.c (nonzero_bits1): Likewise.
17928 (num_sign_bit_copies1): Likewise.
17929 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
17930 (reg_num_sign_bit_copies_general): Likewise.
17931 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
17932 (reg_nonzero_bits_general): Likewise.
17934 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17935 Alan Hayward <alan.hayward@arm.com>
17936 David Sherwood <david.sherwood@arm.com>
17938 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
17939 than in subroutines. Return 1 for non-integer modes.
17940 (cached_num_sign_bit_copies): Change the type of the mode parameter
17941 to scalar_int_mode.
17942 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
17943 classes. Handle CONST_INT_P first and then check whether X also
17944 has a scalar integer mode. Check the same thing for inner registers
17945 of a SUBREG and for values that are being extended or truncated.
17947 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17948 Alan Hayward <alan.hayward@arm.com>
17949 David Sherwood <david.sherwood@arm.com>
17951 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
17952 in subroutines. Return the mode mask for non-integer modes.
17953 (cached_nonzero_bits): Change the type of the mode parameter
17954 to scalar_int_mode.
17955 (nonzero_bits1): Likewise. Remove early exit for other mode
17956 classes. Handle CONST_INT_P first and then check whether X
17957 also has a scalar integer mode.
17959 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17960 Alan Hayward <alan.hayward@arm.com>
17961 David Sherwood <david.sherwood@arm.com>
17963 * expr.c (widest_int_mode_for_size): Make the comment match the code.
17964 Return a scalar_int_mode and assert that the size is greater than
17966 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
17967 (op_by_pieces_d::op_by_pieces_d): Likewise.
17968 (op_by_pieces_d::run): Likewise.
17969 (can_store_by_pieces): Likewise.
17971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17972 Alan Hayward <alan.hayward@arm.com>
17973 David Sherwood <david.sherwood@arm.com>
17975 * combine.c (extract_left_shift): Add a mode argument and update
17977 (make_compound_operation_int): Change the type of the mode parameter
17978 to scalar_int_mode and update the call to extract_left_shift.
17980 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17981 Alan Hayward <alan.hayward@arm.com>
17982 David Sherwood <david.sherwood@arm.com>
17984 * combine.c (simplify_and_const_int): Change the type of the mode
17985 parameter to scalar_int_mode.
17986 (simplify_and_const_int_1): Likewise. Update recursive call.
17988 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17989 Alan Hayward <alan.hayward@arm.com>
17990 David Sherwood <david.sherwood@arm.com>
17992 * combine.c (simplify_compare_const): Check that the mode is a
17993 scalar_int_mode (rather than VOIDmode) before testing its
17995 (simplify_comparison): Move COMPARISON_P handling out of the
17996 loop and restrict the latter part of the loop to scalar_int_modes.
17997 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
17998 and when considering SUBREG_REGs. Use is_int_mode instead of
17999 checking GET_MODE_CLASS against MODE_INT.
18001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18002 Alan Hayward <alan.hayward@arm.com>
18003 David Sherwood <david.sherwood@arm.com>
18005 * combine.c (try_widen_shift_mode): Move check for equal modes to...
18006 (simplify_shift_const_1): ...here. Use scalar_int_mode for
18007 shift_unit_mode and for modes involved in scalar shifts.
18009 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18010 Alan Hayward <alan.hayward@arm.com>
18011 David Sherwood <david.sherwood@arm.com>
18013 * combine.c (force_int_to_mode): New function, split out from...
18014 (force_to_mode): ...here. Keep xmode up-to-date and use it
18015 instead of GET_MODE (x).
18017 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18018 Alan Hayward <alan.hayward@arm.com>
18019 David Sherwood <david.sherwood@arm.com>
18021 * optabs-query.h (extraction_insn::struct_mode): Change type to
18022 opt_scalar_int_mode and update comment.
18023 (extraction_insn::field_mode): Change type to scalar_int_mode.
18024 (extraction_insn::pos_mode): Likewise.
18025 * combine.c (make_extraction): Update accordingly.
18026 * optabs-query.c (get_traditional_extraction_insn): Likewise.
18027 (get_optab_extraction_insn): Likewise.
18028 * recog.c (simplify_while_replacing): Likewise.
18029 * expmed.c (narrow_bit_field_mem): Change the type of the mode
18030 parameter to opt_scalar_int_mode.
18032 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18033 Alan Hayward <alan.hayward@arm.com>
18034 David Sherwood <david.sherwood@arm.com>
18036 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
18037 to a scalar_int_mode instead of a machine_mode.
18038 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
18039 (get_best_mode): Return a boolean and use a pointer argument to store
18040 the selected mode. Replace the limit mode parameter with a bit limit.
18041 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
18042 for the values returned by bit_field_mode_iterator::next_mode.
18043 (store_bit_field): Update call to get_best_mode.
18044 (store_fixed_bit_field): Likewise.
18045 (extract_fixed_bit_field): Likewise.
18046 * expr.c (optimize_bitfield_assignment_op): Likewise.
18047 * fold-const.c (optimize_bit_field_compare): Likewise.
18048 (fold_truth_andor_1): Likewise.
18049 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
18050 Update for new type of m_mode.
18051 (get_best_mode): As above.
18053 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18054 Alan Hayward <alan.hayward@arm.com>
18055 David Sherwood <david.sherwood@arm.com>
18057 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
18058 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
18059 (store_bit_field): Check is_a <scalar_int_mode> before calling
18060 strict_volatile_bitfield_p.
18061 (extract_bit_field): Likewise.
18063 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18064 Alan Hayward <alan.hayward@arm.com>
18065 David Sherwood <david.sherwood@arm.com>
18067 * target.def (cstore_mode): Return a scalar_int_mode.
18068 * doc/tm.texi: Regenerate.
18069 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
18070 * targhooks.h (default_cstore_mode): Likewise.
18071 * targhooks.c (default_cstore_mode): Likewise, using a forced
18073 * expmed.c (emit_cstore): Expect the target of the cstore to be
18076 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18077 Alan Hayward <alan.hayward@arm.com>
18078 David Sherwood <david.sherwood@arm.com>
18080 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
18082 (niter_desc): Likewise mode.
18083 (iv_analyze): Add a mode parameter.
18085 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
18086 and change its type to scalar_int_mode.
18087 * loop-iv.c: Update commentary at head of file.
18088 (iv_constant): Pass the mode paraeter before the rtx it describes
18089 and change its type to scalar_int_mode. Remove VOIDmode handling.
18090 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
18091 (iv_extend): Likewise.
18092 (shorten_into_mode): Likewise.
18093 (iv_add): Use scalar_int_mode.
18094 (iv_mult): Likewise.
18095 (iv_shift): Likewise.
18096 (canonicalize_iv_subregs): Likewise.
18097 (get_biv_step_1): Pass the outer_mode parameter before the rtx
18098 it describes and change its mode to scalar_int_mode. Also change
18099 the type of the returned inner_mode to scalar_int_mode.
18100 (get_biv_step): Likewise, turning outer_mode from a pointer
18101 into a direct parameter. Update call to get_biv_step_1.
18102 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
18103 iv_constant and get_biv_step.
18104 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
18105 and change its type to scalar_int_mode. Don't initialise iv->mode
18106 to VOIDmode and remove later checks for its still being VOIDmode.
18107 Update calls to iv_analyze_op and iv_analyze_expr. Check
18108 is_a <scalar_int_mode> when changing the mode under consideration.
18109 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
18110 Update call to iv_analyze_expr.
18111 (iv_analyze_op): Add a mode parameter. Reject subregs whose
18112 inner register is not also a scalar_int_mode. Update call to
18114 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
18115 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
18116 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
18117 separate mode class checks. Update calls to iv_analyze. Remove
18118 fix-up of VOIDmodes after iv_analyze_biv.
18119 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
18120 don't have a scalar_int_mode. Update call to biv_p.
18122 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18123 Alan Hayward <alan.hayward@arm.com>
18124 David Sherwood <david.sherwood@arm.com>
18126 * cfgexpand.c (convert_debug_memory_address): Use
18127 as_a <scalar_int_mode>.
18128 * combine.c (expand_compound_operation): Likewise.
18129 (make_extraction): Likewise.
18130 (change_zero_ext): Likewise.
18131 (simplify_comparison): Likewise.
18132 * cse.c (cse_insn): Likewise.
18133 * dwarf2out.c (minmax_loc_descriptor): Likewise.
18134 (mem_loc_descriptor): Likewise.
18135 (loc_descriptor): Likewise.
18136 * expmed.c (init_expmed_one_mode): Likewise.
18137 (synth_mult): Likewise.
18138 (emit_store_flag_1): Likewise.
18139 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
18140 of a comparison with size.
18141 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
18142 (reduce_to_bit_field_precision): Likewise.
18143 * function.c (expand_function_end): Likewise.
18144 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
18145 * loop-doloop.c (doloop_modify): Likewise.
18146 * optabs.c (expand_binop): Likewise.
18147 (expand_unop): Likewise.
18148 (expand_copysign_absneg): Likewise.
18149 (prepare_cmp_insn): Likewise.
18150 (maybe_legitimize_operand): Likewise.
18151 * recog.c (const_scalar_int_operand): Likewise.
18152 * rtlanal.c (get_address_mode): Likewise.
18153 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18154 (simplify_cond_clz_ctz): Likewise.
18155 * tree-nested.c (get_nl_goto_field): Likewise.
18156 * tree.c (build_vector_type_for_mode): Likewise.
18157 * var-tracking.c (use_narrower_mode): Likewise.
18159 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18160 Alan Hayward <alan.hayward@arm.com>
18161 David Sherwood <david.sherwood@arm.com>
18163 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
18164 * builtins.c (expand_builtin_signbit): Use it.
18165 * cfgexpand.c (expand_debug_expr): Likewise.
18166 * dojump.c (do_jump): Likewise.
18167 (do_compare_and_jump): Likewise.
18168 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
18169 * expmed.c (make_tree): Likewise.
18170 * expr.c (expand_expr_real_2): Likewise.
18171 (expand_expr_real_1): Likewise.
18172 (try_casesi): Likewise.
18173 * fold-const-call.c (fold_const_call_ss): Likewise.
18174 * fold-const.c (unextend): Likewise.
18175 (extract_muldiv_1): Likewise.
18176 (fold_single_bit_test): Likewise.
18177 (native_encode_int): Likewise.
18178 (native_encode_string): Likewise.
18179 (native_interpret_int): Likewise.
18180 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
18181 * internal-fn.c (expand_addsub_overflow): Likewise.
18182 (expand_neg_overflow): Likewise.
18183 (expand_mul_overflow): Likewise.
18184 (expand_arith_overflow): Likewise.
18185 * match.pd: Likewise.
18186 * stor-layout.c (layout_type): Likewise.
18187 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18188 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
18189 * tree-ssanames.c (get_range_info): Likewise.
18190 * tree-switch-conversion.c (array_value_type) Likewise.
18191 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
18192 (vect_recog_divmod_pattern): Likewise.
18193 (vect_recog_mixed_size_cond_pattern): Likewise.
18194 * tree-vrp.c (extract_range_basic): Likewise.
18195 (simplify_float_conversion_using_ranges): Likewise.
18196 * tree.c (int_fits_type_p): Likewise.
18197 * ubsan.c (instrument_bool_enum_load): Likewise.
18198 * varasm.c (mergeable_string_section): Likewise.
18199 (narrowing_initializer_constant_valid_p): Likewise.
18200 (output_constant): Likewise.
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 * machmode.h (NARROWEST_INT_MODE): New macro.
18207 * expr.c (alignment_for_piecewise_move): Use it instead of
18208 GET_CLASS_NARROWEST_MODE (MODE_INT).
18209 (push_block): Likewise.
18210 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
18212 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18214 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18215 Alan Hayward <alan.hayward@arm.com>
18216 David Sherwood <david.sherwood@arm.com>
18218 * postreload.c (move2add_valid_value_p): Change the type of the
18219 mode parameter to scalar_int_mode.
18220 (move2add_use_add2_insn): Add a mode parameter and use it instead
18222 (move2add_use_add3_insn): Likewise.
18223 (reload_cse_move2add): Update accordingly.
18225 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18226 Alan Hayward <alan.hayward@arm.com>
18227 David Sherwood <david.sherwood@arm.com>
18229 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
18231 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
18232 * optabs.c (expand_unop): Likewise.
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 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
18239 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
18240 (popcount_loc_descriptor): Likewise.
18241 (bswap_loc_descriptor): Likewise.
18242 (rotate_loc_descriptor): Likewise.
18243 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
18244 calling the functions above.
18246 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18247 Alan Hayward <alan.hayward@arm.com>
18248 David Sherwood <david.sherwood@arm.com>
18250 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
18252 (try_combine): Likewise.
18253 (simplify_if_then_else): Likewise.
18254 * cse.c (cse_insn): Likewise.
18255 * dwarf2out.c (mem_loc_descriptor): Likewise.
18256 * emit-rtl.c (gen_lowpart_common): Likewise.
18257 * simplify-rtx.c (simplify_truncation): Likewise.
18258 (simplify_binary_operation_1): Likewise.
18259 (simplify_const_relational_operation): Likewise.
18260 (simplify_ternary_operation): Likewise.
18261 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
18263 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18264 Alan Hayward <alan.hayward@arm.com>
18265 David Sherwood <david.sherwood@arm.com>
18267 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
18268 * reload.c (push_reload): Likewise.
18269 (find_reloads): Likewise.
18271 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18272 Alan Hayward <alan.hayward@arm.com>
18273 David Sherwood <david.sherwood@arm.com>
18275 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
18276 (make_compound_operation_int): Likewise.
18277 (change_zero_ext): Likewise.
18278 * expr.c (convert_move): Likewise.
18279 (convert_modes): Likewise.
18280 * fwprop.c (forward_propagate_subreg): Likewise.
18281 * loop-iv.c (get_biv_step_1): Likewise.
18282 * optabs.c (widen_operand): Likewise.
18283 * postreload.c (move2add_valid_value_p): Likewise.
18284 * recog.c (simplify_while_replacing): Likewise.
18285 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18286 (simplify_binary_operation_1): Likewise. Remove redundant
18287 mode equality check.
18289 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18290 Alan Hayward <alan.hayward@arm.com>
18291 David Sherwood <david.sherwood@arm.com>
18293 * combine.c (combine_simplify_rtx): Add checks for
18294 is_a <scalar_int_mode>.
18295 (simplify_if_then_else): Likewise.
18296 (make_field_assignment): Likewise.
18297 (simplify_comparison): Likewise.
18298 * ifcvt.c (noce_try_bitop): Likewise.
18299 * loop-invariant.c (canonicalize_address_mult): Likewise.
18300 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18302 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18303 Alan Hayward <alan.hayward@arm.com>
18304 David Sherwood <david.sherwood@arm.com>
18306 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
18307 is_a <scalar_int_mode> instead of != BLKmode.
18309 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18310 Alan Hayward <alan.hayward@arm.com>
18311 David Sherwood <david.sherwood@arm.com>
18313 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
18314 instead of != VOIDmode.
18315 * combine.c (if_then_else_cond): Likewise.
18316 (change_zero_ext): Likewise.
18317 * dwarf2out.c (mem_loc_descriptor): Likewise.
18318 (loc_descriptor): Likewise.
18319 * rtlanal.c (canonicalize_condition): Likewise.
18320 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
18322 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18323 Alan Hayward <alan.hayward@arm.com>
18324 David Sherwood <david.sherwood@arm.com>
18326 * simplify-rtx.c (simplify_binary_operation_1): Use
18327 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
18329 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18330 Alan Hayward <alan.hayward@arm.com>
18331 David Sherwood <david.sherwood@arm.com>
18333 * wide-int.h (int_traits<unsigned char>) New class.
18334 (int_traits<unsigned short>) Likewise.
18335 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
18336 Use GET_MODE_UNIT_PRECISION and remove redundant test for
18338 * combine.c (set_nonzero_bits_and_sign_copies): Use
18339 is_a <scalar_int_mode>.
18340 (find_split_point): Likewise.
18341 (combine_simplify_rtx): Likewise.
18342 (simplify_logical): Likewise.
18343 (expand_compound_operation): Likewise.
18344 (expand_field_assignment): Likewise.
18345 (make_compound_operation): Likewise.
18346 (extended_count): Likewise.
18347 (change_zero_ext): Likewise.
18348 (simplify_comparison): Likewise.
18349 * dwarf2out.c (scompare_loc_descriptor): Likewise.
18350 (ucompare_loc_descriptor): Likewise.
18351 (minmax_loc_descriptor): Likewise.
18352 (mem_loc_descriptor): Likewise.
18353 (loc_descriptor): Likewise.
18354 * expmed.c (init_expmed_one_mode): Likewise.
18355 * lra-constraints.c (lra_constraint_offset): Likewise.
18356 * optabs.c (prepare_libcall_arg): Likewise.
18357 * postreload.c (move2add_note_store): Likewise.
18358 * reload.c (operands_match_p): Likewise.
18359 * rtl.h (load_extend_op): Likewise.
18360 * rtlhooks.c (gen_lowpart_general): Likewise.
18361 * simplify-rtx.c (simplify_truncation): Likewise.
18362 (simplify_unary_operation_1): Likewise.
18363 (simplify_binary_operation_1): Likewise.
18364 (simplify_const_binary_operation): Likewise.
18365 (simplify_const_relational_operation): Likewise.
18366 (simplify_subreg): Likewise.
18367 * stor-layout.c (bitwise_mode_for_mode): Likewise.
18368 * var-tracking.c (adjust_mems): Likewise.
18369 (prepare_call_arguments): Likewise.
18371 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18372 Alan Hayward <alan.hayward@arm.com>
18373 David Sherwood <david.sherwood@arm.com>
18375 * machmode.h (is_int_mode): New fuction.
18376 * combine.c (find_split_point): Use it.
18377 (combine_simplify_rtx): Likewise.
18378 (simplify_if_then_else): Likewise.
18379 (simplify_set): Likewise.
18380 (simplify_shift_const_1): Likewise.
18381 (simplify_comparison): Likewise.
18382 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
18383 * cse.c (notreg_cost): Likewise.
18384 (cse_insn): Likewise.
18385 * cselib.c (cselib_lookup_1): Likewise.
18386 * dojump.c (do_jump_1): Likewise.
18387 (do_compare_rtx_and_jump): Likewise.
18388 * dse.c (get_call_args): Likewise.
18389 * dwarf2out.c (rtl_for_decl_init): Likewise.
18390 (native_encode_initializer): Likewise.
18391 * expmed.c (emit_store_flag_1): Likewise.
18392 (emit_store_flag): Likewise.
18393 * expr.c (convert_modes): Likewise.
18394 (store_field): Likewise.
18395 (expand_expr_real_1): Likewise.
18396 * fold-const.c (fold_read_from_constant_string): Likewise.
18397 * gimple-ssa-sprintf.c (get_format_string): Likewise.
18398 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
18399 * optabs.c (expand_binop): Likewise.
18400 (expand_unop): Likewise.
18401 (expand_abs_nojump): Likewise.
18402 (expand_one_cmpl_abs_nojump): Likewise.
18403 * simplify-rtx.c (mode_signbit_p): Likewise.
18404 (val_signbit_p): Likewise.
18405 (val_signbit_known_set_p): Likewise.
18406 (val_signbit_known_clear_p): Likewise.
18407 (simplify_relational_operation_1): Likewise.
18408 * tree.c (vector_type_mode): Likewise.
18410 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18411 Alan Hayward <alan.hayward@arm.com>
18412 David Sherwood <david.sherwood@arm.com>
18414 * machmode.h (smallest_mode_for_size): Fix formatting.
18415 (smallest_int_mode_for_size): New function.
18416 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
18417 instead of smallest_mode_for_size.
18418 * combine.c (make_extraction): Likewise.
18419 * config/arc/arc.c (arc_expand_movmem): Likewise.
18420 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
18421 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
18422 * config/s390/s390.c (s390_expand_insv): Likewise.
18423 * config/sparc/sparc.c (assign_int_registers): Likewise.
18424 * config/spu/spu.c (spu_function_value): Likewise.
18425 (spu_function_arg): Likewise.
18426 * coverage.c (get_gcov_type): Likewise.
18427 (get_gcov_unsigned_t): Likewise.
18428 * dse.c (find_shift_sequence): Likewise.
18429 * expmed.c (store_bit_field_1): Likewise.
18430 * expr.c (convert_move): Likewise.
18431 (store_field): Likewise.
18432 * internal-fn.c (expand_arith_overflow): Likewise.
18433 * optabs-query.c (get_best_extraction_insn): Likewise.
18434 * optabs.c (expand_twoval_binop_libfunc): Likewise.
18435 * stor-layout.c (layout_type): Likewise.
18436 (initialize_sizetypes): Likewise.
18437 * targhooks.c (default_get_mask_mode): Likewise.
18438 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
18440 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18441 Alan Hayward <alan.hayward@arm.com>
18442 David Sherwood <david.sherwood@arm.com>
18444 * machmode.h (opt_mode::else_blk): New function.
18445 (int_mode_for_mode): Declare.
18446 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
18447 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
18449 * cfgexpand.c (expand_debug_expr): Likewise.
18450 * combine.c (gen_lowpart_or_truncate): Likewise.
18451 (gen_lowpart_for_combine): Likewise.
18452 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
18453 * config/avr/avr.c (avr_to_int_mode): Likewise.
18454 (avr_out_plus_1): Likewise.
18455 (avr_out_plus): Likewise.
18456 (avr_out_round): Likewise.
18457 * config/i386/i386.c (ix86_split_to_parts): Likewise.
18458 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
18459 (s390_expand_vcond): Likewise.
18460 * config/spu/spu.c (spu_split_immediate): Likewise.
18461 (spu_expand_mov): Likewise.
18462 * dse.c (get_stored_val): Likewise.
18463 * expmed.c (store_bit_field_1): Likewise.
18464 (convert_extracted_bit_field): Use int_mode_for_mode instead of
18466 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
18467 (extract_low_bits): Likewise.
18468 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
18469 handling rather than repeating the check.
18470 (emit_group_store): Likewise.
18471 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
18472 * optabs.c (expand_absneg_bit): Likewise.
18473 (expand_copysign_absneg): Likewise.
18474 (expand_copysign_bit): Likewise.
18475 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
18476 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18477 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18478 * var-tracking.c (prepare_call_arguments): Likewise.
18479 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
18480 int_mode_for_mode instead of mode_for_size.
18481 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
18483 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18484 Alan Hayward <alan.hayward@arm.com>
18485 David Sherwood <david.sherwood@arm.com>
18487 * machmode.h (int_mode_for_size): New function.
18488 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
18489 instead of mode_for_size.
18490 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
18492 * combine.c (expand_field_assignment): Use int_mode_for_size
18493 instead of mode_for_size.
18494 (make_extraction): Likewise.
18495 (simplify_shift_const_1): Likewise.
18496 (simplify_comparison): Likewise.
18497 * dojump.c (do_jump): Likewise.
18498 * dwarf2out.c (mem_loc_descriptor): Likewise.
18499 * emit-rtl.c (init_derived_machine_modes): Likewise.
18500 * expmed.c (flip_storage_order): Likewise.
18501 (convert_extracted_bit_field): Likewise.
18502 * expr.c (copy_blkmode_from_reg): Likewise.
18503 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
18504 * internal-fn.c (expand_mul_overflow): Likewise.
18505 * lower-subreg.c (simple_move): Likewise.
18506 * optabs-libfuncs.c (init_optabs): Likewise.
18507 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18508 * tree.c (vector_type_mode): Likewise.
18509 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18510 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
18511 * tree-vect-generic.c (expand_vector_parallel): Likewise.
18512 * tree-vect-stmts.c (vectorizable_load): Likewise.
18513 (vectorizable_store): Likewise.
18515 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18516 Alan Hayward <alan.hayward@arm.com>
18517 David Sherwood <david.sherwood@arm.com>
18519 * coretypes.h (pod_mode): New type.
18520 (scalar_int_mode_pod): New typedef.
18521 * machmode.h (pod_mode): New class.
18522 (int_n_data_t::m): Change type to scalar_int_mode_pod.
18523 * genmodes.c (emit_mode_int_n): Update accordingly.
18524 * lower-subreg.h (target_lower_subreg): Change type to
18525 scalar_int_mode_pod.
18526 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
18527 scalar_int_mode_pod.
18529 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18530 Alan Hayward <alan.hayward@arm.com>
18531 David Sherwood <david.sherwood@arm.com>
18533 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
18534 machine_mode to scalar_int_mode.
18535 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
18536 (rs6000_option_override_internal): Remove cast to int.
18537 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
18538 machine_mode to scalar_int_mode.
18539 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
18540 (rs6000_option_override_internal): Remove cast to int.
18541 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
18542 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
18544 * config/s390/s390.c (s390_expand_builtin): Likewise.
18545 * coretypes.h (scalar_int_mode): New type.
18546 (opt_scalar_int_mode): New typedef.
18547 * machmode.h (scalar_int_mode): New class.
18548 (scalar_int_mode::includes_p): New function.
18549 (byte_mode): Change type to scalar_int_mode.
18550 (word_mode): Likewise.
18551 (ptr_mode): Likewise.
18552 * emit-rtl.c (byte_mode): Likewise.
18553 (word_mode): Likewise.
18554 (ptr_mode): Likewise.
18555 (init_derived_machine_modes): Update accordingly.
18556 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
18557 and MODE_PARTIAL_INT.
18558 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
18559 opt_scalar_int_mode.
18561 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18562 Alan Hayward <alan.hayward@arm.com>
18563 David Sherwood <david.sherwood@arm.com>
18565 * target.def (libgcc_floating_mode_supported_p): Take a
18567 * doc/tm.texi: Regenerate.
18568 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
18570 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18571 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
18574 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18575 Alan Hayward <alan.hayward@arm.com>
18576 David Sherwood <david.sherwood@arm.com>
18578 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
18579 * doc/tm.texi: Regenerate.
18580 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
18581 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
18582 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
18583 * targhooks.h (default_floatn_mode): Likewise.
18584 * targhooks.c (default_floatn_mode): Likewise.
18585 * tree.c (build_common_tree_nodes): Update accordingly.
18587 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18588 Alan Hayward <alan.hayward@arm.com>
18589 David Sherwood <david.sherwood@arm.com>
18591 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
18592 (mode_iterator::iterate_p): Likewise.
18593 (mode_iterator::get_wider): Likewise.
18594 * expr.c (init_expr_target): Use opt_scalar_float_mode.
18596 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18597 Alan Hayward <alan.hayward@arm.com>
18598 David Sherwood <david.sherwood@arm.com>
18600 * coretypes.h (opt_scalar_float_mode): New typedef.
18601 * machmode.h (float_mode_for_size): New function.
18602 * emit-rtl.c (double_mode): Delete.
18603 (init_emit_once): Use float_mode_for_size.
18604 * stor-layout.c (layout_type): Likewise.
18605 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
18607 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18608 Alan Hayward <alan.hayward@arm.com>
18609 David Sherwood <david.sherwood@arm.com>
18611 * output.h (assemble_real): Take a scalar_float_mode.
18612 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
18613 * config/arm/arm.md (consttable_4): Likewise.
18614 (consttable_8): Likewise.
18615 (consttable_16): Likewise.
18616 * config/mips/mips.md (consttable_float): Likewise.
18617 * config/s390/s390.c (s390_output_pool_entry): Likewise.
18618 * varasm.c (assemble_real): Take a scalar_float_mode.
18619 (output_constant_pool_2): Update accordingly.
18620 (output_constant): Likewise.
18622 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18623 Alan Hayward <alan.hayward@arm.com>
18624 David Sherwood <david.sherwood@arm.com>
18626 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
18627 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
18628 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
18629 (native_encode_real): Likewise.
18630 (native_interpret_real): Likewise.
18631 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
18632 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18635 Alan Hayward <alan.hayward@arm.com>
18636 David Sherwood <david.sherwood@arm.com>
18638 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
18639 <scalar_float_mode>. Simplify.
18640 (gen_extend_conv_libfunc): Likewise.
18642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18643 Alan Hayward <alan.hayward@arm.com>
18644 David Sherwood <david.sherwood@arm.com>
18646 * coretypes.h (scalar_float_mode): New type.
18647 * machmode.h (mode_traits::from_int): Use machine_mode if
18648 USE_ENUM_MODES is defined.
18649 (is_a): New function.
18651 (dyn_cast): Likewise.
18652 (scalar_float_mode): New class.
18653 (scalar_float_mode::includes_p): New function.
18654 (is_float_mode): Likewise.
18655 * gdbhooks.py (MachineModePrinter): New class.
18656 (build_pretty_printer): Use it for scalar_float_mode.
18657 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
18658 (format_helper::format_helper): Turn into a template.
18659 * genmodes.c (get_mode_class): New function.
18660 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
18661 or machine_mode if none.
18662 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
18663 as_a <scalar_float_mode>.
18664 * dwarf2out.c (mem_loc_descriptor): Likewise.
18665 (insert_float): Likewise.
18666 (add_const_value_attribute): Likewise.
18667 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18668 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
18669 (expand_unop): Update accordingly.
18670 (expand_abs_nojump): Likewise.
18671 (expand_copysign_absneg): Take a scalar_float_mode.
18672 (expand_copysign_bit): Likewise.
18673 (expand_copysign): Update accordingly.
18675 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18676 Alan Hayward <alan.hayward@arm.com>
18677 David Sherwood <david.sherwood@arm.com>
18679 * coretypes.h (opt_mode): New class.
18680 * machmode.h (opt_mode): Likewise.
18681 (opt_mode::else_void): New function.
18682 (opt_mode::require): Likewise.
18683 (opt_mode::exists): Likewise.
18684 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
18685 (GET_MODE_2XWIDER_MODE): Likewise.
18686 (mode_iterator::get_wider): Update accordingly.
18687 (mode_iterator::get_2xwider): Likewise.
18688 (mode_iterator::get_known_wider): Likewise, turning into a template.
18689 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
18690 forcing a wider mode to exist.
18691 * config/cr16/cr16.h (LONG_REG_P): Likewise.
18692 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18693 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
18694 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
18695 * lower-subreg.c (init_lower_subreg): Likewise.
18696 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
18697 on the final iteration.
18698 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
18699 a wider mode exists before asking for a move pattern.
18700 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
18701 forcing a wider mode to exist.
18702 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
18703 returning false if no such mode exists.
18704 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
18705 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
18706 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
18707 Avoid checking for a MODE_INT if we already know the mode is not a
18709 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
18710 forcing a wider mode to exist.
18711 (expmed_mult_highpart_optab): Likewise.
18712 (expmed_mult_highpart): Likewise.
18713 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
18715 * lto-streamer-in.c (lto_input_mode_table): Likewise.
18716 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
18717 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
18718 * internal-fn.c (expand_mul_overflow): Update use of
18719 GET_MODE_2XWIDER_MODE.
18720 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18721 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
18722 GET_MODE_WIDER_MODE.
18723 (convert_plusminus_to_widen): Likewise.
18724 * tree-switch-conversion.c (array_value_type): Likewise.
18725 * var-tracking.c (emit_note_insn_var_location): Likewise.
18726 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18727 Return false inside rather than outside the loop if no wider mode
18729 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
18730 and GET_MODE_2XWIDER_MODE
18731 (can_compare_p): Use else_void.
18732 * gdbhooks.py (OptMachineModePrinter): New class.
18733 (build_pretty_printer): Use it for opt_mode.
18735 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18736 Alan Hayward <alan.hayward@arm.com>
18737 David Sherwood <david.sherwood@arm.com>
18739 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
18740 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
18742 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18743 Alan Hayward <alan.hayward@arm.com>
18744 David Sherwood <david.sherwood@arm.com>
18746 * machmode.h (mode_traits): New structure.
18747 (get_narrowest_mode): New function.
18748 (mode_iterator::start): Likewise.
18749 (mode_iterator::iterate_p): Likewise.
18750 (mode_iterator::get_wider): Likewise.
18751 (mode_iterator::get_known_wider): Likewise.
18752 (mode_iterator::get_2xwider): Likewise.
18753 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
18754 (FOR_EACH_MODE): Likewise.
18755 (FOR_EACH_MODE_FROM): Likewise.
18756 (FOR_EACH_MODE_UNTIL): Likewise.
18757 (FOR_EACH_WIDER_MODE): Likewise.
18758 (FOR_EACH_2XWIDER_MODE): Likewise.
18759 * builtins.c (expand_builtin_strlen): Use new mode iterators.
18760 * combine.c (simplify_comparison): Likewise
18761 * config/i386/i386.c (type_natural_mode): Likewise.
18762 * cse.c (cse_insn): Likewise.
18763 * dse.c (find_shift_sequence): Likewise.
18764 * emit-rtl.c (init_derived_machine_modes): Likewise.
18765 (init_emit_once): Likewise.
18766 * explow.c (hard_function_value): Likewise.
18767 * expmed.c (extract_fixed_bit_field_1): Likewise.
18768 (extract_bit_field_1): Likewise.
18769 (expand_divmod): Likewise.
18770 (emit_store_flag_1): Likewise.
18771 * expr.c (init_expr_target): Likewise.
18772 (convert_move): Likewise.
18773 (alignment_for_piecewise_move): Likewise.
18774 (widest_int_mode_for_size): Likewise.
18775 (emit_block_move_via_movmem): Likewise.
18776 (copy_blkmode_to_reg): Likewise.
18777 (set_storage_via_setmem): Likewise.
18778 (compress_float_constant): Likewise.
18779 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18780 * optabs-query.c (get_best_extraction_insn): Likewise.
18781 * optabs.c (expand_binop): Likewise.
18782 (expand_twoval_unop): Likewise.
18783 (expand_twoval_binop): Likewise.
18784 (widen_leading): Likewise.
18785 (widen_bswap): Likewise.
18786 (expand_parity): Likewise.
18787 (expand_unop): Likewise.
18788 (prepare_cmp_insn): Likewise.
18789 (prepare_float_lib_cmp): Likewise.
18790 (expand_float): Likewise.
18791 (expand_fix): Likewise.
18792 (expand_sfix_optab): Likewise.
18793 * postreload.c (move2add_use_add2_insn): Likewise.
18794 * reg-stack.c (reg_to_stack): Likewise.
18795 * reginfo.c (choose_hard_reg_mode): Likewise.
18796 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18797 * stor-layout.c (mode_for_size): Likewise.
18798 (smallest_mode_for_size): Likewise.
18799 (mode_for_vector): Likewise.
18800 (finish_bitfield_representative): Likewise.
18801 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
18802 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
18803 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18804 * var-tracking.c (prepare_call_arguments): Likewise.
18806 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18807 Alan Hayward <alan.hayward@arm.com>
18808 David Sherwood <david.sherwood@arm.com>
18810 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
18811 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
18812 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
18813 * machmode.h (mode_size): Move earlier in file.
18814 (mode_precision): Likewise.
18815 (mode_inner): Likewise.
18816 (mode_nunits): Likewise.
18817 (mode_unit_size): Likewise.
18818 (unit_unit_precision): Likewise.
18819 (mode_wider): Likewise.
18820 (mode_2xwider): Likewise.
18821 (machine_mode): New class.
18822 (mode_to_bytes): New function.
18823 (mode_to_bits): Likewise.
18824 (mode_to_precision): Likewise.
18825 (mode_to_inner): Likewise.
18826 (mode_to_unit_size): Likewise.
18827 (mode_to_unit_precision): Likewise.
18828 (mode_to_nunits): Likewise.
18829 (GET_MODE_SIZE): Use mode_to_bytes.
18830 (GET_MODE_BITSIZE): Use mode_to_bits.
18831 (GET_MODE_PRECISION): Use mode_to_precision.
18832 (GET_MODE_INNER): Use mode_to_inner.
18833 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
18834 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
18835 (GET_MODE_NUNITS): Use mode_to_nunits.
18836 * system.h (ALWAYS_INLINE): New macro.
18837 * config/powerpcspe/powerpcspe-c.c
18838 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
18839 int for arg1_mode and arg2_mode.
18841 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18842 Alan Hayward <alan.hayward@arm.com>
18843 David Sherwood <david.sherwood@arm.com>
18845 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18846 Prefix mode names with E_ in case statements.
18847 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18848 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
18849 (aarch64_split_simd_move): Likewise.
18850 (aarch64_gen_storewb_pair): Likewise.
18851 (aarch64_gen_loadwb_pair): Likewise.
18852 (aarch64_gen_store_pair): Likewise.
18853 (aarch64_gen_load_pair): Likewise.
18854 (aarch64_get_condition_code_1): Likewise.
18855 (aarch64_constant_pool_reload_icode): Likewise.
18856 (get_rsqrte_type): Likewise.
18857 (get_rsqrts_type): Likewise.
18858 (get_recpe_type): Likewise.
18859 (get_recps_type): Likewise.
18860 (aarch64_gimplify_va_arg_expr): Likewise.
18861 (aarch64_simd_container_mode): Likewise.
18862 (aarch64_emit_load_exclusive): Likewise.
18863 (aarch64_emit_store_exclusive): Likewise.
18864 (aarch64_expand_compare_and_swap): Likewise.
18865 (aarch64_gen_atomic_cas): Likewise.
18866 (aarch64_emit_bic): Likewise.
18867 (aarch64_emit_atomic_swap): Likewise.
18868 (aarch64_emit_atomic_load_op): Likewise.
18869 (aarch64_evpc_trn): Likewise.
18870 (aarch64_evpc_uzp): Likewise.
18871 (aarch64_evpc_zip): Likewise.
18872 (aarch64_evpc_ext): Likewise.
18873 (aarch64_evpc_rev): Likewise.
18874 (aarch64_evpc_dup): Likewise.
18875 (aarch64_gen_ccmp_first): Likewise.
18876 (aarch64_gen_ccmp_next): Likewise.
18877 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18878 (alpha_emit_xfloating_libcall): Likewise.
18879 (emit_insxl): Likewise.
18880 (alpha_arg_type): Likewise.
18881 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
18882 (arc_preferred_simd_mode): Likewise.
18883 (arc_secondary_reload): Likewise.
18884 (get_arc_condition_code): Likewise.
18885 (arc_print_operand): Likewise.
18886 (arc_legitimate_constant_p): Likewise.
18887 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18888 * config/arc/arc.md (casesi_load): Likewise.
18889 (casesi_compact_jump): Likewise.
18890 * config/arc/predicates.md (proper_comparison_operator): Likewise.
18891 (cc_use_register): Likewise.
18892 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18893 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18894 (arm_init_iwmmxt_builtins): Likewise.
18895 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
18896 (neon_expand_vector_init): Likewise.
18897 (arm_attr_length_move_neon): Likewise.
18898 (maybe_get_arm_condition_code): Likewise.
18899 (arm_emit_vector_const): Likewise.
18900 (arm_preferred_simd_mode): Likewise.
18901 (arm_output_iwmmxt_tinsr): Likewise.
18902 (thumb1_output_casesi): Likewise.
18903 (thumb2_output_casesi): Likewise.
18904 (arm_emit_load_exclusive): Likewise.
18905 (arm_emit_store_exclusive): Likewise.
18906 (arm_expand_compare_and_swap): Likewise.
18907 (arm_evpc_neon_vuzp): Likewise.
18908 (arm_evpc_neon_vzip): Likewise.
18909 (arm_evpc_neon_vrev): Likewise.
18910 (arm_evpc_neon_vtrn): Likewise.
18911 (arm_evpc_neon_vext): Likewise.
18912 (arm_validize_comparison): Likewise.
18913 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
18914 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
18915 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
18916 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
18917 (c6x_preferred_simd_mode): Likewise.
18918 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
18919 (epiphany_rtx_costs): Likewise.
18920 * config/epiphany/predicates.md (proper_comparison_operator):
18922 * config/frv/frv.c (condexec_memory_operand): Likewise.
18923 (frv_emit_move): Likewise.
18924 (output_move_single): Likewise.
18925 (output_condmove_single): Likewise.
18926 (frv_hard_regno_mode_ok): Likewise.
18927 (frv_matching_accg_mode): Likewise.
18928 * config/h8300/h8300.c (split_adds_subs): Likewise.
18929 (h8300_rtx_costs): Likewise.
18930 (h8300_print_operand): Likewise.
18931 (compute_mov_length): Likewise.
18932 (output_logical_op): Likewise.
18933 (compute_logical_op_length): Likewise.
18934 (compute_logical_op_cc): Likewise.
18935 (h8300_shift_needs_scratch_p): Likewise.
18936 (output_a_shift): Likewise.
18937 (compute_a_shift_length): Likewise.
18938 (compute_a_shift_cc): Likewise.
18939 (expand_a_rotate): Likewise.
18940 (output_a_rotate): Likewise.
18941 * config/i386/i386.c (classify_argument): Likewise.
18942 (function_arg_advance_32): Likewise.
18943 (function_arg_32): Likewise.
18944 (function_arg_64): Likewise.
18945 (function_value_64): Likewise.
18946 (ix86_gimplify_va_arg): Likewise.
18947 (ix86_legitimate_constant_p): Likewise.
18948 (put_condition_code): Likewise.
18949 (split_double_mode): Likewise.
18950 (ix86_avx256_split_vector_move_misalign): Likewise.
18951 (ix86_expand_vector_logical_operator): Likewise.
18952 (ix86_split_idivmod): Likewise.
18953 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
18954 (ix86_build_const_vector): Likewise.
18955 (ix86_build_signbit_mask): Likewise.
18956 (ix86_match_ccmode): Likewise.
18957 (ix86_cc_modes_compatible): Likewise.
18958 (ix86_expand_branch): Likewise.
18959 (ix86_expand_sse_cmp): Likewise.
18960 (ix86_expand_sse_movcc): Likewise.
18961 (ix86_expand_int_sse_cmp): Likewise.
18962 (ix86_expand_vec_perm_vpermi2): Likewise.
18963 (ix86_expand_vec_perm): Likewise.
18964 (ix86_expand_sse_unpack): Likewise.
18965 (ix86_expand_int_addcc): Likewise.
18966 (ix86_split_to_parts): Likewise.
18967 (ix86_vectorize_builtin_gather): Likewise.
18968 (ix86_vectorize_builtin_scatter): Likewise.
18969 (avx_vpermilp_parallel): Likewise.
18970 (inline_memory_move_cost): Likewise.
18971 (ix86_tieable_integer_mode_p): Likewise.
18972 (x86_maybe_negate_const_int): Likewise.
18973 (ix86_expand_vector_init_duplicate): Likewise.
18974 (ix86_expand_vector_init_one_nonzero): Likewise.
18975 (ix86_expand_vector_init_one_var): Likewise.
18976 (ix86_expand_vector_init_concat): Likewise.
18977 (ix86_expand_vector_init_interleave): Likewise.
18978 (ix86_expand_vector_init_general): Likewise.
18979 (ix86_expand_vector_set): Likewise.
18980 (ix86_expand_vector_extract): Likewise.
18981 (emit_reduc_half): Likewise.
18982 (ix86_emit_i387_round): Likewise.
18983 (ix86_mangle_type): Likewise.
18984 (ix86_expand_round_sse4): Likewise.
18985 (expand_vec_perm_blend): Likewise.
18986 (canonicalize_vector_int_perm): Likewise.
18987 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
18988 (expand_vec_perm_1): Likewise.
18989 (expand_vec_perm_interleave3): Likewise.
18990 (expand_vec_perm_even_odd_pack): Likewise.
18991 (expand_vec_perm_even_odd_1): Likewise.
18992 (expand_vec_perm_broadcast_1): Likewise.
18993 (ix86_vectorize_vec_perm_const_ok): Likewise.
18994 (ix86_expand_vecop_qihi): Likewise.
18995 (ix86_expand_mul_widen_hilo): Likewise.
18996 (ix86_expand_sse2_abs): Likewise.
18997 (ix86_expand_pextr): Likewise.
18998 (ix86_expand_pinsr): Likewise.
18999 (ix86_preferred_simd_mode): Likewise.
19000 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
19001 * config/i386/sse.md (*andnot<mode>3): Likewise.
19002 (<mask_codefor><code><mode>3<mask_name>): Likewise.
19003 (*<code><mode>3): Likewise.
19004 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
19005 (ia64_expand_atomic_op): Likewise.
19006 (ia64_arg_type): Likewise.
19007 (ia64_mode_to_int): Likewise.
19008 (ia64_scalar_mode_supported_p): Likewise.
19009 (ia64_vector_mode_supported_p): Likewise.
19010 (expand_vec_perm_broadcast): Likewise.
19011 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
19012 (iq2000_function_arg_advance): Likewise.
19013 (iq2000_function_arg): Likewise.
19014 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
19015 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
19016 (m68k_libcall_value): Likewise.
19017 (m68k_function_value): Likewise.
19018 (sched_attr_op_type): Likewise.
19019 * config/mcore/mcore.c (mcore_output_move): Likewise.
19020 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
19022 (microblaze_function_arg): Likewise.
19023 * config/mips/mips.c (mips16_build_call_stub): Likewise.
19024 (mips_print_operand): Likewise.
19025 (mips_mode_ok_for_mov_fmt_p): Likewise.
19026 (mips_vector_mode_supported_p): Likewise.
19027 (mips_preferred_simd_mode): Likewise.
19028 (mips_expand_vpc_loongson_even_odd): Likewise.
19029 (mips_expand_vec_unpack): Likewise.
19030 (mips_expand_vi_broadcast): Likewise.
19031 (mips_expand_vector_init): Likewise.
19032 (mips_expand_vec_reduc): Likewise.
19033 (mips_expand_msa_cmp): Likewise.
19034 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
19035 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
19036 (cc_flags_for_mode): Likewise.
19037 * config/msp430/msp430.c (msp430_print_operand): Likewise.
19038 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
19039 (nds32_output_casesi_pc_relative): Likewise.
19040 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19041 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
19042 (nvptx_gen_unpack): Likewise.
19043 (nvptx_gen_pack): Likewise.
19044 (nvptx_gen_shuffle): Likewise.
19045 (nvptx_gen_wcast): Likewise.
19046 (nvptx_preferred_simd_mode): Likewise.
19047 * config/pa/pa.c (pa_secondary_reload): Likewise.
19048 * config/pa/predicates.md (base14_operand): Likewise.
19049 * config/powerpcspe/powerpcspe-c.c
19050 (altivec_resolve_overloaded_builtin): Likewise.
19051 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
19053 (rs6000_preferred_simd_mode): Likewise.
19054 (output_vec_const_move): Likewise.
19055 (rs6000_expand_vector_extract): Likewise.
19056 (rs6000_split_vec_extract_var): Likewise.
19057 (reg_offset_addressing_ok_p): Likewise.
19058 (rs6000_legitimate_offset_address_p): Likewise.
19059 (rs6000_legitimize_address): Likewise.
19060 (rs6000_emit_set_const): Likewise.
19061 (rs6000_const_vec): Likewise.
19062 (rs6000_emit_move): Likewise.
19063 (spe_build_register_parallel): Likewise.
19064 (rs6000_darwin64_record_arg_recurse): Likewise.
19065 (swap_selector_for_mode): Likewise.
19066 (spe_init_builtins): Likewise.
19067 (paired_init_builtins): Likewise.
19068 (altivec_init_builtins): Likewise.
19069 (do_load_for_compare): Likewise.
19070 (rs6000_generate_compare): Likewise.
19071 (rs6000_expand_float128_convert): Likewise.
19072 (emit_load_locked): Likewise.
19073 (emit_store_conditional): Likewise.
19074 (rs6000_output_function_epilogue): Likewise.
19075 (rs6000_handle_altivec_attribute): Likewise.
19076 (rs6000_function_value): Likewise.
19077 (emit_fusion_gpr_load): Likewise.
19078 (emit_fusion_p9_load): Likewise.
19079 (emit_fusion_p9_store): Likewise.
19080 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
19081 (fusion_gpr_mem_load): Likewise.
19082 (fusion_addis_mem_combo_load): Likewise.
19083 (fusion_addis_mem_combo_store): Likewise.
19084 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
19085 (fusion_gpr_mem_load): Likewise.
19086 (fusion_addis_mem_combo_load): Likewise.
19087 (fusion_addis_mem_combo_store): Likewise.
19088 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19090 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
19091 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19092 (rs6000_preferred_simd_mode): Likewise.
19093 (output_vec_const_move): Likewise.
19094 (rs6000_expand_vector_extract): Likewise.
19095 (rs6000_split_vec_extract_var): Likewise.
19096 (reg_offset_addressing_ok_p): Likewise.
19097 (rs6000_legitimate_offset_address_p): Likewise.
19098 (rs6000_legitimize_address): Likewise.
19099 (rs6000_emit_set_const): Likewise.
19100 (rs6000_const_vec): Likewise.
19101 (rs6000_emit_move): Likewise.
19102 (rs6000_darwin64_record_arg_recurse): Likewise.
19103 (swap_selector_for_mode): Likewise.
19104 (paired_init_builtins): Likewise.
19105 (altivec_init_builtins): Likewise.
19106 (rs6000_expand_float128_convert): Likewise.
19107 (emit_load_locked): Likewise.
19108 (emit_store_conditional): Likewise.
19109 (rs6000_output_function_epilogue): Likewise.
19110 (rs6000_handle_altivec_attribute): Likewise.
19111 (rs6000_function_value): Likewise.
19112 (emit_fusion_gpr_load): Likewise.
19113 (emit_fusion_p9_load): Likewise.
19114 (emit_fusion_p9_store): Likewise.
19115 * config/rx/rx.c (rx_gen_move_template): Likewise.
19116 (flags_from_mode): Likewise.
19117 * config/s390/predicates.md (s390_alc_comparison): Likewise.
19118 (s390_slb_comparison): Likewise.
19119 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
19120 (s390_vector_mode_supported_p): Likewise.
19121 (s390_cc_modes_compatible): Likewise.
19122 (s390_match_ccmode_set): Likewise.
19123 (s390_canonicalize_comparison): Likewise.
19124 (s390_emit_compare_and_swap): Likewise.
19125 (s390_branch_condition_mask): Likewise.
19126 (s390_rtx_costs): Likewise.
19127 (s390_secondary_reload): Likewise.
19128 (__SECONDARY_RELOAD_CASE): Likewise.
19129 (s390_expand_cs): Likewise.
19130 (s390_preferred_simd_mode): Likewise.
19131 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
19132 * config/sh/sh.c (sh_print_operand): Likewise.
19133 (dump_table): Likewise.
19134 (sh_secondary_reload): Likewise.
19135 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19136 * config/sh/sh.md (casesi_worker_1): Likewise.
19137 (casesi_worker_2): Likewise.
19138 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
19139 (fcc_comparison_operator): Likewise.
19140 * config/sparc/sparc.c (sparc_expand_move): Likewise.
19141 (emit_soft_tfmode_cvt): Likewise.
19142 (sparc_preferred_simd_mode): Likewise.
19143 (output_cbranch): Likewise.
19144 (sparc_print_operand): Likewise.
19145 (sparc_expand_vec_perm_bmask): Likewise.
19146 (vector_init_bshuffle): Likewise.
19147 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
19148 (spu_vector_mode_supported_p): Likewise.
19149 (spu_expand_insv): Likewise.
19150 (spu_emit_branch_or_set): Likewise.
19151 (spu_handle_vector_attribute): Likewise.
19152 (spu_builtin_splats): Likewise.
19153 (spu_builtin_extract): Likewise.
19154 (spu_builtin_promote): Likewise.
19155 (spu_expand_sign_extend): Likewise.
19156 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
19157 (tilegx_simd_int): Likewise.
19158 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
19159 (tilepro_simd_int): Likewise.
19160 * config/v850/v850.c (const_double_split): Likewise.
19161 (v850_print_operand): Likewise.
19162 (ep_memory_offset): Likewise.
19163 * config/vax/vax.c (vax_rtx_costs): Likewise.
19164 (vax_output_int_move): Likewise.
19165 (vax_output_int_add): Likewise.
19166 (vax_output_int_subtract): Likewise.
19167 * config/visium/predicates.md (visium_branch_operator): Likewise.
19168 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
19169 (visium_print_operand_address): Likewise.
19170 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19171 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
19172 (xtensa_expand_conditional_branch): Likewise.
19173 (xtensa_copy_incoming_a7): Likewise.
19174 (xtensa_output_literal): Likewise.
19175 * dfp.c (decimal_real_maxval): Likewise.
19176 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19178 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19179 Alan Hayward <alan.hayward@arm.com>
19180 David Sherwood <david.sherwood@arm.com>
19182 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
19183 (mode_nunits_inline): Likewise.
19184 (mode_inner_inline): Likewise.
19185 (mode_unit_size_inline): Likewise.
19186 (mode_unit_precision_inline): Likewise.
19187 (emit_insn_modes_h): Likewise. Also emit a #define of the
19189 (emit_mode_wider): Add an E_ prefix to mode names.
19190 (emit_mode_complex): Likewise.
19191 (emit_mode_inner): Likewise.
19192 (emit_mode_adjustments): Likewise.
19193 (emit_mode_int_n): Likewise.
19194 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
19195 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
19196 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
19197 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
19198 (CRC32_BUILTIN, ENTRY): Likewise.
19199 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
19200 (aarch64_pop_regs): Likewise.
19201 (aarch64_process_components): Likewise.
19202 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
19203 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
19204 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
19205 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
19206 * config/arm/arm.c (arm_init_libfuncs): Likewise.
19207 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
19209 * config/i386/i386-builtin.def (pcmpestr): Likewise.
19210 (pcmpistr): Likewise.
19211 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
19212 * config/mmix/mmix.c (mmix_output_condition): Likewise.
19213 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19215 * config/rl78/rl78.c (mduc_regs): Likewise.
19216 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19217 (htm_expand_builtin): Likewise.
19218 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
19219 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
19220 * config/xtensa/xtensa.c (print_operand): Likewise.
19221 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
19222 (NUM_MODE_VECTOR_INT): Likewise.
19223 * genoutput.c (null_operand): Likewise.
19224 (output_operand_data): Likewise.
19225 * genrecog.c (print_parameter_value): Likewise.
19226 * lra.c (debug_operand_data): Likewise.
19228 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19230 * dwarf2out.c (scompare_loc_descriptor_wide)
19231 (scompare_loc_descriptor_narrow): New functions, split out from...
19232 (scompare_loc_descriptor): ...here.
19233 * expmed.c (emit_store_flag_int): New function, split out from...
19234 (emit_store_flag): ...here.
19236 2017-08-30 Richard Biener <rguenther@suse.de>
19238 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
19239 (dwarf2out_early_finish): Move setting of AT_pubnames from
19240 early debug output to early finish.
19242 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19244 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
19245 and -mdata-region to the assembler.
19247 2017-08-30 Richard Biener <rguenther@suse.de>
19249 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
19251 (gen_subprogram_die): Add DW_AT_object_pointer only early.
19252 (dwarf2out_early_global_decl): Only generate a DIE for the
19253 abstract origin if it doesn't already exist or is a declaration DIE.
19254 (resolve_addr): Do not add the linkage name twice when
19255 generating a stub DIE for the DW_TAG_GNU_call_site target.
19257 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19259 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19260 Use machine_mode rather than int for arg1_mode.
19262 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19265 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
19266 that the second argument of the built-in functions to unpack
19267 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
19268 switch statement instead a lot of if statements.
19269 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
19270 Allow 64-bit values to be in Altivec registers as well as
19271 traditional floating point registers.
19272 (pack<mode>, FMOVE128_VSX iterator): Likewise.
19274 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
19276 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
19277 MAX_REGS_PER_ADDRESS == 1.
19279 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
19281 * config/i386/i386.opt (flag_fentry): Do not init to -1.
19282 * config/i386/i386.c (ix86_option_override_internal): Simplify
19283 setting of opts->x_flag_entry.
19285 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19286 Jakub Jelinek <jakub@redhat.com>
19287 Richard Biener <rguenther@suse.de>
19289 PR tree-optimization/81503
19290 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
19291 folded constant fits in the target type; reorder tests for clarity.
19293 2017-08-29 Martin Liska <mliska@suse.cz>
19295 * passes.def: Include pass_lower_switch.
19296 * stmt.c (dump_case_nodes): Remove and move to
19297 tree-switch-conversion.
19298 (case_values_threshold): Likewise.
19299 (expand_switch_as_decision_tree_p): Likewise.
19300 (emit_case_decision_tree): Likewise.
19301 (expand_case): Likewise.
19302 (balance_case_nodes): Likewise.
19303 (node_has_low_bound): Likewise.
19304 (node_has_high_bound): Likewise.
19305 (node_is_bounded): Likewise.
19306 (emit_case_nodes): Likewise.
19307 (struct simple_case_node): New struct.
19308 (add_case_node): Remove.
19309 (emit_case_dispatch_table): Use vector instead of case_list.
19310 (reset_out_edges_aux): Remove.
19311 (compute_cases_per_edge): Likewise.
19312 (expand_case): Build list of simple_case_node.
19313 (expand_sjlj_dispatch_table): Use it.
19314 * tree-switch-conversion.c (struct case_node): Moved from
19315 stmt.c and adjusted.
19316 (emit_case_nodes): Likewise.
19317 (node_has_low_bound): Likewise.
19318 (node_has_high_bound): Likewise.
19319 (node_is_bounded): Likewise.
19320 (case_values_threshold): Likewise.
19321 (reset_out_edges_aux): Likewise.
19322 (compute_cases_per_edge): Likewise.
19323 (add_case_node): Likewise.
19324 (dump_case_nodes): Likewise.
19325 (balance_case_nodes): Likewise.
19326 (expand_switch_as_decision_tree_p): Likewise.
19327 (emit_jump): Likewise.
19328 (emit_case_decision_tree): Likewise.
19329 (try_switch_expansion): Likewise.
19330 (do_jump_if_equal): Likewise.
19331 (emit_cmp_and_jump_insns): Likewise.
19332 (fix_phi_operands_for_edge): New function.
19333 (record_phi_operand_mapping): Likewise.
19334 (class pass_lower_switch): New pass.
19335 (pass_lower_switch::execute): New function.
19336 (make_pass_lower_switch): Likewise.
19337 (conditional_probability):
19338 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
19339 * tree-pass.h: Add make_pass_lower_switch.
19341 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19344 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
19347 2017-08-29 Richard Biener <rguenther@suse.de>
19349 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
19350 we do not add a DW_AT_inline attribute twice.
19351 (gen_subprogram_die): Remove code setting DW_AT_inline on
19352 DECL_ABSTRACT_P nodes.
19354 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
19356 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
19357 calls to internal functions.
19358 (gimplify_modify_expr): Likewise.
19359 * tree-call-cdce.c (use_internal_fn): Likewise.
19360 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
19361 (convert_to_divmod): Set the nothrow flag.
19362 * tree-if-conv.c (predicate_mem_writes): Likewise.
19363 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19364 (vectorizable_call): Likewise.
19365 (vectorizable_store): Likewise.
19366 (vectorizable_load): Likewise.
19367 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
19368 (vect_recog_mask_conversion_pattern): Likewise.
19370 2017-08-29 Martin Liska <mliska@suse.cz>
19373 * gcc.c (driver_handle_option): Add new argument.
19374 * opts-common.c (handle_option): Pass
19375 target_option_override_hook.
19376 * opts-global.c (lang_handle_option): Add new option.
19377 (set_default_handlers): Add new argument.
19378 (decode_options): Likewise.
19379 * opts.c (target_handle_option): Likewise.
19380 (common_handle_option): Call target_option_override_hook.
19381 * opts.h (struct cl_option_handler_func): Add hook for
19382 target option override.
19383 (struct cl_option_handlers): Likewise.
19384 (set_default_handlers): Add new argument.
19385 (decode_options): Likewise.
19386 (common_handle_option): Likewise.
19387 (target_handle_option): Likewise.
19388 * toplev.c (toplev::main): Pass targetm.target_option.override
19391 2017-08-29 Richard Biener <rguenther@suse.de>
19392 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
19394 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
19395 life to the active subtree.
19397 2017-08-28 Jeff Law <law@redhat.com>
19399 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
19400 derive_equivalences.
19401 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
19402 Code moved into....
19403 (edge_info::derive_equivalences): New private member function
19405 * tree-ssa-dom.c (class edge_info): Changed from a struct
19406 to a class. Add ctor/dtor, methods and data members.
19407 (edge_info::edge_info): Renamed from allocate_edge_info.
19408 Initialize additional members.
19409 (edge_info::~edge_info): New.
19410 (free_dom_edge_info): Delete the edge info.
19411 (record_edge_info): Use new class & associated member functions.
19412 Tighten forms for testing for edge equivalences.
19413 (record_temporary_equivalences): Iterate over the simple
19414 equivalences rather than assuming there's only one per edge.
19415 (cprop_into_successor_phis): Iterate over the simple
19416 equivalences rather than assuming there's only one per edge.
19417 (optimize_stmt): Use operand_equal_p rather than pointer
19418 equality for mini-DSE code.
19420 2017-08-28 Nathan Sidwell <nathan@acm.org>
19422 * gcc.c (execute): Fold SIGPIPE handling into switch
19423 statement. Adjust internal error message.
19425 2017-08-28 Richard Biener <rguenther@suse.de>
19428 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
19429 Do nothing for removed DIEs.
19431 2017-08-28 Richard Biener <rguenther@suse.de>
19433 PR tree-optimization/81977
19434 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
19437 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
19440 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
19441 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
19442 using targetm.gen_mem_thread_fence.
19444 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
19447 * config/i386/i386.md (*<btsc><mode>): Change operand 2
19448 predicate to register_operand. Reorder operands.
19449 (*btr<mode>): Ditto.
19450 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
19451 (*btr<mode>_mask): Ditto.
19453 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
19455 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
19456 * config/rs6000/xmmintrin.h: New file.
19457 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
19459 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19462 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
19463 parameter and_insn and return it.
19464 (recombine_lvx_pattern): Insert a copy to ensure availability of
19465 the base register of the copied masking operation at the point of
19466 the instruction replacement.
19467 (recombine_stvx_pattern): Likewise.
19469 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
19471 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
19472 undocumented switches.
19473 (-mpower9-dform-vector): Likewise.
19474 (-mpower9-dform): Likewise.
19475 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19476 comments to delete references to -mpower9-dform* switches.
19477 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
19478 Delete reference to -mpower9-dform* switches, test for
19479 -mpower9-vector instead.
19480 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
19481 (OTHER_P9_VECTOR_MASKS): Likewise.
19482 (POWERPC_MASKS): Likewise.
19483 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
19484 tests against -mpower9-dform* to -mpower9-vector. Delete code
19485 that checked for -mpower9-dform* consistancy with other options.
19486 Add test for -mpower9-misc to enable other power9 switches.
19487 (rs6000_init_hard_regno_mode_ok): Likewise.
19488 (rs6000_option_override_internal): Likewise.
19489 (rs6000_emit_prologue): Likewise.
19490 (rs6000_emit_epilogue): Likewise.
19491 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
19492 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
19493 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
19495 (emit_fusion_p9_store): Likewise.
19496 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
19497 resetting these macros if the assembler does not support ISA 3.0
19499 (TARGET_P9_DFORM_VECTOR): Likewise.
19500 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
19501 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
19503 2017-08-25 Alan Modra <amodra@gmail.com>
19505 PR rtl-optimization/81747
19506 * cse.c (cse_extended_basic_block): Don't attempt to record
19507 equivalences for degenerate conditional branches that branch
19508 to their fall-through.
19510 2017-08-24 Martin Sebor <msebor@redhat.com>
19512 PR middle-end/81908
19513 * gimple-fold.c (size_must_be_zero_p): New function.
19514 (gimple_fold_builtin_memory_op): Call it.
19516 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
19518 * config/rs6000/mm_malloc.h: New file.
19520 2017-08-24 Bin Cheng <bin.cheng@arm.com>
19522 PR tree-optimization/81913
19523 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
19524 analysis when either IVs in condition can wrap.
19526 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
19528 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
19529 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
19531 2017-08-24 Richard Biener <rguenther@suse.de>
19534 * targhooks.c (default_target_can_inline_p): Properly
19535 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
19536 is present and always compare.
19537 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
19538 infer -mfpmath=sse from TARGET_SSE_P.
19539 (ix86_can_inline_p): Properly use target_option_default_node when
19540 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
19542 2017-08-24 Richard Biener <rguenther@suse.de>
19545 * dwarf2out.c (output_die): Handle flag_generate_offload like
19547 (output_comp_unit): Likewise.
19548 (gen_array_type_die): Likewise.
19549 (dwarf2out_early_finish): Likewise.
19550 (note_variable_value_in_expr): Likewise.
19551 (dwarf2out_finish): Likewise. Adjust assert.
19552 * cgraphunit.c (symbol_table::compile): Move setting of
19553 flag_generate_offload earlier ...
19554 (symbol_table::finalize_compilation_unit): ... here, before
19555 early debug finalization.
19557 2017-08-24 Richard Biener <rguenther@suse.de>
19559 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
19560 and ipa-fnsummary.h.
19561 (ix86_can_inline_p): When ix86_fpmath flags do not match
19562 check whether the callee uses FP math at all.
19564 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
19566 PR middle-end/81931
19567 * tree-ssanames.c (get_nonzero_bits): Use element_precision
19568 instead of TYPE_PRECISION.
19570 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
19571 Alan Hayward <alan.hayward@arm.com>
19572 David Sherwood <david.sherwood@arm.com>
19574 * combine.c (make_extraction): Use subreg_offset_from_lsb.
19576 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
19578 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
19580 (ix86_frame::stack_realign_allocate): New field.
19581 (struct machine_frame_state): Modify comments.
19582 (machine_frame_state::sp_realigned_fp_end): New field.
19583 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
19584 layout calculation.
19585 (sp_valid_at): Add assertion to assure no attempt to access invalid
19586 offset of a realigned stack.
19587 (fp_valid_at): Likewise.
19588 (choose_baseaddr): Modify comments.
19589 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
19590 ix86_expand_prologue.
19591 (ix86_expand_prologue): Modify stack realignment and allocation.
19592 (ix86_expand_epilogue): Modify comments.
19593 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
19594 avx2_runtime, avx512f, and avx512f_runtime.
19596 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
19598 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
19599 (mstackrealign): Do not init to -1.
19600 * config/i386/i386.c (ix86_option_override_internal):
19601 Check opts_set, not opts when setting default value of
19602 opts->x_ix86_force_align_arg_pointer.
19604 2017-08-23 Richard Biener <rguenther@suse.de>
19606 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
19607 lang_hooks.decl_printable_name.
19608 * print-rtl-function.c (print_rtx_function): Likewise.
19609 * tree-pretty-print.c (dump_function_header): Likewise.
19611 2017-08-23 Richard Biener <rguenther@suse.de>
19614 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
19615 -g0 at compile-time.
19617 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19619 PR middle-end/19706
19620 * doc/sourcebuild.texi (Other hardware attributes):
19623 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19625 PR middle-end/19706
19626 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
19627 Fix single-use check.
19629 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
19631 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
19633 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19635 * doc/install.texi: Modify to add more details on running selected
19638 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19640 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
19641 is combined with -mabi=ms.
19642 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
19645 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19647 PR tree-optimization/81488
19648 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
19649 and cached_basis fields.
19650 (MAX_SPREAD): New constant.
19651 (alloc_cand_and_find_basis): Initialize new fields.
19652 (clear_visited): New function.
19653 (create_phi_basis_1): Rename from create_phi_basis, set visited
19654 and cached_basis fields.
19655 (create_phi_basis): New wrapper function.
19656 (phi_add_costs_1): Rename from phi_add_costs, add spread
19657 parameter, set visited field, short-circuit when limits reached.
19658 (phi_add_costs): New wrapper function.
19659 (record_phi_increments_1): Rename from record_phi_increments, set
19661 (record_phi_increments): New wrapper function.
19662 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
19663 (phi_incr_cost): New wrapper function.
19664 (all_phi_incrs_profitable_1): Rename from
19665 all_phi_incrs_profitable, set visited field.
19666 (all_phi_incrs_profitable): New wrapper function.
19668 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19669 Alan Hayward <alan.hayward@arm.com>
19670 David Sherwood <david.sherwood@arm.com>
19672 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
19673 that takes the outer and inner modes.
19674 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
19675 comparison as the canonical test for a paradoxical subreg.
19676 * combine.c (simplify_set): Use paradoxical_subreg_p.
19677 (make_extraction): Likewise.
19678 (force_to_mode): Likewise.
19679 (rtx_equal_for_field_assignment_p): Likewise.
19680 (gen_lowpart_for_combine): Likewise.
19681 (simplify_comparison): Likewise.
19682 * cse.c (equiv_constant): Likewise.
19683 * expmed.c (store_bit_field_1): Likewise.
19684 * final.c (alter_subreg): Likewise.
19685 * fwprop.c (propagate_rtx): Likewise.
19686 (forward_propagate_subreg): Likewise.
19687 * ira-conflicts.c (ira_build_conflicts): Likewise.
19688 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
19689 * lra-constraints.c (curr_insn_transform): Likewise.
19690 (split_reg): Likewise.
19691 * lra-eliminations.c (move_plus_up): Likewise.
19692 (lra_eliminate_regs_1): Likewise.
19693 * recog.c (general_operand): Likewise.
19694 * ree.c (combine_reaching_defs): Likewise.
19695 * reload.c (push_reload): Likewise.
19696 (find_reloads): Likewise.
19697 * reload1.c (elimination_effects): Likewise.
19698 (compute_reload_subreg_offset): Likewise.
19699 (choose_reload_regs): Likewise.
19700 * rtlanal.c (subreg_lsb_1): Likewise.
19701 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19702 (simplify_subreg): Likewise.
19703 * var-tracking.c (track_loc_p): Likewise.
19704 * emit-rtl.c (byte_lowpart_offset): Likewise.
19705 (paradoxical_subreg_p): Delete out-of-line definition.
19707 2017-08-22 Jeff Law <law@redhat.com>
19709 PR tree-optimization/81741
19710 PR tree-optimization/71947
19711 * tree-ssa-dom.c: Include tree-inline.h.
19712 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
19713 equivalences if one is more expensive to compute than the other.
19714 * tree-ssa-scopedtables.h (class const_or_copies): Make
19715 record_const_or_copy_raw method private.
19716 (class avail_exprs_stack): New method simplify_binary_operation.
19717 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
19718 avail_exprs_stack::simplify_binary_operation as needed.
19719 (avail_exprs_stack::simplify_binary_operation): New function.
19721 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
19723 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
19724 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
19725 (DOT_SYMBOLS): Likewise.
19726 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
19727 (RELOCATABLE_NEEDS_FIXUP): Likewise.
19728 (RS6000_ABI_NAME): Likewise.
19729 (TARGET_CMODEL): Likewise.
19730 (TOC_SECTION_ASM_OP): Likewise.
19731 (SET_CMODEL): New macro.
19732 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
19734 2017-08-22 Richard Biener <rguenther@suse.de>
19736 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
19737 to allow for free-lang-data replacements similar to verify_type_variant.
19739 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19740 Alan Hayward <alan.hayward@arm.com>
19741 David Sherwood <david.sherwood@arm.com>
19743 * config/aarch64/aarch64.md (casesi): Use DImode rather than
19744 VOIDmode for the LABEL_REF.
19746 2017-08-22 Richard Biener <rguenther@suse.de>
19748 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
19750 2017-08-22 Richard Biener <rguenther@suse.de>
19752 * common.opt (feliminate-dwarf2-dups): Ignore.
19753 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
19754 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
19755 same_die_p_wrap, compute_section_prefix,
19756 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
19757 (comdat_symbol_id, comdat_symbol_number): Likewise.
19758 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
19760 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
19761 (output_die): Mark unreachable path unreachable.
19762 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
19763 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
19764 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
19765 (dwarf2out_early_finish): Likewise.
19767 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
19769 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
19771 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
19774 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
19775 not VAR_P. Filter attribute warnings with OPT_Wattributes.
19776 (avr_attribute_table) <io, io_low, address>: Initialize
19777 .decl_required with true.
19779 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
19781 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
19782 undocumented debugging options.
19783 (-mvsx-scalar-double): Likewise.
19784 (-mallow-df-permute): Likewise.
19785 (-mvectorize-builtins): Likewise.
19786 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19787 (rs6000_builtin_vectorized_function): Likewise.
19788 (rs6000_builtin_md_vectorized_function): Likewise.
19789 (rs6000_opt_vars): Likewise.
19791 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
19794 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
19795 (*btrq_imm): Rename from *btrq.
19796 (*btcq_imm): Rename from *btcq.
19797 (btsc): New code attribute.
19798 (*<btsc><mode>): New insn pattern.
19799 (*btr<mode>): Ditto.
19800 (*<btsc><mode>_mask): New insn_and_split pattern.
19801 (*btr<mode>_mask): Ditto.
19803 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19804 Alan Hayward <alan.hayward@arm.com>
19805 David Sherwood <david.sherwood@arm.com>
19807 * function.c (pad_below): Simplify padding calculation.
19809 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19810 Alan Hayward <alan.hayward@arm.com>
19811 David Sherwood <david.sherwood@arm.com>
19813 * target.def (function_prologue): Remove frame size argument.
19814 (function_epilogue): Likewise.
19815 * doc/tm.texi: Regenerate.
19816 * final.c (final_start_function): Update call to function_prologue.
19817 (final_end_function): Update call to function_epilogue.
19818 (default_function_pro_epilogue): Remove frame size argument.
19819 * output.h (default_function_pro_epilogue): Likewise.
19820 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
19821 (arm_output_function_prologue): Likewise.
19822 * config/frv/frv.c (frv_function_prologue): Likewise.
19823 (frv_function_epilogue): Likewise.
19824 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
19825 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
19826 (ia64_output_function_epilogue): Likewise.
19827 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
19828 (m32r_output_function_epilogue): Likewise.
19829 * config/microblaze/microblaze.c (microblaze_function_prologue)
19830 (microblaze_function_epilogue): Likewise.
19831 * config/mips/mips.c (mips_output_function_prologue): Likewise.
19832 (mips_output_function_epilogue): Likewise.
19833 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
19834 (mmix_target_asm_function_epilogue): Likewise.
19835 * config/msp430/msp430.c (msp430_start_function): Likewise.
19836 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
19837 (nds32_asm_function_epilogue): Likewise.
19838 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
19839 * config/pa/pa.c (pa_output_function_prologue): Likewise.
19840 (pa_output_function_epilogue): Likewise.
19841 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
19842 (rs6000_output_function_epilogue): Likewise.
19843 * config/rl78/rl78.c (rl78_start_function): Likewise.
19844 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
19845 (rs6000_output_function_epilogue): Likewise.
19846 * config/rx/rx.c (rx_output_function_prologue): Likewise.
19847 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
19848 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
19849 (sparc_asm_function_epilogue): Likewise.
19851 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19853 * tree.h (type_has_mode_precision_p): New function.
19854 * convert.c (convert_to_integer_1): Use it.
19855 * expr.c (expand_expr_real_2): Likewise.
19856 (expand_expr_real_1): Likewise.
19857 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
19858 * match.pd: Likewise.
19859 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
19860 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
19861 * tree-tailcall.c (process_assignment): Likewise.
19862 * tree-vect-loop.c (vectorizable_reduction): Likewise.
19863 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
19864 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
19865 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19866 (vectorizable_assignment): Likewise.
19867 (vectorizable_shift): Likewise.
19868 (vectorizable_operation): Likewise.
19869 * tree-vrp.c (register_edge_assert_for_2): Likewise.
19871 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
19873 * match.pd: Add pow (C, x) simplification.
19875 2017-08-21 Richard Biener <rguenther@suse.de>
19877 PR tree-optimization/81900
19878 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
19879 for blocks with abnormal predecessors.
19880 (compute_antic): Do not set visited flag prematurely.
19882 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
19885 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
19887 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19889 * stor-layout.h (vector_type_mode): Move to...
19890 * tree.h (vector_type_mode): ...here.
19891 * stor-layout.c (vector_type_mode): Move to...
19892 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
19894 2017-08-21 Richard Biener <rguenther@suse.de>
19896 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
19897 register_external_die hooks.
19898 (debug_false_tree_charstarstar_uhwistar): Declare.
19899 (debug_nothing_tree_charstar_uhwi): Likewise.
19900 * debug.c (do_nothing_debug_hooks): Adjust.
19901 (debug_false_tree_charstarstar_uhwistar): New do nothing.
19902 (debug_nothing_tree_charstar_uhwi): Likewise.
19903 * dbxout.c (dbx_debug_hooks): Adjust.
19904 (xcoff_debug_hooks): Likewise.
19905 * sdbout.c (sdb_debug_hooks): Likewise.
19906 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19907 * dwarf2out.c (macinfo_label_base): New global.
19908 (dwarf2out_register_external_die): New function for the
19909 register_external_die hook.
19910 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
19911 (dwarf2_debug_hooks): Use them.
19912 (dwarf2_lineno_debug_hooks): Adjust.
19913 (struct die_struct): Add with_offset flag.
19914 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
19915 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
19916 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
19917 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
19918 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
19919 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
19920 defining section names for the early LTO debug variants.
19921 (reset_indirect_string): New helper.
19922 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
19923 (print_dw_val): Add support for offsetted symbol references.
19924 (get_ultimate_context): Split out from is_cxx.
19925 (is_cxx): Use get_ultimate_context.
19926 (is_fortran): Add decl overload.
19927 (compute_comp_unit_symbol): Split out worker from
19928 compute_section_prefix.
19929 (compute_section_prefix): Call compute_comp_unit_symbol and
19930 set comdat_type_p here.
19931 (output_die): Skip DIE symbol output for the LTO added one.
19932 Handle DIE symbol references with offset.
19933 (output_comp_unit): Guard section name mangling properly.
19934 For LTO debug sections emit a symbol at the section beginning
19935 which we use to refer to its DIEs.
19936 (add_abstract_origin_attribute): For DIEs registered via
19937 dwarf2out_register_external_die directly refer to the early
19938 DIE rather than indirectly through the shadow one we created.
19939 Remove obsolete call to dwarf2out_abstract_function for
19940 non-function/block origins.
19941 (gen_array_type_die): When generating early LTO debug do
19942 not emit DW_AT_string_length.
19943 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
19944 late when in LTO. As suggested place a gcc_unreachable for
19945 the DECL_ABSTRACT_P case.
19946 (gen_subprogram_die): Avoid another specification DIE
19947 for early built declarations/definitions for the late LTO case.
19948 (gen_variable_die): Add type references for late duplicated VLA dies
19950 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
19951 we have the abstract instance already.
19952 (process_scope_var): Adjust decl DIE contexts in LTO which
19953 first puts them in limbo.
19954 (gen_decl_die): Do not generate type DIEs late apart from
19955 types for VLAs or for decls we do not yet have a DIE. Do not
19956 call dwarf2out_abstract_function late.
19957 (dwarf2out_early_global_decl): Make sure to create DIEs
19958 for abstract instances of a decl first.
19959 (dwarf2out_late_global_decl): Adjust comment.
19960 (output_macinfo_op): With multiple macro sections use
19961 macinfo_label_base to distinguish labels.
19962 (output_macinfo): Likewise. Update macinfo_label_base.
19963 Pass in the line info label.
19964 (note_variable_value_in_expr): When generating LTO resolve
19965 all variable values here by generating DIEs as needed.
19966 (init_sections_and_labels): Add early LTO debug flag parameter
19967 and generate different sections and names if set. Add generation
19968 counter for the labels so we can have multiple of them.
19969 (reset_dies): Helper to allow DIEs to be output multiple times.
19970 (dwarf2out_finish): When outputting DIEs to the fat part of an
19971 LTO object first reset DIEs.
19972 (dwarf2out_early_finish): Output early DIEs when generating LTO.
19973 (modified_type_die): Check for decl_ultimate_origin being self
19975 (gen_type_die_with_usage): Likewise.
19976 (gen_typedef_die): Allow decl_ultimate_origin being self.
19977 (set_decl_abstract_flags): Remove.
19978 (set_block_abstract_flags): Likewise.
19979 (dwarf2out_abstract_function): Treat the early generated DIEs
19980 as the abstract copy and only add DW_AT_inline and
19981 DW_AT_artificial here and call set_decl_origin_self.
19982 If the DIE has an abstract origin don't do anything.
19983 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
19984 if we have none yet (Go fails to build one, PR78628).
19985 (variably_modified_type_p): Prevent endless recursion for Ada
19986 cyclic pointer types.
19987 * lto-streamer-in.c: Include debug.h.
19988 (dref_queue): New global.
19989 (lto_read_tree_1): Stream in DIE references.
19990 (lto_input_tree): Register DIE references.
19991 (input_function): Stream DECL_DEBUG_ARGS.
19992 * lto-streamer-out.c: Include debug.h.
19993 (lto_write_tree_1): Output DIE references.
19994 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
19995 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
19996 (output_function): Stream DECL_DEBUG_ARGS.
19997 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
19998 Stream DECL_ABSTRACT_ORIGIN.
19999 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
20000 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
20001 DECL_CONTEXT for file-scope decls.
20002 * lto-streamer.h (struct dref_entry): Declare.
20003 (dref_queue): Likewise.
20004 * cfgexpand.c (pass_expand::execute): Do not call the
20005 outlining_inline_function hook here.
20006 * lto-wrapper.c (debug_obj): New global.
20007 (tool_cleanup): Unlink it if required.
20008 (debug_objcopy): New function.
20009 (run_gcc): Handle early debug sections in the IL files by
20010 extracting them to separate files, partially linkin them and
20011 feeding the result back as result to the linker.
20012 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
20013 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
20014 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
20015 sections into a separate segment.
20016 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
20018 (darwin_asm_dwarf_section): Likewise.
20019 (darwin_asm_output_dwarf_offset): Likewise.
20020 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
20022 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20023 Alan Hayward <alan.hayward@arm.com>
20024 David Sherwood <david.sherwood@arm.com>
20026 * read-md.h (md_reader::record_potential_iterator_use): Replace
20027 pointer argument with an rtx and an index.
20028 * read-rtl.c (iterator_group::apply_iterator): Likewise.
20029 (apply_mode_iterator): Likewise.
20030 (apply_code_iterator): Likewise.
20031 (apply_int_iterator): Likewise.
20032 (apply_subst_iterator): Likewise.
20033 (record_iterator_use): Likewise.
20034 (record_attribute_use): Likewise.
20035 (md_reader::record_potential_iterator_use): Likewise. Update calls
20036 to record_iterator_use and apply_iterator.
20037 (iterator_use): Replace ptr with x and index.
20038 (attribute_use): Likewise.
20039 (apply_attribute_uses): Update calls to apply_iterator.
20040 (apply_iterators): Likewise. Update initialization of iterator_use.
20041 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
20042 and record_potential_iterator_use.
20043 (rtx_reader::read_rtx_operand): Likewise.
20045 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20046 Alan Hayward <alan.hayward@arm.com>
20047 David Sherwood <david.sherwood@arm.com>
20049 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
20052 2017-08-21 Richard Biener <rguenther@suse.de>
20054 PR middle-end/81884
20055 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
20056 at struct end conservatively when comparing common bases.
20058 2017-08-21 Richard Biener <rguenther@suse.de>
20060 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
20061 (mem_ref_in_stmt): Remove.
20062 (determine_max_movement): Use ref index to get at the reference.
20063 (invariantness_dom_walker::before_dom_children): Deal with
20064 lim data already initialized.
20065 (gather_mem_refs_stmt): Initialize lim data and record ref index.
20067 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20069 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
20070 (TARGET_ISA_ROUND): Ditto.
20071 (TARGET_ROUND): Ditto.
20072 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
20073 * config/i386/i386.md: Ditto.
20074 * config/i386/sse.md: Ditto.
20075 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
20076 with OPTION_MASK_ISA_SSE4_1.
20078 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20081 * doc/extend.texi (x86 Built-in Functions): Correct the name of
20082 __builtin_ia32_lzcnt_u16.
20084 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
20087 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
20088 (rs6000_set_current_function): Rewrite function to use it.
20090 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
20093 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
20094 and TYPE_WARN_IF_NOT_ALIGN.
20095 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
20096 (handle_warn_if_not_align): New.
20097 (place_union_field): Call handle_warn_if_not_align.
20098 (place_field): Call handle_warn_if_not_align.
20099 Copy TYPE_WARN_IF_NOT_ALIGN.
20100 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
20101 (layout_type): Likewise.
20102 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
20104 (tree_decl_common): Add warn_if_not_align.
20105 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
20106 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
20107 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
20108 (DECL_WARN_IF_NOT_ALIGN): Likewise.
20109 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
20110 * doc/extend.texi: Document warn_if_not_aligned attribute.
20111 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
20113 2017-08-17 Martin Liska <mliska@suse.cz>
20116 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
20117 (get_data_dependence): Use it as pointer type.
20118 (distribute_loop): Likewise.
20120 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20122 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
20123 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
20124 (p8_vmrgow_<mode>_direct): New define_insn.
20125 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
20126 handle endianness for vmrgew and vmrgow permute patterns.
20128 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20130 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
20131 * config/rs6000/rs6000-cpus.def: Remove comment.
20132 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
20133 (POWERPC_MASKS): Likewise.
20134 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
20135 use of TARGET_VSX_TIMODE.
20136 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
20137 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
20138 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
20139 (rs6000_option_override_internal): Remove dead code.
20140 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
20141 (rs6000_legitimize_reload_address): Likewise.
20142 (rs6000_legitimate_address_p): Likewise.
20143 (rs6000_opt_masks): Delete "vsx-timode".
20144 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
20145 from function comment.
20146 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
20147 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
20148 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
20150 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
20151 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
20152 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
20153 (VSX_TI): Likewise.
20155 (define_peephole2): Likewise.
20157 2017-08-17 Martin Sebor <msebor@redhat.com>
20160 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
20161 past the end of an array.
20162 (test_pp_format): Add test cases.
20164 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
20166 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
20167 missing ECF_NOTHROW flags.
20169 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20172 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
20173 operands residing in integer registers.
20174 (*vsx_le_perm_load_<mode>): Likewise.
20175 (*vsx_le_perm_store_<mode>): Likewise.
20176 (define_peephole2): Add peepholes to optimize the above.
20178 2017-08-17 Marek Polacek <polacek@redhat.com>
20180 PR middle-end/81814
20181 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
20182 to mimic what shorten_compare did. Change the return type to bool.
20183 (fold_cond_expr_with_comparison): Update call to
20184 operand_equal_for_comparison_p.
20185 (fold_ternary_loc): Likewise.
20187 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
20189 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
20191 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
20193 2017-08-17 Richard Biener <rguenther@suse.de>
20195 * tree-ssa-structalias.c (solve_graph): When propagating
20196 to successors update the graphs succ edges and avoid duplicate work.
20198 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
20201 * config/i386/i386.c (ix86_option_override_internal): Save target
20202 specific options after ix86_stack_protector_guard_reg was changed.
20204 2017-08-17 Richard Biener <rguenther@suse.de>
20206 PR tree-optimization/81827
20207 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
20208 (new_var_info): Initialize it conservatively.
20209 (get_call_vi): Mark register vars.
20210 (new_scalar_tmp_constraint_exp): Likewise.
20211 (handle_rhs_call): Likewise.
20212 (handle_const_call): Likewise.
20213 (create_function_info_for): Likewise.
20214 (solve_constraints): Sort varinfos to separate register from
20215 non-register vars to pack points-to solution bitmaps during
20218 2017-08-17 Marek Polacek <polacek@redhat.com>
20220 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
20222 2017-08-17 Richard Biener <rguenther@suse.de>
20224 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
20225 to true when overflow is undefined and we saturated the result.
20227 2017-08-17 Alan Modra <amodra@gmail.com>
20230 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
20231 Don't use store multiple if only one reg needs saving.
20232 (interesting_frame_related_regno): New function.
20233 (rs6000_frame_related): Don't emit frame info for regs that
20235 (rs6000_emit_epilogue): Likewise.
20237 2017-08-16 Nathan Sidwell <nathan@acm.org>
20239 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
20240 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
20241 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
20242 * tree.c (free_lang_data_in_type): Use else-if chain. Always
20243 clear TYPE_LANG_1. Remove obsolete member-function stripping.
20244 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
20245 (verify_type): Adjust for TYPE_BINFO move.
20246 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
20247 process TYPE_BINFO directly.
20248 (hash_tree): Likewise.
20249 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20251 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20254 2017-08-16 David Malcolm <dmalcolm@redhat.com>
20256 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
20258 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
20261 * config/i386/i386.md (*anddi_1_btr): Change predicates of
20262 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
20263 Add ix86_binary_operator_ok to insn constraint.
20264 (*iordi_1_bts): Ditto.
20265 (*xordi_1_btc): Ditto.
20266 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
20267 Update corresponding peephole2 pattern.
20271 2017-08-16 Bin Cheng <bin.cheng@arm.com>
20273 PR tree-optimization/81832
20274 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
20275 copy loop header which has IFN_LOOP_DIST_ALIAS call.
20277 2017-08-16 Marek Polacek <polacek@redhat.com>
20280 * fold-const.c (fold_indirect_ref_1): Restore original behavior
20281 regarding size_zero_node.
20283 2017-08-16 Martin Liska <mliska@suse.cz>
20286 * config.gcc: Respect previously set extra_objs in case
20289 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
20291 PR tree-optimization/81835
20292 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
20293 the phi SSA_NAME. Check that the condition in a COND_EXPR does
20294 not depend on the phi.
20296 2017-08-16 Alan Modra <amodra@gmail.com>
20298 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
20301 2017-08-16 Alan Modra <amodra@gmail.com>
20303 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
20304 (save_reg_p): ..into this. Update all callers.
20305 (first_reg_to_save): Simplify.
20307 2017-08-16 Alan Modra <amodra@gmail.com>
20309 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
20312 2017-08-15 Joseph Myers <joseph@codesourcery.com>
20316 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
20317 constant count if that count is less than 32.
20319 2017-08-15 Nathan Sidwell <nathan@acm.org>
20321 * gcc.c (execute): Emit friendlier message if inferior is killed
20322 by an external cause.
20324 2017-08-15 Richard Biener <rguenther@suse.de>
20326 PR tree-optimization/81790
20327 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
20328 CONSTRUCTORs from simplifying and VN.
20330 2017-08-14 Martin Sebor <msebor@redhat.com>
20332 * builtin-attrs.def: Add comments.
20334 2017-08-14 Martin Sebor <msebor@redhat.com>
20337 * doc/extend.texi (attribute nonstring): Document new attribute.
20339 2017-08-14 Martin Sebor <msebor@redhat.com>
20342 * tree-diagnostic.c (default_tree_printer): Handle %G.
20343 * gimple-pretty-print.h (percent_G_format): Declare new function.
20344 * gimple-pretty-print.c (percent_G_format): Define.
20345 * tree-pretty-print.c (percent_K_format): Add argument.
20347 2017-08-14 Martin Sebor <msebor@redhat.com>
20349 PR translation/79998
20350 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
20351 Remove a stray space.
20353 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
20356 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
20357 (*iordi_1_bts): Ditto.
20358 (*xordi_1_btc): Ditto.
20360 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20363 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
20364 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20366 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
20367 quoted strings, and make more translator-friendly.
20368 (darwin_rs6000_override_options): Likewise.
20369 (rs6000_option_override_internal): Likewise.
20370 (rs6000_return_in_memory): Fix overlong line.
20371 (init_cmulative_args): Use quoted strings, and make more
20372 translator-friendly.
20373 (rs6000_pass_by_reference): Fix overlong line.
20374 (def_builtin): Use quoted strings.
20375 (altivec_expand_predicate_builtin): Use quoted strings, and make
20376 more translator-friendly.
20377 (htm_expand_builtin): Use quoted strings.
20378 (cpu_expand_builtin): Use quoted strings, and make more
20379 translator-friendly.
20380 (altivec_expand_builtin): Likewise.
20381 (paired_expand_predicate_builtin): Likewise.
20382 (rs6000_invalid_builtin): Likewise.
20383 (builtin_function_type): Use quoted strings.
20384 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
20385 more translator-friendly.
20386 (rs6000_trampoline_init): Likewise.
20387 (rs6000_handle_altivec_attribute): Likewise.
20388 (rs6000_inner_target_options): Use quoted strings.
20389 (rs6000_disable_incompatible_switches): Likewise.
20390 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
20391 strings, and make more translator-friendly.
20392 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
20394 2017-08-14 Bin Cheng <bin.cheng@arm.com>
20396 PR tree-optimization/81799
20397 * tree-loop-distribution.c (version_loop_by_alias_check): Force
20398 cond_expr to simple gimple operand.
20400 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
20402 PR middle-end/46932
20403 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
20405 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
20409 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
20410 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
20411 TARGET_GASISR_PROLOGUES.
20412 * config/avr/avr.c (avr_option_override): Same.
20413 (avr_pass_pre_proep::execute): Same.
20415 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
20418 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
20419 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
20421 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
20423 * config/i386/i386.md (*load_tp_<mode>): Redefine as
20424 define_insn_and_split. Split to a memory load from 0 in
20425 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
20426 using PTR mode iterator.
20427 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
20428 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
20429 (*add_tp_<mode>): Redefine as define_insn_and_split.
20430 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
20431 address space. Merge with *add_tp_x32 using PTR mode iterator.
20432 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
20433 Split to an add with a memory load from 0 in
20434 DEFAULT_TLS_SEG_REG address space.
20436 2017-08-12 Andrew Pinski <apinski@cavium.com>
20438 * config/aarch64/aarch64-option-extensions.def (rdma):
20439 Fix feature string to what Linux prints out in /proc/cpuinfo.
20441 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
20444 * dwarf2out.c (modified_type_die): For C typedef types that have
20445 an ultimate origin, process the ultimate origin instead of the
20447 (gen_typedef_die): Assert that input DECLs have no ultimate
20449 (gen_type_die_with_usage): For typedef variants that have an
20450 ultimate origin, just call gen_decl_die on the original DECL.
20451 (process_scope_var): Avoid creating DIEs for local typedefs and
20452 concrete static variables.
20454 2017-08-12 Alan Modra <amodra@gmail.com>
20458 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
20459 match gnu-user.h startfile.
20460 (ENDFILE_LINUX_SPEC): Similarly.
20462 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
20465 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
20467 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
20469 2017-08-11 Tamar Christina <tamar.christina@arm.com>
20470 * config/aarch64/aarch64.md (mov<mode>): Change.
20471 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20472 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
20473 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
20475 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
20477 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
20478 for storage order barriers.
20480 2017-08-11 Martin Liska <mliska@suse.cz>
20483 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
20484 variables of void type.
20486 2017-08-11 Martin Liska <mliska@suse.cz>
20488 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
20489 TARGET_SUPPORTS_ALIASES.
20490 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
20491 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
20492 (optimize_weakref): Likewise.
20493 * symtab.c (symtab_node::noninterposable_alias): Likewise.
20494 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
20495 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
20497 2017-08-11 Martin Liska <mliska@suse.cz>
20500 * config/i386/i386.c (make_resolver_func): Do complete
20501 refactoring of the function.
20503 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20506 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
20507 * config/i386/i386.c (ix86_stack_protect_guard): Use
20508 ix86_stack_protect_guard_symbol_str to generate varible declaration.
20509 * doc/invoke.texi (x86 Options): Document
20510 -mstack-protector-guard-symbol= option.
20512 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20514 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
20515 * config/i386/i386.c (ix86_split_stack_guard): New function.
20516 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
20517 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
20518 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
20519 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
20520 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
20521 (split_stack_space_check): Call ix86_split_stack_guard.
20523 2017-08-10 Martin Sebor <msebor@redhat.com>
20525 * print-tree.c (print_node): Print location using the established
20527 Replace spaces with colons.
20528 (debug_raw, debug): Ditto.
20530 2017-08-10 Martin Sebor <msebor@redhat.com>
20533 * pretty-print.c (pp_format): Correct the handling of %s precision.
20535 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
20538 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20540 (ix86_finalize_stack_frame_flags): This. Also clear
20541 frame_pointer_needed if -fno-omit-frame-pointer is used without
20543 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20544 with ix86_finalize_stack_frame_flags.
20545 (ix86_expand_epilogue): Likewise.
20546 (ix86_expand_split_stack_prologue): Likewise.
20547 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
20549 2017-08-10 Martin Liska <mliska@suse.cz>
20552 * c-attribs.c (handle_target_attribute):
20553 Report warning for an empty string argument of target attribute.
20555 2017-08-09 Jakub Jelinek <jakub@redhat.com>
20558 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
20560 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
20561 or DECL_NONLOCAL labels.
20562 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
20563 or DECL_NONLOCAL labels here.
20565 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
20567 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
20568 to indicate when early gimple folding has been disabled.
20569 (rs6000_gimple_fold_builtin): Add debug content.
20570 (rs6000_invalid_builtin): Fix whitespace.
20571 (rs6000_expand_builtin): Fix whitespace.
20572 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
20574 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
20577 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
20578 SAVE_MULTIPLE if not all the registers that saves, should be saved.
20580 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
20582 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
20583 (qdf24xx): Likewise.
20584 * config/aarch64/aarch64.md: Include falkor.md.
20585 * config/aarch64/falkor.md: New.
20587 2017-08-09 Marek Polacek <polacek@redhat.com>
20590 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
20591 * diagnostic.c (emit_diagnostic): Add a comment.
20592 (emit_diagnostic_valist): New function.
20594 2017-08-09 Marek Polacek <polacek@redhat.com>
20597 * input.c (make_location): New overload.
20598 * input.h (make_location): Declare.
20600 2017-08-08 Alan Modra <amodra@gmail.com>
20601 H.J. Lu <hongjiu.lu@intel.com>
20604 * gcc.c (NO_PIE_SPEC): Delete.
20605 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
20607 (LINK_PIE_SPEC): ..to here.
20608 (LINK_COMMAND_SPEC): Support -no-pie.
20609 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
20610 chain of crtbegin*.o selection, update for PIE_SPEC changes and
20612 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
20613 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
20614 (ENDFILE_CRTEND_SPEC): Similarly.
20616 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
20619 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
20620 (mstack-protector-guard-offset=): Ditto.
20621 * config/i386/i386.c (ix86_option_override): Handle
20622 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
20624 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
20625 ix86_stack_protect_guard_offset variables.
20626 (TARGET_STACK_PROTECT_GUARD): Always define.
20627 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
20628 and -mstack-protector-guard-offset= options.
20630 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20632 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
20633 boundary case for the last candidate.
20635 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20637 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
20638 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
20640 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20642 PR middle-end/19706
20643 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
20644 * config/aarch64/aarch64-builtins.c
20645 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
20646 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
20647 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
20649 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20650 Andrew Pinski <pinskia@gmail.com>
20652 PR middle-end/19706
20653 * internal-fn.def (XORSIGN): New.
20654 * optabs.def (xorsign_optab): New.
20655 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
20656 (convert_expand_mult_copysign): New.
20657 (pass_optimize_widening_mul::execute): Call
20658 convert_expand_mult_copysign.
20660 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20662 PR tree-optimization/81354
20663 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
20664 Insert on edges rather than explicitly creating landing pads.
20665 (analyze_candidates_and_replace): Commit edge inserts.
20667 2017-08-08 Richard Biener <rguenther@suse.de>
20669 PR middle-end/81719
20670 * tree-ssa-loop-niter.c: Include tree-dfa.h.
20671 (expand_simple_operations): Also look through ADDR_EXPRs with
20672 MEM_REF bases treating them as POINTER_PLUS_EXPR.
20674 2017-08-08 Richard Biener <rguenther@suse.de>
20676 PR tree-optimization/81723
20677 * tree-vect-slp.c (struct bst_traits): New hash traits.
20678 (bst_fail): New global.
20679 (vect_build_slp_tree_2): New worker, split out from ...
20680 (vect_build_slp_tree): ... this now wrapping it with using
20681 bst_fail set to cache SLP tree build fails. Properly handle
20683 (vect_analyze_slp_instance): Allocate and free bst_fail.
20685 2017-08-08 Martin Liska <mliska@suse.cz>
20688 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
20689 LABEL_DECLs that can be from a different function.
20691 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20693 PR tree-optimization/81744
20694 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
20695 loop's number of iterations.
20697 2017-08-08 Martin Liska <mliska@suse.cz>
20699 * asan.c: Include header files.
20700 * attribs.c (build_decl_attribute_variant): New function moved
20702 (build_type_attribute_qual_variant): Likewise.
20703 (cmp_attrib_identifiers): Likewise.
20704 (simple_cst_list_equal): Likewise.
20705 (omp_declare_simd_clauses_equal): Likewise.
20706 (attribute_value_equal): Likewise.
20707 (comp_type_attributes): Likewise.
20708 (build_type_attribute_variant): Likewise.
20709 (lookup_ident_attribute): Likewise.
20710 (remove_attribute): Likewise.
20711 (merge_attributes): Likewise.
20712 (merge_type_attributes): Likewise.
20713 (merge_decl_attributes): Likewise.
20714 (merge_dllimport_decl_attributes): Likewise.
20715 (handle_dll_attribute): Likewise.
20716 (attribute_list_equal): Likewise.
20717 (attribute_list_contained): Likewise.
20718 * attribs.h (lookup_attribute): New function moved from tree.[ch].
20719 (lookup_attribute_by_prefix): Likewise.
20720 * bb-reorder.c: Include header files.
20721 * builtins.c: Likewise.
20722 * calls.c: Likewise.
20723 * cfgexpand.c: Likewise.
20724 * cgraph.c: Likewise.
20725 * cgraphunit.c: Likewise.
20726 * convert.c: Likewise.
20727 * dwarf2out.c: Likewise.
20728 * final.c: Likewise.
20729 * fold-const.c: Likewise.
20730 * function.c: Likewise.
20731 * gimple-expr.c: Likewise.
20732 * gimple-fold.c: Likewise.
20733 * gimple-pretty-print.c: Likewise.
20734 * gimple.c: Likewise.
20735 * gimplify.c: Likewise.
20736 * hsa-common.c: Likewise.
20737 * hsa-gen.c: Likewise.
20738 * internal-fn.c: Likewise.
20739 * ipa-chkp.c: Likewise.
20740 * ipa-cp.c: Likewise.
20741 * ipa-devirt.c: Likewise.
20742 * ipa-fnsummary.c: Likewise.
20743 * ipa-inline.c: Likewise.
20744 * ipa-visibility.c: Likewise.
20746 * lto-cgraph.c: Likewise.
20747 * omp-expand.c: Likewise.
20748 * omp-general.c: Likewise.
20749 * omp-low.c: Likewise.
20750 * omp-offload.c: Likewise.
20751 * omp-simd-clone.c: Likewise.
20752 * opts-global.c: Likewise.
20753 * passes.c: Likewise.
20754 * predict.c: Likewise.
20755 * sancov.c: Likewise.
20756 * sanopt.c: Likewise.
20757 * symtab.c: Likewise.
20758 * toplev.c: Likewise.
20759 * trans-mem.c: Likewise.
20760 * tree-chkp.c: Likewise.
20761 * tree-eh.c: Likewise.
20762 * tree-into-ssa.c: Likewise.
20763 * tree-object-size.c: Likewise.
20764 * tree-parloops.c: Likewise.
20765 * tree-profile.c: Likewise.
20766 * tree-ssa-ccp.c: Likewise.
20767 * tree-ssa-live.c: Likewise.
20768 * tree-ssa-loop.c: Likewise.
20769 * tree-ssa-sccvn.c: Likewise.
20770 * tree-ssa-structalias.c: Likewise.
20771 * tree-ssa.c: Likewise.
20772 * tree-streamer-in.c: Likewise.
20773 * tree-vectorizer.c: Likewise.
20774 * tree-vrp.c: Likewise.
20775 * tsan.c: Likewise.
20776 * ubsan.c: Likewise.
20777 * varasm.c: Likewise.
20778 * varpool.c: Likewise.
20779 * tree.c: Remove functions moved to attribs.[ch].
20780 * tree.h: Likewise.
20781 * config/aarch64/aarch64.c: Add attrs.h header file.
20782 * config/alpha/alpha.c: Likewise.
20783 * config/arc/arc.c: Likewise.
20784 * config/arm/arm.c: Likewise.
20785 * config/avr/avr.c: Likewise.
20786 * config/bfin/bfin.c: Likewise.
20787 * config/c6x/c6x.c: Likewise.
20788 * config/cr16/cr16.c: Likewise.
20789 * config/cris/cris.c: Likewise.
20790 * config/darwin.c: Likewise.
20791 * config/epiphany/epiphany.c: Likewise.
20792 * config/fr30/fr30.c: Likewise.
20793 * config/frv/frv.c: Likewise.
20794 * config/ft32/ft32.c: Likewise.
20795 * config/h8300/h8300.c: Likewise.
20796 * config/i386/winnt.c: Likewise.
20797 * config/ia64/ia64.c: Likewise.
20798 * config/iq2000/iq2000.c: Likewise.
20799 * config/lm32/lm32.c: Likewise.
20800 * config/m32c/m32c.c: Likewise.
20801 * config/m32r/m32r.c: Likewise.
20802 * config/m68k/m68k.c: Likewise.
20803 * config/mcore/mcore.c: Likewise.
20804 * config/microblaze/microblaze.c: Likewise.
20805 * config/mips/mips.c: Likewise.
20806 * config/mmix/mmix.c: Likewise.
20807 * config/mn10300/mn10300.c: Likewise.
20808 * config/moxie/moxie.c: Likewise.
20809 * config/msp430/msp430.c: Likewise.
20810 * config/nds32/nds32-isr.c: Likewise.
20811 * config/nds32/nds32.c: Likewise.
20812 * config/nios2/nios2.c: Likewise.
20813 * config/nvptx/nvptx.c: Likewise.
20814 * config/pa/pa.c: Likewise.
20815 * config/pdp11/pdp11.c: Likewise.
20816 * config/powerpcspe/powerpcspe.c: Likewise.
20817 * config/riscv/riscv.c: Likewise.
20818 * config/rl78/rl78.c: Likewise.
20819 * config/rx/rx.c: Likewise.
20820 * config/s390/s390.c: Likewise.
20821 * config/sh/sh.c: Likewise.
20822 * config/sol2.c: Likewise.
20823 * config/sparc/sparc.c: Likewise.
20824 * config/spu/spu.c: Likewise.
20825 * config/stormy16/stormy16.c: Likewise.
20826 * config/tilegx/tilegx.c: Likewise.
20827 * config/tilepro/tilepro.c: Likewise.
20828 * config/v850/v850.c: Likewise.
20829 * config/vax/vax.c: Likewise.
20830 * config/visium/visium.c: Likewise.
20831 * config/xtensa/xtensa.c: Likewise.
20833 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20836 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
20837 constraints since the -mupper-regs-* switches have been
20839 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
20840 into a vector from a double word element that was extracted from
20841 another vector, and eliminate extra XXPERMDI instructions.
20842 (vsx_concat_<mode>_2): Likewise.
20843 (vsx_concat_<mode>_3): Likewise.
20844 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
20845 concat to allow optimizing inserts from previous extracts.
20847 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
20849 * config/i386/i386.c (ix86_stack_protect_guard): Generate
20850 memory reference to a SSP offset in TLS address space.
20851 (ix86_print_operand) <case '@'>: Remove.
20852 (ix86_print_operand_punct_valid_p): Remove '@' code.
20853 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
20854 UNSPEC_SP_TLS_TEST.
20855 (stack_tls_protect_set_<mode>): Remove.
20856 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
20857 (stack_tls_protect_test_<mode>): Remove.
20858 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
20860 2017-08-07 Olivier Hainque <hainque@adacore.com>
20863 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
20865 2017-08-07 Douglas Rupp <rupp@adacore.com>
20867 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
20868 variable was referenced as multidir in command.
20870 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20873 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
20876 2017-08-07 Martin Liska <mliska@suse.cz>
20878 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
20879 * config/rl78/rl78.c: Add include of attribs.h.
20880 * config/sh/sh.c: Likewise.
20881 * config/v850/v850.c: Likewise.
20883 2017-08-07 Tom de Vries <tom@codesourcery.com>
20885 PR middle-end/78266
20886 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
20888 2017-08-07 Martin Liska <mliska@suse.cz>
20890 * config/mips/mips.c: Include attribs.h.
20892 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
20895 * doc/invoke.texi: Document change in behvaior for -Ofast for
20898 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
20900 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
20902 (aarch64_pop_regs): Likewise.
20903 (aarch64_gen_load_pair): Likewise.
20904 (aarch64_save_callee_saves): Likewise.
20905 (aarch64_restore_callee_saves): Likewise.
20907 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20909 * config/i386/i386.c: Revert the last change.
20911 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20914 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20916 (ix86_finalize_stack_frame_flags): This. Also clear
20917 frame_pointer_needed if -fno-omit-frame-pointer is used without
20919 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20920 with ix86_finalize_stack_frame_flags.
20921 (ix86_expand_epilogue): Likewise.
20922 (ix86_expand_split_stack_prologue): Likewise.
20924 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20927 * config/i386/i386.c (get_builtin_code_for_version): Set priority
20928 to P_AES for Westmere.
20930 2017-08-07 Jonathan Yong <10walls@gmail.com>
20932 * config/i386/mingw.opt (fset-stack-executable): Removed.
20933 * config/i386/cygming.opt (fset-stack-executable): Moved
20935 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
20937 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
20939 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
20941 2017-08-07 Marek Polacek <polacek@redhat.com>
20943 PR middle-end/81737
20944 * fold-const.c (fold_indirect_ref_1): Check type_domain.
20946 2017-08-07 Martin Liska <mliska@suse.cz>
20948 * attribs.h (canonicalize_attr_name): New function.
20949 (cmp_attribs): Move from c-format.c and adjusted.
20950 (is_attribute_p): Moved from tree.h.
20951 * tree-inline.c: Add new includes.
20952 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
20953 (private_is_attribute_p): Remove.
20954 (private_lookup_attribute): Likewise.
20955 (private_lookup_attribute_by_prefix): Simplify.
20956 (remove_attribute): Use is_attribute_p.
20957 * tree.h: Remove removed declarations.
20959 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20961 PR middle-end/81698
20962 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
20963 instead of computing it in the function. Formatting fix.
20964 (expand_case): Don't rely on default_edge being the first edge,
20965 clear it if removing it, pass default_edge to
20966 emit_case_dispatch_table.
20967 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
20970 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
20972 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
20973 insn in the function, emit NOP after the insn.
20975 2017-08-06 Tom de Vries <tom@codesourcery.com>
20977 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
20980 2017-08-06 Tom de Vries <tom@codesourcery.com>
20982 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
20985 2017-08-04 Yury Gribov <tetra2005@gmail.com>
20987 PR tree-optimization/57371
20988 * match.pd: New pattern.
20990 2017-08-04 Marek Polacek <polacek@redhat.com>
20992 PR middle-end/81695
20993 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
20994 perform the computation in offset_int.
20996 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20998 PR tree-optimization/81136
20999 * tree-vectorizer.h: Include tree-hash-traits.h.
21000 (vec_base_alignments): New typedef.
21001 (vec_info): Add a base_alignments field.
21002 (vect_record_base_alignments): Declare.
21003 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
21005 (DR_IS_CONDITIONAL_IN_STMT): New macro.
21006 (create_data_ref): Add an is_conditional_in_stmt argument.
21007 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
21008 the is_conditional_in_stmt field.
21009 (data_ref_loc): Add an is_conditional_in_stmt field.
21010 (get_references_in_stmt): Set the is_conditional_in_stmt field.
21011 (find_data_references_in_stmt): Update call to create_data_ref.
21012 (graphite_find_data_references_in_stmt): Likewise.
21013 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
21014 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
21015 (vect_record_base_alignment): New function.
21016 (vect_record_base_alignments): Likewise.
21017 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
21018 for nested statements even if we fail to compute a misalignment.
21019 Use pooled base alignments for unconditional references.
21020 (vect_find_same_alignment_drs): Compare base addresses instead
21022 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
21023 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
21025 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21027 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
21028 Add an explicit name for the enum. Use auto_vec for slp_instances
21029 and grouped_stores.
21030 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
21032 (_bb_vec_info): Add a constructor and destructor.
21033 (vinfo_for_stmt): Return NULL for uids of -1 as well.
21034 (destroy_loop_vec_info): Delete.
21035 (vect_destroy_datarefs): Likewise.
21036 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
21037 (vec_info::vec_info): New function.
21038 (vec_info::~vec_info): Likewise.
21039 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
21040 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
21041 destroy_loop_vec_info.
21042 * tree-vect-loop.c (new_loop_vec_info): Replace with...
21043 (_loop_vec_info::_loop_vec_info): ...this.
21044 (destroy_loop_vec_info): Replace with...
21045 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
21046 the stmt_vec_infos. Leave handling of vec_info information to its
21047 destructor. Remove explicit vector releases.
21048 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
21049 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
21050 * tree-vect-slp.c (new_bb_vec_info): Replace with...
21051 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
21052 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
21053 (destroy_bb_vec_info): Replace with...
21054 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
21055 information to its destructor.
21056 (vect_slp_analyze_bb_1): Use new and delete instead of
21057 new_bb_vec_info and destroy_bb_vec_info.
21058 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
21061 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21063 * tree-data-ref.h (subscript): Add access_fn field.
21064 (data_dependence_relation): Add could_be_independent_p.
21065 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
21066 (same_access_functions): Move to tree-data-ref.c.
21067 * tree-data-ref.c (ref_contains_union_access_p): New function.
21068 (access_fn_component_p): Likewise.
21069 (access_fn_components_comparable_p): Likewise.
21070 (dr_analyze_indices): Add a reference to access_fn_component_p.
21071 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
21073 (constant_access_functions): Likewise.
21074 (add_other_self_distances): Likewise.
21075 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
21076 (initialize_data_dependence_relation): Use XCNEW and remove
21077 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
21078 of access functions that have the same type. Allow the
21079 subsequence to end with different bases in some circumstances.
21080 Record the chosen access functions in SUB_ACCESS_FN.
21081 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
21082 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
21083 (subscript_dependence_tester_1): Likewise dra and drb.
21084 (build_classic_dist_vector): Update calls accordingly.
21085 (subscript_dependence_tester): Likewise.
21086 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
21087 DDR_COULD_BE_INDEPENDENT_P.
21088 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
21089 comp_alias_ddrs instead of may_alias_ddrs.
21090 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
21092 (vect_analyze_data_ref_dependence): Use it if
21093 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
21094 distance vectors if that fails.
21095 (dependence_distance_ge_vf): New function.
21096 (vect_prune_runtime_alias_test_list): Use it. Don't clear
21097 LOOP_VINFO_MAY_ALIAS_DDRS.
21099 2017-08-04 Richard Biener <rguenther@suse.de>
21101 PR middle-end/81705
21102 * fold-const.c (fold_binary_loc): Properly restrict
21103 minus_var0 && minus_var1 case when associating undefined overflow
21106 2017-08-04 Tom de Vries <tom@codesourcery.com>
21108 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
21110 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21112 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21113 Don't start diagnostic messages with a capital letter.
21114 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21116 (rs6000_invalid_builtin): Likewise.
21117 (rs6000_trampoline_init): Likewise.
21119 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21122 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
21123 after setting changeable df flags.
21125 2017-08-03 Richard Biener <rguenther@suse.de>
21127 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
21128 up if the use is in USE - X.
21130 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
21132 * toplev.c (dumpfile.h): New include.
21133 (internal_error_reentered): New static function. Use it...
21134 (internal_error_function): ...here to handle reentered internal_error.
21136 2017-08-03 Richard Biener <rguenther@suse.de>
21138 PR middle-end/81148
21139 * fold-const.c (split_tree): Add minus_var and minus_con
21140 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
21141 here but always use minus_*.
21142 (associate_trees): Assert we never associate with MINUS_EXPR
21143 and NULL first operand. Do not recurse for PLUS_EXPR operands
21144 when associating as MINUS_EXPR either.
21145 (fold_binary_loc): Track minus_var and minus_con.
21147 2017-08-03 Tom de Vries <tom@codesourcery.com>
21150 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
21151 ACCEL_COMPILER, apply finish_options on
21152 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
21154 2017-08-03 Tom de Vries <tom@codesourcery.com>
21157 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
21158 function_entry_patch_area_size > 0.
21160 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21163 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
21164 instead of 10??LU, perform unit multiplication in wide_int,
21165 don't change alloc_object_size_limit if the limit is larger
21168 PR tree-optimization/81655
21169 PR tree-optimization/81588
21170 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
21171 the case when ranges[i].low and high are 1 for unsigned type with
21174 PR middle-end/81052
21175 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
21176 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
21178 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21180 * tree-vrp.h: Add include guard.
21182 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
21185 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
21186 (ud2): New insn pattern.
21187 * config/i386/i386.c (ix86_expand_epilogue):
21188 For naked functions, generate ud2 instead of trap insn.
21190 2017-08-02 Marek Polacek <polacek@redhat.com>
21193 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
21195 2017-08-02 Tom de Vries <tom@codesourcery.com>
21196 Cesar Philippidis <cesar@codesourcery.com>
21198 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
21199 Add missing edge probabilities.
21201 2017-08-02 Tamar Christina <tamar.christina@arm.com>
21203 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
21204 Correct endianness.
21206 2017-08-02 Jakub Jelinek <jakub@redhat.com>
21208 PR middle-end/79499
21209 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
21210 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
21211 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
21212 prologue_seq sequences - if any.
21214 2017-08-02 Richard Biener <rguenther@suse.de>
21216 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
21217 via vectors if supported, integer extracts via punning if supported
21218 or otherwise vector extracts.
21220 2017-08-02 Richard Biener <rguenther@suse.de>
21222 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
21224 (bitmap_insert_into_set): ... this.
21226 2017-08-02 Richard Biener <rguenther@suse.de>
21228 PR tree-optimization/81633
21230 2015-08-17 Alan Hayward <alan.hayward@arm.com>
21232 PR tree-optimization/71752
21233 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
21235 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
21237 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
21238 (machine_function::call_ms2sysv_pad_out): Remove field.
21239 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
21240 (ix86_compute_frame_layout): Likewise.
21242 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
21245 * config/i386/i386.c (ix86_set_func_type): Disallow naked
21246 attribute with interrupt attribute.
21248 2017-08-01 Andrew Pinski <apinski@cavium.com>
21250 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
21251 BIT_INSERT_EXPR's operand 1
21252 to see if the types precision matches.
21254 2017-08-01 Martin Liska <mliska@suse.cz>
21256 PR middle-end/70140
21257 * builtins.c (expand_builtin_memcpy_args): Remove.
21258 (expand_builtin_memcpy): Call newly added function
21259 expand_builtin_memory_copy_args.
21260 (expand_builtin_memcpy_with_bounds): Likewise.
21261 (expand_builtin_mempcpy): Remove last argument.
21262 (expand_builtin_mempcpy_with_bounds): Likewise.
21263 (expand_builtin_memory_copy_args): New function created from
21264 expand_builtin_mempcpy_args with small modifications.
21265 (expand_builtin_mempcpy_args): Remove.
21266 (expand_builtin_stpcpy): Remove unused argument.
21267 (expand_builtin): Likewise.
21268 (expand_builtin_with_bounds): Likewise.
21270 2017-08-01 Martin Liska <mliska@suse.cz>
21273 Make mempcpy more optimal (PR middle-end/70140).
21275 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21278 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
21279 __builtin_vec_cmpne verify both arguments are compatible vectors
21280 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
21281 verify arg1_type is a pointer or array type. For __builtin_vec_st,
21282 move computation of aligned to after checking the argument types.
21286 * config/rs6000/vsx.md (vextract_fp_from_shorth,
21287 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
21290 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
21291 Doug Rupp <rupp@adacore.com>
21292 Olivier Hainque <hainque@adacore.com>
21294 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
21295 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
21297 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
21298 for TARGET_OS_CPP_BUILTIN.
21299 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
21300 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
21302 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
21303 passing required abi options to the assembler for EABI configurations.
21304 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
21305 of .text.hot and .text.unlikely sections for kernel modules when
21306 using ARM style exceptions.
21307 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
21308 options. Add EXTRA_CC1_SPEC.
21309 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
21311 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
21313 (ARM_TARGET2_DWARF_FORMAT): Define.
21314 * config/arm/t-vxworks: Adjust multilib control to removal of the
21315 Diab command line options.
21317 2017-08-01 Martin Liska <mliska@suse.cz>
21319 PR gcov-profile/81561
21320 * gcov.c (unblock): Make unblocking safe as we need to preserve
21321 index correspondence of blocks and block_lists.
21323 2017-08-01 Richard Biener <rguenther@suse.de>
21325 PR tree-optimization/81181
21326 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
21327 (compute_antic): ... end of iteration here.
21329 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
21331 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
21332 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
21333 (ftree-slp-vectorize): Likewise.
21334 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
21335 can no longer be set independent of flag_tree_loop_vectorize.
21336 * omp-general.c (emp_max_vf): Likewise.
21337 * opts.c (enable_fdo_optimizations): Remove references to
21338 flag_tree_vectorize, these are now implicit.
21339 (common_handle_option): Remove handling for OPT_ftree_vectorize,
21340 and leave it for the options machinery.
21342 2017-08-01 Martin Liska <mliska@suse.cz>
21344 PR middle-end/70140
21345 * builtins.c (expand_builtin_memcpy_args): Remove.
21346 (expand_builtin_memcpy): Call newly added function
21347 expand_builtin_memory_copy_args.
21348 (expand_builtin_memcpy_with_bounds): Likewise.
21349 (expand_builtin_mempcpy): Remove last argument.
21350 (expand_builtin_mempcpy_with_bounds): Likewise.
21351 (expand_builtin_memory_copy_args): New function created from
21352 expand_builtin_mempcpy_args with small modifications.
21353 (expand_builtin_mempcpy_args): Remove.
21354 (expand_builtin_stpcpy): Remove unused argument.
21355 (expand_builtin): Likewise.
21356 (expand_builtin_with_bounds): Likewise.
21358 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21361 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
21362 print "ds:" only for immediates in generic address space.
21364 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21367 * config/i386/i386.c (ix86_funciton_naked): New prototype.
21368 (ix86_function_ok_for_sibcall): Return false for naked functions.
21370 2017-08-01 Richard Biener <rguenther@suse.de>
21372 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
21373 (compute_antic): Seed worklist with exit block predecessors.
21374 * cfganal.c (dfs_find_deadend): For a cycle return the source
21375 of the edge closing it.
21377 2017-08-01 Tamar Christina <tamar.christina@arm.com>
21379 * config/aarch64/aarch64.c
21380 (aarch64_can_const_movi_rtx_p): Move 0 check.
21382 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21384 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
21385 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
21387 * match.pd: Ditto in address comparison pattern.
21389 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21391 PR tree-optimization/81627
21392 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
21393 closed ssa form for store-store chain.
21395 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21397 PR tree-optimization/81620
21398 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
21399 for store-store chain.
21401 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21403 PR tree-optimization/81588
21404 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
21405 ranges[i].in_p, invert comparison code ccode. For >/>=,
21406 swap rhs1 and rhs2 and comparison code unconditionally,
21407 for </<= don't do that. Don't swap rhs1/rhs2 again if
21408 ranges[i].in_p, instead invert comparison code ccode if
21409 opcode or oe->rank is BIT_IOR_EXPR.
21412 * optabs.def (vec_extract_optab, vec_init_optab): Change from
21413 a direct optab to conversion optab.
21414 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
21415 with GET_MODE_INNER as last argument instead of optab_handler.
21416 * expmed.c (extract_bit_field_1): Likewise. Use vector from
21417 vector extraction if possible and optab is available.
21418 * expr.c (store_constructor): Use convert_optab_handler instead
21419 of optab_handler. Use vector initialization from smaller
21420 vectors if possible and optab is available.
21421 * tree-vect-stmts.c (vectorizable_load): Likewise.
21422 * doc/md.texi (vec_extract, vec_init): Document that the optabs
21423 now have two modes.
21424 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
21425 of vec_init from half-sized vectors with the same element mode.
21426 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
21427 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
21428 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
21429 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
21430 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
21431 after mode in gen_vec_extract* calls.
21432 (vec_extract<mode>): Renamed to ...
21433 (vec_extract<mode><ssescalarmodelower>): ... this.
21434 (vec_extract<mode><ssehalfvecmodelower>): New expander.
21435 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
21436 element mode after mode in gen_vec_init* calls.
21437 (VEC_INIT_HALF_MODE): New mode iterator.
21438 (vec_init<mode>): Renamed to ...
21439 (vec_init<mode><ssescalarmodelower>): ... this.
21440 (vec_init<mode><ssehalfvecmodelower>): New expander.
21441 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
21442 (vec_extractv2sfsf): ... this.
21443 (vec_initv2sf): Renamed to ...
21444 (vec_initv2sfsf): ... this.
21445 (vec_extractv2si): Renamed to ...
21446 (vec_extractv2sisi): ... this.
21447 (vec_initv2si): Renamed to ...
21448 (vec_initv2sisi): ... this.
21449 (vec_extractv4hi): Renamed to ...
21450 (vec_extractv4hihi): ... this.
21451 (vec_initv4hi): Renamed to ...
21452 (vec_initv4hihi): ... this.
21453 (vec_extractv8qi): Renamed to ...
21454 (vec_extractv8qiqi): ... this.
21455 (vec_initv8qi): Renamed to ...
21456 (vec_initv8qiqi): ... this.
21457 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
21458 (vec_init<mode>): Renamed to ...
21459 (vec_init<mode><VEC_base_l>): ... this.
21460 (vec_extract<mode>): Renamed to ...
21461 (vec_extract<mode><VEC_base_l>): ... this.
21462 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
21463 (vec_initv2sfsf): ... this.
21464 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
21465 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21466 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
21467 element mode after mode in gen_vec_init* calls.
21468 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
21469 (vec_init<mode><Vel>): ... this.
21470 (vec_extract<mode>): Renamed to ...
21471 (vec_extract<mode><Vel>): ... this.
21472 * config/aarch64/iterators.md (Vel): New mode attribute.
21473 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
21474 Add element mode after mode in gen_vec_extract* calls.
21475 * config/s390/vector.md (non_vec_l): New mode attribute.
21476 (vec_extract<mode>): Renamed to ...
21477 (vec_extract<mode><non_vec_l>): ... this.
21478 (vec_init<mode>): Renamed to ...
21479 (vec_init<mode><non_vec_l>): ... this.
21480 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
21481 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
21483 * config/arm/iterators.md (V_elem_l): New mode attribute.
21484 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
21485 (vec_extract<mode><V_elem_l>): ... this.
21486 (vec_extractv2di): Renamed to ...
21487 (vec_extractv2didi): ... this.
21488 (vec_init<mode>): Renamed to ...
21489 (vec_init<mode><V_elem_l>): ... this.
21490 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
21491 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
21492 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
21493 Add element mode after gen_vec_extract* calls.
21494 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
21495 (vec_init<mode><unitmode>): ... this.
21496 (vec_extract<mode>): Renamed to ...
21497 (vec_extract<mode><unitmode>): ... this.
21498 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
21499 (vec_init<mode><unitmode>): ... this.
21500 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
21501 (vec_initv2sfsf): ... this.
21502 (vec_extractv2sf): Renamed to ...
21503 (vec_extractv2sfsf): ... this.
21504 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
21505 Add element mode after gen_vec_extract* calls.
21506 * config/mips/mips.md (unitmode): New mode iterator.
21507 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
21508 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
21509 * config/spu/spu.md (inner_l): New mode attribute.
21510 (vec_init<mode>): Renamed to ...
21511 (vec_init<mode><inner_l>): ... this.
21512 (vec_extract<mode>): Renamed to ...
21513 (vec_extract<mode><inner_l>): ... this.
21514 * config/sparc/sparc.md (veltmode): New mode iterator.
21515 (vec_init<VMALL:mode>): Renamed to ...
21516 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
21517 * config/ia64/vect.md (vec_initv2si): Renamed to ...
21518 (vec_initv2sisi): ... this.
21519 (vec_initv2sf): Renamed to ...
21520 (vec_initv2sfsf): ... this.
21521 (vec_extractv2sf): Renamed to ...
21522 (vec_extractv2sfsf): ... this.
21523 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
21524 (vec_init<mode>): Renamed to ...
21525 (vec_init<mode><VEC_base_l>): ... this.
21526 (vec_extract<mode>): Renamed to ...
21527 (vec_extract<mode><VEC_base_l>): ... this.
21528 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
21529 (vec_initv2sfsf): ... this.
21530 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
21531 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21532 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
21533 gen_vec_init* calls.
21535 2017-08-01 Richard Biener <rguenther@suse.de>
21537 PR tree-optimization/81297
21538 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
21539 TREE_OVERFLOW from INTEGER_CSTs.
21541 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
21543 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
21545 2017-07-31 Carl Love <cel@us.ibm.com>
21547 * config/rs6000/rs6000-c: Add support for built-in functions
21548 vector signed char vec_xl_be (signed long long, signed char *);
21549 vector unsigned char vec_xl_be (signed long long, unsigned char *);
21550 vector signed int vec_xl_be (signed long long, signed int *);
21551 vector unsigned int vec_xl_be (signed long long, unsigned int *);
21552 vector signed long long vec_xl_be (signed long long, signed long long *);
21553 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
21554 vector signed short vec_xl_be (signed long long, signed short *);
21555 vector unsigned short vec_xl_be (signed long long, unsigned short *);
21556 vector double vec_xl_be (signed long long, double *);
21557 vector float vec_xl_be (signed long long, float *);
21558 * config/rs6000/altivec.h (vec_xl_be): Add #define.
21559 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
21560 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
21562 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
21563 (altivec_expand_builtin): Add switch statement to call
21564 altivec_expand_xl_be for each builtin.
21565 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
21566 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
21567 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
21568 __builtin_vsx_le_be_v16qi.
21569 * doc/extend.texi: Update the built-in documentation file for the
21570 new built-in functions.
21572 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21575 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
21577 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21579 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21581 * config.gcc: Add z14.
21582 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
21583 CPU model numbers for z13s and z14.
21584 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
21586 * config/s390/s390-opts.h (enum processor_type): Rename
21587 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21588 * config/s390/s390.c (processor_table): Add field for CPU name to
21589 be passed to Binutils.
21590 (s390_asm_output_machine_for_arch): Use the new field in
21591 processor_table for Binutils.
21592 (s390_expand_builtin): Replace arch12 with z14.
21593 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21594 (s390_get_sched_attrmask): Likewise.
21595 (s390_get_unit_mask): Likewise.
21596 * config/s390/s390.opt: Add z14 to processor_type enum.
21598 2017-07-31 Martin Jambor <mjambor@suse.cz>
21601 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
21602 regardless of optimization level.
21604 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
21605 Martin Liska <mliska@suse.cz>
21607 * predict.def: Remove old comment and adjust probability.
21608 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
21609 PREDICT statements.
21611 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21614 * config/i386/i386.c (ix86_function_naked): New function.
21615 (ix86_can_use_return_insn_p): Return false for naked functions.
21616 (ix86_expand_prologue): Skip prologue for naked functions.
21617 (ix86_expand_epilogue): Skip epilogue for naked functions
21618 and emit trap instruction.
21619 (ix86_warn_func_return): New function.
21620 (ix86_attribute_table): Add "naked" attribute specification.
21621 (TARGET_WARN_FUNC_RETURN): Define.
21622 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
21624 2017-07-31 Martin Liska <mliska@suse.cz>
21626 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
21627 (dump_gimple_bb_header): Always dump BB info.
21628 (pp_cfg_jump): Do not append info about BB when dumping a jump.
21630 2017-07-31 Martin Liska <mliska@suse.cz>
21633 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
21634 also with current_function_decl non-null equality.
21636 2017-07-31 Jakub Jelinek <jakub@redhat.com>
21639 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
21640 change type to the element type, instead add eltype variable and
21641 use it where we are interested in the element type.
21643 PR tree-optimization/81603
21644 * ipa-polymorphic-call.c
21645 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
21646 offset arithmetic in offset_int, bail out if the resulting bit offset
21647 doesn't fit into shwi.
21649 2017-07-31 Martin Liska <mliska@suse.cz>
21651 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
21652 (gimplify_save_expr): Fix comment.
21654 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
21657 * config/i386/i386.c (ix86_function_arg): Update arguments for
21659 (ix86_compute_frame_layout): Set the initial stack offset to
21660 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
21661 INCOMING_FRAME_SP_OFFSET.
21662 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
21663 stack before exception handler returns.
21664 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
21665 the 'ERROR_CODE' for exception handler.
21667 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
21669 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
21670 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
21671 (ASM_OUTPUT_REG_POP): Ditto.
21672 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
21673 instead of asm_fprintf to output pure string.
21675 2017-07-29 Jakub Jelinek <jakub@redhat.com>
21677 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
21678 to imported_module_or_decl hook.
21679 (debug_nothing_tree_tree_tree_bool): Remove.
21680 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
21681 * debug.c (do_nothing_debug_hooks): Use
21682 debug_nothing_tree_tree_tree_bool_bool instead of
21683 debug_nothing_tree_tree_tree_bool.
21684 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21685 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
21686 * sdbout.c (sdb_debug_hooks): Likewise.
21687 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
21688 (gen_namespace_die): Add DW_AT_export_symbols attribute if
21690 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
21691 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
21692 attribute, don't add anything.
21694 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21696 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
21697 (fold_build2_stat_loc): Likewise.
21698 (fold_build3_stat_loc): Likewise.
21699 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
21700 (fold_build1_loc): Remove macro.
21701 (fold_build2_loc): Likewise.
21702 (fold_build3_loc): Likewise.
21704 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21706 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
21707 (gimple_build_debug_bind_source_stat): Likewise.
21708 * gimple.h (gimple_build_debug_bind): Remove macro.
21709 (gimple_build_debug_bind_source): Likewise.
21711 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21713 * bitmap.c (bitmap_alloc): Adjust.
21714 (bitmap_gc_alloc): Likewise.
21715 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
21717 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21719 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
21720 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
21721 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
21722 (bitmap_gc_alloc_stat): Likewise.
21723 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
21725 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21727 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
21728 * rtl.h (shallow_copy_rtx): Remove macro.
21730 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21732 * emit-rtl.c (gen_raw_REG): Adjust.
21733 * gengenrtl.c (gendef): Likewise.
21734 * rtl.c (rtx_alloc_stat): Remove _stat from name.
21735 * rtl.h (rtx_alloc): Remove macro.
21737 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21739 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
21740 (build_tree_list_stat): Likewise.
21741 * tree.h (build_tree_list): Remove macro.
21742 (build_tree_list_vec): Likewise.
21744 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21746 * tree.c (make_vector_stat): Remove _stat from name.
21747 (build_vector_stat): Likewise.
21748 * tree.h (make_vector_stat): Remove macro.
21749 (build_vector_stat): Likewise.
21751 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21753 * tree.h (build_var_debug_value): Remove prototype.
21755 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21757 * tree.c (tree_cons_stat): Remove _stat from name.
21758 * tree.h (tree_cons): Remove macro.
21760 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21762 * tree.c (build_vl_exp_stat): Remove _stat from name.
21763 * tree.h (build_vl_exp): Remove macro.
21765 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21767 * tree.c (build_decl_stat): Remove _stat from name.
21768 * tree.h (build_decl): Remove macro.
21770 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21772 * gimple.c (gimple_build_with_ops_stat): Adjust.
21773 (gimple_alloc_stat): Remove _stat from name.
21774 * gimple.h (gimple_alloc): Remove macro.
21776 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21778 * tree.c (make_tree_vec_stat): Remove _stat from name.
21779 (grow_tree_vec_stat): Likewise.
21780 * tree.h (make_tree_vec_stat): Adjust prototype.
21781 (grow_tree_vec_stat): Likewise.
21782 (make_tree_vec): Remove macro.
21783 (grow_tree_vec): Likewise.
21785 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21787 * fold-const.c (fold_build1_stat_loc): Adjust.
21788 (fold_build2_stat_loc): Likewise.
21789 (fold_build3_stat_loc): Likewise.
21790 * tree.c (build0_stat): Remove _stat from name.
21791 (build1_stat): Likewise.
21792 (build2_stat): Likewise.
21793 (build3_stat): Likewise.
21794 (build4_stat): Likewise.
21795 (build5_stat): Likewise.
21796 * tree.h (build1_loc): Remove macro, and rename _stat function
21798 (build2_loc): Likewise.
21799 (build3_loc): Likewise.
21800 (build4_loc): Likewise.
21801 (build5_loc): Likewise.
21803 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21805 * tree.c (make_int_cst_stat): Remove _stat from name.
21806 * tree.h (make_int_cst_stat): Adjust prototype.
21807 (make_int_cst): Remove macro.
21809 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21811 * tree.c (make_tre_binfo_stat): Remove _stat from name.
21812 * tree.h (make_tree_binfo_stat): Adjust prototype.
21813 (make_tree_binfo): Remove.
21815 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21817 * tree.c (copy_node_stat): Rename to copy_node.
21818 (build_distinct_type_copy): Adjust.
21819 * tree.h (copy_node_stat): Adjust prototype.
21820 (copy_node): Remove macro.
21822 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21824 * tree.c (make_node_stat): rename to make_node.
21825 (build_tree_list_stat): Adjust.
21826 (build0_stat): Likewise.
21827 (build2_stat): Likewise.
21828 (build3_stat): Likewise.
21829 (build4_stat): Likewise.
21830 (build5_stat): Likewise.
21831 (build_decl_stat): Likewise.
21832 * tree.h (make_node_stat): Adjust prototype.
21833 (make_node): remove macro.
21835 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
21837 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
21838 (PPC_FEATURE2_SCV): Likewise.
21839 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
21841 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21843 * config/aarch64/aarch64.c
21844 (aarch64_internal_mov_immediate): Add new special pattern.
21845 * config/aarch64/aarch64.md (*movdi_aarch64):
21846 Add reg/32bit const mov case.
21848 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21849 Richard Sandiford <richard.sandiford@linaro.org>
21851 * config/aarch64/aarch64.md (mov<mode>): Generalize.
21852 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21853 Add integer and movi cases.
21854 (movi-split-hf-df-sf split, fp16): New.
21855 (enabled): Added TARGET_FP_F16INST.
21856 * config/aarch64/iterators.md (GPF_HF): New.
21857 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
21859 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21861 * config/aarch64/aarch64.c
21862 (aarch64_simd_container_mode): Add prototype.
21863 (aarch64_expand_mov_immediate): Add HI support.
21864 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
21865 (aarch64_can_const_movi_rtx_p): New.
21866 (aarch64_preferred_reload_class):
21867 Remove restrictions of using FP registers for certain SIMD operations.
21868 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
21869 (aarch64_valid_floating_const): Add integer move validation.
21870 (aarch64_simd_imm_scalar_p): Remove.
21871 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
21872 (aarch64_legitimate_constant_p): Expand list of supported cases.
21873 * config/aarch64/aarch64-protos.h
21874 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
21875 (aarch64_reinterpret_float_as_int): New.
21876 (aarch64_simd_imm_scalar_p): Remove.
21877 * config/aarch64/constraints.md (Uvi): New.
21878 (Dd): Split into Ds and new Dd.
21879 * config/aarch64/aarch64.md (*movsi_aarch64):
21881 (*movdi_aarch64): Add SIMD mov case.
21883 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21885 * tree-predcom.c: (struct chain): Handle store-store chain in which
21886 stores for elimination only store loop invariant values.
21887 (execute_pred_commoning_chain): Ditto.
21888 (prepare_initializers_chain_store_elim): Ditto.
21889 (prepare_finalizers): Ditto.
21890 (is_inv_store_elimination_chain): New function.
21891 (initialize_root_vars_store_elim_1): New function.
21893 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21895 * tree-predcom.c: Revise general description of the pass.
21896 (enum chain_type): New enum type for store elimination.
21897 (struct chain): New field supporting store elimination.
21898 (struct component): Ditto.
21899 (dump_chain): Dump store-stores chain.
21900 (release_chain): Release resources.
21901 (split_data_refs_to_components): Compute and create component
21902 contains only stores for elimination.
21903 (get_chain_last_ref_at): New function.
21904 (make_invariant_chain): Initialization.
21905 (make_rooted_chain): Specify chain type in parameter and record it.
21906 (add_looparound_copies): Skip for store-stores chain.
21907 (determine_roots_comp): Compute type of chain and pass it to
21909 (initialize_root_vars_store_elim_2): New function.
21910 (finalize_eliminated_stores): New function.
21911 (remove_stmt): Handle store for elimination.
21912 (execute_pred_commoning_chain): Execute predictive commoning on
21913 store-store chains.
21914 (determine_unroll_factor): Skip unroll for store-stores chain.
21915 (prepare_initializers_chain_store_elim): New function.
21916 (prepare_initializers_chain): Hanlde store-store chain.
21917 (prepare_finalizers_chain, prepare_finalizers): New function.
21918 (tree_predictive_commoning_loop): Return integer value indicating
21919 if loop is unrolled or lcssa form is corrupted.
21920 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
21922 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21924 * tree-predcom.c (initialize_root): Delete.
21925 (execute_pred_commoning_chain): Initialize root vars and replace
21926 reference of non-combined chain directly, rather than call above
21929 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21931 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
21932 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
21934 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21936 * tree-predcom.c (struct chain): New field init_seq.
21937 (release_chain): Release init_seq.
21938 (prepare_initializers_chain): Record intialization stmts in above
21940 (insert_init_seqs): New function.
21941 (tree_predictive_commoning_loop): Call insert_init_seqs.
21943 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21945 * tree-predcom.c (determine_roots_comp): Skip trivial components.
21947 2017-07-28 Richard Biener <rguenther@suse.de>
21949 * match.pd: Remove superfluous :c.
21950 * genmatch.c (simplify::id): Add member.
21951 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
21953 (current_id): New global.
21954 (dt_node::parent): Move from ...
21955 (dt_operand::parent): ... here. Add for_id member.
21956 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
21957 (decision_tree::find_node): Relax order requirement when
21958 merging DT_TRUE nodes to ones inbetween the current simplify
21959 and the one we try to merge with. Add diagnostic whenever
21960 we need to enforce pattern order by not merging.
21961 (decision_tree::insert): Set current_id.
21962 (decision_tree::print_node): Dump parent node and for_id.
21963 (parser::last_id): Add member.
21964 (parser::push_simplify): Assign unique id.
21965 (parser::parser): Initialize last_id.
21967 2017-07-28 Martin Liska <mliska@suse.cz>
21970 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
21971 gimple_build_debug_bind.
21973 2017-07-28 Richard Biener <rguenther@suse.de>
21975 PR tree-optimization/81502
21976 * match.pd: Add pattern combining BIT_INSERT_EXPR with
21978 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
21980 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
21981 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
21982 for BIT_FIELD_REF args.
21983 * fold-const.c (make_bit_field_ref): Likewise.
21984 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
21986 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21989 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
21990 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
21991 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
21992 Or it into SANITIZER_UNDEFINED.
21993 * ubsan.c: Include gimple-fold.h and varasm.h.
21994 (ubsan_expand_ptr_ifn): New function.
21995 (instrument_pointer_overflow): New function.
21996 (maybe_instrument_pointer_overflow): New function.
21997 (instrument_object_size): Formatting fix.
21998 (pass_ubsan::execute): Call instrument_pointer_overflow
21999 and maybe_instrument_pointer_overflow.
22000 * internal-fn.c (expand_UBSAN_PTR): New function.
22001 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
22002 * sanitizer.def (__ubsan_handle_pointer_overflow,
22003 __ubsan_handle_pointer_overflow_abort): New builtins.
22004 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
22005 * internal-fn.def (UBSAN_PTR): New internal function.
22006 * opts.c (sanitizer_opts): Add pointer-overflow.
22007 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
22008 * fold-const.c (build_range_check): Compute pointer range check in
22009 integral type if pointer arithmetics would be needed. Formatting
22012 2017-07-28 Martin Liska <mliska@suse.cz>
22015 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
22016 parameters that are of a variable-length.
22018 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
22020 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
22022 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
22023 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
22024 (CRT_CALL_STATIC_FUNCTION): Likewise.
22025 (ASM_DEFAULT_SPEC): New define.
22026 (ASM_SPEC32): Likewise.
22027 (ASM_SPEC64): Likewise.
22028 (ASM_SPEC_COMMON): Likewise.
22029 (ASM_SPEC): Likewise.
22030 (INVALID_64BIT): Likewise.
22031 (LINK_OS_DEFAULT_SPEC): Likewise.
22032 (LINK_OS_SPEC32): Likewise.
22033 (LINK_OS_SPEC64): Likewise.
22034 (POWERPC_LINUX): Likewise.
22035 (PTRDIFF_TYPE): Likewise.
22036 (RESTORE_FP_PREFIX): Likewise.
22037 (RESTORE_FP_SUFFIX): Likewise.
22038 (SAVE_FP_PREFIX): Likewise.
22039 (SAVE_FP_SUFFIX): Likewise.
22040 (SIZE_TYPE): Likewise.
22041 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
22042 (TARGET_64BIT): Likewise.
22043 (TARGET_64BIT): Likewise.
22044 (TARGET_AIX): Likewise.
22045 (WCHAR_TYPE_SIZE): Likewise.
22046 (WCHAR_TYPE): Undefine.
22047 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
22048 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
22049 (CPP_OS_RTEMS_SPEC): Delete.
22050 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
22051 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
22053 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
22055 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22057 PR tree-optimization/81578
22058 * tree-parloops.c (build_new_reduction): Bail out if
22059 reduction_code isn't one of the standard OpenMP reductions.
22060 Move the details printing after that decision.
22062 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22064 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
22065 related to reload_in_progress.
22066 (splat_input_operand): Likewise.
22067 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
22069 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
22071 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
22072 (TARGET_INSTANTIATE_DECLS): Likewise.
22073 (legitimate_indexed_address_p): Delete reload_in_progress code.
22074 (rs6000_debug_legitimate_address_p): Likewise.
22075 (rs6000_eliminate_indexed_memrefs): Likewise.
22076 (rs6000_emit_le_vsx_store): Likewise.
22077 (rs6000_emit_move_si_sf_subreg): Likewise.
22078 (rs6000_emit_move): Likewise.
22079 (register_to_reg_type): Likewise.
22080 (rs6000_pre_atomic_barrier): Likewise.
22081 (rs6000_machopic_legitimize_pic_address): Likewise.
22082 (rs6000_allocate_stack_temp): Likewise.
22083 (rs6000_address_for_fpconvert): Likewise.
22084 (rs6000_address_for_altivec): Likewise.
22085 (rs6000_secondary_memory_needed_rtx): Delete function.
22086 (rs6000_check_sdmode): Likewise.
22087 (rs6000_alloc_sdmode_stack_slot): Likewise.
22088 (rs6000_instantiate_decls): Likewise.
22089 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
22090 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
22091 Delete reload_in_progress.
22092 (*vec_reload_and_plus_<mptrsize>): Likewise.
22093 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
22094 (vsx_div_v2di): Likewise.
22095 (vsx_udiv_v2di): Likewise.
22097 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22099 * config/rs6000/rs6000.opt (mlra): Replace with stub.
22100 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
22101 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
22102 (rs6000_debug_reg_global): Delete print of LRA status.
22103 (rs6000_option_override_internal): Delete dead LRA related code.
22104 (rs6000_lra_p): Delete function.
22105 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
22107 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
22109 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
22110 * config/riscv/rtems.h: New file.
22112 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22113 Sudakshina Das <sudi.das@arm.com>
22115 * config/aarch64/aarch64.md
22116 (define_split for and<mode>3nr_compare): Move
22117 non aarch64_logical_operand to a register.
22118 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
22119 register immediate operand to a register.
22120 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
22122 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22124 PR middle-end/81564
22125 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
22127 2017-07-27 Richard Biener <rguenther@suse.de>
22129 PR tree-optimization/81573
22130 PR tree-optimization/81494
22131 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
22132 multi defuse cycle case.
22134 2017-07-27 Richard Biener <rguenther@suse.de>
22136 PR tree-optimization/81571
22137 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
22140 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
22142 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
22143 earlier and only if MASK_FPU is set. Adjust formatting.
22145 2017-07-27 Martin Liska <mliska@suse.cz>
22147 * opt-functions.awk: Add validation of value of Init.
22148 * optc-gen.awk: Pass new argument.
22150 2017-07-27 Martin Liska <mliska@suse.cz>
22152 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
22153 Fix wrong condition.
22155 2017-07-27 Martin Liska <mliska@suse.cz>
22157 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
22158 BBs and edges seen by autoFDO.
22160 2017-07-27 Richard Biener <rguenther@suse.de>
22162 PR tree-optimization/81502
22163 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
22164 with incompatible but same sized type.
22165 (execute_update_addresses_taken): Likewise.
22167 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
22169 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
22170 flag_tree_loop_vectorize rather than flag_tree_vectorize.
22172 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22175 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
22176 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
22177 Change s_operand to memory_operand.
22179 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
22181 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
22182 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
22183 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
22184 Emit instructions rather than returning an expression. Handle TFmode
22185 and KFmode by casting to TImode.
22186 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
22187 (rs6000_emit_le_vsx_store): Likewise.
22188 * config/rs6000/vsx.md (VSX_TI): New iterator.
22189 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
22190 (*vsx_le_undo_permute_<mode>): Likewise.
22191 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
22192 emit the split sequence.
22193 (*vsx_le_perm_store_<mode>): Likewise.
22195 2017-07-27 Jakub Jelinek <jakub@redhat.com>
22197 PR tree-optimization/81555
22198 PR tree-optimization/81556
22199 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
22200 if true, force CHANGED for the recursive invocation.
22201 (reassociate_bb): Remember original length of ops array, pass
22202 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
22204 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
22205 attributes for noipa attribute. For naked attribute use
22206 lookup_attribute first before lookup_attribute_spec.
22207 * final.c (rest_of_handle_final): Disable IPA RA for functions with
22209 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
22210 for functions with noipa attribute.
22211 (cgraph_externally_visible_p): Return true for functions with noipa
22213 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
22214 for functions with noipa attribute.
22215 * doc/extend.texi: Document noipa function attribute.
22216 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
22217 also for functions with noipa attribute.
22218 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
22220 2017-07-26 Andrew Pinski <apinski@cavium.com>
22222 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
22223 vec_unalign_load_cost and vec_unalign_store_cost.
22225 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
22227 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
22228 -mvsx-small-integer option.
22229 (ISA_3_0_MASKS_IEEE): Likewise.
22230 (OTHER_VSX_VECTOR_MASKS): Likewise.
22231 (POWERPC_MASKS): Likewise.
22232 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
22233 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
22234 code, only testing for DImode being allowed in non-VSX floating
22236 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
22237 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
22239 (rs6000_option_override_internal): Delete -mvsx-small-integer.
22240 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
22241 TARGET_P8_VECTOR test.
22242 (rs6000_secondary_reload_simple_move): Likewise.
22243 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
22244 since TARGET_P9_VECTOR was already tested.
22245 (rs6000_opt_masks): Remove -mvsx-small-integer.
22246 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
22247 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22249 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
22250 test for TARGET_VEXTRACTUB was used, and that uses
22252 (p9 extract splitter): Likewise.
22253 (vsx_extract_<mode>_di_p9): Likewise.
22254 (vsx_extract_<mode>_store_p9): Likewise.
22255 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
22256 for TARGET_P9_VECTOR was used. Delete code that is now dead with
22257 the elimination of TARGET_VSX_SMALL_INTEGER.
22258 (vsx_extract_<mode>_p8): Likewise.
22259 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
22260 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
22261 (vsx_set_<mode>_p9): Likewise.
22262 (vsx_set_v4sf_p9): Likewise.
22263 (vsx_set_v4sf_p9_zero): Likewise.
22264 (vsx_insert_extract_v4sf_p9): Likewise.
22265 (vsx_insert_extract_v4sf_p9_2): Likewise.
22266 * config/rs6000/rs6000.md (sign extend splitter): Change
22267 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
22268 (floatsi<mode>2_lfiwax_mem): Likewise.
22269 (floatunssi<mode>2_lfiwzx_mem): Likewise.
22270 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
22271 since a test for TARGET_P9_VECTOR was used.
22272 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22273 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
22274 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22275 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22276 TARGET_P8_VECTOR test.
22277 (fix_trunc<mode>si2_stfiwx): Likewise.
22278 (fix_trunc<mode>si2_internal): Likewise.
22279 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
22280 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22282 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22283 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22284 TARGET_P8_VECTOR test.
22285 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22286 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
22287 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22289 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22290 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
22291 since a test for TARGET_P9_VECTOR was used.
22292 (splitter for loading small constants): Likewise.
22294 2017-07-26 Andrew Pinski <apinski@cavium.com>
22296 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
22299 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
22302 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
22303 (fp_valid_at): Likewise.
22305 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22307 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
22308 (qdf24xx_addrcost_table): Likewise.
22309 (cortexa57_tunings): Update to use generic_branch_cost.
22310 (cortexa72_tunings): Likewise.
22311 (cortexa73_tunings): Likewise.
22312 (qdf24xx_tunings): Likewise.
22314 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22316 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
22317 (thunderx2t99_branch_cost): Likewise.
22318 (cortexa35_tunings): Update to use generic_branch_cost.
22319 (cortexa53_tunings): Likewise.
22320 (cortexa57_tunings): Likewise.
22321 (cortexa72_tunings): Likewise.
22322 (cortexa73_tunings): Likewise.
22323 (thunderx2t99_tunings): Likewise.
22325 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22327 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
22328 (sparc_option_override): Honour MASK_FSMULD.
22329 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
22330 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
22331 * config/sparc/sparc.opt (mfsmuld): New option.
22332 * doc/invoke.texi (mfsmuld): Document option.
22334 2017-07-26 Marek Polacek <polacek@redhat.com>
22336 PR middle-end/70992
22337 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
22339 2017-07-26 Richard Biener <rguenther@suse.de>
22341 * gimple-match-head.c (do_valueize): Return OP if valueize
22343 (get_def): New helper to get at the def stmt of a SSA name
22344 if valueize allows.
22345 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
22346 do_valueize to get at the def stmt.
22347 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
22349 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
22351 PR middle-end/46932
22352 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
22354 2017-07-26 Martin Liska <mliska@suse.cz>
22357 * function.c (expand_function_start): Make expansion of
22358 nonlocal_goto_save_area after parm_birth_insn.
22360 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22362 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
22363 from all CPU target flags enable members.
22365 2017-07-26 Richard Biener <rguenther@suse.de>
22367 * genmatch.c (dt_simplify::gen): Make iterator vars const.
22368 (decision_tree::gen): Make 'type' const.
22369 (write_predicate): Likewise.
22371 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22373 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
22374 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
22375 (rs6000_option_override_internal): Likewise.
22376 (rs6000_expand_vector_set): Likewise.
22377 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
22378 (TARGET_UPPER_REGS_SF): Likewise.
22379 (TARGET_UPPER_REGS_DI): Likewise.
22380 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
22381 (TARGET_DIRECT_MOVE_64BIT): Likewise.
22382 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
22383 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22384 (Splitters for DI constants in Altivec registers): Likewise.
22385 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
22386 (vsx_set_v4sf_p9): Likewise.
22387 (vsx_set_v4sf_p9_zero): Likewise.
22388 (vsx_insert_extract_v4sf_p9): Likewise.
22389 (vsx_insert_extract_v4sf_p9_2): Likewise.
22391 2017-07-25 Carl Love <cel@us.ibm.com>
22393 * doc/extend.texi: Update the built-in documentation file for the
22394 existing built-in functions
22395 vector signed char vec_cnttz (vector signed char);
22396 vector unsigned char vec_cnttz (vector unsigned char);
22397 vector signed short vec_cnttz (vector signed short);
22398 vector unsigned short vec_cnttz (vector unsigned short);
22399 vector signed int vec_cnttz (vector signed int);
22400 vector unsigned int vec_cnttz (vector unsigned int);
22401 vector signed long long vec_cnttz (vector signed long long);
22402 vector unsigned long long vec_cnttz (vector unsigned long long);
22404 2017-07-25 Andrew Pinski <apinski@cavium.com>
22406 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
22407 accesses where the use is for the first operand of a BIT_INSERT.
22409 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22412 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22413 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22415 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22417 * config/i386/gstabs.h: Delete.
22418 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
22420 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
22422 * config/i386/i386.c (ix86_decompose_address): Do not check for
22423 register RTX when looking at index_reg or base_reg.
22424 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
22426 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
22428 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
22429 to update EH info here.
22431 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22433 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
22435 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22437 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
22439 2017-07-25 Torsten Duwe <duwe@suse.de>
22441 * common.opt: Introduce -fpatchable-function-entry
22442 command line option, and its variables function_entry_patch_area_size
22443 and function_entry_patch_area_start.
22444 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
22445 including a two-value parser.
22446 * target.def (print_patchable_function_entry): New target hook.
22447 * targhooks.h (default_print_patchable_function_entry): New function.
22448 * targhooks.c (default_print_patchable_function_entry): Likewise.
22449 * toplev.c (process_options): Switch off IPA-RA if
22450 patchable function entries are being generated.
22451 * varasm.c (assemble_start_function): Look at the
22452 patchable-function-entry command line switch and current
22453 function attributes and maybe generate NOP instructions by
22454 calling the print_patchable_function_entry hook.
22455 * doc/extend.texi: Document patchable_function_entry attribute.
22456 * doc/invoke.texi: Document -fpatchable_function_entry
22457 command line option.
22458 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
22460 * doc/tm.texi: Re-generate.
22462 2017-07-25 Jakub Jelinek <jakub@redhat.com>
22465 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
22466 TARGET_AVX512DQ rather than TARGET_AVX512BW.
22468 2017-07-25 Tamar Christina <tamar.christina@arm.com>
22470 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
22472 2017-07-25 Richard Biener <rguenther@suse.de>
22474 PR tree-optimization/81455
22475 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
22476 not walk in cycles when looking for guards.
22478 2017-07-25 Richard Biener <rguenther@suse.de>
22480 PR tree-optimization/81529
22481 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
22482 when optimizing backedge uses.
22484 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
22486 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
22488 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
22489 to dl_section_ref. On AIX, append an expression to subtract
22490 the size of the section length to dl_section_ref.
22492 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
22494 * configure.ac: If any of the config.* scripts fail, exit 1.
22495 * configure: Regenerate.
22497 2017-07-25 Richard Biener <rguenther@suse.de>
22499 PR middle-end/81546
22500 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
22501 of immediate uses, be more verbose on errors.
22503 2017-07-25 Richard Biener <rguenther@suse.de>
22505 PR tree-optimization/81510
22506 * tree-vect-loop.c (vect_is_simple_reduction): When the
22507 reduction stmt is not inside the loop bail out.
22509 2017-07-25 Richard Biener <rguenther@suse.de>
22511 PR tree-optimization/81303
22512 * tree-vect-loop-manip.c (vect_loop_versioning): Build
22513 profitability check against LOOP_VINFO_NITERSM1.
22515 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22517 * domwalk.c (cmp_bb_postorder): Simplify.
22518 (sort_bbs_postorder): New function. Use it...
22519 (dom_walker::walk): ...here to optimize common cases.
22521 2017-07-25 Martin Liska <mliska@suse.cz>
22524 * ipa-visibility.c (function_and_variable_visibility): Make the
22525 redirection just on target that supports aliasing.
22526 Fix GNU coding style.
22528 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
22531 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
22533 * config/arm/rtems.h: Likewise.
22534 * config/bfin/rtems.h: Likewise.
22535 * config/i386/rtemself.h: Likewise.
22536 * config/lm32/rtems.h: Likewise.
22537 * config/m32c/rtems.h: Likewise.
22538 * config/m68k/rtemself.h: Likewise.
22539 * config/microblaze/rtems.h: Likewise.
22540 * config/mips/rtems.h: Likewise.
22541 * config/moxie/rtems.h: Likewise.
22542 * config/nios2/rtems.h: Likewise.
22543 * config/powerpcspe/rtems.h: Likewise.
22544 * config/rs6000/rtems.h: Likewise.
22545 * config/rtems.h: Likewise.
22546 * config/sh/rtems.h: Likewise.
22547 * config/sh/rtemself.h: Likewise.
22548 * config/sparc/rtemself.h: Likewise.
22550 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22553 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
22554 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
22555 * tree-ssa-structalias.c (alias_get_name): Same.
22557 2017-07-25 Bin Cheng <bin.cheng@arm.com>
22560 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
22561 instructions if no du chain is found.
22563 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22565 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
22567 2017-07-25 Richard Biener <rguenther@suse.de>
22569 PR middle-end/81505
22570 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
22573 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22575 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
22576 upper-regs options.
22577 (ISA_2_7_MASKS_SERVER): Likewise.
22578 (ISA_3_0_MASKS_IEEE): Likewise.
22579 (OTHER_P8_VECTOR_MASKS): Likewise.
22580 (OTHER_VSX_VECTOR_MASKS): Likewise.
22581 (POWERPC_MASKS): Likewise.
22582 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
22583 duplicate list of options.
22584 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
22585 explicit -mupper-regs options.
22586 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
22587 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
22588 alias for -mupper-regs-df.
22589 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
22590 (rs6000_init_hard_regno_mode_ok): Likewise.
22591 (rs6000_option_override_internal): Likewise.
22592 (rs6000_opt_masks): Likewise.
22593 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
22594 options in terms of whether -mvsx or -mpower8-vector was used.
22595 (TARGET_UPPER_REGS_DI): Likewise.
22596 (TARGET_UPPER_REGS_SF): Likewise.
22597 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
22598 -mupper-regs-* options.
22600 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22602 * passes.c (emergency_dump_function): Print some empty lines and a
22603 header before the RTL dump.
22605 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22607 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
22609 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
22612 * config/aarch64/aarch64.c (aarch64_classify_symbol):
22613 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
22615 2017-07-24 Carl Love <cel@us.ibm.com>
22617 * config/rs6000/rs6000-c.c: Add support for built-in functions
22618 vector float vec_extract_fp32_from_shorth (vector unsigned short);
22619 vector float vec_extract_fp32_from_shortl (vector unsigned short);
22620 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
22621 vec_extract_fp_from_shortl): Add defines for the two builtins.
22622 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
22623 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
22625 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
22626 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
22627 * doc/extend.texi: Update the built-in documentation file for the
22628 new built-in function.
22630 2017-07-24 Jakub Jelinek <jakub@redhat.com>
22633 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
22635 * doc/generic.texi: Likewise.
22636 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22637 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22639 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
22641 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
22642 (aarch64_mls_elt_merge<mode>): Likewise.
22644 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
22646 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22647 having __cxa_atexit.
22649 2017-07-23 Michael Collison <michael.collison@arm.com>
22651 * config/arm/arm.c (arm_option_override): Deprecate
22652 use of -mstructure-size-boundary.
22653 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
22654 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
22656 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22659 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22660 Reduce cost estimate for direct moves.
22662 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
22665 * config/i386/i386.c (ix86_option_override_internal): Disable
22666 BMI, BMI2 and TBM instructions for -m16.
22668 2017-07-21 Carl Love <cel@us.ibm.com>
22670 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22671 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22672 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22673 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22674 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22675 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22676 VMULOSW): New enum "unspec" values.
22677 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22678 altivec_vmulosw): New patterns.
22679 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22680 VMULOSW): Add definitions.
22682 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
22684 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
22685 (qdf24xx): Likewise.
22686 * config/aarch64/aarch64-options-extensions.def (rdma); New.
22687 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
22688 (AARCH64_FL_V8_1): Renumber.
22689 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
22690 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
22691 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
22692 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
22693 rdma to feature modifiers list.
22695 2017-07-21 Yury Gribov <tetra2005@gmail.com>
22697 PR middle-end/56727
22698 * ipa-visibility (function_and_variable_visibility): Convert
22699 recursive PLT call to direct call if appropriate.
22701 2017-07-21 Andrew Pinski <apinski@cavium.com>
22703 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
22704 operand 1 to see if the types precision matches.
22705 * fold-const.c (operand_equal_p): Likewise.
22707 2017-07-21 Richard Biener <rguenther@suse.de>
22709 PR tree-optimization/81303
22710 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
22711 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
22712 (vect_peeling_hash_get_lowest_cost): Adjust.
22713 (vect_enhance_data_refs_alignment): Likewise. Use
22714 vect_get_peeling_costs_all_drs to compute the penalty for no
22715 peeling to match up costs.
22717 2017-07-21 Richard Biener <rguenther@suse.de>
22719 PR tree-optimization/81500
22720 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
22721 we didn't identify a reduction path.
22723 2017-07-21 Tom de Vries <tom@codesourcery.com>
22724 Cesar Philippidis <cesar@codesourcery.com>
22726 PR gcov-profile/81442
22727 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
22730 2017-07-21 Tom de Vries <tom@codesourcery.com>
22733 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
22735 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
22736 nvptx_override_options_after_change.
22738 2017-07-21 Ulrich Drepper <drepper@redhat.com>
22740 * dwarf2out.c (output_file_names): Avoid double testing for
22741 dwarf_version >= 5.
22743 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
22745 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
22747 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22749 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
22751 (try_crossjump_to_edge): Do not punt on partitioned functions.
22753 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22755 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
22756 Put all BBs reachable only via paths crossing cold region to cold
22758 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
22760 2016-07-21 Richard Biener <rguenther@suse.de>
22762 PR tree-optimization/81303
22763 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
22764 into account prologue and epilogue iterations when raising
22765 min_profitable_iters to sth at least covering one vector iteration.
22767 2017-07-21 Tamar Christina <tamar.christina@arm.com>
22769 * config/arm/arm.c (arm_test_cpu_arch_dat):
22772 2017-07-20 Nathan Sidwell <nathan@acm.org>
22774 Remove TYPE_METHODS.
22775 * tree.h (TYPE_METHODS): Delete.
22776 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
22777 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
22778 (dbxout_type_methods): Scan TYPE_FIELDS.
22779 (dbxout_type): Don't check TYPE_METHODS here.
22780 * function.c (use_register_for_decl): Always ignore register for
22781 class types when not optimizing.
22782 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
22783 * tree.c (free_lang_data_in_type): Stitch out member functions and
22784 templates from TYPE_FIELDS.
22785 (build_distinct_type_copy, verify_type_variant,
22786 verify_type): Member fns are on TYPE_FIELDS.
22787 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
22788 * tree-pretty-print.c (dump_generic_node): Likewise.
22790 2017-07-20 Jakub Jelinek <jakub@redhat.com>
22793 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
22794 V2TImode and V4TImode.
22795 (ix86_expand_vector_extract): Likewise.
22796 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
22797 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
22798 (ssescalarmode): Handle V4TImode and V2TImode.
22799 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
22800 (*vec_extractv2ti, *vec_extractv4ti): New insns.
22801 (VEXTRACTI128_MODE): New mode iterator.
22802 (splitter for *vec_extractv?ti first element): New.
22803 (VEC_INIT_MODE): New mode iterator.
22804 (vec_init<mode>): Consolidate 3 expanders into one using
22805 VEC_INIT_MODE mode iterator.
22807 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22809 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
22810 non_spilled_static_chain_regno_p.
22812 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22814 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
22816 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
22818 * bb-reorder.c (connect_traces): Allow copying of blocks within
22821 2017-07-20 Richard Biener <rguenther@suse.de>
22823 * gimple.h (gimple_phi_result): Add gphi * overload.
22824 (gimple_phi_result_ptr): Likewise.
22825 (gimple_phi_arg): Likewise. Adjust index assert to only
22826 allow actual argument accesses rather than all slots available
22828 (gimple_phi_arg_def): Add gphi * overload.
22829 * tree-phinodes.c (make_phi_node): Initialize only actual
22831 (resize_phi_node): Clear memory not covered by old node,
22832 do not initialize excess argument slots.
22833 (reserve_phi_args_for_new_edge): Initialize new argument slot
22836 2017-07-20 Bin Cheng <bin.cheng@arm.com>
22838 PR tree-optimization/81388
22840 2016-07-21 Bin Cheng <bin.cheng@arm.com>
22842 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
22843 by removing computation of may_be_zero.
22845 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22846 Tom de Vries <tom@codesourcery.com>
22848 PR middle-end/81030
22849 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
22850 when gimple level profile disagrees with what RTL expander did.
22852 2017-07-20 Richard Biener <rguenther@suse.de>
22854 PR tree-optimization/61171
22855 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
22856 (vect_analyze_stmt): Add slp instance parameter.
22857 (vectorizable_reduction): Likewise.
22858 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
22859 (vect_is_simple_reduction): Deal with chains not detected
22860 as SLP reduction chain, specifically not properly associated
22861 chains containing a mix of plus/minus.
22862 (get_reduction_op): Remove.
22863 (get_initial_defs_for_reduction): Simplify, pass in whether
22864 this is a reduction chain, pass in the SLP node for the PHIs.
22865 (vect_create_epilog_for_reduction): Get the SLP instance as
22867 (vectorizable_reduction): Get the SLP instance as arg.
22868 During analysis remember the SLP node with the PHIs in the
22869 instance. Simplify getting at the vectorized reduction PHIs.
22870 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
22871 through SLP instance.
22872 (vect_slp_analyze_operations): Likewise.
22873 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
22874 (vect_transform_stmt): Likewise.
22876 2017-07-20 Tom de Vries <tom@codesourcery.com>
22878 PR tree-optimization/81489
22879 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
22880 read of phi arg location to before loop that modifies phi.
22882 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22884 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
22887 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22889 PR middle-end/81331
22890 * except.c (execute): Fix ordering issue.
22892 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22894 PR rtl-optimization/81423
22895 * combine.c (make_compound_operation_int): Don't try to optimize
22896 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
22898 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22900 PR rtl-optimization/81423
22901 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
22902 with a constant that is -1 in the truncated to mode.
22904 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22906 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
22907 (determine_unlikely_bbs): ... here.
22908 * predict.h (propagate_unlikely_bbs_forward): Declare.
22909 * cfgexpand.c (pass_expand::execute): Use it.
22910 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
22912 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
22913 propagate_unlikely_bbs_forward.
22915 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22917 PR middle-end/81331
22918 * except.c (maybe_add_nop_after_section_switch): New function.
22921 2017-07-19 Tom de Vries <tom@codesourcery.com>
22923 * gimple.h (gimple_phi_set_arg): Make assert more strict.
22925 2017-07-19 Tom de Vries <tom@codesourcery.com>
22927 * gimple.h (gimple_phi_arg): Make assert more strict.
22929 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
22931 * config.gcc (powerpc*-*-*): Add mmintrin.h.
22932 * config/rs6000/mmintrin.h: New file.
22933 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
22935 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22937 PR tree-optimization/81346
22938 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
22940 2017-07-19 Tom de Vries <tom@codesourcery.com>
22942 * config/nvptx/nvptx.md (VECIM): Add V2DI.
22944 2017-07-19 Tom de Vries <tom@codesourcery.com>
22946 * config/nvptx/nvptx-modes.def: Add V2DImode.
22947 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
22948 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
22949 (nvptx_output_mov_insn): Handle lack of mov.b128.
22950 (nvptx_print_operand): Handle 'H' and 'L' codes.
22951 (nvptx_vector_mode_supported): Allow V2DImode.
22952 (nvptx_preferred_simd_mode): New function.
22953 (nvptx_data_alignment): New function.
22954 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
22955 nvptx_preferred_simd_mode.
22956 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
22958 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
22960 2017-07-19 Tom de Vries <tom@codesourcery.com>
22962 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
22963 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
22964 (nvptx_vector_mode_supported): New function. Allow V2SImode.
22965 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
22966 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
22967 (mov<VECIM>_insn): New define_insn.
22968 (define_expand "mov<VECIM>): New define_expand.
22970 2017-07-19 Tom de Vries <tom@codesourcery.com>
22972 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
22974 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22976 PR tree-optimization/81346
22977 * fold-const.h (fold_div_compare, range_check_type): Declare.
22978 * fold-const.c (range_check_type): New function.
22979 (build_range_check): Use range_check_type.
22980 (fold_div_compare): No longer static, rewritten into
22981 a match.pd helper function.
22982 (fold_comparison): Don't call fold_div_compare here.
22983 * match.pd (X / C1 op C2): New optimization using fold_div_compare
22984 as helper function.
22986 2017-07-19 Nathan Sidwell <nathan@acm.org>
22988 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
22989 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
22990 * tree.c (find_decls_types_r, verify_type): Use
22991 TYPE_{MIN,MAX}_VALUE_RAW.
22992 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
22993 (hash_tree): Likewise.
22994 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
22996 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
22999 2017-07-18 Tom de Vries <tom@codesourcery.com>
23001 PR middle-end/81464
23002 * omp-expand.c (expand_omp_for_static_chunk): Handle
23003 equal-argument loop exit phi.
23005 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
23008 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
23009 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
23010 operand 2 predicate.
23011 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
23012 operand 2 predicate.
23013 (ror,rol -> rorx splitters): Use const_int_operand as
23014 operand 2 predicate.
23016 2017-06-18 Richard Biener <rguenther@suse.de>
23018 PR tree-optimization/81410
23019 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
23020 the gap in the ! slp_perm SLP case after each group.
23022 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23024 PR middle-end/81463
23025 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
23028 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23030 PR middle-end/81462
23031 * predict.c (set_even_probabilities): Cleanup; do not affect
23032 probabilities that are already known.
23033 (combine_predictions_for_bb): Call even when count is set.
23035 2017-07-18 Nathan Sidwell <nathan@acm.org>
23037 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
23040 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23043 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
23044 optimization for loop niter analysis.
23046 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
23049 * config/avr/avr.c (avr_optimize_casesi): Don't use
23050 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
23052 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
23054 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
23055 body_cost_vec from _vect_peel_extended_info.
23056 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
23057 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
23060 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23062 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
23064 2017-07-18 Richard Biener <rguenther@suse.de>
23066 PR tree-optimization/80620
23067 PR tree-optimization/81403
23068 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
23069 info when re-using a VN table entry.
23071 2017-07-18 Richard Biener <rguenther@suse.de>
23073 PR tree-optimization/81418
23074 * tree-vect-loop.c (vectorizable_reduction): Properly compute
23075 vectype_in. Verify that with lane-reducing reduction operations
23076 we have a single def-use cycle.
23078 2017-07-17 Carl Love <cel@us.ibm.com>
23080 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
23082 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23083 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23084 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23085 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23086 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23087 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23088 VMULOSW): New enum "unspec" values.
23089 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23090 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23091 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23092 altivec_vmulosw): New patterns.
23093 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23094 VMULOSW): Add definitions.
23096 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
23098 * config/alpha/alpha.c: Include predict.h.
23100 2017-07-17 Yury Gribov <tetra2005@gmail.com>
23102 * tree-vrp.c (compare_assert_loc): Fix comparison function
23103 to return predictable results.
23105 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23107 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
23109 (subdi3): Likewise.
23110 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
23111 * doc/invoke.texi (mexpand-adddi): Update text.
23113 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23115 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
23116 that also clobbers the CC register. The old expand code is moved
23118 (*arc_clzsi2): ... here.
23119 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
23120 the CC register. The old expand code is moved to ...
23121 (arc_ctzsi2): ... here.
23123 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23125 * config/arc/arc.opt (mindexed-loads): Use initial value
23126 TARGET_INDEXED_LOADS_DEFAULT.
23127 (mauto-modify-reg): Use initial value
23128 TARGET_AUTO_MODIFY_REG_DEFAULT.
23129 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23130 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23131 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23132 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23134 2017-07-17 Martin Liska <mliska@suse.cz>
23137 * opts.c (finish_options): Do not allow -fgnu-tm
23138 w/ -fsanitize={kernel-,}address. Say sorry.
23140 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23143 * tree-loop-distribution.c (classify_partition): Only assert on
23144 numer of iterations.
23145 (merge_dep_scc_partitions): Delete prameter. Update function call.
23146 (distribute_loop): Remove code handling loop with unknown niters.
23147 (pass_loop_distribution::execute): Skip loop with unknown niters.
23149 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23152 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
23153 function sort_partitions_by_post_order.
23155 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23157 PR tree-optimization/81374
23158 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
23159 the max index of basic blocks, rather than number of basic blocks.
23161 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23163 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
23165 (arc_legitimate_pic_operand_p): Likewise.
23166 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
23168 (arc_needs_pcl_p): Likewise.
23169 (arc_legitimate_pc_offset_p): Likewise.
23170 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
23171 function is also used in constrains.md.
23172 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
23173 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
23174 PLUS. Only return true/false in known cases, otherwise assert.
23175 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
23176 is already called in arc_legitimate_constant_p.
23177 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
23179 (LEGITIMATE_PIC_OPERAND_P): Use
23180 arc_raw_symbolic_reference_mentioned_p function.
23181 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
23186 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23187 Andrew Burgess <andrew.burgess@embecosm.com>
23189 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
23190 (arc_return_address_register): New function.
23191 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
23192 (arc_handle_fndecl_attribute): Add naked attribute.
23193 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
23194 (TARGET_WARN_FUNC_RETURN): Likewise.
23195 (arc_allocate_stack_slots_for_args): New function.
23196 (arc_warn_func_return): Likewise.
23197 (machine_function): Change type fn_type.
23198 (arc_compute_function_type): Consider new naked function type,
23199 change function return type.
23200 (arc_must_save_register): Adapt to handle new
23201 arc_compute_function_type's return type.
23202 (arc_expand_prologue): Likewise.
23203 (arc_expand_epilogue): Likewise.
23204 (arc_return_address_regs): Delete.
23205 (arc_return_address_register): New function.
23206 (arc_epilogue_uses): Use above function.
23207 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
23208 (arc_function_type): Change encoding, add naked type.
23209 (ARC_INTERRUPT_P): Change to handle the new encoding.
23210 (ARC_FAST_INTERRUPT_P): Likewise.
23211 (ARC_NORMAL_P): Define.
23212 (ARC_NAKED_P): Likewise.
23213 (arc_compute_function_type): Delete prototype.
23214 * config/arc/arc.md (in_ret_delay_slot): Use
23215 arc_return_address_register function.
23216 (simple_return): Likewise.
23217 (p_return_i): Likewise.
23219 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23221 PR tree-optimization/81428
23222 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
23223 can't be built for those types.
23225 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23227 Remove stuff dead since r239246.
23229 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
23230 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
23231 (avr_inform_devices): Remove dead stuff.
23233 2017-07-17 Tamar Christina <tamar.christina@arm.com>
23235 * config/arm/arm_neon.h: Fix softp typo.
23237 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23239 PR tree-optimization/81365
23240 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
23241 aggregate moves onto bb predecessor edges, make sure there are no
23242 loads that could alias the lhs in between the start of bb and the
23245 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23248 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
23249 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
23250 [LSHIFTRT, outer_code = TRUNCATE]: Same.
23252 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23254 PR tree-optimization/81396
23255 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
23256 (init_symbolic_number): Initialize it to 1.
23257 (perform_symbolic_merge): Add n_ops from both operands into the new
23259 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
23260 without base_addr as useless if they need more than one operation.
23261 (bswap_replace): Handle !bswap case for NULL base_addr.
23263 2017-07-17 Tom de Vries <tom@codesourcery.com>
23266 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
23269 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
23271 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
23272 conditional builtin define __FIX_LEON3FT_B2BST.
23274 2017-07-17 Daniel Cederman <cederman@gaisler.com>
23276 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
23277 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
23280 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23282 PR rtl-optimization/81424
23283 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
23284 to remove potential trapping from operands if -fnon-call-exceptions.
23286 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23288 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
23289 profile_proability for scalling.
23290 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
23292 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23294 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
23296 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23298 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
23299 fixpoint arithmetics.
23301 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23303 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
23304 fixpoint arithmetics.
23306 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23308 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
23309 fixpoint arithmetics.
23311 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23313 * profile-count.h (profile_probability::from_reg_br_prob_note,
23314 profile_probability::to_reg_br_prob_note): New functions.
23315 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
23316 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
23317 * predict.c (probability_reliable_p): Update.
23318 (edge_probability_reliable_p): Update.
23319 (br_prob_note_reliable_p): Update.
23320 (invert_br_probabilities): Update.
23321 (add_reg_br_prob_note): New function.
23322 (combine_predictions_for_insn): Update.
23323 * asan.c (asan_clear_shadow): Update.
23324 * cfgbuild.c (compute_outgoing_frequencies): Update.
23325 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23326 (update_br_prob_note): Update.
23327 (rtl_verify_edges): Update.
23328 (purge_dead_edges): Update.
23329 (fixup_reorder_chain): Update.
23330 * emit-rtl.c (try_split): Update.
23331 * ifcvt.c (cond_exec_process_insns): Update.
23332 (cond_exec_process_if_block): Update.
23333 (dead_or_predicable): Update.
23334 * internal-fn.c (expand_addsub_overflow): Update.
23335 (expand_neg_overflow): Update.
23336 (expand_mul_overflow): Update.
23337 * loop-doloop.c (doloop_modify): Update.
23338 * loop-unroll.c (compare_and_jump_seq): Update.
23339 * optabs.c (emit_cmp_and_jump_insn_1): Update.
23340 * predict.h: Update.
23341 * reorg.c (mostly_true_jump): Update.
23343 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
23344 * config/alpha/alpha.c (emit_unlikely_jump): Update.
23345 * config/arc/arc.c: (emit_unlikely_jump): Update.
23346 * config/arm/arm.c: (emit_unlikely_jump): Update.
23347 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
23348 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
23349 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
23350 (ix86_print_operand): Update.
23351 (ix86_split_fp_branch): Update.
23352 (predict_jump): Update.
23353 * config/ia64/ia64.c (ia64_print_operand): Update.
23354 * config/mmix/mmix.c (mmix_print_operand): Update.
23355 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
23356 (rs6000_expand_split_stack_prologue): Update.
23357 * config/rs6000/rs6000.c: Update.
23358 * config/s390/s390.c (s390_expand_vec_strlen): Update.
23359 (s390_expand_vec_movstr): Update.
23360 (s390_expand_cs_tdsi): Update.
23361 (s390_expand_split_stack_prologue): Update.
23362 * config/sh/sh.c (sh_print_operand): Update.
23363 (expand_cbranchsi4): Update.
23364 (expand_cbranchdi4): Update.
23365 * config/sparc/sparc.c (output_v9branch): Update.
23366 * config/spu/spu.c (get_branch_target): Update.
23367 (ea_load_store_inline): Update.
23368 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
23369 * config/tilepro/tilepro.c: Update.
23371 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23373 * gimplify.c (mostly_copy_tree_r): Revert latest change.
23374 (gimplify_save_expr): Likewise.
23376 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23378 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
23380 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23382 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
23384 * timevar.def (TV_IPA_FNSUMMARY): Define.
23386 2017-07-16 Daniel Cederman <cederman@gaisler.com>
23388 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
23389 to back store errata sensitive sequence from being generated.
23390 (sqrtdf2_fix): Likewise.
23392 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23394 * tree-ssa-threadupdate.c (compute_path_counts,
23395 update_joiner_offpath_counts): Use profile_probability.
23397 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
23400 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23402 * config/arm/arm-c.c (arm_cpu_builtins): Define
23403 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23405 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23407 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23408 array entries to represent __ieee128 versions of the
23409 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
23410 scalar_extract_sig, and scalar_insert_exp built-in functions.
23411 (altivec_resolve_overloaded_builtin): Add special case handling
23412 for the __builtin_scalar_insert_exp function, as represented by
23413 the P9V_BUILTIN_VEC_VSIEDP constant.
23414 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
23415 exponent support for __ieee128 argument.
23416 (VSESQP): Add scalar extract signature support for __ieee128
23418 (VSTDCNQP): Add scalar test negative support for __ieee128
23420 (VSIEQP): Add scalar insert exponent support for __int128 argument
23421 with __ieee128 result.
23422 (VSIEQPF): Add scalar insert exponent support for __ieee128
23423 argument with __ieee128 result.
23424 (VSTDCQP): Add scalar test data class support for __ieee128
23426 (VSTDCNQP): Add overload support for scalar test negative with
23427 __ieee128 argument.
23428 (VSTDCQP): Add overload support for scalar test data class
23429 __ieee128 argument.
23430 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
23431 UNSPEC_VSX_SXSIGDP.
23432 (UNSPEC_VSX_SIEXPQP): New constant.
23433 (xsxexpqp): New insn for VSX scalar extract exponent quad
23435 (xsxsigqp): New insn for VSX scalar extract significand quad
23437 (xsiexpqpf): New insn for VSX scalar insert exponent quad
23438 precision with floating point argument.
23439 (xststdcqp): New expand for VSX scalar test data class quad
23441 (xststdcnegqp): New expand for VSX scalar test negative quad
23443 (xststdcqp): New insn to match expansions for VSX scalar test data
23444 class quad precision and VSX scalar test negative quad precision.
23445 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
23446 special case operand checking to enforce that second operand of
23447 VSX scalar test data class with quad precision argument is a 7-bit
23449 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23450 prototypes and descriptions of __ieee128 versions of
23451 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
23452 scalar_test_data_class, and scalar_test_neg built-in functions.
23454 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23456 PR tree-optimization/81162
23457 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
23458 replace a negate with an add.
23460 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
23462 * doc/invoke.texi (arm/-mcpu): Document +crypto.
23464 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23466 * config/arm/arm-c.c (arm_cpu_builtins): Define
23467 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23469 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23471 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
23472 (armv8-r): Set ARM Cortex-R52 as default CPU.
23473 * config/arm/arm-tables.opt: Regenerate.
23474 * config/arm/arm-tune.md: Regenerate.
23475 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23477 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
23478 extension for -mcpu=cortex-r52.
23480 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23482 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
23483 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
23484 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
23485 (fp-armv8): Define it as FP_ARMv8 only.
23486 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
23487 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
23489 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
23490 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
23491 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
23492 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
23493 than TARGET_FPU_ARMV8.
23494 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
23495 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
23496 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
23498 * config/arm/neon.md (neon_vrint): Likewise.
23499 (neon_vcvt): Likewise.
23500 (neon_<fmaxmin_op><mode>): Likewise.
23501 (<fmaxmin><mode>3): Likewise.
23502 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
23503 * config/arm/predicates.md (arm_cond_move_operator): Check against
23504 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
23506 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
23508 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
23509 to top of function.
23511 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23513 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
23514 loop in comment with memset.
23516 2017-07-14 Martin Liska <mliska@suse.cz>
23518 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
23519 * dwarf2out.c (is_java): Remove the function.
23520 (output_pubname): Remove usage of the function.
23521 (lower_bound_default): Remove usage of DW_LANG_Java.
23522 (gen_compile_unit_die): Likewise.
23523 * gcc.c: Remove compiler defaults for .java and .zip files.
23524 * gimple-expr.c (remove_suffix): Change as there's no longer
23525 extension than 4-letter one.
23526 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
23527 (gimplify_save_expr): Likewise.
23528 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
23529 as it's possible even for other languages than Java.
23530 * langhooks.h (struct lang_hooks): Remove Java from a comment.
23531 * lto-opts.c (lto_write_options): Remove reference to Java.
23532 * opts.c (strip_off_ending): Update file extension handling.
23533 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
23534 * tree-eh.c (lower_resx): Likewise.
23535 * tree.c (free_lang_data_in_type): Remove dead code.
23536 (find_decls_types_r): Likewise.
23537 (build_common_builtin_nodes): Remove Java from a comment.
23538 (verify_type): Remove dead code.
23539 * varasm.c (assemble_external): Remove Java from a comment.
23541 2017-07-14 Martin Liska <mliska@suse.cz>
23543 * opts.c (finish_options): Add quotes.
23544 (common_handle_option): Likewise.
23546 2017-07-14 Martin Liska <mliska@suse.cz>
23548 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
23549 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
23550 Remove N_SO_PASCAL.
23551 * dwarf2out.c (lower_bound_default): Do not handle
23553 (gen_compile_unit_die): Likewise.
23554 * gcc.c: Remove default extension binding for GNU Pascal.
23555 * stmt.c: Remove Pascal language from a comment.
23556 * xcoffout.c: Likewise.
23558 2017-07-13 David Malcolm <dmalcolm@redhat.com>
23561 * diagnostic-show-locus.c (fixit_cmp): New function.
23562 (layout::layout): Sort m_fixit_hints.
23563 (column_range::column_range): Assert that the values are valid.
23564 (struct char_span): New struct.
23565 (correction::overwrite): New method.
23566 (struct source_line): New struct.
23567 (line_corrections::add_hint): Add assertions. Reimplement memcpy
23568 calls in terms of classes source_line and char_span, and
23569 correction::overwrite.
23570 (selftest::test_overlapped_fixit_printing_2): New function.
23571 (selftest::diagnostic_show_locus_c_tests): Call it.
23573 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
23575 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
23576 early if there is no lhs.
23578 2017-07-13 Martin Liska <mliska@suse.cz>
23580 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
23581 (gen_reference_type_die): Likewise.
23582 * stor-layout.c: Remove Pascal-related comment.
23584 2017-07-13 Martin Liska <mliska@suse.cz>
23586 * opts.c (finish_options): Add quotes to error messages.
23587 (parse_sanitizer_options): Likewise.
23589 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
23591 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
23593 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
23595 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
23597 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23599 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
23601 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
23603 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
23606 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
23607 provides the hardware capability bits, define the macro
23608 __BUILTIN_CPU_SUPPORTS__.
23609 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
23610 if GLIBC does not provide the hardware capability bits. Add a
23611 gcc_unreachable call if the built-in cpu function is neither
23612 __builtin_cpu_is nor __builtin_cpu_supports.
23613 (rs6000_get_function_versions_dispatcher): Change the warning
23614 that an old GLIBC is used which does not export the capability
23615 bits to be an error.
23616 * doc/extend.texi (target_clones attribute): Document the
23617 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
23618 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
23619 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
23620 the macros defined by GCC if the newer GLIBC is available.
23622 2017-07-12 Jeff Law <law@redhat.com>
23624 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
23625 remaining includes slightly.
23626 * config/riscv/riscv-builtins.c: Include profile-count.h.
23628 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23631 * config/avr/avr.c (avr_set_current_function): In diagnostic
23632 messages: Quote keywords and (parts of) identifiers.
23633 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
23636 2017-07-12 Carl Love <cel@us.ibm.com>
23638 * config/rs6000/rs6000-c.c: Add support for built-in functions
23639 vector bool char vec_revb (vector bool char);
23640 vector bool short vec_revb (vector short char);
23641 vector bool int vec_revb (vector bool int);
23642 vector bool long long vec_revb (vector bool long long);
23643 * doc/extend.texi: Update the built-in documentation file for the
23644 new built-in functions.
23646 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23648 * config/s390/s390.md: Remove movcc splitter.
23650 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23652 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
23653 load/store on condition.
23655 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23658 * config/avr/avr.c (avr_encode_section_info)
23659 [progmem && !TREE_READONLY]: Error if progmem object needs
23662 2017-07-11 Michael Collison <michael.collison@arm.com>
23664 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
23667 2017-07-11 Carl Love <cel@us.ibm.com>
23669 * config/rs6000/rs6000-c.c: Add support for builtins
23670 vector unsigned int vec_parity_lsbb (vector signed int);
23671 vector unsigned int vec_parity_lsbb (vector unsigned int);
23672 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
23673 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
23674 vector unsigned long long vec_parity_lsbb (vector signed long long);
23675 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
23676 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
23677 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
23678 * doc/extend.texi: Update the built-in documentation file for the
23679 new built-in functions.
23681 2017-07-11 David Malcolm <dmalcolm@redhat.com>
23683 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
23684 (layout::m_primary_loc): New field.
23685 (layout::layout): Initialize new field. Move location filtering
23686 logic from here to...
23687 (layout::maybe_add_location_range): ...this new method. Add
23688 support for filtering to just the lines already specified by other
23690 (layout::will_show_line_p): New method.
23691 (gcc_rich_location::add_location_if_nearby): New method.
23692 (selftest::test_add_location_if_nearby): New test function.
23693 (selftest::diagnostic_show_locus_c_tests): Call it.
23694 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
23697 2017-07-11 Tom de Vries <tom@codesourcery.com>
23699 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
23700 (bb_first_real_insn): New function.
23701 (nvptx_single): Add extra initialization of broadcasted condition
23704 2017-07-11 Nathan Sidwell <nathan@acm.org>
23706 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
23708 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
23710 * doc/extend.texi (AVR Function Attributes): Remove weblink to
23711 Binutils doc as TEXI will mess them up.
23712 * doc/invoke.texi (AVR Options): Same here.
23714 2017-07-11 Daniel Cederman <cederman@gaisler.com>
23716 * config/sparc/sparc.opt (mfix-ut700): New option.
23717 (mfix-gr712rc): Likewise.
23718 (sparc_fix_b2bst): New variable.
23719 * doc/invoke.texi (SPARC options): Document them.
23720 (ARM options): Fix warnings.
23721 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
23722 instructions to prevent sequences that can trigger the store-store
23723 errata for certain LEON3FT processors.
23724 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
23725 (sparc_option_override): Set sparc_fix_b2bst appropriately.
23726 * config/sparc/sparc.md (fix_b2bst): New attribute.
23727 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
23729 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
23732 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
23734 (*rsqrtsf2_sse): Ditto.
23736 (div<mode>3): Macroize insn from divdf3 and divsf3
23737 using MODEF mode iterator.
23739 2017-07-10 Martin Sebor <msebor@redhat.com>
23741 PR tree-optimization/80397
23742 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
23743 instead of testing for equality to INTEGER_TYPE.
23745 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
23747 * config.gcc: Remove uclibc from arc target spec.
23749 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
23751 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
23753 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23756 * lto-wrapper.c (remove_option): New function.
23757 (merge_and_complain): Merge PIC/PIE options more realistically.
23759 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23761 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
23765 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
23766 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
23767 * config.in: Regenerate.
23768 * configure: Regenerate.
23769 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
23770 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
23771 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
23772 (TARGET_GASISR_PROLOGUES): ...target mask.
23773 * common/config/avr/avr-common.c
23774 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
23775 Set -mgas-isr-prologues.
23776 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
23777 INSERT_PASS_BEFORE for it.
23778 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
23779 * config/avr/avr.c (avr_option_override)
23780 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
23781 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
23782 (avr_attribute_table) <no_gccisr>: Add new function attribute.
23783 (avr_set_current_function) <is_no_gccisr>: Init machine field.
23784 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
23786 (make_avr_pass_pre_proep): New function.
23787 (emit_push_sfr) <treg>: Add argument to function and use it
23788 instead of TMP_REG.
23789 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
23790 and set machine->gasisr.yes.
23791 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
23792 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
23793 __gcc_isr.n_pushed to .L__stack_usage.
23794 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
23795 (avr_asm_final_postscan_insn): ...this new static function.
23796 * config/avr/avr.h (machine_function)
23797 <is_no_gccisr, use_L__stack_usage>: New fields.
23798 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
23799 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
23800 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
23801 (gasisr, *gasisr): New expander and insn.
23802 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23803 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
23804 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
23806 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
23808 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
23811 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23813 Move jump-tables out of .text again.
23816 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
23817 (ASM_OUTPUT_ADDR_VEC): New function.
23818 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
23819 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
23820 INSN_ADDRESSes as asm comment.
23821 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
23822 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
23823 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
23824 * config/avr/avr.md (*tablejump): Adjust comment.
23825 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
23826 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
23828 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
23829 (avr_output_addr_vec): New proto.
23830 (avr_log_t) <insn_addresses>: New field.
23832 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
23835 * config/i386/i386.c (ix86_function_arg_advance): Set
23836 outgoing_args_on_stack to true if there are outgoing arguments
23838 (ix86_function_arg): Likewise.
23839 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
23840 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
23841 * config/i386/i386.h (machine_function): Add
23842 outgoing_args_on_stack.
23844 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
23846 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23847 supporting pthreds.
23848 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
23850 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
23852 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
23853 (REAL_H): Remove $(MACHMODE_H).
23854 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
23856 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
23857 $(MACHMODE_H) and double-int.h.
23858 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
23860 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
23863 2017-07-07 Andrew Pinski <apinski@cavium.com>
23865 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
23866 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
23868 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23870 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
23871 Add warning if GCC was not configured to link against a GLIBC that
23872 exports the hardware capability bits.
23873 (make_resolver_func): Make resolver function private and not a
23874 COMDAT function. Create the name with clone_function_name instead
23875 of make_unique_name.
23878 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
23879 correct operand in doing the split.
23881 2017-07-07 Carl Love <cel@us.ibm.com>
23883 * config/rs6000/rs6000-c: Add support for built-in function
23884 vector unsigned short vec_pack_to_short_fp32 (vector float,
23886 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
23887 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
23888 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
23889 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
23890 (convert_4f32_8i16): Add define_expand.
23891 * doc/extend.texi: Update the built-in documentation file for the
23892 new built-in function.
23894 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23896 * config/sparc/m8.md: New file.
23897 * config/sparc/sparc.md: Include m8.md.
23899 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23901 * config/sparc/sparc.opt: New option -mvis4b.
23902 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
23903 (sparc_option_override): Handle VIS4B.
23904 (enum sparc_builtins): Define
23905 SPARC_BUILTIN_DICTUNPACK{8,16,32},
23906 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
23907 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
23908 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
23909 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
23910 (check_constant_argument): New function.
23911 (sparc_vis_init_builtins): Define builtins
23912 __builtin_vis_dictunpack{8,16,32},
23913 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
23914 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
23915 __builtin_vis_fpcmpde{8,16,32}shl and
23916 __builtin_vis_fpcmpur{8,16,32}shl.
23917 (sparc_expand_builtin): Check that the constant operands to
23918 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
23919 constant and in range.
23920 * config/sparc/sparc-c.c (sparc_target_macros): Handle
23922 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
23923 (SPARC_IMM5_P): Likewise.
23924 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
23925 (enabled): Handle vis4b.
23926 (UNSPEC_DICTUNPACK): New unspec.
23927 (UNSPEC_FPCMPSHL): Likewise.
23928 (UNSPEC_FPUCMPSHL): Likewise.
23929 (UNSPEC_FPCMPDESHL): Likewise.
23930 (UNSPEC_FPCMPURSHL): Likewise.
23931 (cpu_feature): New CPU feature `vis4b'.
23932 (dictunpack{8,16,32}): New insns.
23933 (FPCSMODE): New mode iterator.
23934 (fpcscond): New code iterator.
23935 (fpcsucond): Likewise.
23936 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
23937 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
23938 (fpcmpde{8,16,32}{si,di}shl): Likewise.
23939 (fpcmpur{8,16,32}{si,di}shl): Likewise.
23940 * config/sparc/constraints.md: Define constraints `q' for unsigned
23941 2-bit integer constants and `t' for unsigned 5-bit integer
23943 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
23945 (imm5_operand_dictunpack16): Likewise.
23946 (imm5_operand_dictunpack32): Likewise.
23947 (imm2_operand): Likewise.
23948 * doc/invoke.texi (SPARC Options): Document -mvis4b.
23949 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
23950 ditunpack* and fpcmp*shl builtins.
23952 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23954 * config.gcc: Handle m8 in --with-{cpu,tune} options.
23955 * config.in: Add HAVE_AS_SPARC6 define.
23956 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
23958 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
23960 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
23961 (CPP_CPU_SPEC): Handle m8.
23962 (ASM_CPU_SPEC): Likewise.
23963 * config/sparc/sparc-opts.h (enum processor_type): Add
23965 * config/sparc/sparc.c (m8_costs): New struct.
23966 (sparc_option_override): Handle TARGET_CPU_m8.
23967 (sparc32_initialize_trampoline): Likewise.
23968 (sparc64_initialize_trampoline): Likewise.
23969 (sparc_issue_rate): Likewise.
23970 (sparc_register_move_cost): Likewise.
23971 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
23972 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
23973 (ASM_CPU64_DEFAULT_SPEC): Likewise.
23974 (CPP_CPU_SPEC): Handle M8.
23975 (ASM_CPU_SPEC): Likewise.
23976 (AS_M8_FLAG): Define.
23977 * config/sparc/sparc.md: Add m8 to the cpu attribute.
23978 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
23979 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
23981 * configure: Regenerate.
23982 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
23985 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23987 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
23989 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
23990 ("*movdi_insn_sp32"): Do not set v3pipe.
23991 ("*movsi_insn"): Likewise.
23992 ("*movdi_insn_sp64"): Likewise.
23993 ("*movsf_insn"): Likewise.
23994 ("*movdf_insn_sp32"): Likewise.
23995 ("*movdf_insn_sp64"): Likewise.
23996 ("*zero_extendsidi2_insn_sp64"): Likewise.
23997 ("*sign_extendsidi2_insn"): Likewise.
23998 ("*mov<VM32:mode>_insn"): Likewise.
23999 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24000 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24001 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24002 ("<vlop:code><VL:mode>3"): Likewise.
24003 ("*not_<vlop:code><VL:mode>3"): Likewise.
24004 ("*nand<VL:mode>_vis"): Likewise.
24005 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
24006 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
24007 ("one_cmpl<VL:mode>2"): Likewise.
24008 ("faligndata<VM64:mode>_vis"): Likewise.
24009 ("alignaddrsi_vis"): Likewise.
24010 ("alignaddrdi_vis"): Likweise.
24011 ("alignaddrlsi_vis"): Likewise.
24012 ("alignaddrldi_vis"): Likewise.
24013 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24014 ("bmaskdi_vis"): Likewise.
24015 ("bmasksi_vis"): Likewise.
24016 ("bshuffle<VM64:mode>_vis"): Likewise.
24017 ("cmask8<P:mode>_vis"): Likewise.
24018 ("cmask16<P:mode>_vis"): Likewise.
24019 ("cmask32<P:mode>_vis"): Likewise.
24020 ("pdistn<P:mode>_vis"): Likewise.
24021 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24023 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24025 * config/sparc/sparc.md ("subtype"): New insn attribute.
24026 ("*wrgsr_sp64"): Set insn subtype.
24027 ("*rdgsr_sp64"): Likewise.
24028 ("alignaddrsi_vis"): Likewise.
24029 ("alignaddrdi_vis"): Likewise.
24030 ("alignaddrlsi_vis"): Likewise.
24031 ("alignaddrldi_vis"): Likewise.
24032 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24033 ("fexpand_vis"): Likewise.
24034 ("fpmerge_vis"): Likewise.
24035 ("faligndata<VM64:mode>_vis"): Likewise.
24036 ("bshuffle<VM64:mode>_vis"): Likewise.
24037 ("cmask8<P:mode>_vis"): Likewise.
24038 ("cmask16<P:mode>_vis"): Likewise.
24039 ("cmask32<P:mode>_vis"): Likewise.
24040 ("fchksm16_vis"): Likewise.
24041 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
24042 ("fmean16_vis"): Likewise.
24043 ("fp<plusminus_insn>64_vis"): Likewise.
24044 ("<plusminus_insn>v8qi3"): Likewise.
24045 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24046 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
24047 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
24048 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
24049 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
24050 ("*movqi_insn"): Likewise.
24051 ("*movhi_insn"): Likewise.
24052 ("*movsi_insn"): Likewise.
24053 ("movsi_pic_gotdata_op"): Likewise.
24054 ("*movdi_insn_sp32"): Likewise.
24055 ("*movdi_insn_sp64"): Likewise.
24056 ("movdi_pic_gotdata_op"): Likewise.
24057 ("*movsf_insn"): Likewise.
24058 ("*movdf_insn_sp32"): Likewise.
24059 ("*movdf_insn_sp64"): Likewise.
24060 ("*zero_extendhisi2_insn"): Likewise.
24061 ("*zero_extendqihi2_insn"): Likewise.
24062 ("*zero_extendqisi2_insn"): Likewise.
24063 ("*zero_extendqidi2_insn"): Likewise.
24064 ("*zero_extendhidi2_insn"): Likewise.
24065 ("*zero_extendsidi2_insn_sp64"): Likewise.
24066 ("ldfsr"): Likewise.
24067 ("prefetch_64"): Likewise.
24068 ("prefetch_32"): Likewise.
24069 ("tie_ld32"): Likewise.
24070 ("tie_ld64"): Likewise.
24071 ("*tldo_ldub_sp32"): Likewise.
24072 ("*tldo_ldub1_sp32"): Likewise.
24073 ("*tldo_ldub2_sp32"): Likewise.
24074 ("*tldo_ldub_sp64"): Likewise.
24075 ("*tldo_ldub1_sp64"): Likewise.
24076 ("*tldo_ldub2_sp64"): Likewise.
24077 ("*tldo_ldub3_sp64"): Likewise.
24078 ("*tldo_lduh_sp32"): Likewise.
24079 ("*tldo_lduh1_sp32"): Likewise.
24080 ("*tldo_lduh_sp64"): Likewise.
24081 ("*tldo_lduh1_sp64"): Likewise.
24082 ("*tldo_lduh2_sp64"): Likewise.
24083 ("*tldo_lduw_sp32"): Likewise.
24084 ("*tldo_lduw_sp64"): Likewise.
24085 ("*tldo_lduw1_sp64"): Likewise.
24086 ("*tldo_ldx_sp64"): Likewise.
24087 ("*mov<VM32:mode>_insn"): Likewise.
24088 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24089 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24091 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24093 * config/sparc/sparc.md ("type"): New insn type viscmp.
24094 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
24096 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
24097 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
24098 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24099 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
24101 ("n7_vis_logical_11cycle"): Likewise.
24102 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
24103 * config/sparc/niagara2.md ("niag3_vis": Likewise.
24104 * config/sparc/niagara.md ("niag_vis"): Likewise.
24105 * config/sparc/ultra3.md ("us3_fga"): Likewise.
24106 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
24108 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24110 * config/sparc/sparc.md: New instruction type `bmask'.
24111 (bmaskdi_vis): Use the `bmask' type.
24112 (bmasksi_vis): Likewise.
24113 * config/sparc/ultra3.md (us3_array): Likewise.
24114 * config/sparc/niagara7.md (n7_array): Likewise.
24115 * config/sparc/niagara4.md (n4_array): Likewise.
24116 * config/sparc/niagara2.md (niag2_vis): Likewise.
24117 (niag3_vis): Likewise.
24118 * config/sparc/niagara.md (niag_vis): Likewise.
24120 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24122 * ipa-comdats.c: Remove optimize check from gate.
24123 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
24124 for functions not optimized.
24125 (ipa_fn_summary_read): Skip optimize check.
24126 (ipa_fn_summary_write): Likewise.
24127 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
24129 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
24131 (can_inline_edge_p): Check flag_pcc_struct_return for match.
24132 (check_callers): Give up on caller which is not optimized.
24133 (inline_small_functions): Likewise.
24134 (ipa_inline): Do not give up when not optimizing.
24135 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
24136 away unoptimizes cdtors.
24137 (whole_program_function_and_variable_visibility): Do
24138 ipa_discover_readonly_nonaddressable_vars in LTO mode.
24139 * ipa.c (process_references): Do not check optimize.
24140 (symbol_table::remove_unreachable_nodes): Update optimize check.
24141 (set_writeonly_bit): Update optimize check.
24142 (pass_ipa_cdtor_merge::gate): Do not check optimize.
24143 (pass_ipa_single_use::gate): Remove.
24145 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24147 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
24148 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
24149 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
24150 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
24151 permute_load, permute_store, adjust_extract, adjust_splat,
24152 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
24153 replace_swap_with_copy, dump_swap_insn_table,
24154 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
24155 recombine_lvx_pattern, recombine_stvx_pattern,
24156 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
24157 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
24158 to file rs6000-p8swap.c.
24159 * config/rs6000/rs6000-p8swap.c: New file.
24160 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
24161 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
24162 and rs6000*-*-* targets.
24164 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24166 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
24168 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24170 * lto-wrapper.c (merge_and_complain): Do not merge
24171 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
24172 fsigned_zeros, ftrapping_math, fwrapv.
24173 (append_compiler_options): Do not track these options.
24174 (append_linker_options): Likewie
24176 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24178 * cgraphunit.c (cgraph_node::finalize_function): When
24179 !flag_toplevel_reorde set no_reorder flag.
24180 (varpool_node::finalize_decl): Likewise.
24181 (symbol_table::compile): Drop no toplevel reorder path.
24183 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24185 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
24186 edges; zero probability is not better than uninitialized.
24188 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24190 * asan.h (asan_sanitize_allocas_p): Declare.
24191 * asan.c (asan_sanitize_allocas_p): New function.
24192 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
24193 (handle_builtin_alloca): Likewise.
24194 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
24195 if !asan_sanitize_allocas_p.
24196 * params.def (asan-instrument-allocas): Add new option.
24197 * params.h (ASAN_PROTECT_ALLOCAS): Define.
24198 * opts.c (common_handle_option): Disable allocas sanitization for
24201 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24203 * asan.c: Include gimple-fold.h.
24204 (get_last_alloca_addr): New function.
24205 (handle_builtin_stackrestore): Likewise.
24206 (handle_builtin_alloca): Likewise.
24207 (asan_emit_allocas_unpoison): Likewise.
24208 (get_mem_refs_of_builtin_call): Add new parameter, remove const
24209 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
24210 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
24211 (instrument_builtin_call): Pass gimple iterator to
24212 get_mem_refs_of_builtin_call.
24213 (last_alloca_addr): New global.
24214 * asan.h (asan_emit_allocas_unpoison): Declare.
24215 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
24216 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
24217 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
24218 if function calls alloca.
24219 * gimple-fold.c (replace_call_with_value): Remove static keyword.
24220 * gimple-fold.h (replace_call_with_value): Declare.
24221 * internal-fn.c: Include asan.h.
24222 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
24223 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
24225 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24227 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
24228 (C_SELFTEST_FLAGS): New.
24229 (CPP_SELFTEST_FLAGS): New.
24230 (SELFTEST_DEPS): New, from deps of s-selftest.
24231 (C_SELFTEST_DEPS): New, from deps of s-selftest.
24232 (CPP_SELFTEST_DEPS): New.
24233 (selftest): Add dependency on s-selftest-c++.
24234 (s-selftest): Rename to...
24235 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
24236 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
24237 than SELFTEST_FLAGS.
24238 (selftest-gdb): Rename to...
24239 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
24241 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
24242 (selftest-valgrind): Rename to...
24243 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
24245 (selftest-valgrind): Reintroduce as an alias for
24246 selftest-c-valgrind.
24247 (s-selftest-c++): New.
24248 (selftest-c++-gdb): New.
24249 (selftest-c++-valgrind): New.
24251 2017-07-06 Olivier Hainque <hainque@adacore.com>
24253 * gcc.c (process_command): When deciding if undefined variables
24254 should be ignored when processing specs, accept "gcc -v" as well.
24256 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24258 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
24259 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
24261 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24263 * config/arm/arm-cpus.in (armv8-r): Add new entry.
24264 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
24265 * config/arm/arm-tables.opt: Regenerate.
24266 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
24268 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
24270 2017-07-06 Carl Love <cel@us.ibm.com>
24272 * ChangeLog: Clean up from mid air collision
24274 2017-07-06 Carl Love <cel@us.ibm.com>
24276 * config/rs6000/rs6000-c.c: Add support for built-in functions
24277 vector signed int vec_subc (vector signed int, vector signed int);
24278 vector signed __int128 vec_subc (vector signed __int128,
24279 vector signed __int128);
24280 vector unsigned __int128 vec_subc (vector unsigned __int128,
24281 vector unsigned __int128);
24282 vector signed int vec_sube (vector signed int, vector signed int,
24283 vector signed int);
24284 vector unsigned int vec_sube (vector unsigned int,
24285 vector unsigned int,
24286 vector unsigned int);
24287 vector signed __int128 vec_sube (vector signed __int128,
24288 vector signed __int128,
24289 vector signed__int128);
24290 vector unsigned __int128 vec_sube (vector unsigned __int128,
24291 vector unsigned __int128,
24292 vector unsigned __int128);
24293 vector signed int vec_subec (vector signed int, vector signed int,
24294 vector signed int);
24295 vector unsigned int vec_subec (vector unsigned int,
24296 vector unsigned int,
24297 vector unsigned int);
24298 vector signed __int128 vec_subec (vector signed __int128,
24299 vector signed __int128,
24300 vector signed__int128);
24301 vector unsigned __int128 vec_subec (vector unsigned __int128,
24302 vector unsigned __int128,
24303 vector unsigned __int128);
24304 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
24305 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
24306 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
24307 BU_ALTIVEC_OVERLOAD_X definitions.
24308 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
24309 * doc/extend.texi: Update the built-in documentation file for the new
24310 built-in functions.
24312 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24315 * diagnostic-show-locus.c (layout::layout): Use start and finish
24316 spelling location for the start and finish of each range.
24317 * genmatch.c (linemap_client_expand_location_to_spelling_point):
24318 Add unused aspect param.
24319 * input.c (expand_location_1): Add "aspect" param, and use it
24320 to access the correct part of the location.
24321 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
24323 (expand_location_to_spelling_point): Likewise.
24324 (linemap_client_expand_location_to_spelling_point): Add "aspect"
24325 param, and pass it to expand_location_1.
24327 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
24329 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
24330 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
24331 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
24332 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
24333 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
24334 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
24335 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
24336 _mm_maskz_getmant_ss): New intrinsics.
24337 (__builtin_ia32_getexpss128_mask): Changed to ...
24338 __builtin_ia32_getexpss128_round ... this.
24339 (__builtin_ia32_getexpsd128_mask): Changed to ...
24340 __builtin_ia32_getexpsd128_round ... this.
24341 * config/i386/i386-builtin-types.def
24342 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
24343 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
24344 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
24345 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
24346 __builtin_ia32_getmantss_mask_round): New builtins.
24347 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
24348 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
24349 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
24350 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
24351 * config/i386/sse.md
24352 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
24353 avx512f_sgetexp<mode><mask_scalar_name>
24354 <round_saeonly_scalar_name> ... this.
24355 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
24356 %0, %1, %2<round_saeonly_op3>}): Changed to ...
24357 vgetexp<ssescalarmodesuffix>
24358 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24359 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
24360 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
24361 avx512f_vgetmant<mode><mask_scalar_name>
24362 <round_saeonly_scalar_name> ... this.
24363 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24364 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
24365 vgetmant<ssescalarmodesuffix>
24366 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
24367 %0<mask_scalar_operand4>, %1, %2
24368 <round_saeonly_scalar_mask_op4>, %3} ... this.
24369 * config/i386/subst.md (mask_scalar_operand4,
24370 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
24371 round_saeonly_scalar_nimm_predicate): New subst attributes.
24373 2017-07-06 Julia Koval <julia.koval@intel.com>
24375 * config/i386/i386.c (ix86_erase_embedded_rounding):
24376 Remove code for old rounding pattern.
24378 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
24380 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
24382 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
24384 * doc/sourcebuild.texi (Test Directives, Variants of
24385 dg-require-support): Add documentation for dg-require-stack-check.
24387 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
24389 * config/i386/subst.md (mask_scalar, round_scalar,
24390 round_saeonly_scalar): New meta-templates.
24391 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
24392 round_scalar_mask_operand3, round_scalar_mask_op3,
24393 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
24394 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
24395 round_saeonly_scalar_constraint,
24396 round_saeonly_scalar_prefix): New subst attribute.
24397 * config/i386/sse.md
24398 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
24399 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
24400 <round_scalar_name> ... this.
24401 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
24402 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
24403 <round_scalar_name> ... this.
24404 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
24405 <sse>_vm<code><mode>3<mask_scalar_name>
24406 <round_saeonly_scalar_name> ... this.
24407 (v<plusminus_mnemonic><ssescalarmodesuffix>
24408 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24409 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24410 v<plusminus_mnemonic><ssescalarmodesuffix>
24411 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24412 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24413 (v<multdiv_mnemonic><ssescalarmodesuffix>
24414 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24415 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24416 v<multdiv_mnemonic><ssescalarmodesuffix>
24417 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24418 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24419 (v<maxmin_float><ssescalarmodesuffix>
24420 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
24421 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
24422 v<maxmin_float><ssescalarmodesuffix>
24423 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24424 %0<mask_scalar_operand3>, %1, %<iptr>2
24425 <round_saeonly_scalar_mask_op3>} ... this.
24427 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
24429 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
24430 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
24432 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
24433 Alan Hayward <alan.hayward@arm.com>
24434 David Sherwood <david.sherwood@arm.com>
24436 * combine.c (simplify_if_then_else): Remove "enum" before
24438 * compare-elim.c (can_eliminate_compare): Likewise.
24439 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
24441 (aarch64_lookup_simd_builtin_type): Likewise.
24442 (aarch64_simd_builtin_type): Likewise.
24443 (aarch64_init_simd_builtin_types): Likewise.
24444 (aarch64_simd_expand_args): Likewise.
24445 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
24447 (aarch64_reverse_mask): Likewise.
24448 (aarch64_simd_emit_reg_reg_move): Likewise.
24449 (aarch64_gen_adjusted_ldpstp): Likewise.
24450 (aarch64_ccmp_mode_to_code): Likewise.
24451 (aarch64_operands_ok_for_ldpstp): Likewise.
24452 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24453 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
24455 (aarch64_min_divisions_for_recip_mul): Likewise.
24456 (aarch64_reassociation_width): Likewise.
24457 (aarch64_get_condition_code_1): Likewise.
24458 (aarch64_simd_emit_reg_reg_move): Likewise.
24459 (aarch64_simd_attr_length_rglist): Likewise.
24460 (aarch64_reverse_mask): Likewise.
24461 (aarch64_operands_ok_for_ldpstp): Likewise.
24462 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24463 (aarch64_gen_adjusted_ldpstp): Likewise.
24464 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
24466 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
24467 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
24468 (arm_lookup_simd_builtin_type): Likewise.
24469 (arm_simd_builtin_type): Likewise.
24470 (arm_init_simd_builtin_types): Likewise.
24471 (arm_expand_builtin_args): Likewise.
24472 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
24473 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
24474 (ft32_setup_incoming_varargs): Likewise.
24475 (ft32_function_arg): Likewise.
24476 (ft32_function_arg_advance): Likewise.
24477 (ft32_pass_by_reference): Likewise.
24478 (ft32_arg_partial_bytes): Likewise.
24479 (ft32_valid_pointer_mode): Likewise.
24480 (ft32_addr_space_pointer_mode): Likewise.
24481 (ft32_addr_space_legitimate_address_p): Likewise.
24482 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
24484 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
24485 (ix86_emit_outlined_ms2sysv_restore): Likewise.
24486 (iamcu_alignment): Likewise.
24487 (canonicalize_vector_int_perm): Likewise.
24488 (ix86_noce_conversion_profitable_p): Likewise.
24489 (ix86_mpx_bound_mode): Likewise.
24490 (ix86_operands_ok_for_move_multiple): Likewise.
24491 * config/microblaze/microblaze-protos.h
24492 (microblaze_expand_conditional_branch_reg): Likewise.
24493 * config/microblaze/microblaze.c
24494 (microblaze_expand_conditional_branch_reg): Likewise.
24495 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
24497 (rs6000_reassociation_width): Likewise.
24498 (rs6000_invalid_binary_op): Likewise.
24499 (fusion_p9_p): Likewise.
24500 (emit_fusion_p9_load): Likewise.
24501 (emit_fusion_p9_store): Likewise.
24502 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
24504 (riscv_hard_regno_mode_ok_p): Likewise.
24505 (riscv_address_insns): Likewise.
24506 (riscv_split_symbol): Likewise.
24507 (riscv_legitimize_move): Likewise.
24508 (riscv_function_value): Likewise.
24509 (riscv_hard_regno_nregs): Likewise.
24510 (riscv_expand_builtin): Likewise.
24511 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
24512 (riscv_build_integer): Likewise.
24513 (riscv_split_integer): Likewise.
24514 (riscv_legitimate_constant_p): Likewise.
24515 (riscv_cannot_force_const_mem): Likewise.
24516 (riscv_regno_mode_ok_for_base_p): Likewise.
24517 (riscv_valid_base_register_p): Likewise.
24518 (riscv_valid_offset_p): Likewise.
24519 (riscv_valid_lo_sum_p): Likewise.
24520 (riscv_classify_address): Likewise.
24521 (riscv_legitimate_address_p): Likewise.
24522 (riscv_address_insns): Likewise.
24523 (riscv_load_store_insns): Likewise.
24524 (riscv_force_binary): Likewise.
24525 (riscv_split_symbol): Likewise.
24526 (riscv_force_address): Likewise.
24527 (riscv_legitimize_address): Likewise.
24528 (riscv_move_integer): Likewise.
24529 (riscv_legitimize_const_move): Likewise.
24530 (riscv_legitimize_move): Likewise.
24531 (riscv_address_cost): Likewise.
24532 (riscv_subword): Likewise.
24533 (riscv_output_move): Likewise.
24534 (riscv_canonicalize_int_order_test): Likewise.
24535 (riscv_emit_int_order_test): Likewise.
24536 (riscv_function_arg_boundary): Likewise.
24537 (riscv_pass_mode_in_fpr_p): Likewise.
24538 (riscv_pass_fpr_single): Likewise.
24539 (riscv_pass_fpr_pair): Likewise.
24540 (riscv_get_arg_info): Likewise.
24541 (riscv_function_arg): Likewise.
24542 (riscv_function_arg_advance): Likewise.
24543 (riscv_arg_partial_bytes): Likewise.
24544 (riscv_function_value): Likewise.
24545 (riscv_pass_by_reference): Likewise.
24546 (riscv_setup_incoming_varargs): Likewise.
24547 (riscv_print_operand): Likewise.
24548 (riscv_elf_select_rtx_section): Likewise.
24549 (riscv_save_restore_reg): Likewise.
24550 (riscv_for_each_saved_reg): Likewise.
24551 (riscv_register_move_cost): Likewise.
24552 (riscv_hard_regno_mode_ok_p): Likewise.
24553 (riscv_hard_regno_nregs): Likewise.
24554 (riscv_class_max_nregs): Likewise.
24555 (riscv_memory_move_cost): Likewise.
24556 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
24557 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
24558 (rl78_addr_space_address_mode): Likewise.
24559 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
24561 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
24562 (rs6000_reassociation_width): Likewise.
24563 (rs6000_invalid_binary_op): Likewise.
24564 (fusion_p9_p): Likewise.
24565 (emit_fusion_p9_load): Likewise.
24566 (emit_fusion_p9_store): Likewise.
24567 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
24568 (ok_for_simple_move_operands): Likewise.
24569 (ok_for_simple_move_strict_operands): Likewise.
24570 (ok_for_simple_arith_logic_operands): Likewise.
24571 (visium_legitimize_reload_address): Likewise.
24572 (visium_select_cc_mode): Likewise.
24573 (output_cbranch): Likewise.
24574 (visium_split_double_move): Likewise.
24575 (visium_expand_copysign): Likewise.
24576 (visium_expand_int_cstore): Likewise.
24577 (visium_expand_fp_cstore): Likewise.
24578 * config/visium/visium.c (visium_pass_by_reference): Likewise.
24579 (visium_function_arg): Likewise.
24580 (visium_function_arg_advance): Likewise.
24581 (visium_libcall_value): Likewise.
24582 (visium_setup_incoming_varargs): Likewise.
24583 (visium_legitimate_constant_p): Likewise.
24584 (visium_legitimate_address_p): Likewise.
24585 (visium_legitimize_address): Likewise.
24586 (visium_secondary_reload): Likewise.
24587 (visium_register_move_cost): Likewise.
24588 (visium_memory_move_cost): Likewise.
24589 (prepare_move_operands): Likewise.
24590 (ok_for_simple_move_operands): Likewise.
24591 (ok_for_simple_move_strict_operands): Likewise.
24592 (ok_for_simple_arith_logic_operands): Likewise.
24593 (visium_function_value_1): Likewise.
24594 (rtx_ok_for_offset_p): Likewise.
24595 (visium_legitimize_reload_address): Likewise.
24596 (visium_split_double_move): Likewise.
24597 (visium_expand_copysign): Likewise.
24598 (visium_expand_int_cstore): Likewise.
24599 (visium_expand_fp_cstore): Likewise.
24600 (visium_split_cstore): Likewise.
24601 (visium_select_cc_mode): Likewise.
24602 (visium_split_cbranch): Likewise.
24603 (output_cbranch): Likewise.
24604 (visium_print_operand_address): Likewise.
24605 * expmed.c (flip_storage_order): Likewise.
24606 * expmed.h (emit_cstore): Likewise.
24607 (flip_storage_order): Likewise.
24608 * genrecog.c (validate_pattern): Likewise.
24609 * hsa-gen.c (gen_hsa_addr): Likewise.
24610 * internal-fn.c (expand_arith_overflow): Likewise.
24611 * ira-color.c (allocno_copy_cost_saving): Likewise.
24612 * lra-assigns.c (find_hard_regno_for_1): Likewise.
24613 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
24614 (process_invariant_for_inheritance): Likewise.
24615 * lra-eliminations.c (move_plus_up): Likewise.
24616 * omp-low.c (lower_oacc_reductions): Likewise.
24617 * simplify-rtx.c (simplify_subreg): Likewise.
24618 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
24619 (TARGET_CHKP_BOUND_MODE): Likewise..
24620 * targhooks.c (default_chkp_bound_mode): Likewise.
24621 (default_setup_incoming_vararg_bounds): Likewise.
24622 * targhooks.h (default_chkp_bound_mode): Likewise.
24623 (default_setup_incoming_vararg_bounds): Likewise.
24624 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
24625 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
24626 (have_whole_vector_shift): Likewise.
24627 * tree-vect-stmts.c (vectorizable_load): Likewise.
24628 * doc/tm.texi: Regenerate.
24630 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24632 Graceful degrade if Binutils PR21472 is not available.
24635 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
24636 .rodata in flash test fails.
24637 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
24638 * confgure: Regenerate.
24639 * config.in: Regenerate.
24640 * config/avr/avr.c (avr_asm_named_section)
24641 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
24642 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
24643 (avr_asm_init_sections): Same.
24645 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24647 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
24648 (fma<VH:mode>4_intrinsic): Likewise.
24649 (*fmsub<VCVTF:mode>4): Likewise.
24650 (*fmsub<VH:mode>4_intrinsic): Likewise.
24652 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24655 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
24656 Don't depend on "optimize > 0".
24657 (out_movhi_r_mr, out_movqi_mr_r): Same.
24658 (out_movhi_mr_r, out_movqi_r_mr): Same.
24659 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
24660 io_address_operand on "optimize > 0".
24662 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24664 * tree-loop-distribution.c: Add general explanantion on the pass.
24665 (generate_loops_for_partition): Mark distributed loop.
24666 (pg_add_dependence_edges): New parameter. Handle alias data
24667 dependence specially and record it in the parameter if asked.
24668 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
24669 (init_partition_graph_vertices, add_partition_graph_edge): New.
24670 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
24671 (free_partition_graph_vdata, build_partition_graph): New.
24672 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
24673 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
24674 (data_ref_segment_size, latch_dominated_by_data_ref): New.
24675 (compute_alias_check_pairs, version_loop_by_alias_check): New.
24676 (version_for_distribution_p, finalize_partitions): New.
24677 (distribute_loop): Handle alias data dependence specially. Factor
24678 out loop fusion code as functions and call these functions.
24680 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24682 * tree-loop-distribution.c (classify_partition): New parameter and
24683 better handle reduction statement.
24684 (rdg_build_partitions): Revise comment.
24685 (distribute_loop): Compute statements in all partitions and pass it
24686 to classify_partition.
24688 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24690 * tree-loop-distribution.c (enum partition_type): New.
24691 (struct partition): New field type.
24692 (partition_merge_into): Add parameter. Update partition type.
24693 (data_dep_in_cycle_p, update_type_for_merge): New functions.
24694 (build_rdg_partition_for_vertex): Compute partition type.
24695 (rdg_build_partitions): Dump partition type.
24696 (distribute_loop): Update calls to partition_merge_into.
24698 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24700 * tree-loop-distribution.c (struct ddr_hasher): New.
24701 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
24703 (classify_partition): Call get_data_dependence.
24704 (pg_add_dependence_edges): Ditto.
24705 (distribute_loop): Release data dependence hash table.
24707 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24709 * tree-loop-distribution.c (ref_base_address): Delete.
24710 (similar_memory_accesses): Rename ...
24711 (share_memory_accesses): ... to this. Check if partitions access
24712 the same memory reference.
24713 (distribute_loop): Call share_memory_accesses.
24715 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24717 * tree-loop-distribution.c (struct partition): New field recording
24718 its data reference.
24719 (partition_alloc, partition_free): Init and release data refs.
24720 (partition_merge_into): Merge data refs.
24721 (build_rdg_partition_for_vertex): Collect data refs for partition.
24722 (pg_add_dependence_edges): Change parameters from vector to bitmap.
24724 (distribute_loop): Remve data refs from vertice data of partition
24727 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24729 * tree-loop-distribution.c (params.h): Include header file.
24730 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
24731 (datarefs_vec): New global var.
24732 (create_rdg_vertices): Use datarefs_vec directly.
24733 (free_rdg): Don't free data references.
24734 (build_rdg): Update use. Don't free data references.
24735 (distribute_loop): Compute global variable for data references.
24736 Bail out if there are too many data references.
24738 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24740 * tree-loop-distribution.c (loop_nest): New global var.
24741 (build_rdg): Use loop directly, rather than loop nest.
24742 (pg_add_dependence_edges): Remove loop nest parameter. Use global
24744 (distribute_loop): Compute global variable loop nest. Update use.
24746 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24748 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
24749 (partition_merge_into): New parameter. Dump reason for fusion.
24750 (distribute_loop): Update use of partition_merge_into.
24752 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24754 * tree-loop-distribution.c (bb_top_order_index): New.
24755 (bb_top_order_index_size, bb_top_order_cmp): New.
24756 (stmts_from_loop): Use topological order.
24757 (pass_loop_distribution::execute): Compute and release topological
24758 order for basic blocks.
24760 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24762 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
24765 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24767 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
24768 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
24769 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
24770 * internal-fn.def (LOOP_DIST_ALIAS): New.
24771 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
24772 (fold_loop_internal_call): ... this.
24773 (vect_loop_dist_alias_call): New function.
24774 (set_uid_loop_bbs): Call fold_loop_internal_call.
24775 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
24778 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24781 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
24782 Require dead FLAGS_REG at the beginning of a peephole.
24784 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24787 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
24788 arguments in the call to __builtin_ia32_sbb_u32.
24789 (_subborrow_u64): Swap _X and _Y arguments in the call to
24790 __builtin_ia32_sbb_u64.
24792 2017-07-04 Jakub Jelinek <jakub@redhat.com>
24795 * tree-vrp.c (compare_assert_loc): Turn into a function template
24796 with stable template parameter. Only test if a->e is NULL,
24797 !a->e == !b->e has been verified already. Use e == NULL or
24798 e != NULL instead of e or ! e tests. If stable is true, don't use
24799 iterative_hash_expr, on the other side allow a or b or both NULL
24800 and sort the NULLs last.
24801 (process_assert_insertions): Sort using compare_assert_loc<false>
24802 instead of compare_assert_loc, later sort using
24803 compare_assert_loc<true> before calling process_assert_insertions_for
24804 in a loop. Use break instead of continue once seen NULL pointer.
24806 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24808 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24809 Cortex-R7 and Cortex-R8 processors.
24811 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24813 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
24814 uninitialized while src is not.
24816 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
24818 * common/config/arm/arm-common.c: Adjust include path for
24820 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
24821 (arm-cpu.h): Create in build directory. Adjust dependency rules.
24822 (arm-cpu-data.h): Likewise.
24823 (arm-cpu-cdata.h): Likewise.
24824 * config/arm/arm-cpu.h: Delete.
24825 * config/arm/arm-cpu-cdata.h: Delete.
24826 * config/arm/arm-cpu-data.h: Delete.
24828 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
24830 * config/arm/arm-cpus.in (cortex-a55): New.
24831 (cortex-a75): Likewise.
24832 (cortex-a75.cortex-a55): Likewise.
24833 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
24835 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
24836 * config/arm/arm-cpu-cdata.h: Regenerate.
24837 * config/arm/arm-cpu-data.h: Regenerate.
24838 * config/arm/arm-cpu.h: Regenerate.
24839 * config/arm/arm-tables.opt: Regenerate.
24840 * config/arm/arm-tune.md: Regenerate.
24842 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24844 * haifa-sched.c (sched_create_recovery_edges): Update profile.
24846 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24848 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
24851 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
24853 PR tree-optimization/81292
24854 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
24855 full_string_p, also call adjust_related_strinfos if the adjustment
24856 is simple, otherwise invalidate related strinfos.
24858 2017-07-04 Martin Liska <mliska@suse.cz>
24861 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
24862 newly created variable as DECL_IGNORED_P.
24864 2017-07-04 Martin Liska <mliska@suse.cz>
24867 * ipa-inline.c (inline_small_functions):
24868 Use xstrdup_for_dump.
24870 2017-07-04 Tom de Vries <tom@codesourcery.com>
24872 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
24874 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24877 * config/darwin.c (darwin_function_switched_text_sections):
24880 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
24882 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
24884 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24886 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
24888 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24890 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
24891 min_profitable_iters, and th as inclusive lower bounds.
24892 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
24893 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
24894 for min_profitable_iters and min_profitable_estimate.
24895 (vect_transform_loop): Treat th as an inclusive lower bound.
24896 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24898 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24901 * config/darwin.c (darwin_function_switched_text_sections):
24902 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
24903 in two pieces, and suppress the use of buf.
24905 2017-07-03 Nathan Sidwell <nathan@acm.org>
24907 * hash-table.h (hash_table_mod1): Fix indentation.
24909 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24911 PR middle-end/81290
24912 * predict.c (force_edge_cold): Be more careful about propagation
24914 * profile-count.h (profile_probability::guessed,
24915 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
24917 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
24919 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
24921 * doc/invoke.texi (rcpc architecture extension): Document it.
24923 2017-07-03 Richard Biener <rguenther@suse.de>
24925 PR tree-optimization/60510
24926 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
24927 the scalar reduction PHI and use it.
24928 (vectorizable_reduction): Properly guard the single_defuse_cycle
24929 path for non-SLP reduction chains where we cannot use it.
24930 Rework reduc_def/index and vector type deduction. Rework
24931 vector operand gathering during reduction op code-gen.
24932 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
24933 chains dissolve the chain and leave it to non-SLP reduction
24936 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24938 * tree-data-ref.h (dr_alignment): Declare.
24939 * tree-data-ref.c (dr_alignment): New function.
24940 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
24941 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
24943 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
24945 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24947 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
24948 and base_misalignment fields.
24949 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
24950 * tree-data-ref.c: Include builtins.h.
24951 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
24952 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
24953 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
24954 * tree-vect-data-refs.c: Include tree-cfg.h.
24955 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
24956 fields instead of calculating an alignment here.
24957 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
24958 innermost_loop_behavior fields.
24960 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24962 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
24964 (DR_STEP_ALIGNMENT): New macro.
24965 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
24966 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
24967 (create_data_ref): Print it.
24968 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
24969 to tell whether the step preserves vector (mis)alignment.
24970 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24971 Move the check for an integer step and generalise to all INTEGER_CST.
24972 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
24973 Print the outer step alignment.
24975 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24977 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
24978 with offset_alignment.
24979 (DR_ALIGNED_TO): Delete.
24980 (DR_OFFSET_ALIGNMENT): New macro.
24981 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
24982 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
24983 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
24984 (create_data_ref): Likewise.
24985 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24986 (vect_analyze_data_refs): Likewise.
24987 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
24988 creating dummy innermost behavior.
24990 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24992 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
24993 with a "innermost_loop_behavior *" and refeence tree.
24994 * tree-data-ref.c (dr_analyze_innermost): Likewise.
24995 (create_data_ref): Update call accordingly.
24996 * tree-predcom.c (find_looparound_phi): Likewise.
24998 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25000 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
25001 fields with dr_wrt_vec_loop.
25002 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
25003 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
25004 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
25005 (vect_dr_behavior): New function.
25006 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25007 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
25008 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
25009 track whether the step preserves the misalignment.
25010 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25011 Use vect_dr_behavior.
25012 (vect_setup_realignment): Update call accordingly.
25013 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
25014 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
25015 call to vect_create_addr_base_for_vector_ref.
25016 (vect_create_cond_for_align_checks): Likewise.
25017 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
25018 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
25019 (vect_recog_mask_conversion_pattern): Likewise.
25020 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
25021 (new_stmt_vec_info): Remove redundant zeroing.
25023 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25025 * common/config/arm/arm-common.c (arm_be8_option): New function.
25026 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
25027 (ISA_ARMv6): Add isa_bit_be8.
25028 * config/arm/arm.h (arm_be8_option): Add prototype.
25029 (BE8_SPEC_FUNCTION): New define.
25030 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
25031 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
25032 (mlittle-endian): Similarly.
25033 (mbe8, mbe32): New options.
25034 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
25035 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
25037 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25039 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
25041 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25043 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
25044 (cleanup_tree_cfg_bb): Use it.
25045 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
25047 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
25049 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25052 * loop-doloop.c (add_test): Update profile.
25054 2017-07-03 Martin Liska <mliska@suse.cz>
25057 * sanopt.c (rewrite_usage_of_param): New function.
25058 (sanitize_rewrite_addressable_params): Likewise.
25059 (pass_sanopt::execute): Call rewrite_usage_of_param.
25061 2017-07-03 Richard Biener <rguenther@suse.de>
25063 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
25064 back to using VIEW_CONVERT_EXPR.
25066 2017-07-03 Martin Liska <mliska@suse.cz>
25069 * doc/extend.texi: Document when a resolver function is
25070 generated for target_clones.
25072 2017-07-03 Martin Liska <mliska@suse.cz>
25074 * asan.c (asan_emit_stack_protection): Unpoison just red zones
25075 and shadow memory of auto variables which are subject of
25076 use-after-scope sanitization.
25077 (asan_expand_mark_ifn): Add do set only when is_poison.
25079 2016-07-03 Richard Biener <rguenther@suse.de>
25081 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
25083 (vect_force_simple_reduction): Record reduction def -> phi mapping.
25084 (vectorizable_reduction): Perform reduction PHI creation when
25085 visiting a reduction PHI and adjust and simplify code generation
25086 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
25087 (vect_transform_loop): Visit reduction PHIs.
25088 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
25089 defs into the SLP tree.
25090 (vect_build_slp_tree): Reduction defs terminate the recursion.
25091 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
25093 (vect_get_vec_defs_for_stmt_copy): Export.
25094 (vect_get_vec_defs): Likewise.
25095 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
25097 (vect_get_vec_defs_for_stmt_copy): Declare.
25098 (vect_get_vec_defs): Likewise.
25100 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25102 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
25103 parameter with a "loop" parameter and use it instead of the
25104 loop containing DR_STMT. Don't check simple_iv when doing
25105 BB analysis. Describe the two analysis modes in the comment.
25107 2017-07-03 Tom de Vries <tom@codesourcery.com>
25109 PR tree-optimization/69468
25110 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
25111 (find_same_succ_bb): Handle ignore_edge_flags.
25113 2017-07-03 Tom de Vries <tom@codesourcery.com>
25115 PR tree-optimization/81192
25116 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
25118 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
25120 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
25122 2017-07-03 Tom de Vries <tom@codesourcery.com>
25124 PR tree-optimization/81192
25125 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
25126 BB_SAME_SUCC (bb) == NULL.
25128 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25130 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
25133 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25135 * dumpfile.c: Include profile-count.h
25136 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
25138 (insert_cond_bb): Update profile.
25139 * tree-cfg.h (insert_cond_bb): Update prototype.
25140 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
25141 * tree-dump.c: Do not include tree-cfg.
25143 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25145 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
25147 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25149 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
25152 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25154 * tree-complex.c (expand_complex_div_wide): update profile.
25156 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25157 Alan Hayward <alan.hayward@arm.com>
25158 David Sherwood <david.sherwood@arm.com>
25160 * Makefile.in (MACHMODE_H): Remove insn-modes.h
25161 (CORETYPES_H): New define.
25162 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
25163 (insn-modes-inline.h, s-modes-inline-h): New rules.
25164 (generated_files): Add insn-modes-inline.h.
25165 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
25166 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
25167 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
25168 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
25169 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
25170 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
25171 (build/gencodes.o, build/genconditions.o): Likewise.
25172 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
25173 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
25174 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
25175 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
25176 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
25177 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
25178 * coretypes.h: Include everything up to real.h for generators.
25179 Include insn-modes.h first. Include wide-int-print.h after
25180 wide-int.h. Include insn-modes-inline.h and then machmode.h.
25181 * machmode.h: Don't include insn-modes.h here.
25182 * function-tests.c: Remove includes of signop.h, machmode.h,
25183 double-int.h and wide-int.h.
25185 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
25187 * optc-save-gen.awk: Likewise.
25188 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
25189 * godump.c: Remove include of wide-int-print.h.
25190 * pretty-print.h: Likewise.
25191 * wide-int-print.cc: Likewise.
25192 * wide-int.cc: Likewise.
25193 * hash-map-tests.c: Remove include of signop.h.
25194 * hash-set-tests.c: Likewise.
25195 * rtl-tests.c: Likewise.
25196 * mkconfig.sh: Remove include of machmode.h.
25197 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
25199 (emit_insn_modes_inline_h): ...this new function. Emit the code
25200 into an insn-modes-inline.h header file, adding appropriate
25201 include guards and end comments.
25202 (emit_insn_modes_c_header): Remove include of machmode.h.
25203 (emit_min_insn_modes_c_header): Include coretypes.h rather than
25205 (main): Handle -i flag and call emit_insn_modes_inline_h when
25208 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25210 * tree-ssa-strlen.c (strinfo): Rename the length field to
25211 nonzero_chars. Add a full_string_p field.
25212 (compare_nonzero_chars, zero_length_string_p): New functions.
25213 (get_addr_stridx): Add an offset_out parameter.
25214 Use compare_nonzero_chars.
25215 (get_stridx): Update accordingly. Use compare_nonzero_chars.
25216 (new_strinfo): Update after above changes to strinfo.
25217 (set_endptr_and_length): Set full_string_p.
25218 (get_string_length): Update after above changes to strinfo.
25219 (unshare_strinfo): Update call to new_strinfo.
25220 (maybe_invalidate): Likewise.
25221 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
25222 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
25223 as a uhwi instead of an shwi. Update after above changes to
25224 strinfo and new_strinfo.
25225 (zero_length_string): Assert that chainsi contains full strings.
25226 Use zero_length_string_p. Update call to new_strinfo.
25227 (adjust_related_strinfos): Update after above changes to strinfo.
25228 Copy full_string_p from origsi.
25229 (adjust_last_stmt): Use zero_length_string_p.
25230 (handle_builtin_strlen): Update after above changes to strinfo and
25231 new_strinfo. Install the lhs as the string length if the previous
25232 entry didn't describe a full string.
25233 (handle_builtin_strchr): Update after above changes to strinfo
25235 (handle_builtin_strcpy): Likewise.
25236 (handle_builtin_strcat): Likewise.
25237 (handle_builtin_malloc): Likewise.
25238 (handle_pointer_plus): Likewise.
25239 (handle_builtin_memcpy): Likewise. Track nonzero characters
25240 that aren't necessarily followed by a nul terminator.
25241 (handle_char_store): Likewise.
25243 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25245 PR tree-optimization/80769
25246 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
25247 for malloc and calloc. Document the new invariant that all related
25248 strinfos have delayed lengths or none do.
25249 (verify_related_strinfos): Move earlier in file.
25250 (set_endptr_and_length): New function, split out from...
25251 (get_string_length): ...here. Also set the lengths of related
25253 (zero_length_string): Assert that chainsi has known (rather than
25255 (adjust_related_strinfos): Likewise.
25257 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25259 PR tree-optimization/81136
25260 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
25261 assert that two references with the same misalignment have the same
25262 compile-time misalignment if those compile-time misalignments
25265 2017-07-01 Andi Kleen <ak@linux.intel.com>
25267 * print-tree.c (print_node): Print all attributes.
25269 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25271 * cfg.c (scale_bbs_frequencies): New function.
25272 * cfg.h (scale_bbs_frequencies): Declare it.
25273 * cfgloopanal.c (single_likely_exit): Cleanup.
25274 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
25276 (scale_loop_profile): Likewise.
25277 (loop_version): Likewise.
25278 (create_empty_loop_on_edge): Update.
25279 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
25280 scale_loop_frequencies, scale_loop_profile, loopify,
25281 loop_version): Update prototypes.
25282 * modulo-sched.c (sms_schedule): Update.
25283 * predict.c (unlikely_executed_edge_p): Also check probability.
25284 (probably_never_executed_edge_p): Fix typo.
25285 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25286 * tree-parloops.c (gen_parallel_loop): Update.
25287 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
25288 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25289 * tree-ssa-loop-split.c (split_loop): Update.
25290 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25291 * tree-vect-loop-manip.c (vect_do_peeling): Update.
25292 (vect_loop_versioning): Update.
25293 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25295 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25297 * trans-mem.c (split_bb_make_tm_edge): Update profile.
25299 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25301 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
25302 to keep profile consistent.
25304 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25306 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
25307 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
25308 * profile-count.h (max_safe_multiplier): Make unsigned.
25309 (profile_count::guessed_zero): New.
25311 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25313 * bb-reorder.c (fix_up_crossing_landing_pad,
25314 fix_crossing_conditional_branches): Use make_single_succ_edge
25315 to keep profile consistent.
25317 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25319 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
25322 2017-07-01 Jakub Jelinek <jakub@redhat.com>
25325 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
25326 the right scopes, make sure cond_jump isn't preserved between multiple
25327 iterations. Search for fallthru edge whenever there are 3+ edges and
25328 use find_fallthru_edge for it.
25330 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25332 Patch by Alexander Monakov <amonakov@ispras.ru>
25333 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
25334 probabilities consistently.
25336 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25338 * pa.c (pa_expand_compare_and_swap_loop): Update call of
25339 emit_cmp_and_jump_insns.
25341 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25344 * tree-inline.c (expand_call_inline): Combine profile statuses.
25346 2017-06-30 Andrew Pinski <apinski@cavium.com>
25348 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
25349 fold_stmt returned true.
25351 2017-06-30 Nathan Sidwell <nathan@acm.org>
25353 * ggc.h (empty_string): Delete.
25354 * cfgexpand.c (expand_asm_stmt): Use plain "".
25355 * optabs.c (expand_asm_memory_barrier): Likewise.
25356 * stringpool.c (empty_string): Delete.
25357 (digit_vector, digit_string): Delete.
25358 (ggc_alloc_string): Use plain "", don't optimize single digit
25359 strings. Use ggc_alloc_atomic.
25361 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
25363 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
25364 comparison set and one other set, use the cost of the non-comparison
25367 2017-06-30 Nathan Sidwell <nathan@acm.org>
25369 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
25372 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
25374 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
25375 loops. Remove now unneeded calls to gimple_switch_set_label() that
25376 just set removed labels to NULL_TREE.
25378 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
25380 * tree-ssanames.c (set_range_info_raw): Abstract from ...
25381 (set_range_info): ...here. Only call set_range_info_raw if domain
25383 (set_nonzero_bits): Call set_range_info_raw.
25384 * tree-ssanames.h (set_range_info_raw): New.
25386 2017-06-30 Jakub Jelinek <jakub@redhat.com>
25389 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
25390 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
25391 of nonimmediate_operand and <store_mask_constraint> instead of m
25392 for the input operand. For V8FI iterator, always split if input
25393 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
25394 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
25395 <store_mask_predicate> instead of register_operand and
25396 <store_mask_constraint> instead of v for the input operand. Make
25397 sure both operands aren't MEMs for if not <mask_applied>.
25399 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
25401 * lto-wrapper.c (copy_file) Close both file descriptors before
25404 2017-06-30 Martin Liska <mliska@suse.cz>
25407 * multiple_target.c (create_dispatcher_calls): Make ifunc
25408 also for function that don't have calls or are not referenced.
25410 2017-06-30 Richard Biener <rguenther@suse.de>
25412 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
25413 analyze the first scalar stmt. Move vector type computation
25414 for the BB case here from ...
25415 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
25416 live operation processing in the SLP case properly.
25418 2017-06-30 Richard Biener <rguenther@suse.de>
25420 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
25422 2017-06-30 Martin Liska <mliska@suse.cz>
25425 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
25426 before BUILT_IN_UNWIND_RESUME when ASAN is used.
25428 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
25430 * doc/invoke.texi (AArch64): Add missing options and remove redundant
25433 2017-06-30 Richard Biener <rguenther@suse.de>
25435 PR tree-optimization/81249
25436 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
25437 condition reduction result to original scalar type.
25439 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25441 * profile-count.h (enum profile_quality): Fix typos and whitespace
25444 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25446 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
25447 type for branch probabilities.
25449 2017-06-29 Julian Brown <julian@codesourcery.com>
25450 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25452 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
25453 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
25454 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
25455 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
25457 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25459 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
25460 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
25461 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
25462 CC usage from generic code to here.
25463 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
25464 CC usage into the target macros.
25466 2017-06-29 Maya Rashish <coypu@sdf.org>
25468 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
25471 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25473 * arm/arm-builtins.c: Include profile-count.h
25474 * except.c (sjlj_emit_function_enter): Use
25475 profile_probability::unlikely.
25477 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25479 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
25480 and tocrel_offset be pointer args rather than implicitly using
25482 (legitimate_constant_pool_address_p, rs6000_emit_move,
25483 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
25484 tocrel_offset and use in toc_relative_expr_p call.
25485 (print_operand, print_operand_address): Use static tocrel_base_oac
25486 and tocrel_offset_oac.
25487 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
25490 2017-06-29 Maya Rashish <coypu@sdf.org>
25492 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
25494 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
25496 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
25497 objects, take into account only the alignment of 'op0' and 'mode1' if
25500 2017-06-29 Steve Ellcey <sellcey@cavium.com>
25502 * ccmp.c (ccmp_tree_comparison_p): New function.
25503 (ccmp_candidate_p): Update to use above function.
25504 (get_compare_parts): New function.
25505 (expand_ccmp_next): Update to use new functions.
25506 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
25508 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
25509 take mode as argument.
25510 * ccmp.h (expand_ccmp_expr): Add mode as argument.
25511 * expr.c (expand_expr_real_1): Pass mode as argument.
25513 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
25515 * combine.c (combine_instructions): Print insns to dump_file, together
25518 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25520 * asan.c (asan_emit_stack_protection): Update.
25521 (create_cond_insert_point): Update.
25522 * auto-profile.c (afdo_propagate_circuit): Update.
25523 * basic-block.h (struct edge_def): Turn probability to
25524 profile_probability.
25525 (EDGE_FREQUENCY): Update.
25526 * bb-reorder.c (find_traces_1_round): Update.
25527 (better_edge_p): Update.
25528 (sanitize_hot_paths): Update.
25529 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
25530 (make_single_succ_edge): Update.
25531 (check_bb_profile): Update.
25532 (dump_edge_info): Update.
25533 (update_bb_profile_for_threading): Update.
25534 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
25535 probabilitycount to 0.
25536 * cfgbuild.c (compute_outgoing_frequencies): Update.
25537 * cfgcleanup.c (try_forward_edges): Update.
25538 (outgoing_edges_match): Update.
25539 (try_crossjump_to_edge): Update.
25540 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
25541 (expand_gimple_tailcall): Update.
25542 (construct_init_block): Use make_single_succ_edge.
25543 (construct_exit_block): Use make_single_succ_edge.
25544 * cfghooks.c (verify_flow_info): Update.
25545 (redirect_edge_succ_nodup): Update.
25546 (split_edge): Update.
25547 (account_profile_record): Update.
25548 * cfgloopanal.c (single_likely_exit): Update.
25549 * cfgloopmanip.c (scale_loop_profile): Update.
25550 (set_zero_probability): Remove.
25551 (duplicate_loop_to_header_edge): Update.
25552 * cfgloopmanip.h (loop_version): Update prototype.
25553 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
25554 (force_nonfallthru_and_redirect): Update.
25555 (update_br_prob_note): Update.
25556 (rtl_verify_edges): Update.
25557 (purge_dead_edges): Update.
25558 (rtl_lv_add_condition_to_bb): Update.
25559 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
25560 * cgraphunit.c (init_lowered_empty_function): Update.
25561 (cgraph_node::expand_thunk): Update.
25562 * cilk-common.c: Include profile-count.h
25563 * dojump.c (inv): Remove.
25564 (jumpifnot): Update.
25565 (jumpifnot_1): Update.
25566 (do_jump_1): Update.
25568 (do_jump_by_parts_greater_rtx): Update.
25569 (do_compare_rtx_and_jump): Update.
25570 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
25571 do_jump_1. do_compare_rtx_and_jump): Update prototype.
25572 * dwarf2cfi.c: Include profile-count.h
25573 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
25574 (sjlj_emit_dispatch_table): Likewise.
25575 * explow.c: Include profile-count.h
25576 * expmed.c (emit_store_flag_force): Update.
25577 (do_cmp_and_jump): Update.
25578 * expr.c (compare_by_pieces_d::generate): Update.
25579 (compare_by_pieces_d::finish_mode): Update.
25580 (emit_block_move_via_loop): Update.
25581 (store_expr_with_bounds): Update.
25582 (store_constructor): Update.
25583 (expand_expr_real_2): Update.
25584 (expand_expr_real_1): Update.
25585 * expr.h (try_casesi, try_tablejump): Update prototypes.
25586 * gimple-pretty-print.c (dump_probability): Update.
25587 (dump_profile): New.
25588 (dump_gimple_label): Update.
25589 (dump_gimple_bb_header): Update.
25590 * graph.c (draw_cfg_node_succ_edges): Update.
25591 * hsa-gen.c (convert_switch_statements): Update.
25592 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
25593 (find_if_case_1): Update.
25594 (find_if_case_2): Update.
25595 * internal-fn.c (expand_arith_overflow_result_store): Update.
25596 (expand_addsub_overflow): Update.
25597 (expand_neg_overflow): Update.
25598 (expand_mul_overflow): Update.
25599 (expand_vector_ubsan_overflow): Update.
25600 * ipa-cp.c (good_cloning_opportunity_p): Update.
25601 * ipa-split.c (split_function): Use make_single_succ_edge.
25602 * ipa-utils.c (ipa_merge_profiles): Update.
25603 * loop-doloop.c (add_test): Update.
25604 (doloop_modify): Update.
25605 * loop-unroll.c (compare_and_jump_seq): Update.
25606 (unroll_loop_runtime_iterations): Update.
25607 * lra-constraints.c (lra_inheritance): Update.
25608 * lto-streamer-in.c (input_cfg): Update.
25609 * lto-streamer-out.c (output_cfg): Update.
25610 * mcf.c (adjust_cfg_counts): Update.
25611 * modulo-sched.c (sms_schedule): Update.
25612 * omp-expand.c (expand_omp_for_init_counts): Update.
25613 (extract_omp_for_update_vars): Update.
25614 (expand_omp_ordered_sink): Update.
25615 (expand_omp_for_ordered_loops): Update.
25616 (expand_omp_for_generic): Update.
25617 (expand_omp_for_static_nochunk): Update.
25618 (expand_omp_for_static_chunk): Update.
25619 (expand_cilk_for): Update.
25620 (expand_omp_simd): Update.
25621 (expand_omp_taskloop_for_outer): Update.
25622 (expand_omp_taskloop_for_inner): Update.
25623 * omp-simd-clone.c (simd_clone_adjust): Update.
25624 * optabs.c (expand_doubleword_shift): Update.
25625 (expand_abs): Update.
25626 (emit_cmp_and_jump_insn_1): Update.
25627 (expand_compare_and_swap_loop): Update.
25628 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
25629 * predict.c (predictable_edge_p): Update.
25630 (edge_probability_reliable_p): Update.
25631 (set_even_probabilities): Update.
25632 (combine_predictions_for_insn): Update.
25633 (combine_predictions_for_bb): Update.
25634 (propagate_freq): Update.
25635 (estimate_bb_frequencies): Update.
25636 (force_edge_cold): Update.
25637 * profile-count.c (profile_count::dump): Add missing space into dump.
25638 (profile_count::debug): Add newline.
25639 (profile_count::differs_from_p): Explicitly convert to unsigned.
25640 (profile_count::stream_in): Update.
25641 (profile_probability::dump): New member function.
25642 (profile_probability::debug): New member function.
25643 (profile_probability::differs_from_p): New member function.
25644 (profile_probability::differs_lot_from_p): New member function.
25645 (profile_probability::stream_in): New member function.
25646 (profile_probability::stream_out): New member function.
25647 * profile-count.h (profile_count_quality): Rename to ...
25648 (profile_quality): ... this one.
25649 (profile_probability): New.
25650 (profile_count): Update.
25651 * profile.c (compute_branch_probabilities): Update.
25652 * recog.c (peep2_attempt): Update.
25653 * sched-ebb.c (schedule_ebbs): Update.
25654 * sched-rgn.c (find_single_block_region): Update.
25655 (compute_dom_prob_ps): Update.
25656 (schedule_region): Update.
25657 * sel-sched-ir.c (compute_succs_info): Update.
25658 * stmt.c (struct case_node): Update.
25659 (do_jump_if_equal): Update.
25660 (get_outgoing_edge_probs): Update.
25661 (conditional_probability): Update.
25662 (emit_case_dispatch_table): Update.
25663 (expand_case): Update.
25664 (expand_sjlj_dispatch_table): Update.
25665 (emit_case_nodes): Update.
25666 * targhooks.c: Update.
25667 * tracer.c (better_p): Update.
25668 (find_best_successor): Update.
25669 * trans-mem.c (expand_transaction): Update.
25670 * tree-call-cdce.c: Update.
25671 * tree-cfg.c (gimple_split_edge): Upate.
25672 (move_sese_region_to_fn): Upate.
25673 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
25674 * tree-eh.c (lower_resx): Upate.
25675 (cleanup_empty_eh_move_lp): Upate.
25676 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25677 * tree-inline.c (copy_edges_for_bb): Update.
25678 (copy_cfg_body): Update.
25679 * tree-parloops.c (gen_parallel_loop): Update.
25680 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
25681 (gimple_gen_time_profiler): Update.
25682 * tree-ssa-dce.c (remove_dead_stmt): Update.
25683 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25684 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
25685 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
25686 (unloop_loops): Update.
25687 (try_peel_loop): Update.
25688 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25689 * tree-ssa-loop-split.c (connect_loops): Update.
25690 (split_loop): Update.
25691 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25692 (hoist_guard): Update.
25693 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
25694 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
25695 (value_replacement): Update.
25696 * tree-ssa-reassoc.c (branch_fixup): Update.
25697 * tree-ssa-tail-merge.c (replace_block_by): Update.
25698 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
25699 (create_edge_and_update_destination_phis): Update.
25700 (compute_path_counts): Update.
25701 (recompute_probabilities): Update.
25702 (update_joiner_offpath_counts): Update.
25703 (freqs_to_counts_path): Update.
25704 (duplicate_thread_path): Update.
25705 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
25706 (struct switch_conv_info): Update.
25707 (gen_inbound_check): Update.
25708 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25709 (vect_do_peeling): Update.
25710 (vect_loop_versioning): Update.
25711 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25712 (optimize_mask_stores): Update.
25713 * ubsan.c (ubsan_expand_null_ifn): Update.
25714 * value-prof.c (gimple_divmod_fixed_value): Update.
25715 (gimple_divmod_fixed_value_transform): Update.
25716 (gimple_mod_pow2): Update.
25717 (gimple_mod_pow2_value_transform): Update.
25718 (gimple_mod_subtract): Update.
25719 (gimple_mod_subtract_transform): Update.
25720 (gimple_ic): Update.
25721 (gimple_stringop_fixed_value): Update.
25722 (gimple_stringops_transform): Update.
25723 * value-prof.h: Update.
25725 2017-06-29 Carl Love <cel@us.ibm.com>
25727 * config/rs6000/rs6000-c.c: Add support for built-in functions
25728 vector signed int vec_signed (vector float);
25729 vector signed long long vec_signed (vector double);
25730 vector signed int vec_signed2 (vector double, vector double);
25731 vector signed int vec_signede (vector double);
25732 vector signed int vec_signedo (vector double);
25733 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
25734 instruction generator.
25735 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25736 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
25737 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
25739 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
25740 vunsignede_v2df): Add define_expands.
25741 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
25742 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
25743 VEC_UNSIGNEDO): Add definitions.
25744 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25745 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
25746 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
25747 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
25748 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
25749 * config/rs6000/altivec.h (vec_signed, vec_signed2,
25750 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
25751 vec_unsignede, vec_unsignedo): Add builtin defines.
25752 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
25754 * doc/extend.texi: Update the built-in documentation file for the
25755 new built-in functions.
25757 2017-06-29 Richard Biener <rguenther@suse.de>
25759 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
25760 reduction chains to LOOP_VINFO_REDUCTIONS.
25761 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
25762 SLP reductions after processing reduction chains.
25764 2017-06-29 Nathan Sidwell <nathan@acm.org>
25766 * builtins.c (fold_builtin_FUNCTION): Use
25767 lang_hooks.decl_printable_name.
25769 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
25771 PR middle-end/81194
25772 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
25773 with only one label.
25774 * stmt.c (expand_case): Assert NCASES is greater than one.
25776 2017-06-29 Richard Biener <rguenther@suse.de>
25778 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
25780 (group_case_labels): Likewise.
25781 (find_taken_edge): Push sanity checking on val to workers...
25782 (find_taken_edge_cond_expr): ... here
25783 (find_taken_edge_switch_expr): ... and here, handle cases
25784 with just a default label.
25785 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
25786 (group_case_labels): Likewise.
25787 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
25788 group_case_labels does anything cleanup the CFG again.
25790 2017-06-29 Bin Cheng <bin.cheng@arm.com>
25792 PR tree-optimization/81196
25793 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
25794 exit condition comparing two IVs.
25796 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
25798 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
25799 profile to the dummy entry at the end of the list of architectures.
25800 * config/arm/arm-cpu-cdata.h: Regenerated.
25802 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25803 Michael Collison <michael.collison@arm.com>
25806 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
25808 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
25809 (*aarch64_reg_<mode>3_minus_mask): New pattern.
25810 (*aarch64_<optab>_reg_di3_mask2): New pattern.
25811 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
25812 of shift when the shift amount is masked with constant equal to
25813 the size of the mode.
25814 * config/aarch64/predicates.md (subreg_lowpart_operator): New
25817 2017-06-29 Martin Liska <mliska@suse.cz>
25819 * config/i386/i386.opt: Change range from [1,5] to [0,5].
25821 2017-06-29 Yury Gribov <tetra2005@gmail.com>
25824 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
25827 (edge_growth_cache_entry::edge_growth_cache_entry): New
25829 (reset_edge_growth_cache): Update to use constructor.
25831 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25833 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
25834 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25835 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25837 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
25839 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
25840 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
25842 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
25844 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
25845 (*-linux-uclibc*): Add t-uclibc tmake_file.
25846 * config/t-musl: New.
25847 * config/t-uclibc: New.
25849 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
25851 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
25853 (gen_comm_data): Emit architectural setting of arch_prof.
25854 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
25856 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
25857 (armv8-m.base, armv8-m.main): Likewise.
25858 * arm-protos.h (arm_build_target): Add profile field.
25859 (arch_option): Likewise.
25860 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
25862 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
25863 arm_active_target.profile.
25865 2017-06-28 Richard Biener <rguenther@suse.de>
25867 PR middle-end/81227
25868 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
25869 TYPE_OVERFLOW_WRAPS.
25870 * match.pd (negate_expr_p): Likewise.
25871 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
25872 fold_build2, not fold_binary.
25874 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25876 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
25877 Convert memory address to Pmode.
25878 (aarch64_print_operand): Assert MEM operands are always Pmode.
25880 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25883 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
25884 Remove redundant if.
25885 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
25886 * config/arm/aarch-common-protos.h
25887 (aarch_forward_to_shift_is_not_shifted_re): Remove.
25888 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
25890 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
25893 * multiple_target.c (create_dispatcher_calls): Set the default
25894 clone to be static, not public.
25896 2017-06-28 Richard Biener <rguenther@suse.de>
25898 * tree-vect-loop.c (vectorizable_reduction): Move special
25899 cond reduction IV var creation ...
25900 (vect_create_epilog_for_reduction): ... here. Remove induction_index
25901 parameter. Use STMT_VINFO_VECTYPE.
25902 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
25905 2017-06-28 Martin Liska <mliska@suse.cz>
25908 * ipa-visibility.c (non_local_p): Handle visibility.
25910 2017-06-28 Martin Liska <mliska@suse.cz>
25913 * common.opt: Add IntegerRange to various options.
25914 * opt-functions.awk (integer_range_info): New function.
25915 * optc-gen.awk: Add integer_range_info to cl_options struct.
25916 * opts-common.c (decode_cmdline_option): Handle
25917 CL_ERR_INT_RANGE_ARG.
25918 (cmdline_handle_error): Likewise.
25919 * opts.c (print_filtered_help): Show valid interval in
25920 when --help is provided.
25921 * opts.h (struct cl_option): Add range_min and range_max fields.
25922 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
25924 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
25926 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
25927 (x * C EQ/NE y * C): New transformation.
25929 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
25931 * genmultilib (combination_space): Accept '+' in option names.
25933 2017-06-28 Martin Liska <mliska@suse.cz>
25936 * asan.c (instrument_derefs): Bail out inner references
25937 that are hard register variables.
25939 2017-06-28 Jakub Jelinek <jakub@redhat.com>
25942 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
25943 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
25945 2017-06-28 Richard Biener <rguenther@suse.de>
25947 * tree-vectorizer.h (vect_get_vec_defs): Remove.
25948 (vect_get_slp_defs): Adjust.
25949 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
25951 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
25953 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
25954 get_initial_defs_for_reduction instead of vect_get_vec_defs.
25955 (vectorizable_reduction): Adjust.
25956 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
25958 (vect_get_slp_defs): Likewise.
25959 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
25960 (vectorizable_bswap): Adjust.
25961 (vectorizable_call): Likewise.
25962 (vectorizable_conversion): Likewise.
25963 (vectorizable_assignment): Likewise.
25964 (vectorizable_shift): Likewise.
25965 (vectorizable_operation): Likewise.
25966 (vectorizable_store): Likewise.
25967 (vectorizable_condition): Likewise.
25968 (vectorizable_comparison): Likewise.
25970 2017-06-28 Michael Collison <michael.collison@arm.com>
25973 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
25975 (arm_gen_movmemqi): Removed unused variable 'i'.
25976 Convert 'for' loop into 'while' loop.
25977 (arm_expand_prologue): Remove last unnecessary set of insn.
25978 (thumb_pop): Remove unused variable 'pushed_words'.
25979 (thumb_exit): Remove last unnecessary set of regs_to_pop.
25981 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25983 * config/s390/predicates.md: Use s390_rel_address_ok_p.
25984 * config/s390/s390-protos.h: Add prototype of
25985 s390_rel_address_ok_p.
25986 * config/s390/s390.c (s390_got_symbol): New function.
25987 (s390_rel_address_ok_p): New function.
25988 (legitimize_pic_address): Use s390_rel_address_ok_p.
25989 (s390_load_got): Use s390_got_symbol.
25990 (s390_option_override): Issue error if
25991 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
25992 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
25994 * config/s390/s390.opt: New option mpic-data-is-text-relative.
25996 2017-06-27 Andrew Pinski <apinski@cavium.com>
25998 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
25999 (X * copysign (1.0, X)): New pattern.
26000 (X * copysign (1.0, -X)): New pattern.
26001 (copysign (-1.0, CST)): New pattern.
26003 2017-06-27 Joseph Myers <joseph@codesourcery.com>
26005 * genmultilib (combination_space): Remove variable.
26006 Validate reuse rules against regular expression for any sequence
26007 of multilib options in any order.
26009 2017-06-27 Michael Collison <michael.collison@arm.com>
26011 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
26012 call aarch64_split_simd_combine.
26013 * (aarch64_combine_internal<mode>): Delete pattern.
26014 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
26015 Allow register and subreg operands.
26017 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26019 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
26020 specific need, just fallback on defaults.
26021 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
26023 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26024 Olivier Hainque <hainque@adacore.com>
26026 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
26028 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
26029 targets. Pick a default if no particular attempt applied.
26030 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
26033 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26035 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
26036 (x86_64-wrs-vxworks7): Likewise.
26038 2017-06-27 Marek Polacek <polacek@redhat.com>
26041 * ubsan.c (instrument_null): Check get_base_address's result for null.
26043 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26045 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
26047 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26049 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
26050 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
26051 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
26052 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
26053 New function types.
26054 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
26055 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
26056 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
26057 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
26058 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
26059 BUILT_IN_FEUPDATEENV): New builtins.
26060 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
26061 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
26062 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
26063 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
26065 (builtin_structptr_types): Adjust size.
26066 * tree.c (builtin_structptr_types): Add four entries.
26068 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26069 Olivier Hainque <hainque@adacore.com>
26071 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
26072 (TLS_SYM): New local macro, forcing reference to __tls__ on
26073 link command lines for VxWorks 7 RTPs, triggering initialization
26075 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
26076 OS features TLS support, true for RTPs on VxWorks 7.
26077 * config/vxworks.c (vxworks_override_options): Setup emutls
26080 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26082 * predict.c (test_prediction_value_range): Use -1U instead of -1
26083 to avoid narrowing conversion warning.
26084 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
26085 to avoid narrowing conversion warning.
26086 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
26088 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
26090 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26092 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
26093 64bit configurations.
26094 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
26095 (SIZE_TYPE): Likewise.
26096 * config/vxworks.c (vxworks_emutls_var_fields): Use
26097 long_unsigned_type_node instead of unsigned_type_node as the offset
26098 field type, which is "pointer" mode in emutls.c.
26100 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26103 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
26105 PR middle-end/81207
26106 * gimple-fold.c (replace_call_with_call_and_fold): Handle
26107 gimple_vuse copying separately from gimple_vdef copying.
26109 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26111 * value-prof.c (free_hist): Remove call to memset and the enclosing if
26114 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26115 Olivier Hainque <hainque@adacore.com>
26117 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
26118 for all vxworks7 targets.
26119 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
26120 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
26121 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
26122 variations for VX6/VX7 and 32/64bits later on in ...
26123 (VXWORKS_LIB_SPEC): Leverage new macros.
26124 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
26125 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
26127 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26129 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
26130 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
26132 2017-06-26 Carl Love <cel@us.ibm.com>
26134 * config/rs6000/rs6000-c.c: Add support for built-in functions
26135 vector bool char vec_reve (vector bool char);
26136 vector signed char vec_reve (vector signed char);
26137 vector unsigned char vec_reve (vector unsigned char);
26138 vector bool int vec_reve (vector bool int);
26139 vector signed int vec_reve (vector signed int);
26140 vector unsigned int vec_reve (vector unsigned int);
26141 vector bool long long vec_reve (vector bool long long);
26142 vector signed long long vec_reve (vector signed long long);
26143 vector unsigned long long vec_reve (vector unsigned long long);
26144 vector bool short vec_reve (vector bool short);
26145 vector signed short vec_reve (vector signed short);
26146 vector double vec_reve (vector double);
26147 vector float vec_reve (vector float);
26148 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
26149 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
26150 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
26151 (altivec_vreve): New pattern.
26152 * config/rs6000/altivec.h (vec_reve): New define.
26153 * doc/extend.texi (vec_rev): Update the built-in documentation file
26154 for the new built-in functions.
26156 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26158 PR tree-optimization/71815
26159 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
26161 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
26163 (slsr_process_phi): Likewise.
26164 (replace_uncond_cands_and_profitable_phis): Don't replace a
26165 multiply candidate with a stride of 1 (copy or cast).
26166 (phi_incr_cost): Call uses_consumed_by_stmt rather than
26168 (lowest_cost_path): Likewise.
26169 (total_savings): Likewise.
26171 2017-06-26 Richard Biener <rguenther@suse.de>
26174 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
26175 Use def_builtin_pure for all gather builtins.
26177 2017-06-26 Richard Biener <rguenther@suse.de>
26179 PR tree-optimization/81203
26180 * tree-tailcall.c (find_tail_calls): Do not move stmts into
26181 non-dominating BBs.
26183 2017-06-26 Marek Polacek <polacek@redhat.com>
26186 * doc/invoke.texi: Document -Wmultistatement-macros.
26188 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
26190 * doc/sourcebuild.texi (ARM-specific attributes): Document new
26191 arm_neon_ok_no_float_abi effective target.
26193 2017-06-26 Richard Biener <rguenther@suse.de>
26195 PR tree-optimization/80928
26196 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
26197 (copy_bbs): Set BB_DUPLICATED flag early.
26198 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
26200 (execute_on_shrinking_pred): Likewise.
26201 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
26202 BB_DUPLICATED blocks.
26203 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
26204 iterate over all PHIs considering removal of *gsi.
26206 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
26208 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
26211 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26213 * config/rs6000/rs6000-string.c: (expand_block_clear,
26214 do_load_for_compare, select_block_compare_mode,
26215 compute_current_alignment, expand_block_compare,
26216 expand_strncmp_align_check, expand_strn_compare,
26217 expand_block_move, rs6000_output_load_multiple)
26218 Move functions related to string/block move/compare
26219 to a separate file.
26220 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
26221 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
26222 for this function which is now used in two files.
26223 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
26224 * config.gcc: Add rs6000-string.o to extra_objs for
26225 targets powerpc*-*-* and rs6000*-*-*.
26227 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
26230 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
26231 32-bit, since indexed is not valid for DImode.
26232 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
26233 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
26234 (define_peephole2 for Altivec d-form load): Add 32-bit support.
26235 (define_peephole2 for Altivec d-form store): Likewise.
26238 * multiple_target.c (create_dispatcher_calls): Only create the
26239 dispatcher call if the function is the default clone of a
26240 versioned function.
26242 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
26244 PR middle-end/80902
26245 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
26246 a call, force the call to not be a tail call.
26248 2017-06-23 Jeff Law <law@redhat.com>
26250 * doc/contrib.texi: Add entry for Steven Pemberton's work on
26253 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
26255 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
26256 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
26257 handling for early expansion of vector shifts (sl,sr,sra,rl).
26258 (builtin_function_type): Add vector shift right instructions
26259 to the unsigned argument list.
26261 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
26263 rtl-optimizatoin/79286
26264 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
26265 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
26266 trap. PIC register plus a const unspec without offset can never trap.
26268 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
26270 * tree.h (builtin_structptr_type): New type.
26271 (builtin_structptr_types): Declare new array.
26272 * tree.c (builtin_structptr_types): New array.
26273 (free_lang_data, build_common_tree_nodes): Use it.
26275 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
26278 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
26281 2017-06-22 Matt Turner <mattst88@gmail.com>
26283 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
26284 Lake models to skylake case. Assume skylake for unknown
26285 models with clflushopt.
26287 2017-06-22 Jeff Law <law@redhat.com>
26289 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
26290 frame sizes that do not satisfy aarch64_uimm12_shift.
26292 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
26294 * profile-count.h (apply_probability,
26295 apply_scale, probability_in): Fix checks for zero.
26297 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26299 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
26300 * doc/cppdiropts.texi (-I @var{dir}): Document it.
26302 2016-06-22 Richard Biener <rguenther@suse.de>
26304 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26305 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
26306 REDUC_MAX_EXPR support.
26307 (vectorizable_reduction): Likewise.
26308 (vect_create_epilog_for_reduction): Likewise.
26310 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
26312 * match.pd (A / (1 << B) -> A >> B): New.
26313 * generic-match-head.c: Include optabs-tree.h.
26314 * gimple-match-head.c: Likewise.
26315 * optabs-tree.h (target_supports_op_p): New.
26316 * optabs-tree.c (target_supports_op_p): New.
26318 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26320 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
26321 $gcc_cv_ld --help output.
26322 (gcc_cv_ld_demangle): Likewise.
26323 (gcc_cv_ld_eh_frame_hdr): Likewise.
26324 (gcc_cv_ld_pie): Likewise.
26325 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
26326 (gcc_cv_ld_buildid): Likewise.
26327 (gcc_cv_ld_sysroot): Likewise.
26328 (ld_bndplt_support): Likewise.
26329 (ld_pushpopstate_support): Likewise.
26330 * configure: Regenerate.
26331 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
26333 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26336 * config/i386/sse.md (round<mode>2): Renumber match_dup and
26337 operands indexes to avoid gap between operands and match_dups.
26339 2017-06-21 Andrew Pinski <apinski@cavium.com>
26341 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
26342 Increment Arith_shift and Arith_shift_reg by 1.
26343 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
26345 * config/aarch64/aarch64.c (thunderx_tunings): Enable
26346 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
26347 (aarch64_strip_extend): Add new argument and test for it.
26348 (aarch64_cheap_mult_shift_p): New function.
26349 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
26350 add a cost if it is true.
26351 Update calls to aarch64_strip_extend.
26352 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
26354 2017-06-21 Andrew Pinski <apinski@cavium.com>
26356 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
26358 (thunderxt88): Likewise.
26359 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
26360 (thunderx_prefetch_tune): New variable.
26361 (thunderx2t99_prefetch_tune): Update for the correct values.
26362 (thunderxt88_tunings): New variable.
26363 (thunderx_tunings): Use thunderx_prefetch_tune instead of
26364 generic_prefetch_tune.
26365 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
26367 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26369 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
26370 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
26371 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
26372 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
26373 (aarch64_atomic_cas<mode>, GPI): Likewise.
26375 2017-06-21 Martin Liska <mliska@suse.cz>
26377 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
26378 statements on cold and hot labels.
26379 * predict.c (tree_estimate_probability_bb): Remove the
26380 prediction from this place.
26382 2017-06-21 Martin Liska <mliska@suse.cz>
26384 PR tree-optimization/79489
26385 * gimplify.c (maybe_add_early_return_predict_stmt): New
26387 (gimplify_return_expr): Call the function.
26388 * predict.c (tree_estimate_probability_bb): Remove handling
26390 * predict.def: Update comment about early return predictor.
26391 * gimple-predict.h (is_gimple_predict): New function.
26392 * predict.def: Change default value of early return to 66.
26393 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
26395 * passes.def: Put pass_strip_predict_hints to the beginning of
26398 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
26400 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
26401 FUNCTION_DECL declarations.
26402 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
26404 (dwaf2out_decl): Likewise.
26405 * godump.c (go_early_global_decl): Skip call to the real debug hook
26406 for FUNCTION_DECL declarations.
26407 * passes.c (rest_of_decl_compilation): Skip call to the
26408 early_global_decl debug hook for FUNCTION_DECL declarations, unless
26409 -fdump-go-spec is passed.
26411 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26413 * config/i386/i386.c (struct builtin_isa): New field pure_p.
26414 Reorder for compactness.
26415 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
26416 (def_builtin_pure, def_builtin_pure2): New functions.
26417 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
26419 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26421 * match.pd (nop_convert): New predicate.
26422 ((A +- CST1) +- CST2): Allow some NOP conversions.
26424 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26427 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
26428 with ctors/dtors if GOVD_SHARED is set.
26430 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26432 * config/aarch64/aarch64.md (movti_aarch64):
26433 Emit mov rather than orr.
26434 (movtf_aarch64): Likewise.
26435 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
26436 Emit mov rather than orr.
26438 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26440 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
26441 Swap alternatives, make integer dup more expensive.
26443 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26445 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
26446 Return true for non-tls symbols.
26448 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
26450 * config/aarch64/aarch64-cores.def (cortex-a55): New.
26451 (cortex-a75): Likewise.
26452 (cortex-a75.cortex-a55): Likewise.
26453 * config/aarch64/aarch64-tune.md: Regenerate.
26454 * doc/invoke.texi (-mtune): Document new values for -mtune.
26456 2017-06-21 Tom de Vries <tom@codesourcery.com>
26458 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
26459 stack_size feature.
26460 (Effective-Target Keywords, Other attributes): Suggest using
26461 dg-add-options stack_size feature to get stack limit in stack_size
26462 effective target documentation.
26464 2017-06-21 Julian Brown <julian@codesourcery.com>
26465 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26467 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
26468 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
26469 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
26471 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
26472 attribute type list for neon_multiply.
26473 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
26474 attribute type list for neon_multiply.
26475 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
26476 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
26477 attribute type list for neon_multiply.
26478 * config/arm/types.md (crypto_pmull): Add.
26479 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
26480 attribute type list.
26482 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
26484 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
26487 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26489 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
26490 to make sure not to dereference a NULL cost_classes_ptr pointer.
26492 2017-06-20 Carl Love <cel@us.ibm.com>
26494 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26495 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
26496 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
26497 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
26498 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
26499 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
26500 VMULOSW): New enum "unspec" values.
26501 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
26502 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
26503 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
26504 altivec_vmulosw): New patterns.
26505 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
26506 VMULOSW): Add definitions.
26508 2017-06-20 Julia Koval <julia.koval@intel.com>
26510 * config/i386/i386.c: Fix rounding expand for new pattern.
26511 * config/i386/subst.md: Fix pattern (parallel -> unspec).
26513 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26515 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
26516 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
26518 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26520 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
26523 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26525 * config/aarch64/aarch64-cores.def: Rearrange to sort by
26526 architecture, then by implementer ID.
26527 * config/aarch64/aarch64-tune.md: Regenerate.
26529 2017-06-20 Richard Biener <rguenther@suse.de>
26531 PR middle-end/81097
26532 * fold-const.c (split_tree): Fold to type before negating.
26534 2017-06-20 David Malcolm <dmalcolm@redhat.com>
26536 * diagnostic-show-locus.c
26537 (selftest::test_fixit_deletion_affecting_newline): New function.
26538 (selftest::diagnostic_show_locus_c_tests): Call it.
26540 2017-06-20 Andreas Schwab <schwab@suse.de>
26543 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
26546 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26548 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
26549 __ARM_FEATURE_COPROC according to support.
26551 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26553 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26554 Rewritten to avoid overflow for > 32-bit pointers.
26557 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
26558 by removing enum keyword.
26559 (ubsan_type_descriptor): Likewise. Formatting fix.
26562 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
26563 splitter): Require TARGET_SSE2 in the condition.
26565 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
26568 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
26569 for doing vector set of SFmode on ISA 3.0.
26570 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
26571 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
26573 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
26574 SFmode value into a V4SF variable that was extracted from another
26575 V4SF variable without converting the element to double precision
26576 and back to single precision vector format.
26577 (vsx_insert_extract_v4sf_p9_2): Likewise.
26579 2017-06-19 Jakub Jelinek <jakub@redhat.com>
26581 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
26582 in UWHI to avoid undefined overflow.
26585 * ubsan.h (enum ubsan_encode_value_phase): New.
26586 (ubsan_encode_value): Change second argument to
26587 enum ubsan_encode_value_phase with default value of
26588 UBSAN_ENCODE_VALUE_GENERIC.
26589 * ubsan.c (ubsan_encode_value): Change second argument to
26590 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
26591 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
26592 create_tmp_var_raw instead of create_tmp_var and use a
26594 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
26595 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
26596 ubsan_encode_value callers.
26599 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
26600 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
26601 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
26603 2017-06-19 Richard Biener <rguenther@suse.de>
26605 PR middle-end/81118
26606 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
26607 estimates if we changed anything.
26609 2017-06-19 Richard Biener <rguenther@suse.de>
26611 PR tree-optimization/80887
26612 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
26613 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
26614 simplified lookups, then reset mprts_hook.
26615 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
26617 (try_to_simplify): Likewise.
26619 2017-06-19 Martin Liska <mliska@suse.cz>
26622 * gimplify.c (gimplify_switch_expr):
26623 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
26625 2017-06-19 Martin Liska <mliska@suse.cz>
26627 * doc/install.texi: Document that PGO runs in 4 stages.
26629 2017-06-19 Martin Liska <mliska@suse.cz>
26632 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
26633 to dispatcher function name.
26634 * multiple_target.c (replace_function_decl): New function.
26635 (create_dispatcher_calls): Redirect both edges and references.
26637 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
26639 * profile-count.c (profile_count::dump): Dump quality.
26640 (profile_count::differs_from_p): Update for unsigned val.
26641 * profile-count.h (profile_count_quality): New enum.
26642 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
26644 2017-06-19 Richard Biener <rguenther@suse.de>
26646 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
26647 struct function as arg.
26648 (estimate_numbers_of_iterations): Export overload with loop arg.
26649 (free_numbers_of_iterations_estimates_loop): Use an overload of
26650 free_numbers_of_iterations_estimates instead.
26651 * tree-cfg.c (remove_bb): Adjust.
26652 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
26653 * tree-parloops.c (gen_parallel_loop): Likewise.
26654 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
26656 (tree_unroll_loops_completely): Likewise.
26657 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
26658 Use an overload instead and export.
26659 (estimated_loop_iterations): Adjust.
26660 (max_loop_iterations): Likewise.
26661 (likely_max_loop_iterations): Likewise.
26662 (estimate_numbers_of_iterations): Take struct function as arg
26664 (loop_exits_before_overflow): Adjust.
26665 (free_numbers_of_iterations_estimates_loop): Use an overload.
26666 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
26667 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
26669 2017-06-19 Richard Biener <rguenther@suse.de>
26672 * ipa-prop.c (find_constructor_constant_at_offset): Handle
26673 RANGE_EXPR conservatively.
26675 2017-06-16 Carl Love <cel@us.ibm.com>
26677 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26678 definitions for vec_float, vec_float2, vec_floato,
26679 vec_floate built-ins.
26680 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
26681 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
26683 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
26684 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
26685 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
26686 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
26687 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
26688 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
26689 vec_floato): Add builtin defines.
26690 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
26691 Update the built-in documentation file for the new built-in
26694 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26696 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
26697 (mthumb): Mark as the negative of -marm.
26699 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26701 * doc/invoke.texi (ARM Options, -mcpu): Document supported
26703 (ARM Options, -mtune): Document that this accepts the same
26704 extension options as -mcpu.
26705 (ARM Options, -mfpu): Document addition of -mfpu=auto.
26707 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26709 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
26710 permitted extensions.
26712 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26714 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
26715 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
26716 (armv8-m.main): Add option +nodsp.
26717 * config/arm/arm-cpu-cdata.h: Regenerated.
26719 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26721 * config/arm/t-fuchsia: New file.
26722 * config.gcc (arm*-*-fuchsia*): Use it.
26724 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26726 * config/arm/t-symbian: Rewrite for new option infrastructure.
26728 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26730 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
26731 (MULTILIB_REQUIRED): Likewise.
26733 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26735 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
26736 (MULTILIB_RESUE): Likewise.
26737 (MULTILIB_MATCHES): Likewise.
26738 (MULTLIB_REQUIRED): Likewise.
26740 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26742 * config/arm/t-rtems: Rewrite for new option framework.
26744 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26746 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
26747 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
26748 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
26749 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
26750 * config/arm/t-multilib: ... here.
26751 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
26752 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
26753 armv7-a and armv8*-a when A-profile libraries have not been built.
26754 * config/arm/t-rmprofile: Rewrite.
26756 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26758 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
26759 with a backslash. Remove the backslash after substituting unescaped
26761 * doc/fragments.texi (MULTILIB_REUSE): Document it.
26763 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26765 * config.gcc: (arm*-*-*): When building a-profile libraries, force
26766 the driver to pass through the default setting of -mfloat-abi.
26767 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
26769 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
26770 (all_feat_combs): New rule.
26771 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
26773 * config/arm/t-aprofile: Rewrite.
26775 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26777 * config/arm/arm.h (FPUTYPE_AUTO): Define.
26778 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
26779 fpu is not specified by the user/command-line.
26780 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
26781 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
26782 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
26783 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
26784 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
26785 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
26787 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26789 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
26790 * config/arm/t-arm-elf: Rewritten.
26792 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26794 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
26795 have some floating-point instructions.
26796 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
26797 (TARGET_MAYBE_HARD_FLOAT): New macro.
26798 * config/arm/arm-builtins.c (arm_init_builtins): Use
26799 TARGET_MAYBE_HARD_FLOAT.
26800 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
26802 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26804 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
26805 (configargs.h): Include it.
26806 (arm_print_hint_for_fpu_option): New function.
26807 (arm_parse_fpu_option): New function.
26808 (candidate_extension): New class.
26809 (arm_canon_for_multilib): New function.
26810 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
26811 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
26812 (ARCH_CANONICAL_SPECS): New macro.
26813 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
26815 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26817 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
26818 are set after handling multilib fragments. Set target_cpu_default2
26821 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26823 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
26825 (arm*-*-*): Set target_cpu_default2 to a quoted string.
26826 * config/arm/parsecpu.awk (check_cpu): Validate any extension
26828 (check_arch): Likewise.
26829 * config/arm/arm.c (arm_configure_build_target): Handle
26830 TARGET_CPU_DEFAULT being a string constant. Scan any feature
26831 options in the default.
26833 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26835 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
26836 when an option is an alias of another.
26837 * config/arm/parsecpu.awk (optalias): New parser token.
26838 (gen_comm_data): Mark non-alias options as such. Emit entries
26839 for extension aliases.
26840 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
26841 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
26842 (armv6kz, armv6zk, armv6t2): Likewise.
26843 (armv7): Make vfpv3-d16 an alias.
26844 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
26846 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
26847 Sort in canonical order.
26848 (armv8-a): Sort in canonical order.
26849 (armv8.1-a, armv8.2-a): Likewise.
26850 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
26852 (cortex-a9): Sort in canonical order.
26853 * config/arm/arm.c (selftests.h): Include it.
26854 (arm_test_cpu_arch_data): New function.
26855 (arm_run_self_tests): New function.
26856 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
26857 (targetm): Move declaration to the end of the file.
26858 * arm-cpu-cdata.h: Regenerated.
26860 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26862 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
26863 call to target_mode_check describing the type of option passed.
26864 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
26865 (arm_target_thumb_only): Use arm_parse_arch_option_name or
26866 arm_parse_cpu_option_name to match parameters against list of
26868 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
26869 arm_arch_core_flags data structure.
26870 * config/arm/arm-cpu_cdata.h: Regenerated.
26872 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26874 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
26876 (arm_print_hint_for_cpu_option): Likewise.
26877 (arm_print_hint_for_arch_option): Likewise.
26878 (arm_parse_cpu_option_name): Likewise.
26879 (arm_parse_arch_option_name): Likewise.
26880 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
26881 of entries in the all_fpus list.
26882 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
26883 (arm_parse_cpu_option_name): Declare.
26884 (arm_parse_arch_option_name): Declare.
26885 (arm_parse_option_features): Declare.
26886 (arm_intialize_isa): Declare.
26887 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
26889 (gen_comm_data): ... here. Make definitions non-static.
26890 * config/arm/arm-cpu-data.h: Regenerated.
26891 * config/arm/arm-cpu-cdata.h: Regenerated.
26893 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26895 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
26896 (cpu_arch_extension): New structure.
26897 (cpu_arch_option, arch_option, cpu_option): New structures.
26898 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
26899 architecture types.
26900 (gen_data): Generate new format data tables.
26901 * config/arm/arm.c (cpu_tune): New structure.
26902 (cpu_option, processors): Delete.
26903 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
26904 (arm_print_hint_for_cpu_option): ... this and ...
26905 (arm_print_hint_for_arch_option): ... this.
26906 (arm_parse_arch_cpu_name): Delete. Replace with ...
26907 (arm_parse_cpu_option_name): ... this and ...
26908 (arm_parse_arch_option_name): ... this.
26909 (arm_unrecognized_feature): Change type of target parameter to
26911 (arm_parse_arch_cpu_features): Delete. Replace with ...
26912 (arm_parse_option_features): ... this.
26913 (arm_configure_build_target): Rework to use new configuration data
26915 (arm_print_tune_info): Rework for new configuration data tables.
26916 * config/arm/arm-cpu-data.h: Regenerated.
26917 * config/arm/arm-cpu.h: Regenerated.
26919 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26921 * Makefile.in (OBJS): Move sbitmap.o from here ...
26922 (OBJS-libcommon): ... to here.
26924 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26926 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
26927 (ISA_ALL_CRYPTO): New macro.
26928 (ISA_ALL_SIMD): New macro
26929 (ISA_ALL_FP): New macro.
26930 * config/arm/arm.c (fpu_bitlist): Update initializer.
26931 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
26933 (arm9e): Add fpu. Add option for nofp
26934 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
26935 (arm926ej-s, arm1026ej-s): Likewise.
26936 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
26937 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
26938 neon-fp16, neon-vfpv4, nofp and nosimd.
26939 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
26940 (cortex-a8): Add fpu. Add option for nofp.
26941 (cortex-a9): Add fpu. Add options for nosimd and nofp.
26942 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
26943 (cortex-r4f): Add fpu.
26944 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
26945 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
26947 (cortex-r8): Likewise.
26948 (cortex-m4): Add fpu. Add option for nofp.
26949 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
26950 (cortex-a17.cortex-a7): Likewise.
26951 (cortex-a32): Add fpu. Add options for crypto and nofp.
26952 (cortex-a35, cortex-a53): Likewise.
26953 (cortex-a57): Add fpu. Add option for crypto.
26954 (cortex-a72, cortex-a73): Likewise.
26955 (exynos-m1): Likewise.
26956 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
26957 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
26958 (cortex-m33): Add fpu. Add option for nofp.
26959 * config/arm/arm-cpu-cdata.h: Regenerated
26960 * config/arm/arm-cpu-data.h: Regenerated.
26962 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26964 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
26965 (armv5te, armv5tej): Likewise.
26966 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
26967 (armv7): Add options fp and vfpv3-d16.
26968 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
26969 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
26971 (armv7ve): Likewise.
26972 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
26973 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
26974 (armv8-a): Add nocrypto option.
26975 (armv8.1-a, armv8.2-a): Likewise.
26976 (armv8-m.main): add options fp, fp.dp and nofp.
26978 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26980 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
26982 (armv8-a+crc): Delete.
26983 (armv8.1-a): Add options simd, crypto and nofp.
26984 (armv8.2-a): Add options fp16, simd, crypto and nofp.
26985 (armv8.2-a+fp16): Delete.
26986 (armv8-m.main): Add option dsp.
26987 (armv8-m.main+dsp): Delete.
26988 (cortex-a8): Add fpu. Add nofp option.
26989 (cortex-a9): Add fpu. Add nofp and nosimd options.
26990 * config/arm/parsecpu.awk (gen_data): Generate option tables and
26991 link to main cpu and architecture data structures.
26992 (gen_comm_data): Only put isa attributes from the main architecture
26994 (option): New statement for architecture and CPU entries.
26995 * arm.c (struct cpu_option): New structure.
26996 (struct processors): Add entry for options.
26997 (arm_unrecognized_feature): New function.
26998 (arm_parse_arch_cpu_name): Ignore any characters after the first
27000 (arm_parse_arch_cpu_feature): New function.
27001 (arm_configure_build_target): Separate out any CPU and architecture
27002 features and parse separately. Don't error out if -mfpu=auto is
27003 used with only an architecture string.
27004 (arm_print_asm_arch_directives): New function.
27005 (arm_file_start): Call it.
27006 * config/arm/arm-cpu-cdata.h: Regenerated.
27007 * config/arm/arm-cpu-data.h: Likewise.
27008 * config/arm/arm-tables.opt: Likewise.
27010 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27012 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
27013 assembler when it is not -mfpu=auto.
27015 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27017 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
27018 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
27019 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
27020 (ASM_CPU_SPEC): Rewrite.
27021 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
27022 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
27023 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
27024 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
27025 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
27026 copied string is NUL-terminated. Also strip any characters prefixed
27028 (arm_rewrite_selected_arch): New function.
27029 (arm_rewrite_march): New function.
27031 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27033 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
27034 (x_arm_cpu_string, x_arm_tune_string): Likewise.
27035 (march, mcpu, mtune): Convert to string-based options.
27036 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
27037 (arm_parse_arch_cpu_name): New function.
27038 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
27039 identify selected architecture or CPU.
27040 (arm_option_save): New function.
27041 (TARGET_OPTION_SAVE): Redefine.
27042 (arm_option_restore): Restore string options.
27043 (arm_option_print): Print string options.
27045 2017-06-16 Martin Sebor <msebor@redhat.com>
27047 PR tree-optimization/80933
27048 PR tree-optimization/80934
27049 * builtins.c (fold_builtin_3): Do not handle bcmp here.
27050 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
27051 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
27052 (gimple_fold_builtin): Call them.
27054 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27056 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
27057 as unlikely; update profile.
27059 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27061 * predict.c (force_edge_cold): Handle declaring edges impossible
27064 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27066 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
27068 (try_unroll_loop_completely): Fix reporting.
27070 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27072 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
27074 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
27077 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
27078 if given a non-constant argument for an intrinsic which requires a
27081 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27083 * profile.c (compare_freqs): New function.
27084 (branch_prob): Sort edge list.
27085 (find_spanning_tree): Assume that the list is priority sorted.
27087 2017-06-16 Richard Biener <rguenther@suse.de>
27089 PR tree-optimization/81090
27090 * passes.def (pass_record_bounds): Remove.
27091 * tree-pass.h (make_pass_record_bounds): Likewise.
27092 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
27093 make_pass_record_bounds): Likewise.
27094 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
27095 not free niter estimates at the beginning but at the end.
27096 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
27098 2017-06-16 Richard Biener <rguenther@suse.de>
27100 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
27101 initializer to workaround ICE in host GCC 4.8.
27103 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27105 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
27107 (clone_inlined_nodes): Update.
27109 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27111 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
27112 prefetch settings, and enable prefetching by default at -O3.
27114 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27116 * config/aarch64/aarch64.c (aarch64_override_options_internal):
27117 Set flag_prefetch_loop_arrays according to tuning data.
27119 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27121 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
27122 New tune structure.
27123 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
27124 [Unrelated to main purpose of the patch] Place the pointer field last
27125 to enable type checking errors when tune structure are wrongly merged.
27126 * config/aarch64/aarch64.c (generic_prefetch_tune,)
27127 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
27128 (thunderx2t99_prefetch_tune): New tune constants.
27129 (tune_params *_tunings): Update all tunings (no functional change).
27130 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
27131 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
27132 from tunings structures.
27134 2017-06-16 Jakub Jelinek <jakub@redhat.com>
27137 * ubsan.c (instrument_null): Add T argument, use it instead
27138 of computing it based on IS_LHS.
27139 (instrument_object_size): Likewise.
27140 (pass_ubsan::execute): Adjust instrument_null and
27141 instrument_object_size callers to pass gimple_get_lhs or
27142 gimple_assign_rhs1 result to it. Use instrument_null instead of
27143 calling get_base_address and instrument_mem_ref. Handle
27144 aggregate call arguments for object-size sanitization.
27146 2017-06-16 Yury Gribov <tetra2005@gmail.com>
27148 PR tree-optimization/81089
27149 * tree-vrp.c (is_masked_range_test): Validate operands of
27152 2017-06-15 Martin Sebor <msebor@redhat.com>
27155 * dumpfile.c (dump_register): Avoid calling memset to initialize
27156 a class with a default ctor.
27157 * gcc.c (struct compiler): Remove const qualification.
27158 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
27159 * hash-table.h: Ditto.
27160 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
27162 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
27163 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
27165 * params.h (struct param_info): Make struct members non-const.
27166 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
27167 with default initialization.
27168 * vec.h (vec_copy_construct, vec_default_construct): New helper
27170 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
27171 with vec_copy_construct.
27172 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
27173 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
27174 * doc/invoke.texi (-Wclass-memaccess): Document.
27176 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27178 * emit-rtl.h (is_leaf): Update comment about local
27179 register allocator.
27181 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27184 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
27185 for a variable to have a section before checking if the section has a
27187 Set section to.persistent if persistent attribute is set.
27188 Warn if .persistent attribute is used on an automatic variable.
27190 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
27192 PR rtl-optimization/80474
27193 * reorg.c (update_block): Do not ignore instructions in a delay slot.
27195 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
27197 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
27200 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
27202 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
27203 (casesi): Emit bounds checking as RTL.
27204 (casesi_internal_mips16_<mode>): Remove bounds checking.
27206 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
27208 * config/xtensa/xtensa.c (xtensa_option_override): Append
27209 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
27210 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
27211 xtensa_doloop_hooks): Define unconditionally.
27212 (xtensa_reorg_loops): Only call reorg_loops in the presence of
27214 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
27215 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
27216 for it in xtensa_option_override.
27217 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
27218 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
27220 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
27222 * doc/cppopts.texi: Document '-' special value to -MF.
27224 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
27226 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
27227 (cortex_a53_fconst): Likewise.
27228 (cortex_a53_fpmul): Likewise.
27229 (cortex_a53_f_load_64): Likewise.
27230 (cortex_a53_f_load_many): Likewise.
27231 (cortex_a53_advsimd_alu): Likewise.
27232 (cortex_a53_advsimd_alu_q): Likewise.
27233 (cortex_a53_advsimd_mul): Likewise.
27234 (cortex_a53_advsimd_mul_q): Likewise.
27235 (fpmac bypass): Add new bypass for fpmac-fpmac case.
27236 Add missing fmul, r2f_cvt and fconst cases.
27238 2017-06-14 Richard Biener <rguenther@suse.de>
27240 PR middle-end/81088
27241 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
27243 (fold_binary_loc): When associating do not treat pre-existing
27244 TREE_OVERFLOW on literal constants as a reason to allow
27245 TREE_OVERFLOW on associated literal constants.
27247 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27249 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
27250 (MASK_FEATURES): New macro.
27251 * config/sparc/sparc.c (sparc_option_override): Remove the special
27252 handling of -mfpu and generalize it to all MASK_FEATURES switches.
27254 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27256 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
27257 a division of 0 if non-call exceptions are enabled.
27259 2017-06-14 Andrew Pinski <apinski@cavium.com>
27260 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27263 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
27264 Improve vector initialization code gen for only variable case.
27266 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27268 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
27270 2017-06-14 Richard Biener <rguenther@suse.de>
27272 PR tree-optimization/81083
27273 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
27276 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27278 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
27279 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
27280 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
27281 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
27282 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
27283 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
27285 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27287 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
27288 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
27290 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27292 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
27294 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27296 * config/rs6000/t-rtems: Don't handle SPE.
27298 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27300 * config/rs6000/t-linux: Don't handle SPE.
27302 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27304 * config/rs6000/eabispe.h: Delete file.
27306 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27308 * config/rs6000/t-spe: Delete file.
27310 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27312 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
27313 (rs6000_legitimate_offset_address_p): Return false for anything in
27314 V2SImode or V2SFmode.
27316 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27318 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
27319 except V2SF and V2SI. Rearrange the vector modes, and add comments.
27320 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
27322 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
27323 (rs6000_legitimate_offset_address_p): Ditto.
27324 (rs6000_emit_move): Ditto.
27325 (rs6000_init_builtins): Remove V4HI_type_node.
27327 2017-06-13 Martin Liska <mliska@suse.cz>
27330 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
27331 (gate_asan): Likewise.
27332 * asan.h (asan_no_sanitize_address_p): Remove the function.
27333 (sanitize_flags_p): New function.
27334 * builtins.def: Fix coding style.
27335 * common.opt: Use renamed enum value.
27336 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
27337 * doc/extend.texi: Document no_sanitize attribute.
27338 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
27339 to SANITIZE_UNDEFINED_NONDEFAULT.
27340 * gcc.c (sanitize_spec_function): Use the renamed enum value.
27341 * gimple-fold.c (optimize_atomic_compare_exchange_p):
27342 Use sanitize_flags_p.
27343 * gimplify.c (gimplify_function_tree): Likewise.
27344 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
27345 * opts.c (parse_no_sanitize_attribute): New function.
27346 (common_handle_option): Use renamed enum value.
27347 * opts.h (parse_no_sanitize_attribute): Declare.
27348 * tree.c (sanitize_flags_p): New function.
27349 * tree.h: Declared here.
27350 * tsan.c: Use sanitize_flags_p.
27351 * ubsan.c (ubsan_expand_null_ifn): Likewise.
27352 (instrument_mem_ref): Likewise.
27353 (instrument_bool_enum_load): Likewise.
27354 (do_ubsan_in_current_function): Remove the function.
27355 (pass_ubsan::execute): Use sanitize_flags_p.
27356 * ubsan.h: Remove do_ubsan_in_current_function
27357 * tree-cfg.c (print_no_sanitize_attr_value): New function.
27358 (dump_function_to_file): Use it here.
27360 2017-06-13 Martin Jambor <mjambor@suse.cz>
27362 PR tree-optimization/80803
27363 PR tree-optimization/81063
27364 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
27365 (propagate_subaccesses_across_link): Enqueue subtree whenever
27366 necessary instead of relying on the caller.
27368 2017-06-13 Martin Jambor <mjambor@suse.cz>
27370 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
27371 that have a first_link.
27372 (sort_and_splice_var_accesses): Do not check first_link before
27374 (subtree_mark_written_and_enqueue): Likewise.
27375 (propagate_all_subaccesses): Likewise and do not stop at first
27376 parent with a first_link.
27378 2017-06-13 Martin Jambor <mjambor@suse.cz>
27380 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
27383 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27385 * match.pd: New pattern.
27387 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27389 * tree-vrp.c (is_masked_range_test): New function.
27390 (register_edge_assert_for): Determine ranges for
27393 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27395 PR tree-optimization/67328
27396 * fold-const.c (maskable_range_p): New function.
27397 (build_range_check): Generate bittests if possible.
27399 2017-06-13 Martin Liska <mliska@suse.cz>
27401 * gimple-pretty-print.c (dump_probability): Add new argument.
27402 (dump_edge_probability): Dump both probability and count.
27403 (dump_gimple_label): Likewise.
27404 (dump_gimple_bb_header): Likewise.
27406 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
27409 * config/avr/avr-devices.c: Fix indentation.
27410 * config/avr/gen-avr-mmcu-specs.c: Dito.
27412 2017-06-13 Richard Biener <rguenther@suse.de>
27414 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
27415 instead get vector type from stmt_info.
27416 (vectorizable_reduction): Adjust. Remove dead code.
27418 2017-06-13 Richard Biener <rguenther@suse.de>
27420 PR middle-end/81065
27421 * fold-const.c (extract_muldiv_1): Remove bogus distribution
27422 case of C * (x * C2 + C3).
27423 (fold_addr_of_array_ref_difference): Properly fold index difference.
27425 2017-06-12 David S. Miller <davem@davemloft.net>
27428 * config/sparc/sparc.md (return expander): Emit frame blockage if
27429 function uses alloca.
27431 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
27433 * combine.c (make_field_assignment): Check len rather than the mode
27434 precision when calling force_to_mode.
27436 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
27438 Support multilibs and devices that see flash in RAM address range.
27441 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
27442 (avr_mcu_t) <flash_pm_offset>: New field.
27443 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
27444 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
27445 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
27446 (AVR_TINY_PM_OFFSET): Remove macro.
27447 * config/avr/avr.opt (-mshort-calls): New option.
27448 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
27449 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
27450 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
27451 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
27452 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
27453 instead of avr_arch->have_jmp_call.
27454 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
27455 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
27456 avr_arch->flash_pm_offset to define.
27457 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
27458 new field flash_pm_offset. Add entry for avrxmega3.
27459 (avr_texinfo): Add entry for avrxmega3.
27460 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
27461 attiny212, attiny214,
27462 attiny412, attiny414, attiny416, attiny417,
27463 attiny814, attiny816, attiny817,
27464 attiny1614, attiny1616, attiny1617,
27465 attiny3214, attiny3216, attiny3217.
27466 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
27467 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
27468 (avr_print_operand_address) [AVR_TINY]: Same.
27469 (avr_asm_init_sections) <readonly_data_section>: Only patch
27470 callback if avr_arch->flash_pm_offset = 0.
27471 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
27472 for rodata if avr_arch->flash_pm_offset != 0.
27473 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
27474 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
27475 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
27476 (m_options): Append opt_rcall.
27477 (m_dirnames): Append dir_rcall.
27478 * config/avr/t-multilib: Regenerate.
27480 * configure.ac [target=avr]: Check whether avrxmega3 default
27481 linker description file works as needed.
27482 * configure: Regenerate.
27483 * doc/avr-mmcu.texi: Regenerate.
27484 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
27485 <__AVR_ARCH__>: Document avrxmega3 and 103.
27486 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
27487 <__AVR_SHORT_CALLS__>: Document it.
27488 <__AVR_PM_BASE_ADDRESS__>: Document it.
27489 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
27490 (AVR Variable Attributes) <progmem>: Document this is
27491 not needed for avrxmega3.
27492 (AVR Named Address Spaces) <__flash>: Dito.
27494 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
27496 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
27498 2017-06-12 Doug Rupp <rupp@adacore.com>
27500 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
27501 Append vxworks-stdint.h to the tm_file list.
27502 * config/vxworks-stdint.h: New file.
27504 2017-06-12 Martin Liska <mliska@suse.cz>
27506 PR tree-optimization/81041
27507 * tree-profile.c (gimple_gen_ic_func_profiler):
27508 Create an extra BB in profile-generate
27509 (gimple_gen_time_profiler): Likewise.
27511 2017-06-12 Jakub Jelinek <jakub@redhat.com>
27513 PR tree-optimization/81003
27514 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
27515 (update_range_test): Use it instead of force_gimple_operand_gsi.
27517 2017-06-12 Richard Biener <rguenther@suse.de>
27519 PR tree-optimization/81053
27520 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
27521 with backedge value not defined in loop. Simplify def stmt
27524 2017-06-11 Tom de Vries <tom@codesourcery.com>
27527 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
27529 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
27530 nvptx_cannot_force_const_mem.
27532 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27534 * opts.c (finish_options): Move test for flag_split_stack after
27535 it has been initialized.
27537 2017-06-11 Jason Merrill <jason@redhat.com>
27539 * tree.h (id_equal): New.
27540 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
27541 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
27542 instead of strcmp of IDENTIFIER_POINTER.
27544 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27546 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
27547 (mark_all_inlined_calls_cdtor): Fix formating.
27548 (inline_transform): Rescale profile before inlining.
27550 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27552 * cgraph.h (cgraph_edge::clone): Update prototype.
27553 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
27554 (cgraph_node::create_clone): Update.
27555 (cgraph_node::create_version_clone): Update.
27556 * tree-inline.c (copy_bb): Update.
27557 (expand_call_inline): Update.
27559 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27561 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
27562 factored out from ...
27563 (rs6000_emit_prologue): ... here.
27565 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27567 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
27568 factored out from ...
27569 (rs6000_emit_prologue): ... here.
27571 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27573 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
27575 (handle_missing_profiles): Fix computation of tp_first_run.
27576 (counts_to_freqs): Do not touch freqs when count is 0.
27578 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27580 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
27583 2017-06-10 Tom de Vries <tom@codesourcery.com>
27585 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27586 attributes): Document signal effective target.
27588 2017-06-10 Tom de Vries <tom@codesourcery.com>
27590 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27591 Document effective target stack_size.
27593 2017-06-09 David Malcolm <dmalcolm@redhat.com>
27595 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
27596 to the edit_context if they can be auto-applied.
27598 2017-06-9 Ian Lance Taylor <iant@golang.org>
27600 * opts.c (finish_options): If -fsplit-stack, disable implicit
27601 -forder-blocks-and-partition.
27602 * doc/invoke.texi (Optimize Options): Document that when using
27603 -fsplit-stack -forder-blocks-and-partition is not implicitly
27606 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27608 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
27609 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
27610 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
27611 * builtins.def (abort, trap, unreachable): Declare cold.
27612 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
27613 * tree-core.h (ECF_COLD): New.
27614 * tree.c (set_call_expr_flags): Handle ECF_COLD.
27615 (build_common_builtin_nodes): Mark unreachable and abort as cold.
27617 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27619 * predict.c (unlikely_executed_stmt_p): Cleanup.
27621 2017-06-09 Richard Biener <rguenther@suse.de>
27623 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
27624 model if the ref is always written to.
27626 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27628 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
27630 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27632 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
27635 2017-06-09 Tom de Vries <tom@codesourcery.com>
27638 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
27639 "target cannot support label values" when encountering LABEL_REF.
27641 2017-06-09 Martin Liska <mliska@suse.cz>
27643 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
27644 (gimple_gen_ic_func_profiler): Emit direct comparison
27645 of __gcov_indirect_call_callee with NULL.
27646 (gimple_gen_time_profiler): Change probability from
27647 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
27649 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27651 * profile.c (edge_gcov_counts): Turn to pointer.
27652 (compute_branch_probabilities, compute_branch_probabilities): Update.
27653 (branch_prob): Do not clear edge_gcov_count.
27654 * profile.h (edge_gcov_counts): Turn to pointer.
27655 (edge_gcov_count): Update.
27657 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27659 * gimple.h (gimple_check_failed): Mark cold.
27661 2017-06-09 Richard Biener <rguenther@suse.de>
27663 PR tree-optimization/66623
27664 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
27665 refactor check_reduction into two parts, properly computing
27666 whether we have to check reduction validity for outer loop
27669 2017-06-09 Richard Biener <rguenther@suse.de>
27671 PR tree-optimization/79483
27672 * graphite-scop-detection.c (order): New global.
27673 (get_order): Compute bb to order mapping that satisfies code
27674 generation constraints.
27675 (cmp_pbbs): New helper.
27676 (build_scops): Start domwalk at entry block, sort generated
27679 2017-06-09 Richard Biener <rguenther@suse.de>
27681 PR middle-end/81007
27682 * ipa-polymorphic-call.c
27683 (ipa_polymorphic_call_context::restrict_to_inner_class):
27684 Skip FIELD_DECLs with error_mark_node type.
27685 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
27688 2017-06-09 Martin Liska <mliska@suse.cz>
27690 * predict.c (struct branch_predictor): New struct.
27691 (test_prediction_value_range): New test.
27692 (predict_c_tests): New function.
27693 * selftest-run-tests.c (selftest::run_tests): Run the function.
27694 * selftest.h: Declare new tests.
27696 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
27699 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
27700 gen_add3_insn did not fail.
27701 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
27702 r0, construct that number in a temporary reg and add that reg to r0.
27703 If asked to put the result in r0 as well, fail.
27705 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
27707 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27708 for early expansion of vec_eqv.
27710 2017-06-08 Jakub Jelinek <jakub@redhat.com>
27712 PR middle-end/81005
27713 * ubsan.c (instrument_null): Avoid pointless code temporary.
27714 (pass_ubsan::execute): Instrument aggregate arguments of calls.
27716 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
27720 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
27723 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
27726 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27728 * predict.c (unlikely_executed_edge_p): Move ahead.
27729 (probably_never_executed_edge_p): Use it.
27731 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27733 PR middle-end/79988
27734 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
27735 gimple_call_builtin_p call.
27737 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27739 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
27740 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
27741 rtl_check_failed_type2, rtl_check_failed_code1,
27742 rtl_check_failed_code2, rtl_check_failed_code_mode,
27743 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
27744 rtvec_check_failed_bounds, rtl_check_failed_flag,
27745 _fatal_insn_not_found, _fatal_insn): Likewise.
27746 * tree.h (tree_contains_struct_check_failed,
27747 tree_check_failed, tree_not_check_failed,
27748 tree_class_check_failed, tree_range_check_failed,
27749 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
27750 tree_vec_elt_check_failed, phi_node_elt_check_failed,
27751 tree_operand_check_failed, omp_clause_check_failed,
27752 omp_clause_operand_check_failed, omp_clause_range_check_failed):
27755 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27757 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
27758 flag_branch_probabilities.
27759 * ipa-inline.c (edge_badness): Likewise.
27760 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
27761 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
27762 * predict.c (maybe_hot_frequency_p): Likewise.
27763 (probably_never_executed): Likewise.
27764 * sched-ebb.c (schedule_ebbs): Likewise.
27765 * sched-rgn.c (find_single_block_region): Likewise.
27766 * tracer.c (tail_duplicate): Likewise.
27768 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27770 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
27771 longer requires x_flag_profile_use.
27773 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27775 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
27776 instead of flag_reorder_blocks_and_partition.
27777 * dbxout.c (dbxout_function_end): Likewise.
27778 * dwarf2out.c (gen_subprogram_die): Likewise.
27779 * haifa-sched.c (sched_create_recovery_edges): Likewise.
27780 * hw-doloop.c (reorg_loops): Likewise.
27781 * varasm.c (assemble_start_function,
27782 assemble_end_function): Likewise.
27783 (decide_function_section): Do not check for
27784 flag_reorder_blocks_and_partition.
27786 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27788 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27790 (chkp_get_hard_register_fake_addr_expr): Ditto.
27791 (chkp_build_addr_expr): Add check for hard reg case.
27792 (chkp_parse_array_and_component_ref): Ditto.
27793 (chkp_find_bounds_1): Ditto.
27794 (chkp_process_stmt): Don't generate bounds store for
27797 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27799 * predict.c (maybe_hot_bb_p): Do not check profile status.
27800 (maybe_hot_edge_p): Likewise.
27801 (probably_never_executed): Check for zero counts even if profile
27803 (unlikely_executed_edge_p): New function.
27804 (unlikely_executed_stmt_p): New function.
27805 (unlikely_executed_bb_p): New function.
27806 (set_even_probabilities): Use unlikely predicates.
27807 (combine_predictions_for_bb): Likewise.
27808 (predict_paths_for_bb): Likewise.
27809 (predict_paths_leading_to_edge): Likewise.
27810 (determine_unlikely_bbs): New function.
27811 (estimate_bb_frequencies): Use it.
27812 (compute_function_frequency): Use zero counts even if profile is
27814 * profile-count.h: Fix typo.
27816 2017-08-08 Julia Koval <julia.koval@intel.com>
27818 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
27819 _mm512_mask_cvtsepi16_storeu_epi8,
27820 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
27821 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
27822 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
27823 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
27824 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
27825 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
27826 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
27827 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
27828 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
27829 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
27830 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
27831 __builtin_ia32_pmovuswb256mem_mask,
27832 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
27833 __builtin_ia32_pmovwb512mem_mask): New builtins.
27835 2017-08-08 Julia Koval <julia.koval@intel.com>
27837 PR target/73350,80862
27838 * config/i386/subst.md (round): Fix round pattern.
27839 * config/i386/i386.c (ix86_erase_embedded_rounding):
27840 Fix erasing rounding for the fixed pattern.
27842 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27844 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
27846 2017-06-08 Martin Liska <mliska@suse.cz>
27848 PR gcov-profile/80911
27849 * gcov.c (block_info::block_info): New constructor.
27851 2017-06-07 Carl Love <cel@us.ibm.com>
27853 * config/rs6000/rs6000-c: The return type of the following
27854 built-in functions was implemented as int not long long. Fix sign
27855 of return value for the unsigned version of vec_mulo and vec_mule.
27856 vector unsigned long long vec_bperm (vector unsigned long long,
27857 vector unsigned char)
27858 vector signed long long vec_mule (vector signed int,
27860 vector unsigned long long vec_mule (vector unsigned int,
27861 vector unsigned int)
27862 vector signed long long vec_mulo (vector signed int,
27864 vector unsigned long long vec_mulo (vector unsigned int,
27865 vector unsigned int)
27866 * doc/extend.texi: Fix the documentation for the built-in
27869 2017-06-07 Carl Love <cel@us.ibm.com>
27872 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
27875 2017-06-07 Carl Love <cel@us.ibm.com>
27877 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
27878 support, Generate doublehv for signed int/float for BE case only.
27880 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
27882 * doc/invoke.texi (mcx16): Rewrite.
27884 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27886 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
27887 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
27888 *mov<mode>_softfloat, and an anonymous splitter): Use
27889 nonimmediate_operand instead of rs6000_nonimmediate_operand.
27891 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27893 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
27895 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27896 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
27897 (rs6000_debug_reg_global): Adjust.
27898 (rs6000_init_hard_regno_mode_ok): Adjust.
27899 (rs6000_dbx_register_number): Adjust.
27900 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
27901 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27902 Remove SPE_ACC and SPEFSCR.
27903 (REG_ALLOC_ORDER): Ditto.
27904 (FRAME_POINTER_REGNUM): Change to 111.
27905 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
27906 (REG_CLASS_NAMES): Ditto.
27907 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
27908 (REGISTER_NAMES): Ditto.
27909 (ADDITIONAL_REG_NAMES): Ditto.
27910 (rs6000_reg_names): Ditto.
27911 * config/rs6000/rs6000.md: Renumber some register number
27914 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27916 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
27918 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27919 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
27921 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
27922 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27923 Delete the SPE high registers.
27924 (REG_ALLOC_ORDER): Ditto.
27925 (enum reg_class): Remove SPE_HIGH_REGS.
27926 (REG_CLASS_NAMES): Ditto.
27927 (REG_CLASS_CONTENTS): Delete the SPE high registers.
27928 (REGISTER_NAMES): Ditto.
27929 (rs6000_reg_names): Ditto.
27930 * doc/tm.texi.in: Remove SPE as example.
27931 * doc/tm.texi: Regenerate.
27933 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27935 * config/rs6000/8540.md (ppc8540_brinc): Delete.
27936 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
27937 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
27938 * config/rs6000/rs6000.md (type): Remove "brinc".
27940 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27942 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
27943 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
27944 * config/rs6000/linuxspe.h: Delete file.
27945 * config/rs6000/rs6000.md: Don't include spe.md.
27946 * config/rs6000/spe.h: Delete file.
27947 * config/rs6000/spe.md: Delete file.
27948 * config/rs6000/t-rs6000: Remove spe.md.
27950 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27952 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
27953 (reg_or_none500mem_operand): Delete.
27954 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
27955 instead of reg_or_none500mem_operand.
27957 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27959 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27960 handling of SPE flags.
27961 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
27963 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27965 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
27967 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
27968 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
27969 paired_divv2sf3): Similar.
27970 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
27971 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
27972 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
27974 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
27975 Rename the paired_* instruction patterns.
27976 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
27978 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
27979 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
27980 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
27981 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
27982 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
27983 PAIRED_VECTOR_MODE.
27984 (struct machine_function): Delete field spe_insn_chain_scanned_p.
27985 (spe_func_has_64bit_regs_p): Delete.
27986 (spe_expand_predicate_builtin): Delete.
27987 (spe_expand_evsel_builtin): Delete.
27988 (TARGET_DWARF_REGISTER_SPAN): Do not define.
27989 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
27990 (invalid_e500_subreg): Delete.
27991 (rs6000_legitimize_address): Always force_reg op2 as well, for
27992 paired single memory accesses.
27993 (rs6000_member_type_forces_blk): Delete.
27994 (rs6000_spe_function_arg): Delete.
27995 (rs6000_expand_unop_builtin): Delete SPE handling.
27996 (rs6000_expand_binop_builtin): Ditto.
27997 (spe_expand_stv_builtin): Delete.
27998 (bdesc_2arg_spe): Delete.
27999 (spe_expand_builtin): Delete.
28000 (spe_expand_predicate_builtin): Delete.
28001 (spe_expand_evsel_builtin): Delete.
28002 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
28003 (spe_init_builtins): Delete.
28004 (spe_func_has_64bit_regs_p): Delete.
28005 (savres_routine_name): Delete "info" parameter. Adjust callers.
28006 (rs6000_emit_stack_reset): Ditto.
28007 (rs6000_dwarf_register_span): Delete.
28008 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
28009 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
28010 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
28012 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
28014 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
28015 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
28016 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
28017 mulv2sf3, divv2sf3): Delete expanders.
28019 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28021 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
28023 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28025 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
28026 * config/rs6000/rs6000.c: Ditto.
28028 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28030 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
28031 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
28032 comparison_operator.
28034 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28036 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
28037 * config/rs6000/rs6000.opt: Ditto.
28038 * config/rs6000/t-rtems: Ditto.
28040 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28042 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
28043 TARGET_E500_SINGLE by 0, simplify.
28044 * config/rs6000/rs6000.c: Ditto.
28045 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
28046 (spe_build_register_parallel): Delete.
28047 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
28048 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
28049 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
28050 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
28051 (E500_CONVERT): Delete.
28052 * config/rs6000/spe.md: Remove many patterns and all define_constants.
28054 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28056 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
28057 * config/rs6000/dfp.md: Ditto.
28058 (negdd2, *negdd2_fpr): Merge.
28059 (absdd2, *absdd2_fpr): Merge.
28060 (negtd2, *negtd2_fpr): Merge.
28061 (abstd2, *abstd2_fpr): Merge.
28062 * config/rs6000/e500.h: Delete file.
28063 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
28064 TARGET_FPRS by 1 and simplify.
28065 * config/rs6000/rs6000-c.c: Ditto.
28066 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
28067 TARGET_DF_SPE by 0.
28068 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
28070 * config/rs6000/rs6000.md: Ditto.
28071 (floatdidf2, *floatdidf2_fpr): Merge.
28072 (move_from_CR_gt_bit): Delete.
28073 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
28074 (E500_CR_IOR_COMPARE): Delete.
28075 (All patterns that require !TARGET_FPRS): Delete.
28076 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
28078 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28080 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
28082 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28084 * graphds.c (add_edge): Intitialize edge's attached data.
28085 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
28086 pointer parameter. Call pointed function on each edge during
28087 graph traversing. Skip traversing the edge when the function
28089 (graphds_dfs, graphds_scc): Ditto.
28090 (for_each_edge): New parameter. Pass the new parameter to callback
28092 * graphds.h (skip_edge_callback): New function pointer type.
28093 (graphds_dfs, graphds_scc): New function pointer parameter.
28094 (graphds_edge_callback, for_each_edge): New parameter.
28096 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28098 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
28099 out code checking if runtime alias check is possible to below ...
28100 Call the new function.
28101 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
28102 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
28104 2017-06-07 Marek Polacek <polacek@redhat.com>
28107 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
28108 TYPE_OVERFLOW_WRAPS check.
28110 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28112 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
28113 if versioning is required.
28114 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
28115 peeling with the check for versioning.
28117 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28119 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
28120 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
28121 Set true to new parameter if new ssa variable is defined.
28122 (vect_gen_vector_loop_niters): Refactor. Set range information
28123 for the new vector loop bound variable.
28124 (vect_do_peeling): Ditto.
28126 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28128 * tree-affine.c (ssa.h): Include header file.
28129 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
28130 has wrapping overflow behavior.
28132 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28134 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
28136 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28138 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
28139 (tree_to_aff_combination): ... here.
28141 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28143 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
28144 reg_pressure model function.
28145 (ivopts_global_cost_for_size): Delete.
28146 (determine_set_costs, iv_ca_recount_cost): Call new model function
28147 ivopts_estimate_reg_pressure.
28149 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28151 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
28152 expensive than udiv. Remove floating point cases from mod.
28154 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28156 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
28157 Increase idiv cost.
28159 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28161 * config/aarch64/aarch64.md
28162 (copysignsf3): Fix mask generation.
28164 2017-06-07 Jakub Jelinek <jakub@redhat.com>
28166 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
28168 (class dump_manager): Add register_dumps method.
28169 * dumpfile.c: Include langhooks.h.
28170 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
28171 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
28172 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
28173 (dump_manager::dump_register): Start with 512 entries instead of 32.
28174 (dump_manager::register_dumps): New method.
28175 * toplev.c (general_init): Instead of invoking register_dumps
28176 langhook, invoke register_dumps method on the dump manager.
28177 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
28180 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
28182 * doc/md.texi: Clarify the restrictions on a define_insn condition.
28183 Say that # requires an associated define_split to exist, and that
28184 the define_split must be suitable for use after register allocation.
28186 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28188 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
28189 (compute_outgoing_frequencies): Also initialize zero counts.
28190 (find_many_sub_basic_blocks): Do not produce uninitialized profile
28191 around loops; preserve more of profile when nothing changes.
28193 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
28195 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
28197 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
28198 * config/arm/arm-cpu-cdata.h: Regenerate.
28199 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
28200 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
28201 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
28202 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
28203 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
28205 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
28207 * config/arm/t-rmprofile: Likewise.
28208 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
28210 2017-06-06 David S. Miller <davem@davemloft.net>
28213 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
28214 blockage if function uses alloca.
28216 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28218 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
28219 New "uid" fields to hold pretty-print IDs of group and ref.
28220 Memory references are now identified as <group_id>:<ref_id>
28221 instead of using [random] addresses.
28222 (dump_mem_details): Simplify, no functional change.
28223 (dump_mem_ref): Simplify and make output more concise.
28224 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
28225 (find_or_create_group): Initialize group uid.
28226 (record_ref): Initialize ref uid. Improve debug output.
28227 (prune_group_by_reuse, should_issue_prefetch_p,)
28228 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
28229 (mark_nontemporal_store, determine_loop_nest_reuse):
28230 Improve debug output.
28232 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28234 * dbgcnt.def (prefetch): New debug counter.
28235 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
28236 (schedule_prefetches): Stop issueing prefetches if debug counter
28239 2017-06-06 Tom de Vries <tom@codesourcery.com>
28241 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
28242 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
28244 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28246 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
28247 Use aarch64_reg_or_zero predicate for operand 4.
28248 (aarch64_compare_and_swap<mode> define_insn_and_split):
28249 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
28250 (aarch64_store_exclusive<mode>): Likewise for operand 2.
28252 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
28254 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
28255 (arm_compute_save_core_reg_mask): This.
28256 (thumb1_compute_save_reg_mask): Rename into ...
28257 (thumb1_compute_save_core_reg_mask): This.
28258 (arm_compute_save_reg0_reg12_mask): Adapt comment.
28259 (arm_compute_frame_layout): Likewise.
28261 2017-06-06 Richard Biener <rguenther@suse.de>
28263 PR tree-optimization/80974
28264 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
28265 keep or clear leaders SSA info.
28267 2017-06-06 Tom de Vries <tom@codesourcery.com>
28269 * config/nvptx/nvptx.c (split_mode_p): New function.
28270 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
28272 2017-06-06 Tom de Vries <tom@codesourcery.com>
28274 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
28276 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28279 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
28282 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28284 * shrink-wrap.c (handle_simple_exit): Update profile.
28285 (try_shrink_wrapping): Upate profile.
28287 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28289 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
28290 (tree_guess_outgoing_edge_probabilities): New.
28291 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
28292 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
28294 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28296 * ipa-split.c (split_function): Initialize return bb profile.
28298 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28300 * profile.c (compute_branch_probabilities): Also initialize
28301 EXIT_BLOCK profile.
28303 2017-06-06 Richard Biener <rguenther@suse.de>
28305 PR tree-optimization/80928
28306 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
28307 (vect_analyze_loop_operations): Properly guard analysis for
28309 (vect_transform_loop): Likewise.
28310 (vect_analyze_loop_2): Also reset SLP type on PHIs.
28311 (vect_model_induction_cost): Do not cost for pure SLP.
28312 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
28313 of induction in inner loop vectorization.
28314 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
28315 (vect_get_and_check_slp_defs): Handle vect_induction_def.
28316 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
28318 (vect_analyze_slp_cost_1): Cost induction.
28319 (vect_detect_hybrid_slp_stmts): Handle PHIs.
28320 (vect_get_slp_vect_defs): Likewise.
28321 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
28322 (vect_transform_stmt): Handle SLP reductions.
28323 * tree-vectorizer.h (vectorizable_induction): Adjust.
28325 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28327 * config/rs6000/rs6000.c (make_resolver_func): Update
28328 init_lowered_empty_function call.
28330 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
28332 * doc/invoke.texi: Document the -fprofile-abs-path option.
28333 * common.opt (fprofile-abs-path): New option.
28334 * gcov-io.h (gcov_write_filename): Declare.
28335 * gcov-io.c (gcov_write_filename): New function.
28336 * coverage.c (coverage_begin_function): Use gcov_write_filename.
28337 * profile.c (output_location): Likewise.
28339 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28341 * shring-wrap.c: Revert accidental commit.
28343 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
28345 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
28347 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28349 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
28351 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
28352 profile in callgraph edge.
28353 * profile-count.h (apply_probability): If THIS is 0, then result is 0
28354 (apply_scale): Likewise.
28355 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
28356 Also scale profile when inlining function with zero profile.
28357 (initialize_cfun): Update exit block profile even when it is zero.
28358 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
28359 when profile is read.
28361 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28363 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
28364 (CLONE_*): New constants to define the processors we can generate
28365 code for with the target_clone attribute.
28366 (rs6000_clone_map): New array to identify which clone processors
28367 the current program is running on.
28368 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
28369 target_clone attribute.
28370 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
28371 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
28372 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
28373 (cpu_expand_builtin): Add support for target_clone attribute.
28374 (rs6000_valid_attribute_p): Allow "default" attribute.
28375 (get_decl_name): New debug function to simplify printing the
28376 current function name in debugging statements.
28377 (rs6000_clone_priority): New functions to support the target_clone
28378 attribute, and be able to generate code to switch between ISA 2.05
28379 through ISA 3.0 (power6 through power9).
28380 (rs6000_compare_version_priority): Likewise.
28381 (rs6000_get_function_versions_dispatcher): Likewise.
28382 (make_resolver_func): Likewise.
28383 (add_condition_to_bb): Likewise.
28384 (dispatch_function_versions): Likewise.
28385 (rs6000_generate_version_dispatcher_body): Likewise.
28386 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
28387 (fusion_gpr_load_p): Fix a spacing issue.
28388 * doc/extend.texi (Common Function Attributes): Document that the
28389 PowerPC supports the target_clone attribute.
28391 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28393 * config/arm/arm.h: explain F symbol found in description of ARM
28394 register allocation in its legend.
28396 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28398 * config/mips/frame-header-opt.c: Include profile-count.h.
28399 * config/riscv/riscv.c: Include profile-count.h
28401 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28403 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
28405 (sm_set_flag_if_changed): Add bbs field.
28406 (execute_sm_if_changed_flag_set): Pass BBS.
28407 (execute_sm): Update.
28409 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28411 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
28414 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28416 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
28417 (peephole2): New peephole2 to emit the above.
28418 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
28420 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28422 * config/aarch64/aarch64.c (define_peephole2 above
28423 *sub_<shift>_<mode>): New peephole.
28425 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28427 * config/i386/i386.c (make_resolver_func): Update.
28428 * Makefile.in: Add profile-count.h and profile-count.o
28429 * auto-profile.c (afdo_indirect_call): Update to new API.
28430 (afdo_set_bb_count): Update.
28431 (afdo_propagate_edge): Update.
28432 (afdo_propagate_circuit): Update.
28433 (afdo_calculate_branch_prob): Update.
28434 (afdo_annotate_cfg): Update.
28435 * basic-block.h: Include profile-count.h
28436 (struct edge_def): Turn count to profile_count.
28437 (struct basic_block_def): Likewie.
28438 (REG_BR_PROB_BASE): Move to profile-count.h
28439 (RDIV): Move to profile-count.h
28440 * bb-reorder.c (max_entry_count): Turn to profile_count.
28441 (find_traces): Update.
28442 (rotate_loop):Update.
28443 (connect_traces):Update.
28444 (sanitize_hot_paths):Update.
28445 * bt-load.c (migrate_btr_defs): Update.
28446 * cfg.c (RDIV): Remove.
28447 (init_flow): Use alloc_block.
28448 (alloc_block): Uninitialize count.
28449 (unchecked_make_edge): Uninitialize count.
28450 (check_bb_profile): Update.
28451 (dump_edge_info): Update.
28452 (dump_bb_info): Update.
28453 (update_bb_profile_for_threading): Update.
28454 (scale_bbs_frequencies_int): Update.
28455 (scale_bbs_frequencies_gcov_type): Update.
28456 (scale_bbs_frequencies_profile_count): New.
28457 * cfg.h (update_bb_profile_for_threading): Update.
28458 (scale_bbs_frequencies_profile_count): Declare.
28459 * cfgbuild.c (compute_outgoing_frequencies): Update.
28460 (find_many_sub_basic_blocks): Update.
28461 * cfgcleanup.c (try_forward_edges): Update.
28462 (try_crossjump_to_edge): Update.
28463 * cfgexpand.c (expand_gimple_tailcall): Update.
28464 (construct_exit_block): Update.
28465 * cfghooks.c (verify_flow_info): Update.
28466 (dump_bb_for_graph): Update.
28467 (split_edge): Update.
28468 (make_forwarder_block): Update.
28469 (duplicate_block): Update.
28470 (account_profile_record): Update.
28471 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
28472 (get_estimated_loop_iterations): Update.
28473 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
28474 (single_likely_exit): Update.
28475 * cfgloopmanip.c (scale_loop_profile): Update.
28477 (set_zero_probability): Update.
28478 (lv_adjust_loop_entry_edge): Update.
28479 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
28480 (purge_dead_edges): Update.
28481 (rtl_account_profile_record): Update.
28482 * cgraph.c (cgraph_node::create): Uninitialize count.
28483 (symbol_table::create_edge): Uninitialize count.
28484 (cgraph_update_edges_for_call_stmt_node): Update.
28485 (cgraph_edge::dump_edge_flags): Update.
28486 (cgraph_node::dump): Update.
28487 (cgraph_edge::maybe_hot_p): Update.
28488 * cgraph.h: Include profile-count.h
28489 (create_clone), create_edge, create_indirect_edge): Update.
28490 (cgraph_node): Turn count to profile_count.
28491 (cgraph_edge0: Likewise.
28492 (make_speculative, clone): Update.
28493 (create_edge): Update.
28494 (init_lowered_empty_function): Update.
28495 * cgraphclones.c (cgraph_edge::clone): Update.
28496 (duplicate_thunk_for_node): Update.
28497 (cgraph_node::create_clone): Update.
28498 * cgraphunit.c (cgraph_node::analyze): Update.
28499 (cgraph_node::expand_thunk): Update.
28500 * final.c (dump_basic_block_info): Update.
28501 * gimple-streamer-in.c (input_bb): Update.
28502 * gimple-streamer-out.c (output_bb): Update.
28503 * graphite.c (print_global_statistics): Update.
28504 (print_graphite_scop_statistics): Update.
28505 * hsa-brig.c: Include basic-block.h.
28506 * hsa-dump.c: Include basic-block.h.
28507 * hsa-gen.c (T sum_slice): Update.
28508 (convert_switch_statements):Update.
28509 * hsa-regalloc.c: Include basic-block.h.
28510 * ipa-chkp.c (chkp_produce_thunks): Update.
28511 * ipa-cp.c (struct caller_statistics): Update.
28512 (init_caller_stats): Update.
28513 (gather_caller_stats): Update.
28514 (ipcp_cloning_candidate_p): Update.
28515 (good_cloning_opportunity_p): Update.
28516 (get_info_about_necessary_edges): Update.
28517 (dump_profile_updates): Update.
28518 (update_profiling_info): Update.
28519 (update_specialized_profile): Update.
28520 (perhaps_add_new_callers): Update.
28521 (decide_about_value): Update.
28522 (ipa_cp_c_finalize): Update.
28523 * ipa-devirt.c (struct odr_type_warn_count): Update.
28524 (struct decl_warn_count): Update.
28525 (struct final_warning_record): Update.
28526 (possible_polymorphic_call_targets): Update.
28527 (ipa_devirt): Update.
28528 * ipa-fnsummary.c (redirect_to_unreachable): Update.
28529 * ipa-icf.c (sem_function::merge): Update.
28530 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
28531 * ipa-inline.c (compute_uninlined_call_time): Update.
28532 (compute_inlined_call_time): Update.
28533 (want_inline_small_function_p): Update.
28534 (want_inline_self_recursive_call_p): Update.
28535 (edge_badness): Update.
28536 (lookup_recursive_calls): Update.
28537 (recursive_inlining): Update.
28538 (inline_small_functions): Update.
28539 (dump_overall_stats): Update.
28540 (dump_inline_stats): Update.
28541 * ipa-profile.c (ipa_profile_generate_summary): Update.
28542 (ipa_propagate_frequency): Update.
28543 (ipa_profile): Update.
28544 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28545 * ipa-utils.c (ipa_merge_profiles): Update.
28546 * loop-doloop.c (doloop_modify): Update.
28547 * loop-unroll.c (report_unroll): Update.
28548 (unroll_loop_runtime_iterations): Update.
28549 * lto-cgraph.c (lto_output_edge): Update.
28550 (lto_output_node): Update.
28551 (input_node): Update.
28552 (input_edge): Update.
28553 (merge_profile_summaries): Update.
28554 * lto-streamer-in.c (input_cfg): Update.
28555 * lto-streamer-out.c (output_cfg): Update.
28556 * mcf.c (create_fixup_graph): Update.
28557 (adjust_cfg_counts): Update.
28558 (sum_edge_counts): Update.
28559 * modulo-sched.c (sms_schedule): Update.
28560 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
28561 * predict.c (maybe_hot_count_p): Update.
28562 (probably_never_executed): Update.
28563 (dump_prediction): Update.
28564 (combine_predictions_for_bb): Update.
28565 (propagate_freq): Update.
28566 (handle_missing_profiles): Update.
28567 (counts_to_freqs): Update.
28568 (rebuild_frequencies): Update.
28569 (force_edge_cold): Update.
28570 * predict.h: Include profile-count.h
28571 (maybe_hot_count_p, counts_to_freqs): UPdate.
28572 * print-rtl-function.c: Do not include cfg.h
28573 * print-rtl.c: Include basic-block.h
28574 * profile-count.c: New file.
28575 * profile-count.h: New file.
28576 * profile.c (is_edge_inconsistent): Update.
28577 (correct_negative_edge_counts): Update.
28578 (is_inconsistent): Update.
28579 (set_bb_counts): Update.
28580 (read_profile_edge_counts): Update.
28581 (compute_frequency_overlap): Update.
28582 (compute_branch_probabilities): Update; Initialize and deinitialize
28584 (branch_prob): Update.
28585 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
28586 (edge_gcov_count): New.
28587 (bb_gcov_count): New.
28588 * shrink-wrap.c (try_shrink_wrapping): Update.
28589 * tracer.c (better_p): Update.
28590 * trans-mem.c (expand_transaction): Update.
28591 (ipa_tm_insert_irr_call): Update.
28592 (ipa_tm_insert_gettmclone_call): Update.
28593 * tree-call-cdce.c: Update.
28594 * tree-cfg.c (gimple_duplicate_sese_region): Update.
28595 (gimple_duplicate_sese_tail): Update.
28596 (gimple_account_profile_record): Update.
28597 (execute_fixup_cfg): Update.
28598 * tree-inline.c (copy_bb): Update.
28599 (copy_edges_for_bb): Update.
28600 (initialize_cfun): Update.
28601 (freqs_to_counts): Update.
28602 (copy_cfg_body): Update.
28603 (expand_call_inline): Update.
28604 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
28605 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
28606 (try_unroll_loop_completely): Update.
28607 (try_peel_loop): Update.
28608 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
28609 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
28610 * tree-ssa-loop-split.c (connect_loops): Update.
28611 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
28612 * tree-ssa-reassoc.c (branch_fixup): Update.
28613 * tree-ssa-tail-merge.c (replace_block_by): Update.
28614 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
28615 (compute_path_counts): Update.
28616 (update_profile): Update.
28617 (recompute_probabilities): Update.
28618 (update_joiner_offpath_counts): Update.
28619 (estimated_freqs_path): Update.
28620 (freqs_to_counts_path): Update.
28621 (clear_counts_path): Update.
28622 (ssa_fix_duplicate_block_edges): Update.
28623 (duplicate_thread_path): Update.
28624 * tree-switch-conversion.c (case_bit_test_cmp): Update.
28625 (struct switch_conv_info): Update.
28626 * tree-tailcall.c (decrease_profile): Update.
28627 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
28628 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
28629 * value-prof.c (check_counter): Update.
28630 (gimple_divmod_fixed_value): Update.
28631 (gimple_mod_pow2): Update.
28632 (gimple_mod_subtract): Update.
28633 (gimple_ic_transform): Update.
28634 (gimple_stringop_fixed_value): Update.
28635 * value-prof.h (gimple_ic): Update.
28637 2017-06-02 Carl Love <cel@us.ibm.com>
28639 * config/rs6000/rs6000-c: Add support for built-in functions
28640 vector double vec_doublee (vector signed int);
28641 vector double vec_doublee (vector unsigned int);
28642 vector double vec_doublee (vector float);
28643 vector double vec_doubleh (vector signed int);
28644 vector double vec_doubleh (vector unsigned int);
28645 vector double vec_doubleh (vector float);
28646 vector double vec_doublel (vector signed int);
28647 vector double vec_doublel (vector unsigned int);
28648 vector double vec_doublel (vector float);
28649 vector double vec_doubleo (vector signed int);
28650 vector double vec_doubleo (vector unsigned int);
28651 vector double vec_doubleo (vector float);.
28652 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
28653 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
28655 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
28656 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
28657 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
28659 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
28660 vec_doublel, vec_doubleh.
28661 * doc/extend.texi: Update the built-in documentation file for the
28662 new built-in functions.
28664 2017-06-02 David Malcolm <dmalcolm@redhat.com>
28667 * ipa-inline-analysis.c (free_growth_caches): Set
28668 edge_removal_hook_holder to NULL after removing it.
28670 2017-06-02 Sudakshina Das <sudi.das@arm.com>
28672 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
28673 comparision with zero.
28675 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28676 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28677 for early expansion of vec_min and vec_max builtins.
28678 (builtin_function_type): Add min/max unsigned variants to those
28679 identified as having unsigned arguments.
28681 2017-06-02 Olivier Hainque <hainque@adacore.com>
28683 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
28685 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28687 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
28688 Use VALL_F16 iterator rather than VALL.
28690 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28692 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
28693 Emit CBNZ inside loop when doing a strong exchange and comparing
28694 against zero. Generate the CC flags after the loop.
28696 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
28698 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
28699 (dl_section_ref): New.
28700 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
28701 On AIX, append an expression to subtract the size of the
28702 section length to dl_section_ref.
28704 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28706 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28707 for early expansion of vector absolute builtins.
28709 2017-06-02 Richard Biener <rguenther@suse.de>
28711 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
28712 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
28714 2017-06-02 Richard Biener <rguenther@suse.de>
28716 PR tree-optimization/80948
28717 * tree-tailcall.c (find_tail_calls): Track stmts to move in
28718 stmt order as well.
28720 2017-06-02 Richard Biener <rguenther@suse.de>
28722 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
28724 * tree-vect-stmts.c (process_use): Do not mark backedge defs
28725 for inductions as relevant.
28727 2017-06-02 Richard Biener <rguenther@suse.de>
28729 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
28730 (vectorizable_induction): ... this. Remove dead code.
28732 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
28734 * builtins. (expand_builtin_alloca): Remove second parameter and
28735 infer its value from the first parameter instead.
28736 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
28738 2017-06-02 Jakub Jelinek <jakub@redhat.com>
28740 PR rtl-optimization/80903
28741 * loop-doloop.c (add_test): Unshare sequence.
28743 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
28745 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
28747 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
28749 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
28751 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
28752 xlogue_layout::get_instance, logue_layout::xlogue_layout,
28753 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
28754 (xlogue_layout::get_stub_rtx): Make static.
28755 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
28756 (xlogue_layout::compute_stub_managed_regs): Rename to...
28757 (xlogue_layout::count_stub_managed_regs): ...this.
28758 (xlogue_layout::is_stub_managed_reg): New function.
28759 (xlogue_layout::m_stub_names): Rename to...
28760 (xlogue_layout::s_stub_names): ...this, make static.
28761 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
28762 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
28763 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
28764 xlogue_layout::s_stub_names): Instantiate statics.
28765 (stub_managed_regs): Remove.
28766 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
28767 (disable_call_ms2sysv_xlogues): Rename to...
28768 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
28769 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
28771 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
28772 change after reload_completed.
28773 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
28775 (ix86_expand_prologue): Likewise.
28776 (ix86_expand_epilogue): Likewise.
28777 (ix86_expand_split_stack_prologue): Likewise.
28778 (ix86_compute_frame_layout): Remove frame parameter ...
28779 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
28780 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
28782 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
28783 (ix86_frame): Move from here ...
28784 * config/i386/i386.h (ix86_frame): ... to here.
28785 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
28786 complete ix86_frame data structure instead. Remove some_ld_name.
28788 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28790 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
28791 symbols that hold a DECL_VALUE_EXPR.
28793 2017-06-01 Martin Jambor <mjambor@suse.cz>
28795 PR tree-optimization/80898
28796 * tree-sra.c (process_subtree_disqualification): Removed.
28797 (disqualify_candidate): Do not acll
28798 process_subtree_disqualification.
28799 (subtree_mark_written_and_enqueue): New function.
28800 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
28801 RHS has been disqualified and re-queue LHS if necessary. Apart
28802 from that, ignore disqualified RHS.
28804 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28806 * config/s390/s390.c (s390_emit_epilogue): Disable early return
28807 address fetch for z10 or later.
28809 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28811 * config/arc/arc.md (tst_movb): Add guard when splitting.
28813 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28815 * config/arc/arc.c (arc_can_eliminate): Test against
28816 arc_frame_pointer_needed.
28818 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28820 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
28821 to prevent store reordering.
28822 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
28823 (type): Add block type.
28824 (stack_tie): Define special instruction to be used in
28827 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28829 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
28830 constraint. It is not valid for the pattern.
28831 (noncommutative_binary_comparison): Likewise.
28833 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28835 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
28838 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28840 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
28841 be used by the reg-alloc.
28843 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28845 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
28846 reg-alloc when having mul64 or mul32x16 instructions.
28847 (mulsidi3): Likewise.
28848 (umulsidi3): Likewise.
28849 (mulsi32x16): New pattern.
28850 (mulsi64): Likewise.
28851 (mulsidi64): Likewise.
28852 (umulsidi64): Likewise.
28853 (MUL32x16_REG): Define.
28854 (mul64_600): Use MUL32x16_REG.
28855 (mac64_600): Likewise.
28856 (umul64_600): Likewise.
28857 (umac64_600): Likewise.
28859 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28861 * config/arc/arc.md (mulsi3_700): Make it commutative.
28863 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
28865 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
28867 (*sign_extendsidi2_insn): Likewise for movstosw.
28869 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28871 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
28872 the type of the input discriminant value. Convert the
28873 discriminant value of signedness vary.
28875 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
28877 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
28878 Add to -Wall section.
28880 2017-06-01 Richard Biener <rguenther@suse.de>
28882 PR middle-end/66313
28883 * fold-const.c (fold_plusminus_mult_expr): If the factored
28884 factor may be zero use a wrapping type for the inner operation.
28885 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
28886 and handle moved defs.
28887 (process_assignment): Properly guard the unary op case. Return a
28888 tri-state indicating that moving the stmt before the call may allow
28889 to continue. Pass through to_move.
28890 (find_tail_calls): Handle moving unrelated defs before
28893 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
28896 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
28897 splitter result in the canonical way.
28899 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
28901 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
28902 also for 32bit target. Update insn attributes.
28903 (zero-extendsidi2 splitter): Allow all registers for operand 1.
28905 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
28907 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
28908 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
28909 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
28910 (_mm_maskz_min_ss): New intrinsics.
28912 2017-05-31 Martin Liska <mliska@suse.cz>
28914 * tree-vect-loop.c (vect_create_epilog_for_reduction):
28915 Change comment style to one we normally use.
28916 (vectorizable_reduction): Likewise.
28917 (vectorizable_induction): Likewise.
28918 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
28919 (vectorizable_call): Likewise.
28920 (vectorizable_simd_clone_call): Likewise.
28921 (vectorizable_conversion): Likewise.
28922 (vectorizable_assignment): Likewise.
28923 (vectorizable_shift): Likewise.
28924 (vectorizable_operation): Likewise.
28925 (vectorizable_store): Likewise.
28926 (vectorizable_load): Likewise.
28927 * tree-vectorizer.h: Likewise.
28929 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
28931 * passes.c (emergency_dump_function): New.
28932 * tree-pass.h (emergency_dump_function): Declare.
28933 * plugin.c (plugins_internal_error_function): Remove.
28934 * plugin.h (plugins_internal_error_function): Remove declaration.
28935 * toplev.c (internal_error_function): New static function. Use it...
28936 (general_init): ...here.
28938 2017-05-31 Graham Markall <graham.markall@embecosm.com>
28940 * config/arc/arc.c (arc_print_operand): Handle constant operands.
28941 (arc_rtx_costs): Add costs for new patterns.
28942 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
28943 * config/arc/predicates.md: Add _1_2_3_operand predicate.
28945 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28947 * tree-ssa-strlen.c (get_next_strinfo): New function.
28948 (get_stridx_plus_constant): Use it.
28949 (zero_length_string): Likewise.
28950 (adjust_related_strinfos): Likewise.
28951 (adjust_last_stmt): Likewise.
28953 2017-05-31 Richard Biener <rguenther@suse.de>
28956 * config/i386/i386.c (ix86_expand_builtin): Remove assert
28957 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
28959 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28961 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
28962 loop_vinfo argument and use of dependence distance vectors.
28963 Check instead whether the two references differ only in their
28964 initial value and assume that they have the same alignment if the
28965 difference is a multiple of the vector alignment.
28966 (vect_analyze_data_refs_alignment): Update call accordingly.
28968 2017-05-31 Martin Liska <mliska@suse.cz>
28971 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
28973 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28975 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
28976 (create_intersect_range_checks): Move from ...
28977 * tree-data-ref.c (create_intersect_range_checks_index)
28978 (create_intersect_range_checks): ... to here.
28979 (create_runtime_alias_checks): New function factored from ...
28980 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
28981 here. Call above function.
28982 * tree-data-ref.h (create_runtime_alias_checks): New function.
28984 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28986 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
28987 segment length for dr_b and compute it in wide_int.
28989 2017-05-31 Richard Biener <rguenther@suse.de>
28991 PR tree-optimization/80906
28992 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
28993 and pass through iv_map.
28994 (copy_bb_and_scalar_dependences): Adjust.
28995 (translate_pending_phi_nodes): Likewise.
28996 (copy_loop_close_phi_args): Handle code-generating IVs instead
28999 2017-05-30 David Malcolm <dmalcolm@redhat.com>
29001 * diagnostic-color.c (color_dict): Add "type-diff".
29002 (parse_gcc_colors): Update comment.
29003 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
29004 -fdiagnostics-show-template-tree and -fno-elide-type.
29005 (GCC_COLORS): Add type-diff to example.
29007 (-fdiagnostics-show-template-tree): New.
29008 (-fno-elide-type): New.
29009 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
29010 the pp_format_decoder callback. Call any m_format_postprocessor's
29012 (pretty_printer::pretty_printer): Initialize
29013 m_format_postprocessor.
29014 (pretty_printer::~pretty_printer): Delete any
29015 m_format_postprocessor.
29016 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
29017 (class format_postprocessor): New class.
29018 (struct pretty_printer::format_decoder): Document the new parameters.
29019 (struct pretty_printer::m_format_postprocessor): New field.
29020 * tree-diagnostic.c (default_tree_printer): Update for new
29021 bool and const char ** params.
29022 * tree-diagnostic.h (default_tree_printer): Likewise.
29024 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
29026 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
29027 (lwa_operand): Delete rs6000_gen_cell_microcode test.
29028 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29029 rs6000_gen_cell_microcode code.
29030 (rs6000_final_prescan_insn): Delete.
29031 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
29032 "warn-cell-microcode" entries.
29033 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
29034 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
29035 throughout. Change cc_reg_not_micro_cr0_operand to
29036 cc_reg_not_cr0_operand throughout.
29037 (*extendhi<mode>2_noload): Delete.
29038 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
29039 (mwarn-cell-microcode): Delete.
29040 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
29041 -mgen-cell-microcode and -mwarn-cell-microcode.
29043 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
29046 * config/i386/constraints.md (Yd): New constraint.
29048 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
29049 and (?Yd, r) alternatives. Update insn attributes.
29050 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
29051 and (?*Yd, r) alternatives. Update insn attributes.
29052 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
29054 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
29056 * gimplify.c (gimplify_modify_expr): Don't create a
29057 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
29060 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
29062 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
29064 2017-05-30 Richard Biener <rguenther@suse.de>
29066 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
29067 and reduc_def fields.
29068 (STMT_VINFO_REDUC_TYPE): New define.
29069 (STMT_VINFO_REDUC_DEF): Likewise.
29070 (vect_force_simple_reduction): Adjust prototype.
29071 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
29072 (vect_is_simple_reduction): Remove check_reduction argument.
29073 (vect_force_simple_reduction): Adjust and set
29074 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29075 (vectorizable_reduction): Do not re-do reduction analysis
29076 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29077 * tree-parloops.c (gather_scalar_reductions): Adjust.
29079 2017-05-30 Richard Biener <rguenther@suse.de>
29081 PR middle-end/80901
29082 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
29085 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
29087 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
29088 Introduce unknown_misalignment parameter and remove vf.
29089 (vect_peeling_hash_get_lowest_cost):
29090 Pass unknown_misalignment parameter.
29091 (vect_enhance_data_refs_alignment):
29092 Fix unsupportable data ref treatment.
29094 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29096 * tree-vect-data-refs.c (vect_get_data_access_cost):
29097 Workaround for SLP handling.
29098 (vect_enhance_data_refs_alignment):
29099 Compute costs for doing no peeling at all, compare to the best
29100 peeling costs so far and avoid peeling if cheaper.
29102 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29104 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
29105 Return peeling info and set costs to zero for unlimited cost
29107 (vect_enhance_data_refs_alignment): Also inspect all datarefs
29108 with unknown misalignment. Compute and costs for unknown
29109 misalignment, compare them to the costs for known misalignment
29110 and choose the cheapest for peeling.
29112 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29114 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
29115 (vect_get_peeling_costs_all_drs): Create function.
29116 (vect_peeling_hash_get_lowest_cost):
29117 Use vect_get_peeling_costs_all_drs.
29118 (vect_peeling_supportable): Create function.
29119 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
29121 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29123 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
29124 DR_HAS_NEGATIVE_STEP.
29125 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
29126 (vect_enhance_data_refs_alignment): Use.
29127 (vect_duplicate_ssa_name_ptr_info): Use.
29128 * tree-vectorizer.h (dr_misalignment): Use.
29129 (known_alignment_for_access_p): Use.
29131 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
29134 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
29136 (has_section_name): New function.
29138 2017-05-30 Martin Liska <mliska@suse.cz>
29141 * auto-profile.c (get_function_decl_from_block): Fix
29144 2017-05-30 Richard Biener <rguenther@suse.de>
29146 PR middle-end/80876
29147 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
29149 2017-05-30 Martin Liska <mliska@suse.cz>
29151 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
29152 * dumpfile.h (struct dump_file_info): Remove ctors.
29154 2017-05-30 Martin Liska <mliska@suse.cz>
29156 * predict.def: Fix GNU coding style.
29158 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29160 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29161 Mark 'to' argument with ATTRIBUTE_UNUSED.
29163 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29165 * config/xtensa/xtensa.c (xtensa_emit_call): Use
29166 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
29167 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
29170 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
29172 * doc/install.texi (Options specification): Restore entry of
29173 --enable-sjlj-exceptions.
29175 2017-05-27 Michael Eager <eager@eagercon.com>
29178 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
29180 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
29182 * config/microblaze/microblaze.h
29183 (FIXED_REGISTERS): Update in macro.
29184 (CALL_USED_REGISTERS): Update in macro.
29186 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
29188 * doc/install.texi: Add links to macOS binary distributions.
29190 2017-05-27 Jakub Jelinek <jakub@redhat.com>
29194 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29196 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29198 2017-05-26 Martin Liska <mliska@suse.cz>
29200 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
29202 2017-05-26 Martin Liska <mliska@suse.cz>
29204 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
29205 always leading ';; '.
29206 (dump_bb_info): Likewise.
29207 (brief_dump_cfg): Likewise.
29208 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
29209 * dumpfile.c: Remove usage of TDF_VERBOSE.
29210 * dumpfile.h (enum dump_kind): Likewise.
29211 (dump_gimple_bb_header): Do not use TDF_COMMENT.
29212 * print-tree.c (debug_verbose): Remove.
29213 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
29214 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
29215 * tree-diagnostic.c (default_tree_printer): Replace
29216 TDF_DIAGNOSTIC with TDF_SLIM.
29218 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29220 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
29221 in parameter loop, rather than loop_vinfo.
29222 (create_intersect_range_checks): Ditto.
29223 (vect_create_cond_for_alias_checks): Update call to above functions.
29225 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29227 PR tree-optimization/80815
29228 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
29229 for merging runtime alias checks. Handle negative DR_STEPs.
29231 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29233 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
29235 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
29236 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
29237 out code pruning runtime alias checks.
29238 * tree-data-ref.c (prune_runtime_alias_test_list): New function
29239 factored out from above.
29240 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29242 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29244 (prune_runtime_alias_test_list): New decalaration.
29246 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29248 * tree-vect-data-refs.c (compare_tree): Rename and move ...
29249 * tree-data-ref.c (data_ref_compare_tree): ... to here.
29250 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
29251 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
29252 (operator==, comp_dr_with_seg_len_pair): Ditto.
29253 (vect_prune_runtime_alias_test_list): Ditto.
29255 2017-05-26 Martin Liska <mliska@suse.cz>
29258 * params.def: Bound partial-inlining-entry-probability param.
29260 2017-05-26 Marek Polacek <polacek@redhat.com>
29263 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
29266 2017-05-26 Richard Biener <rguenther@suse.de>
29268 PR tree-optimization/80842
29269 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
29272 2017-05-26 Richard Biener <rguenther@suse.de>
29274 PR tree-optimization/80844
29275 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
29277 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
29279 * doc/md.texi (Machine Constraints): Update x86 family
29280 machine constraints section to match 'config/i386/constraints.md'.
29282 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
29284 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
29286 2017-05-25 Nathan Sidwell <nathan@acm.org>
29288 * doc/invoke.texi (--enable-languages): Update documentation.
29290 2017-05-25 Martin Liska <mliska@suse.cz>
29292 * dumpfile.c: Add TDF_FOLDING.
29293 * dumpfile.h (enum dump_kind): Likewise.
29294 * genmatch.c (dt_simplify::gen_1): Use it.
29296 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29298 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
29300 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29302 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29303 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
29305 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29307 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
29308 * match.pd (X == C): Rewrite it here.
29309 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
29310 with_certain_nonzero_bits2): New predicates.
29311 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
29313 2017-05-24 Nathan Sidwell <nathan@acm.org>
29315 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
29318 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
29321 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29323 * config/powerpcspe: New port. Files are copied from the rs6000
29324 port, with "rs6000" in filenames replaced by "powerpcspe".
29326 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
29328 PR rtl-optimization/80754
29329 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
29331 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
29333 * config/sparc/sparc.md (length): Return the correct value for -mflat
29334 sibcalls to match output_sibcall.
29336 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29340 * config/rs6000/rs6000.c (struct machine_function): Add new field
29342 (rs6000_get_separate_components): Init that field, use it.
29343 (rs6000_components_for_bb): Use the field.
29345 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
29347 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
29349 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
29351 PR middle-end/80823
29352 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
29354 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29357 * config/s390/s390.c (s390_check_qrst_address): Check incoming
29358 address against address_operand predicate.
29359 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
29361 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
29363 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
29364 parameters passed indirectly.
29366 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
29368 * config/i386/i386.md (*movdi_internal): Remove SSE4
29369 alternative 18 (?r, *v). Update insn attributes.
29370 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
29371 Update insn attributes.
29372 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
29373 Update insn attributes.
29374 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
29375 alternative 1 (r, v). Remove isa attribute.
29376 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
29377 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
29378 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
29380 2017-05-23 Tom de Vries <tom@codesourcery.com>
29382 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
29385 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29387 * cgraphunit.c (symbol_table::process_new_functions): Update.
29388 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
29389 (inline_generate_summary): Rename to ...
29390 (ipa_fn_summary_generate): ... this one.
29391 (inline_read_summary): Rename to ...
29392 (ipa_fn_summary_read): ... this one.
29393 (inline_write_summary): Rename to ...
29394 (ipa_fn_summary_write): ... this one.
29395 (inline_free_summary): Rename to ...
29396 (ipa_free_fn_summary): ... this one.
29397 (pass_data_local_fn_summary, pass_local_fn_summary,
29398 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
29399 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
29400 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
29401 make_pass_ipa_fn_summary): New.
29402 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
29403 inline_write_summary, inline_free_summary): Remove.
29404 (ipa_free_fn_summary) : New.
29405 * ipa-inline.c (ipa_inline): Update.
29406 (pass_ipa_inline): Do not generate summaries.
29407 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
29409 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
29410 and add pass_ipa_fn_summary.
29411 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
29413 (make_pass_inline_parameters): Remove.
29415 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
29417 * omp-low.c (struct omp_context): Remove "default_kind" member.
29420 * omp-offload.c (execute_oacc_device_lower): Remove the
29421 parallelism dimensions function attributes for unparallelized
29422 OpenACC kernels constructs.
29424 2017-05-23 Martin Liska <mliska@suse.cz>
29426 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
29428 (cgraph_edge::make_speculative): Likewise.
29429 (cgraph_edge::resolve_speculation): Likewise.
29430 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29431 (cgraph_node::dump): Likewise.
29432 * cgraph.h: Likewise.
29433 * cgraphunit.c (analyze_functions): Likewise.
29434 (symbol_table::compile): Likewise.
29435 * ipa-cp.c (print_all_lattices): Likewise.
29436 (determine_versionability): Likewise.
29437 (initialize_node_lattices): Likewise.
29438 (ipcp_verify_propagated_values): Likewise.
29439 (estimate_local_effects): Likewise.
29440 (update_profiling_info): Likewise.
29441 (create_specialized_node): Likewise.
29442 (perhaps_add_new_callers): Likewise.
29443 (decide_about_value): Likewise.
29444 (decide_whether_version_node): Likewise.
29445 (identify_dead_nodes): Likewise.
29446 (ipcp_store_bits_results): Likewise.
29447 * ipa-devirt.c (dump_targets): Likewise.
29448 (ipa_devirt): Likewise.
29449 * ipa-icf.c (sem_item::dump): Likewise.
29450 (sem_function::equals): Likewise.
29451 (sem_variable::equals): Likewise.
29452 (sem_item_optimizer::read_section): Likewise.
29453 (sem_item_optimizer::execute): Likewise.
29454 (congruence_class::dump): Likewise.
29455 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
29456 (dump_inline_summary): Likewise.
29457 (estimate_node_size_and_time): Likewise.
29458 (inline_analyze_function): Likewise.
29459 * ipa-inline-transform.c (inline_call): Likewise.
29460 * ipa-inline.c (report_inline_failed_reason): Likewise.
29461 (want_early_inline_function_p): Likewise.
29462 (edge_badness): Likewise.
29463 (update_edge_key): Likewise.
29464 (inline_small_functions): Likewise.
29465 * ipa-profile.c (ipa_profile): Likewise.
29466 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
29467 (ipa_make_edge_direct_to_target): Likewise.
29468 (remove_described_reference): Likewise.
29469 (ipa_impossible_devirt_target): Likewise.
29470 (propagate_controlled_uses): Likewise.
29471 (ipa_print_node_params): Likewise.
29472 (ipcp_transform_function): Likewise.
29473 * ipa-pure-const.c (pure_const_read_summary): Likewise.
29474 (propagate_pure_const): Likewise.
29475 * ipa-reference.c (generate_summary): Likewise.
29476 (read_write_all_from_decl): Likewise.
29477 (propagate): Likewise.
29478 (ipa_reference_read_optimization_summary): Likewise.
29479 * ipa-utils.c (ipa_merge_profiles): Likewise.
29480 * ipa.c (walk_polymorphic_call_targets): Likewise.
29481 (symbol_table::remove_unreachable_nodes): Likewise.
29482 (ipa_single_use): Likewise.
29483 * passes.c (execute_todo): Likewise.
29484 * predict.c (drop_profile): Likewise.
29485 * symtab.c (symtab_node::get_dump_name): New function.
29486 (symtab_node::dump_name): Likewise.
29487 (symtab_node::dump_asm_name): Likewise.
29488 (symtab_node::dump_references): Likewise.
29489 (symtab_node::dump_referring): Likewise.
29490 (symtab_node::dump_base): Likewise.
29491 (symtab_node::debug_symtab): Likewise.
29492 * tree-sra.c (convert_callers_for_node): Likewise.
29493 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
29494 * value-prof.c (init_node_map): Likewise.
29496 2017-05-23 Martin Liska <mliska@suse.cz>
29498 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
29499 and symtab_node::debug_symtab to symbol_table::debug.
29500 * cgraphunit.c (analyze_functions): Use the renamed function.
29501 (symbol_table::compile): Likewise.
29502 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
29503 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
29504 * passes.c (execute_todo): Likewise.
29505 * symtab.c (symbol_table::dump): New function.
29506 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
29508 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29510 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
29511 that nonconst implies exec.
29513 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29515 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29516 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29517 (inline_edge_summary_vec): Turn into ...
29518 (ipa_call_summaries): ... this one.
29519 (redirect_to_unreachable, edge_set_predicate,
29520 evaluate_properties_for_edge, inline_summary_alloc,
29521 reset_ipa_call_summary, reset_inline_summary,
29522 inline_summary_t::duplicate): Update.
29523 (inline_edge_duplication_hook): Turn to ...
29524 (ipa_call_summary_t::duplicate): ... this one.
29525 (inline_edge_removal_hook): Turn to ...
29526 (ipa_call_summary_t::remove): ... this one.
29527 (dump_inline_edge_summary): Turn to ...
29528 (dump_ipa_call_summary): ... this one.
29529 (estimate_function_body_sizes): Update.
29530 (inline_update_callee_summaries): Update.
29531 (remap_edge_change_prob): Update.
29532 (remap_edge_summaries): Update.
29533 (inline_merge_summary): Update.
29534 (do_estimate_edge_time): Update.
29535 (inline_generate_summary): Update.
29536 (inline_read_section): Update.
29537 (inline_read_summary): Update.
29538 (inline_free_summary): Update.
29539 * ipa-inline.c (can_inline_edge_p): Update.
29540 (compute_inlined_call_time): Update.
29541 (want_inline_small_function_p): Update.
29542 (edge_badness): Update.
29543 (early_inliner): Update.
29544 * ipa-inline.h (inline_edge_summary): Turn to ...
29545 (ipa_call_summary): ... this one.
29546 (ipa_call_summary_t): New class.
29547 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29548 (ipa_call_summaries): New.
29549 (inline_edge_summary): Remove.
29550 (estimate_edge_growth): Update.
29551 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29552 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29553 * ipa-split.c (execute_split_functions): Update.
29554 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29556 2017-05-23 Tom de Vries <tom@codesourcery.com>
29558 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29559 attributes): Document rdrand effective target.
29561 2017-05-23 Tom de Vries <tom@codesourcery.com>
29563 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29564 attributes): Sort alphabetically.
29566 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
29568 * config/avr/genmultilib.awk: Use gsub instead of gensub.
29570 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29573 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
29574 V2DF/V2DI splat into two separate patterns, one that handles
29575 registers, and the other that only handles memory. Drop support
29576 for splatting from a GPR on ISA 2.07 and then splitting the
29577 splat into direct move and splat.
29578 (vsx_splat_<mode>_reg): Likewise.
29579 (vsx_splat_<mode>_mem): Likewise.
29581 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
29583 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
29585 2017-05-22 Jakub Jelinek <jakub@redhat.com>
29587 PR middle-end/80809
29588 * omp-low.c (finish_taskreg_remap): New function.
29589 (finish_taskreg_scan): If unit size of ctx->record_type
29590 is non-constant, unshare the size expression and replace
29591 decls in it with possible outer var refs.
29593 PR middle-end/80809
29594 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
29595 GOVD_SHARED rather than GOVD_PRIVATE with it.
29596 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
29597 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
29599 PR middle-end/80853
29600 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
29601 as last argument to build_outer_var_ref for pointer bases of array
29602 section reductions.
29604 2017-05-19 Martin Sebor <msebor@redhat.com>
29606 * print-tree.c (print_node): Print DECL_READ_P flag.
29608 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29610 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
29611 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
29612 * cgraph.c: Likewise.
29613 * cgraphunit.c: Likewise.
29614 * gengtype.c: Likewise.
29615 * ipa-cp.c: Likewise.
29616 * ipa-devirt.c: Likewise.
29617 * ipa-icf.c: Likewise.
29618 * ipa-predicate.c: Likewise.
29619 * ipa-profile.c: Likewise.
29620 * ipa-prop.c: Likewise.
29621 * ipa-split.c: Likewise.
29623 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
29624 edge_predicate_pool, dump_inline_hints,
29625 inline_summary::account_size_time, redirect_to_unreachable,
29626 edge_set_predicate, set_hint_predicate,
29627 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
29628 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
29629 inline_summary_t::remove, remap_hint_predicate_after_duplication,
29630 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
29631 ipa_call_summary_t::remove, initialize_growth_caches,
29632 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
29633 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
29634 mark_modified, unmodified_parm_1, unmodified_parm,
29635 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
29636 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
29637 compute_bb_predicates, will_be_nonconstant_expr_predicate,
29638 will_be_nonconstant_predicate, record_modified_bb_info,
29639 get_minimal_bb, record_modified, param_change_prob,
29640 phi_result_unknown_predicate, predicate_for_phi_result,
29641 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
29642 estimate_function_body_sizes, compute_inline_parameters,
29643 compute_inline_parameters_for_curren, pass_data_inline_parameters,
29644 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
29645 inline_update_callee_summaries, remap_edge_change_prob,
29646 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
29647 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
29648 inline_analyze_function, inline_summary_t::insert,
29649 inline_generate_summary, read_ipa_call_summary, inline_read_section,
29650 inline_read_summary, write_ipa_call_summary, inline_write_summary,
29651 inline_free_summary): Move to ipa-fnsummary.h
29652 (predicate_t): Remove.
29653 * ipa-fnsummary.c: New file.
29654 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
29655 (enum inline_hints_vals, inline_hints, agg_position_info,
29656 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
29657 inline_summaries, ipa_call_summary, ipa_call_summary_t,
29658 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
29659 dump_inline_summary, dump_inline_hints, inline_generate_summary,
29660 inline_read_summary, inline_write_summary, inline_free_summary,
29661 inline_analyze_function, initialize_inline_failed,
29662 inline_merge_summary, inline_update_overall_summary,
29663 compute_inline_parameters): Move to ipa-fnsummary.h
29664 * ipa-fnsummary.h: New file.
29665 * ipa-inline-transform.h: Include ipa-inline.h.
29666 * ipa-inline.c: LIkewise.
29668 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29670 * ipa-inline.c (edge_badness): Use inlined_time instead of
29671 inline_summaries->get.
29673 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29675 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
29677 2017-05-22 Nathan Sidwell <nathan@acm.org>
29679 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
29680 (fdump-lang): Document 'raw' option.
29681 * dumpfile.h (TDI_tu): Delete.
29682 * dumpfile.c (dump_files): Remove translation-unit.
29683 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
29685 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
29687 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
29688 command option from $(AWK) call.
29689 * config/avr/genmultilib.awk: Simplify and rewrite so that it
29690 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
29691 [FORMAT]: Remove handling of variable.
29692 * config/avr/t-multilib: Regenerate.
29694 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29696 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
29698 (dump_inline_summary): Do not print self_time.
29699 (estimate_function_body_sizes): Do not set self_time.
29700 (compute_inline_parameters): Likewise.
29701 (inline_read_section, inline_write_summary): Do not stream self_time.
29702 * ipa-inline.h (inline_summary): Drop self_time.
29704 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29706 * ipa-inline-analysis.c (account_size_time): Rename to ...
29707 (inline_summary::account_size_time): ... this one.
29708 (reset_ipa_call_summary): Turn to ...
29709 (ipa_call_summary::reset): ... this one.
29710 (reset_inline_summary): Turn to ...
29711 (inline_summary::reset): ... this one.
29712 (inline_summary_t::remove): Update.
29713 (inline_summary_t::duplicate): Update.
29714 (ipa_call_summary_t::remove): Update.
29715 (dump_inline_summary): Update.
29716 (estimate_function_body_sizes): Update.
29717 (compute_inline_parameters): Update.
29718 (estimate_node_size_and_time): Update.
29719 (inline_merge_summary): Update.
29720 (inline_update_overall_summary): Update.
29721 (inline_read_section): Update.
29722 (inline_write_summary): Update.
29723 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
29724 add account_size_time and reset member functions.
29725 (ipa_call_summary): Add reset function.
29726 * ipa-predicate.h (predicate::operator &): Constify.
29728 2017-05-22 Richard Biener <rguenther@suse.de>
29730 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
29732 2017-05-19 Jason Merrill <jason@redhat.com>
29734 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
29736 2017-05-19 Marek Polacek <polacek@redhat.com>
29739 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
29740 TYPE_OVERFLOW_WRAPS checks.
29742 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
29744 * tree-core.h (enum omp_clause_default_kind): Add
29745 "OMP_CLAUSE_DEFAULT_PRESENT".
29746 * tree-pretty-print.c (dump_omp_clause): Handle it.
29747 * gimplify.c (enum gimplify_omp_var_data): Add
29748 "GOVD_MAP_FORCE_PRESENT".
29749 (gimplify_adjust_omp_clauses_1): Map it to
29750 "GOMP_MAP_FORCE_PRESENT".
29751 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
29753 * gimplify.c (oacc_default_clause): Clarify.
29755 2017-05-19 Nathan Sidwell <nathan@acm.org>
29757 LANG_HOOK_REGISTER_DUMPS
29758 * toplev.c (general_init): Call register dump lang hook.
29759 * doc/invoke.texi: Document -fdump-lang option family.
29760 * dumpfile.c (dump_files): Remove class dump here.
29761 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
29762 * dumpfile.h (tree_dump_index): Remove TDI_class.
29763 * langhooks-def.h (lhd_register_dumps): Declare.
29764 (LANG_HOOKS_REGISTER_DUMPS): Define.
29765 (LANG_HOOKS_INITIALIZER): Add it.
29766 * langhooks.c (lhd_register_dumps): Define.
29767 * langhooks.h (struct lang_hooks): Add register_dumps.
29769 2017-05-19 Nathan Sidwell <nathan@acm.org>
29771 * context.h (context::set_passes): New.
29772 * context.c (context::context): Do not create pass manager.
29773 * toplev.c (general_init): Create pass manager here.
29775 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
29777 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
29778 use this splitter if two add or or instructions would also work for
29779 the constant we want to generate.
29781 2017-05-19 Richard Biener <rguenther@suse.de>
29784 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
29785 predicate evaluation.
29787 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29789 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
29791 * ipa-inline.c (want_inline_small_function_p): Do not cast to
29794 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29796 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29797 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29798 (inline_edge_summary_vec): Turn into ...
29799 (ipa_call_summaries): ... this one.
29800 (redirect_to_unreachable, edge_set_predicate,
29801 evaluate_properties_for_edge, inline_summary_alloc,
29802 reset_ipa_call_summary, reset_inline_summary,
29803 inline_summary_t::duplicate): Update.
29804 (inline_edge_duplication_hook): Turn to ...
29805 (ipa_call_summary_t::duplicate): ... this one.
29806 (inline_edge_removal_hook): Turn to ...
29807 (ipa_call_summary_t::remove): ... this one.
29808 (dump_inline_edge_summary): Turn to ...
29809 (dump_ipa_call_summary): ... this one.
29810 (estimate_function_body_sizes): Update.
29811 (inline_update_callee_summaries): Update.
29812 (remap_edge_change_prob): Update.
29813 (remap_edge_summaries): Update.
29814 (inline_merge_summary): Update.
29815 (do_estimate_edge_time): Update.
29816 (inline_generate_summary): Update.
29817 (inline_read_section): Update.
29818 (inline_read_summary): Update.
29819 (inline_free_summary): Update.
29820 * ipa-inline.c (can_inline_edge_p): Update.
29821 (compute_inlined_call_time): Update.
29822 (want_inline_small_function_p): Update.
29823 (edge_badness): Update.
29824 (early_inliner): Update.
29825 * ipa-inline.h (inline_edge_summary): Turn to ...
29826 (ipa_call_summary): ... this one.
29827 (ipa_call_summary_t): New class.
29828 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29829 (ipa_call_summaries): New.
29830 (inline_edge_summary): Remove.
29831 (estimate_edge_growth): Update.
29832 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29833 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29834 * ipa-split.c (execute_split_functions): Update.
29835 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29837 2017-05-19 Richard Biener <rguenther@suse.de>
29839 PR middle-end/80764
29840 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
29842 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
29844 * config/rs6000/rs6000.c (struct machine_function): Add field
29845 fpr_is_wrapped_separately.
29846 (rs6000_get_separate_components): Use 64 components. Handle the
29847 new FPR components.
29848 (rs6000_components_for_bb): Handle the FPR components.
29849 (rs6000_emit_prologue_components): Handle the FPR components.
29850 (rs6000_emit_epilogue_components): Handle the FPR components.
29851 (rs6000_set_handled_components): Handle the FPR components.
29852 (rs6000_emit_prologue): Don't output prologue code for those FPRs
29853 that are already separately shrink-wrapped.
29854 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
29855 that are already separately shrink-wrapped.
29857 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29860 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
29863 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
29864 (define_peephole2 for Altivec d-form load): Add peepholes to catch
29865 cases where the register allocator uses a move and an offsettable
29866 memory operation to/from a FPR register on ISA 2.06/2.07.
29867 (define_peephole2 for Altivec d-form store): Likewise.
29869 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
29872 * config/i386/mmx.md (*mov<mode>_internal): Enable
29873 alternatives 11, 12, 13 and 14 also for 32bit targets.
29874 Remove alternatives 15, 16, 17 and 18.
29875 * config/i386/sse.md (vec_concatv2di): Change
29876 alternative (!x, *y) to (x, ?!*Yn).
29878 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
29880 * dumpfile.h (enum dump_kind): Remove stray comma.
29882 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29884 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
29885 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
29886 predicate::num_conditions
29887 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
29888 (CHANGED): turn into predicate::changed.
29889 (agg_position_info): Move to ipa-predicate.h
29890 (add_condition, predicate::add_clause, predicate::operator &=,
29891 predicate::or_with, predicate::evaluate, predicate::probability,
29892 dump_condition, dump_clause, predicate::dump,
29893 predicate::remap_after_duplication, predicate::remap_after_inlining,
29894 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
29895 (evaluate_conditions_for_known_args): Update.
29896 (set_cond_stmt_execution_predicate): Update.
29897 * ipa-inline.h: Include ipa-predicate.h
29898 (condition, inline_param_summary, conditions, agg_position_info,
29899 predicate): Move to ipa-predicate.h
29900 * ipa-predicate.c: New file.
29901 * ipa-predicate.h: New file.
29903 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
29905 * final.c (leaf_function_p): Check we are not in a sequence.
29907 2017-05-18 Martin Liska <mliska@suse.cz>
29909 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
29910 * dumpfile.c (dump_register): Use new enum dump_kind.
29911 (get_dump_file_name): Likewise.
29912 (dump_enable_all): Likewise.
29913 (dump_switch_p_1): Likewise.
29914 (enable_rtl_dump_file): Remove usage of TDF_RTL.
29915 * dumpfile.h (enum dump_kind): New enum type.
29916 (struct dump_file_info): Create constructor and
29917 format fields and comments.
29918 * passes.c (pass_manager::register_one_dump_file):
29920 * statistics.c (statistics_early_init): Likewise.
29921 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
29922 TDF_TREE with TDF_SLIM.
29923 (gather_memory_references_ref): Likewise.
29925 2017-05-18 Martin Liska <mliska@suse.cz>
29927 * vec.h (struct vnull): Use it.
29929 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29931 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
29932 (true_predicate, false_predicate, true_predicate_p,
29933 false_predicate_p): Remove.
29934 (single_cond_predicate, not_inlined_predicate): Turn to member function
29936 (add_condition): Update.
29937 (add_clause): Turn to...
29938 (predicate::add_clause): ... this one; update; allow passing NULL
29940 (and_predicates): Turn to ...
29941 (predicate::operator &=): ... this one.
29942 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
29943 (or_predicates): Turn to ...
29944 (predicate::or_with): ... this one.
29945 (evaluate_predicate): Turn to ...
29946 (predicate::evaluate): ... this one.
29947 (predicate_probability): Turn to ...
29948 (predicate::probability): ... this one.
29949 (dump_condition): Update.
29950 (dump_predicate): Turn to ...
29951 (predicate::dump): ... this one.
29952 (account_size_time): Update.
29953 (edge_set_predicate): Update.
29954 (set_hint_predicate): UPdate.
29955 (evaluate_conditions_for_known_args): Update.
29956 (evaluate_properties_for_edge): Update.
29957 (remap_predicate_after_duplication): Turn to...
29958 (predicate::remap_after_duplication): ... this one.
29959 (remap_hint_predicate_after_duplication): Update.
29960 (inline_summary_t::duplicate): UPdate.
29961 (dump_inline_edge_summary): Update.
29962 (dump_inline_summary): Update.
29963 (set_cond_stmt_execution_predicate): Update.
29964 (set_switch_stmt_execution_predicate): Update.
29965 (compute_bb_predicates): Update.
29966 (will_be_nonconstant_expr_predicate): Update.
29967 (will_be_nonconstant_predicate): Update.
29968 (phi_result_unknown_predicate): Update.
29969 (predicate_for_phi_result): Update.
29970 (array_index_predicate): Update.
29971 (estimate_function_body_sizes): Update.
29972 (estimate_node_size_and_time): Update.
29973 (estimate_ipcp_clone_size_and_time): Update.
29974 (remap_predicate): Rename to ...
29975 (predicate::remap_after_inlining): ... this one.
29976 (remap_hint_predicate): Update.
29977 (inline_merge_summary): Update.
29978 (inline_update_overall_summary): Update.
29979 (estimate_size_after_inlining): Update.
29980 (read_predicate): Rename to ...
29981 (predicate::stream_in): ... this one.
29982 (read_inline_edge_summary): Update.
29983 (write_predicate): Rename to ...
29984 (predicate::stream_out): ... this one.
29985 (write_inline_edge_summary): Update.
29986 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
29987 (clause_t): Turn to uint32_t
29988 (predicate): Turn to class; implement constructor and operators
29990 (size_time_entry): Update.
29991 (inline_summary): Update.
29992 (inline_edge_summary): Update.
29994 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
29996 * fold-const.c (fold_binary_loc): Move transformation...
29997 * match.pd (C - X CMP X): ... here.
29999 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
30001 * config/sparc/sparc.c (sparc_option_override): Set function
30002 alignment for -mcpu=niagara7 to 64 to match the I$ line.
30003 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
30005 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
30007 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
30009 2017-05-18 Marek Polacek <polacek@redhat.com>
30012 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
30013 (pass_ubsan::execute): Call gimple_assign_single_p instead of
30014 gimple_assign_load_p.
30016 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
30018 PR middle-end/80692
30019 * real.c (do_compare): Give decimal_do_compare preference over
30020 comparing just the signs.
30022 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
30024 * doc/md.texi (Canonicalization of Instructions): Describe the
30025 canonical form of instructions that inherently set a condition
30028 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
30030 PR middle-end/80775
30031 * tree-cfg.c: Move deletion of unreachable case statements to after
30032 the merging of consecutive case labels.
30034 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
30036 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
30037 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
30038 restoring of callee-saved registers.
30040 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
30042 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
30043 * config/visium/visium.c (single_set_and_flags): Likewise.
30044 * config/visium/visium.md (Substitutions): Likewise.
30046 2017-05-17 Martin Liska <mliska@suse.cz>
30048 * cfg.c: Introduce dump_flags_t type and
30049 use it instead of int type.
30051 * cfghooks.c: Likewise.
30052 * cfghooks.h (struct cfg_hooks): Likewise.
30053 * cfgrtl.c: Likewise.
30054 * cfgrtl.h: Likewise.
30055 * cgraph.c (cgraph_node::get_body): Likewise.
30056 * coretypes.h: Likewise.
30057 * domwalk.c: Likewise.
30058 * domwalk.h: Likewise.
30059 * dumpfile.c (struct dump_option_value_info): Likewise.
30060 (dump_enable_all): Likewise.
30061 (dump_switch_p_1): Likewise.
30062 (opt_info_switch_p): Likewise.
30063 * dumpfile.h (enum tree_dump_index): Likewise.
30064 (struct dump_file_info): Likewise.
30065 * genemit.c: Likewise.
30066 * generic-match-head.c: Likewise.
30067 * gengtype.c (open_base_files): Likewise.
30068 * gimple-pretty-print.c: Likewise.
30069 * gimple-pretty-print.h: Likewise.
30070 * graph.c (print_graph_cfg): Likewise.
30071 * graphite-scop-detection.c (dot_all_sese): Likewise.
30072 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
30073 * loop-unroll.c (report_unroll): Likewise.
30074 * passes.c (pass_manager::register_one_dump_file): Likewise.
30075 * print-tree.c: Likewise.
30076 * statistics.c: Likewise.
30077 * tree-cfg.c: Likewise.
30078 * tree-cfg.h: Likewise.
30079 * tree-dfa.c: Likewise.
30080 * tree-dfa.h: Likewise.
30081 * tree-dump.c (dump_function): Likewise.
30082 * tree-dump.h (struct dump_info): Likewise.
30083 * tree-pretty-print.c: Likewise.
30084 * tree-pretty-print.h: Likewise.
30085 * tree-ssa-live.c: Likewise.
30086 * tree-ssa-live.h: Likewise.
30087 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
30088 * tree-vect-loop.c: Likewise.
30089 * tree-vect-slp.c: Likewise.
30091 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
30092 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30094 PR tree-optimization/80457
30095 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
30096 of all arguments to a statement as scalar_to_vec operations.
30097 (vectorizable_call): Adjust call to vect_model_simple_cost for
30099 (vectorizable_conversion): Likewise.
30100 (vectorizable_assignment): Likewise.
30101 (vectorizable_shift): Likewise.
30102 (vectorizable_operation): Likewise.
30103 (vectorizable_comparison): Likewise.
30104 (vect_is_simple_cond): Record the def types for operands.
30105 (vectorizable_condition): Likewise, call vect_model_simple_cost.
30106 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
30107 for statement argument count.
30109 2017-05-16 Carl Love <cel@us.ibm.com>
30111 * config/rs6000/rs6000-c: Add support for built-in functions
30112 vector unsigned long long vec_bperm (vector unsigned long long,
30113 vector unsigned char)
30114 vector signed long long vec_mule (vector signed int,
30116 vector unsigned long long vec_mule (vector unsigned int,
30117 vector unsigned int)
30118 vector signed long long vec_mulo (vector signed int,
30120 vector unsigned long long vec_mulo (vector unsigned int,
30121 vector unsigned int)
30122 vector signed char vec_sldw (vector signed char,
30123 vector signed char,
30125 vector unsigned char vec_sldw (vector unsigned char,
30126 vector unsigned char,
30128 vector signed short vec_sldw (vector signed short,
30129 vector signed short,
30131 vector unsigned short vec_sldw (vector unsigned short,
30132 vector unsigned short,
30134 vector signed int vec_sldw (vector signed int,
30137 vector unsigned int vec_sldw (vector unsigned int,
30138 vector unsigned int,
30140 vector signed long long vec_sldw (vector signed long long,
30141 vector signed long long,
30143 vector unsigned long long vec_sldw (vector unsigned long long,
30144 vector unsigned long long,
30146 * config/rs6000/rs6000-c: Add support for built-in functions
30147 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
30148 * config/rs6000/altivec.h: Add defintion for vec_sldw.
30149 * doc/extend.texi: Update the built-in documentation for the
30150 new built-in functions.
30152 2017-05-16 Marek Polacek <polacek@redhat.com>
30156 * tree.c (save_expr): Don't fold the expression.
30158 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
30160 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
30161 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
30162 and (?*y,m). Update insn attributes.
30164 2017-05-16 Martin Liska <mliska@suse.cz>
30166 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
30167 flags argument of print_gimple_stmt, print_gimple_expr,
30168 print_generic_stmt and print_generic_expr.
30169 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
30170 * coretypes.h: Likewise.
30171 * except.c (dump_eh_tree): Likewise.
30172 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
30173 * gimple-pretty-print.h: Likewise.
30174 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
30175 (backprop::push_to_worklist): Likewise.
30176 (backprop::pop_from_worklist): Likewise.
30177 (backprop::process_use): Likewise.
30178 (backprop::intersect_uses): Likewise.
30179 (note_replacement): Likewise.
30180 * gimple-ssa-store-merging.c
30181 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
30182 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
30183 (pass_store_merging::execute): Likewise.
30184 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
30185 (ssa_base_cand_dump_callback): Likewise.
30186 (dump_incr_vec): Likewise.
30187 (replace_refs): Likewise.
30188 (replace_mult_candidate): Likewise.
30189 (create_add_on_incoming_edge): Likewise.
30190 (create_phi_basis): Likewise.
30191 (insert_initializers): Likewise.
30192 (all_phi_incrs_profitable): Likewise.
30193 (introduce_cast_before_cand): Likewise.
30194 (replace_one_candidate): Likewise.
30195 * gimplify.c (gimplify_expr): Likewise.
30196 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
30197 (set_rename): Likewise.
30198 (rename_uses): Likewise.
30199 (copy_loop_phi_nodes): Likewise.
30200 (add_close_phis_to_merge_points): Likewise.
30201 (copy_loop_close_phi_args): Likewise.
30202 (copy_cond_phi_args): Likewise.
30203 (graphite_copy_stmts_from_block): Likewise.
30204 (translate_pending_phi_nodes): Likewise.
30205 * graphite-poly.c (print_pdr): Likewise.
30206 (dump_gbb_cases): Likewise.
30207 (dump_gbb_conditions): Likewise.
30208 (print_scop_params): Likewise.
30209 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
30210 (build_cross_bb_scalars_use): Likewise.
30211 (gather_bbs::before_dom_children): Likewise.
30212 * hsa-dump.c (dump_hsa_immed): Likewise.
30213 * ipa-cp.c (print_ipcp_constant_value): Likewise.
30214 (get_replacement_map): Likewise.
30215 * ipa-inline-analysis.c (dump_condition): Likewise.
30216 (estimate_function_body_sizes): Likewise.
30217 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
30218 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
30219 * ipa-prop.c (ipa_dump_param): Likewise.
30220 (ipa_print_node_jump_functions_for_edge): Likewise.
30221 (ipa_modify_call_arguments): Likewise.
30222 (ipa_modify_expr): Likewise.
30223 (ipa_dump_param_adjustments): Likewise.
30224 (ipa_dump_agg_replacement_values): Likewise.
30225 (ipcp_modif_dom_walker::before_dom_children): Likewise.
30226 * ipa-pure-const.c (check_stmt): Likewise.
30227 (pass_nothrow::execute): Likewise.
30228 * ipa-split.c (execute_split_functions): Likewise.
30229 * omp-offload.c (dump_oacc_loop_part): Likewise.
30230 (dump_oacc_loop): Likewise.
30231 * trans-mem.c (tm_log_emit): Likewise.
30232 (tm_memopt_accumulate_memops): Likewise.
30233 (dump_tm_memopt_set): Likewise.
30234 (dump_tm_memopt_transform): Likewise.
30235 * tree-cfg.c (gimple_verify_flow_info): Likewise.
30236 (print_loop): Likewise.
30237 * tree-chkp-opt.c (chkp_print_addr): Likewise.
30238 (chkp_gather_checks_info): Likewise.
30239 (chkp_get_check_result): Likewise.
30240 (chkp_remove_check_if_pass): Likewise.
30241 (chkp_use_outer_bounds_if_possible): Likewise.
30242 (chkp_reduce_bounds_lifetime): Likewise.
30243 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
30244 (chkp_mark_completed_bounds): Likewise.
30245 (chkp_register_incomplete_bounds): Likewise.
30246 (chkp_mark_invalid_bounds): Likewise.
30247 (chkp_maybe_copy_and_register_bounds): Likewise.
30248 (chkp_build_returned_bound): Likewise.
30249 (chkp_get_bound_for_parm): Likewise.
30250 (chkp_build_bndldx): Likewise.
30251 (chkp_get_bounds_by_definition): Likewise.
30252 (chkp_generate_extern_var_bounds): Likewise.
30253 (chkp_get_bounds_for_decl_addr): Likewise.
30254 * tree-chrec.c (chrec_apply): Likewise.
30255 * tree-data-ref.c (dump_data_reference): Likewise.
30256 (dump_subscript): Likewise.
30257 (dump_data_dependence_relation): Likewise.
30258 (analyze_overlapping_iterations): Likewise.
30259 * tree-inline.c (expand_call_inline): Likewise.
30260 (tree_function_versioning): Likewise.
30261 * tree-into-ssa.c (dump_defs_stack): Likewise.
30262 (dump_currdefs): Likewise.
30263 (dump_names_replaced_by): Likewise.
30264 (dump_update_ssa): Likewise.
30265 (update_ssa): Likewise.
30266 * tree-object-size.c (pass_object_sizes::execute): Likewise.
30267 * tree-parloops.c (build_new_reduction): Likewise.
30268 (try_create_reduction_list): Likewise.
30269 (ref_conflicts_with_region): Likewise.
30270 (oacc_entry_exit_ok_1): Likewise.
30271 (oacc_entry_exit_single_gang): Likewise.
30272 * tree-pretty-print.h: Likewise.
30273 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
30274 (get_scalar_evolution): Likewise.
30275 (add_to_evolution): Likewise.
30276 (get_loop_exit_condition): Likewise.
30277 (analyze_evolution_in_loop): Likewise.
30278 (analyze_initial_condition): Likewise.
30279 (analyze_scalar_evolution): Likewise.
30280 (instantiate_scev): Likewise.
30281 (number_of_latch_executions): Likewise.
30282 (gather_chrec_stats): Likewise.
30283 (final_value_replacement_loop): Likewise.
30284 (scev_const_prop): Likewise.
30285 * tree-sra.c (dump_access): Likewise.
30286 (disqualify_candidate): Likewise.
30287 (create_access): Likewise.
30288 (reject): Likewise.
30289 (maybe_add_sra_candidate): Likewise.
30290 (create_access_replacement): Likewise.
30291 (analyze_access_subtree): Likewise.
30292 (analyze_all_variable_accesses): Likewise.
30293 (sra_modify_assign): Likewise.
30294 (initialize_constant_pool_replacements): Likewise.
30295 (find_param_candidates): Likewise.
30296 (decide_one_param_reduction): Likewise.
30297 (replace_removed_params_ssa_names): Likewise.
30298 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
30299 * tree-ssa-copy.c (dump_copy_of): Likewise.
30300 (copy_prop_visit_cond_stmt): Likewise.
30301 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
30302 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
30303 (record_equivalences_from_stmt): Likewise.
30304 * tree-ssa-dse.c (compute_trims): Likewise.
30305 (delete_dead_call): Likewise.
30306 (delete_dead_assignment): Likewise.
30307 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
30308 (forward_propagate_into_cond): Likewise.
30309 (pass_forwprop::execute): Likewise.
30310 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
30311 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
30313 (move_computations_worker): Likewise.
30314 (execute_sm): Likewise.
30315 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
30316 (remove_exits_and_undefined_stmts): Likewise.
30317 (remove_redundant_iv_tests): Likewise.
30318 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
30319 (adjust_iv_update_pos): Likewise.
30320 * tree-ssa-math-opts.c (bswap_replace): Likewise.
30321 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
30322 (value_replacement): Likewise.
30323 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
30324 * tree-ssa-pre.c (print_pre_expr): Likewise.
30325 (get_representative_for): Likewise.
30326 (create_expression_by_pieces): Likewise.
30327 (insert_into_preds_of_block): Likewise.
30328 (eliminate_insert): Likewise.
30329 (eliminate_dom_walker::before_dom_children): Likewise.
30330 (eliminate): Likewise.
30331 (remove_dead_inserted_code): Likewise.
30332 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
30333 * tree-ssa-reassoc.c (get_rank): Likewise.
30334 (eliminate_duplicate_pair): Likewise.
30335 (eliminate_plus_minus_pair): Likewise.
30336 (eliminate_not_pairs): Likewise.
30337 (undistribute_ops_list): Likewise.
30338 (eliminate_redundant_comparison): Likewise.
30339 (update_range_test): Likewise.
30340 (optimize_range_tests_var_bound): Likewise.
30341 (optimize_vec_cond_expr): Likewise.
30342 (rewrite_expr_tree): Likewise.
30343 (rewrite_expr_tree_parallel): Likewise.
30344 (linearize_expr): Likewise.
30345 (break_up_subtract): Likewise.
30346 (linearize_expr_tree): Likewise.
30347 (attempt_builtin_powi): Likewise.
30348 (attempt_builtin_copysign): Likewise.
30349 (transform_stmt_to_copy): Likewise.
30350 (transform_stmt_to_multiply): Likewise.
30351 (dump_ops_vector): Likewise.
30352 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
30353 (print_scc): Likewise.
30354 (set_ssa_val_to): Likewise.
30355 (visit_reference_op_store): Likewise.
30356 (visit_use): Likewise.
30357 (sccvn_dom_walker::before_dom_children): Likewise.
30358 (run_scc_vn): Likewise.
30359 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
30361 (expr_hash_elt::print): Likewise.
30362 (const_and_copies::pop_to_marker): Likewise.
30363 (const_and_copies::record_const_or_copy_raw): Likewise.
30364 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
30365 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
30366 (dump_predicates): Likewise.
30367 (find_uninit_use): Likewise.
30368 (warn_uninitialized_phi): Likewise.
30369 (pass_late_warn_uninitialized::execute): Likewise.
30370 * tree-ssa.c (verify_vssa): Likewise.
30371 (verify_ssa): Likewise.
30372 (maybe_optimize_var): Likewise.
30373 * tree-vrp.c (dump_value_range): Likewise.
30374 (dump_all_value_ranges): Likewise.
30375 (dump_asserts_for): Likewise.
30376 (register_edge_assert_for_2): Likewise.
30377 (vrp_visit_cond_stmt): Likewise.
30378 (vrp_visit_switch_stmt): Likewise.
30379 (vrp_visit_stmt): Likewise.
30380 (vrp_visit_phi_node): Likewise.
30381 (simplify_cond_using_ranges_1): Likewise.
30382 (fold_predicate_in): Likewise.
30383 (evrp_dom_walker::before_dom_children): Likewise.
30384 (evrp_dom_walker::push_value_range): Likewise.
30385 (evrp_dom_walker::pop_value_range): Likewise.
30386 (execute_early_vrp): Likewise.
30388 2017-05-16 Richard Biener <rguenther@suse.de>
30390 * dwarf2out.c (loc_list_from_tree_1): Do not create
30391 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
30393 2017-05-16 Richard Biener <rguenther@suse.de>
30395 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
30397 (note_variable_value_in_expr): If we resolved the decl ref
30398 do not push to the stack.
30400 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
30402 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
30403 operations in fast-math mode.
30404 (vaddq_f16): Likewise.
30405 (vmul_f16): Likewise.
30406 (vmulq_f16): Likewise.
30407 (vsub_f16): Likewise.
30408 (vsubq_f16): Likewise.
30409 * config/arm/neon.md (add<mode>3): New.
30411 (fma:<VH:mode>3): New. Also remove outdated comment.
30414 2017-05-16 Martin Liska <mliska@suse.cz>
30418 * ipa-devirt.c (warn_types_mismatch): Fix typo.
30419 (odr_types_equivalent_p): Likewise.
30421 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
30423 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
30425 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
30428 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
30429 non-interunit SSE move alternatives with '?'.
30430 (zero-extendsidi peephole2): New peephole to skip intermediate
30431 general register in SSE zero-extend sequence.
30433 2017-05-15 Jeff Law <law@redhat.com>
30435 * reorg.c (relax_delay_slots): Create a new variable to hold
30436 the temporary target rather than clobbering TARGET_LABEL.
30438 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
30439 missing argument to extract_bit_field call.
30440 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
30442 2017-05-15 Martin Liska <mliska@suse.cz>
30445 * gcc.c (process_command): Do not allow empty argument of -o option.
30447 2017-05-15 Renlin Li <renlin.li@arm.com>
30449 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
30450 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
30451 * config/aarch64/constraints.md (Usf): Add long call check.
30452 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
30453 (call_value): Likewise.
30454 (sibcall): Likewise.
30455 (sibcall_value): Likewise.
30457 (call_value_insn): New.
30458 (sibcall_insn): Update rtx pattern.
30459 (sibcall_value_insn): Likewise.
30460 (call_internal): Remove.
30461 (call_value_internal): Likewise.
30462 (sibcall_internal): Likewise.
30463 (sibcall_value_internal): Likewise.
30464 (call_reg): Likewise.
30465 (call_symbol): Likewise.
30466 (call_value_reg): Likewise.
30467 (call_value_symbol): Likewise.
30469 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
30472 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
30474 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
30476 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
30477 compatible with CCGOCmode and with CCZmode.
30479 2017-05-14 Martin Sebor <msebor@redhat.com>
30481 PR middle-end/77671
30482 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
30483 (gimple_fold_builtin_snprintf): Same.
30484 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
30485 (gimple_fold_builtin_snprintf): Same.
30486 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
30487 of character types.
30488 (is_call_safe): New function.
30489 (try_substitute_return_value): Call it.
30490 (try_simplify_call): New function.
30491 (pass_sprintf_length::handle_gimple_call): Call it.
30493 2017-05-14 Martin Sebor <msebor@redhat.com>
30495 PR middle-end/80669
30496 * builtins.c (expand_builtin_stpncpy): Simplify.
30498 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
30500 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
30501 * config/i386/i386.h
30502 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
30503 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
30504 (struct machine_function): Add new members call_ms2sysv,
30505 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
30506 (struct machine_frame_state): New fields sp_realigned and
30507 sp_realigned_offset.
30508 * config/i386/i386.c
30509 (enum xlogue_stub): New enum.
30510 (enum xlogue_stub_sets): New enum.
30511 (class xlogue_layout): New class.
30512 (struct ix86_frame): New fields stack_realign_allocate_offset,
30513 stack_realign_offset and outlined_save_offset. Modify comments to
30514 detail stack layout when using out-of-line stubs.
30515 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
30516 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
30517 -mcall-ms2sysv-xlogues.
30518 (stub_managed_regs): New static variable.
30519 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
30520 registers managed by out-of-line stub.
30521 (disable_call_ms2sysv_xlogues): New function.
30522 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
30523 m->call_ms2sysv when appropriate and compute frame layout for
30525 (sp_valid_at, fp_valid_at): New inline functions.
30526 (choose_basereg): New function.
30527 (choose_baseaddr): Add align parameter, use choose_basereg and modify
30529 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
30530 Use align parameter of choose_baseaddr to generated aligned SSE movs
30532 (pro_epilogue_adjust_stack): Modify to track
30533 machine_frame_state::sp_realigned.
30534 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
30535 (ix86_nsaved_sseregs): Likewise.
30536 (ix86_emit_save_regs): Likewise.
30537 (ix86_emit_save_regs_using_mov): Likewise.
30538 (ix86_emit_save_sse_regs_using_mov): Likewise.
30539 (get_scratch_register_on_entry): Likewise.
30540 (gen_frame_set): New function.
30541 (gen_frame_load): Likewise.
30542 (gen_frame_store): Likewise.
30543 (emit_outlined_ms2sysv_save): Likewise.
30544 (emit_outlined_ms2sysv_restore): Likewise.
30545 (ix86_expand_prologue): Modify stack re-alignment code and call
30546 emit_outlined_ms2sysv_save when appropriate.
30547 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
30548 parameter rtx_insn *insn, which allows the function to be used to only
30549 generate the notes.
30550 (ix86_expand_epilogue): Modify validity checks of frame and stack
30551 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
30552 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
30553 * config/i386/predicates.md
30554 (save_multiple): New predicate.
30555 (restore_multiple): Likewise.
30556 * config/i386/sse.md
30557 (save_multiple<mode>): New pattern.
30558 (save_multiple_realign<mode>): Likewise.
30559 (restore_multiple<mode>): Likewise.
30560 (restore_multiple_and_return<mode>): Likewise.
30561 (restore_multiple_leave_return<mode>): Likewise.
30562 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
30564 2017-05-14 Julia Koval <julia.koval@intel.com>
30566 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
30567 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
30568 (__builtin_ia32_xsetbv): New builtins.
30569 * config/i386/i386.c (ix86_expand_special_args_builtin):
30571 (ix86_expand_builtin): Special expand for new intrinsics.
30572 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
30573 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
30574 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
30576 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30578 * cfganal.c (inverted_post_order_compute): Change argument type
30580 * cfganal.h (inverted_post_order_compute): Adjust prototype.
30581 * df-core.c (rest_of_handle_df_initialize): Adjust.
30582 (rest_of_handle_df_finish): Likewise.
30583 (df_analyze_1): Likewise.
30584 (df_analyze): Likewise.
30585 (loop_inverted_post_order_compute): Change argument to be a vec *.
30586 (df_analyze_loop): Adjust.
30587 (df_get_n_blocks): Likewise.
30588 (df_get_postorder): Likewise.
30589 * df.h (struct df_d): Change field to be a vec.
30590 * lcm.c (compute_laterin): Adjust.
30591 (compute_available): Likewise.
30592 * lra-lives.c (lra_create_live_ranges_1): Likewise.
30593 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
30594 * tree-ssa-pre.c (compute_antic): Likewise.
30596 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30598 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
30599 (depth_first_search::depth_first_search): Change structure init
30600 function to this constructor.
30601 (depth_first_search::add_bb): Rename function to this member.
30602 (depth_first_search::execute): Likewise.
30603 (flow_dfs_compute_reverse_finish): Adjust.
30605 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30607 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
30608 (longest_simple_path): Likewise.
30609 * shrink-wrap.c (spread_components): Likewise.
30610 (disqualify_problematic_components): Likewise.
30611 (emit_common_heads_for_components): Likewise.
30612 (emit_common_tails_for_components): Likewise.
30613 (insert_prologue_epilogue_for_components): Likewise.
30615 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30617 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
30620 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30622 * df-core.c (df_set_blocks): Start using auto_bitmap.
30623 (df_compact_blocks): Likewise.
30624 * df-problems.c (df_rd_confluence_n): Likewise.
30625 * df-scan.c (df_insn_rescan_all): Likewise.
30626 (df_process_deferred_rescans): Likewise.
30627 (df_update_entry_block_defs): Likewise.
30628 (df_update_exit_block_uses): Likewise.
30629 (df_entry_block_bitmap_verify): Likewise.
30630 (df_exit_block_bitmap_verify): Likewise.
30631 (df_scan_verify): Likewise.
30632 * lra-constraints.c (lra_constraints): Likewise.
30633 (undo_optional_reloads): Likewise.
30634 (lra_undo_inheritance): Likewise.
30635 * lra-remat.c (calculate_gen_cands): Likewise.
30636 (do_remat): Likewise.
30637 * lra-spills.c (assign_spill_hard_regs): Likewise.
30638 (spill_pseudos): Likewise.
30639 * tree-ssa-pre.c (bitmap_set_and): Likewise.
30640 (bitmap_set_subtract_values): Likewise.
30642 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30644 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
30645 management with auto_bitmap.
30646 (fix_inter_tick): Likewise.
30647 (fix_recovery_deps): Likewise.
30648 * ira.c (add_store_equivs): Likewise.
30649 (find_moveable_pseudos): Likewise.
30650 (split_live_ranges_for_shrink_wrap): Likewise.
30651 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
30652 (rtx_reuse_manager::seen_def_p): Likewise.
30653 (rtx_reuse_manager::set_seen_def): Likewise.
30654 * print-rtl.h (class rtx_reuse_manager): Likewise.
30656 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30658 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
30660 (migrate_btr_def): Likewise.
30661 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
30662 * df-core.c (loop_post_order_compute): Likewise.
30663 (loop_inverted_post_order_compute): Likewise.
30664 * hsa-common.h: Likewise.
30665 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
30666 * init-regs.c (initialize_uninitialized_regs): Likewise.
30667 * ipa-inline.c (resolve_noninline_speculation): Likewise.
30668 (inline_small_functions): Likewise.
30669 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
30670 * ira.c (combine_and_move_insns): Likewise.
30671 (build_insn_chain): Likewise.
30672 * loop-invariant.c (find_invariants): Likewise.
30673 * lower-subreg.c (propagate_pseudo_copies): Likewise.
30674 * predict.c (tree_predict_by_opcode): Likewise.
30675 (predict_paths_leading_to): Likewise.
30676 (predict_paths_leading_to_edge): Likewise.
30677 (estimate_loops_at_level): Likewise.
30678 (estimate_loops): Likewise.
30679 * shrink-wrap.c (try_shrink_wrapping): Likewise.
30680 (spread_components): Likewise.
30681 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
30682 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
30683 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
30684 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
30685 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
30686 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
30687 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
30688 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
30689 (mark_threaded_blocks): Likewise.
30690 (thread_through_all_blocks): Likewise.
30691 * tree-ssa.c (verify_ssa): Likewise.
30692 (execute_update_addresses_taken): Likewise.
30693 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
30695 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30697 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
30699 (post_order_compute): Likewise.
30700 (inverted_post_order_compute): Likewise.
30701 (pre_and_rev_post_order_compute_fn): Likewise.
30703 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30705 * genrecog.c (int_set::int_set): Explicitly construct our
30706 auto_vec base class.
30707 * vec.h (auto_vec::auto_vec): New constructor.
30709 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30711 * bitmap.h (class auto_bitmap): New constructor taking
30712 bitmap_obstack * argument.
30714 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30716 * bitmap.h (class auto_bitmap): Change type of m_bits to
30717 bitmap_head, and adjust ctor / dtor and member operators.
30719 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
30721 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
30722 when returned register mode doesn't match original mode.
30724 2017-05-12 Jeff Law <law@redhat.com>
30725 Jakub Jelinek <jakub@redhat.com>
30727 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
30728 we look for cc setter after the compare-elim changes.
30729 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
30730 within the vector to match what compare-elim now expects.
30731 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
30732 (xorsi3_flags, one_cmplsi2_flags): Likewise.
30734 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
30735 after the compare-elim changes.
30736 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
30737 the vector to match what compare-elim now expects.
30738 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
30739 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
30740 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
30741 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
30742 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
30744 * config/visium/visium.c (single_set_and_flags): Fix where
30745 we look for cc setter after the compare-elim changes.
30746 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
30747 with the vector to match what compare-elim now expects.
30748 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
30749 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
30750 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
30751 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
30752 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
30753 (neg<mode>2_insn_set_overflow): Likewise.
30755 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
30757 PR middle-end/79794
30758 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
30759 maybe_expand_insn call, set ops[0].target. If still set after call,
30760 set alt_rtl. Add extra arg to recursive calls.
30761 (extract_bit_field): Add alt_rtl argument. Pass to
30763 * expmed.h (extract_bit_field): Fix prototype.
30764 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
30765 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
30766 to extract_bit_field_calls.
30767 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
30768 Pass alt_rtl to extract_bit_field calls.
30769 * calls.c (store_unaligned_arguments_into_psuedos)
30770 load_register_parameters): Pass extra NULL to extract_bit_field calls.
30771 * optabs.c (maybe_legitimize_operand): Clear op->target when call
30773 * optabs.h (struct expand_operand): Add target bitfield.
30775 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30777 * compare-elim.c (try_eliminate_compare): Canonicalize
30778 operation with embedded compare to
30779 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
30780 (set (reg) (operation)].
30782 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
30784 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30787 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
30788 cost of adding a carry flag for ADC instruction.
30789 [case MINUS]: Ignore the cost of subtracting a carry flag
30790 for SBB instruction.
30792 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
30794 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
30796 * config/rs6000/bmiintrin.h: New file.
30797 * config/rs6000/bmi2intrin.h: New file.
30798 * config/rs6000/x86intrin.h: New file.
30800 2017-05-12 Jeff Law <law@redhat.com>
30802 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
30805 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
30807 PR middle-end/80707
30808 * tree-cfg.c: Remove cfg edges of unreachable case statements.
30810 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30812 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30813 early expansion of vector divide builtins.
30814 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
30815 builtins identified as having unsigned arguments.
30817 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30819 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
30820 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
30821 expansion of vector logical operations (and, andc, or, xor,
30824 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30826 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
30827 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
30829 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30831 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30832 early GIMPLE expansion of vector multiplies.
30834 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30836 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
30837 TARGET_HAVE_MOVT conditional.
30838 (movt splitter): Likewise.
30840 2017-05-12 Richard Biener <rguenther@suse.de>
30842 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
30843 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30844 Fold all stmts not inplace.
30846 2017-05-12 Richard Biener <rguenther@suse.de>
30848 PR tree-optimization/80713
30849 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
30850 inserted_exprs bit for not removed stmts.
30852 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
30854 PR middle-end/69921
30855 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
30856 parallelized" attribute for parallelized OpenACC kernels.
30857 * omp-offload.c (execute_oacc_device_lower): Use it.
30859 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
30860 Set "oacc kernels" attribute.
30861 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
30862 parameter. Adjust all users.
30863 (oacc_fn_attrib_kernels_p): Remove function.
30864 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
30865 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
30866 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
30867 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
30868 assert "oacc kernels" attribute is set.
30870 2017-05-11 Carl Love <cel@us.ibm.com>
30872 * config/rs6000/rs6000-c: Add support for built-in functions
30873 vector unsigned char vec_popcnt (vector signed char)
30874 vector unsigned char vec_popcnt (vector unsigned char)
30875 vector unsigned short vec_popcnt (vector signed short)
30876 vector unsigned short vec_popcnt (vector unsigned short)
30877 vector unsigned int vec_popcnt (vector signed int)
30878 vector unsigned int vec_popcnt (vector unsigned int)
30879 vector unsigned long long vec_popcnt (vector signed long long)
30880 vector unsigned long long vec_popcnt (vector unsigned long long)
30881 vector signed long long vec_slo (vector signed long long,
30882 vector signed char)
30883 vector signed long long vec_slo (vector signed long long,
30884 vector unsigned char)
30885 vector unsigned long long vec_slo (vector unsigned long long,
30886 vector signed char)
30887 vector unsigned long long vec_slo (vector unsigned long long,
30888 vector unsigned char)
30889 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
30890 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
30891 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
30892 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
30893 * doc/extend.texi: Update the built-in documentation file for the
30894 new built-in functions.
30896 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
30898 * attribs.h (sorted_attr_string): Move machine independent
30899 functions for target clone support from the i386 port to common
30900 code. Rename ix86_function_versions to common_function_versions.
30901 Rename make_name to make_unique_name.
30902 (common_function_versions): Likewise.
30903 (make_unique_name): Likewise.
30904 (make_dispatcher_decl): Likewise.
30905 (is_function_default_version): Likewise.
30906 * attribs.c (attr_strcmp): Likewise.
30907 (sorted_attr_string): Likewise.
30908 (common_function_versions): Likewise.
30909 (make_unique_name): Likewise.
30910 (make_dispatcher_decl): Likewise.
30911 (is_function_default_version): Likewise.
30912 * config/i386/i386.c (attr_strcmp): Likewise.
30913 (sorted_attr_string): Likewise.
30914 (ix86_function_versions): Likewise.
30915 (make_name): Likewise.
30916 (make_dispatcher_decl): Likewise.
30917 (is_function_default_version): Likewise.
30918 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
30920 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30923 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
30924 Account for direct move costs for vec_construct of integer
30927 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
30930 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
30931 (UNSPEC_STX_ATOMIC): Ditto.
30932 (loaddi_via_sse): New insn.
30933 (storedi_via_sse): Ditto.
30934 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
30935 Update corresponding peephole2 patterns.
30936 (atomic_storedi_fpu): Ditto.
30938 2017-05-11 Julia Koval <julia.koval@intel.com>
30940 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
30941 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
30943 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
30944 (__builtin_ia32_rsqrt14ss_mask): New builtins.
30945 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
30947 2017-05-11 Nathan Sidwell <nathan@acm.org>
30949 * graphite-poly.c: Include dumpfile.h.
30951 * dumpfle.h (dump_function): Declare here ...
30952 * tree-dump.h (dump_function): ... not here.
30953 * dumpfile.c: #include tree-cfg.h.
30954 (dump_function): Move here from ...
30955 * tree-dump.c (dump_function): ... here.
30956 * gimplify.c: #include splay-tree.h, not tree-dump.h.
30957 * graphite-poly.c: Don't include tree-dump.h.
30958 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
30959 * print-tree.c: Likewise.
30960 * stor-layout.c: Likewise.
30961 * tree-nested.c: Likewise.
30963 * dumpfile.c (dump_start): Use TDF_FLAGS.
30964 (dump_enable_all): Fix TDF_KIND check thinko.
30966 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
30968 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30969 array entries to represent two legal parameterizations of the
30970 overloaded __builtin_cmpb function, as represented by the
30971 P6_OV_BUILTIN_CMPB constant.
30972 (altivec_resolve_overloaded_builtin): Add special case handling
30973 for the __builtin_cmpb function, as represented by the
30974 P6_OV_BUILTIN_CMPB constant.
30975 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
30976 (BU_P6_64BIT_2): New macro.
30977 (BU_P6_OVERLOAD_2): New macro
30978 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
30979 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
30980 (CMPB): Add overload support to represent both 32-bit and 64-bit
30981 compare-bytes function.
30982 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
30983 support for TARGET_CMPB.
30984 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
30985 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30986 documentation of the __builtin_cmpb overloaded built-in function.
30988 2017-05-11 Richard Biener <rguenther@suse.de>
30990 PR tree-optimization/80705
30991 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
30992 bases are not vectorizable.
30994 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30996 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
30997 when counting register pressure.
30999 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31001 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
31002 (iv_ca_more_deps): Renamed to ...
31003 (iv_ca_compare_deps): ... this.
31004 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
31006 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31008 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
31010 (determine_group_iv_costs): ... here.
31011 (find_inv_vars_cb): Record inv var if it's not recorded before.
31013 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31015 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
31016 (get_shiftadd_cost): Ditto.
31018 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31020 * tree-ssa-address.c: Include header file.
31021 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
31023 (add_to_parts): Refactor.
31024 (addr_to_parts): New parameter. Update use of move_hint_to_base.
31025 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
31028 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31030 PR tree-optimization/53090
31031 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
31033 (extract_cond_operands): Detect condition with IV on both sides
31034 and return COMP_IV_EXPR_2.
31035 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
31036 (rewrite_use_compare): Simplify by removing call to function
31037 extract_cond_operands.
31039 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31041 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
31042 (extract_cond_operands): Detect condition comparing against non-
31043 invariant bound and return appropriate enum value.
31044 (find_interesting_uses_cond): Update use of extract_cond_operands.
31045 Handle its return value accordingly.
31046 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
31048 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31050 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
31051 nonlinear iv_use computation in loop invariant sensitive way.
31053 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31055 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
31056 (find_iv_candidates): Call relate_compare_use_with_all_cands.
31058 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31060 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
31061 (dump_cand): Support iv_cand.inv_exprs.
31062 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
31064 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
31067 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31069 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
31071 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
31072 as local function. Include necessary header files.
31073 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
31075 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31077 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
31079 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31081 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
31082 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
31083 RSHIFT_EXPR and BIT_NOT_EXPR.
31085 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31087 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
31088 (adjust_setup_cost): New parameter supporting round up adjustment.
31089 (struct address_cost_data): Delete.
31090 (force_expr_to_var_cost): Don't bound cost with spill_cost.
31091 (split_address_cost, ptr_difference_cost): Delete.
31092 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
31093 (struct ainc_cost_data): New struct.
31094 (get_address_cost_ainc): New function.
31095 (get_address_cost, get_computation_cost): Reimplement.
31096 (determine_group_iv_cost_address): Record inv_expr for all uses of
31098 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
31099 (iv_ca_has_deps): Reimplemented to ...
31100 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
31102 (iv_ca_extend): Call iv_ca_more_deps.
31104 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31106 * tree-ssa-address.c (struct mem_address): Move to header file.
31107 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
31108 * tree-ssa-address.h (struct mem_address): Move from C file.
31109 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
31111 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31113 * tree-affine.h (aff_combination_type): New interface.
31114 (aff_combination_zero_p): Remove static.
31115 (aff_combination_const_p): New interface.
31116 (aff_combination_singleton_var_p): New interfaces.
31118 2017-05-11 Richard Biener <rguenther@suse.de>
31120 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31121 Skip unreachable blocks and destinations.
31122 (eliminate): Move stmt removal and fixup ...
31123 (fini_eliminate): ... here. Skip inserted exprs.
31124 (pass_pre::execute): Move fini_pre after fini_eliminate.
31125 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
31126 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
31127 PRE to get rid of dead code that has invalid SSA form and
31128 split critical edges again.
31130 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31132 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31134 2017-05-11 Richard Biener <rguenther@suse.de>
31136 * passes.c (execute_function_todo): Verify loops if they are
31137 said to be up-to-date.
31138 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
31139 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
31141 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
31144 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
31145 handle calling assemble_external ourself.
31148 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
31149 modes with zero size. Enhance comment.
31151 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31153 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
31154 built-ins for vec_xl and vec_xst with short and char pointer
31157 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31159 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
31160 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
31161 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
31162 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
31163 (_mm_maskz_min_round_ss): New intrinsics.
31164 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
31165 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
31166 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
31167 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
31168 (__builtin_ia32_minss_mask_round): New builtins.
31169 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31170 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31171 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
31173 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
31174 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
31176 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
31179 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31181 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
31182 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
31183 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
31184 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
31185 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
31186 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
31187 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
31188 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31189 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31190 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
31191 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
31192 (__builtin_ia32_mulss_mask_round): New builtins.
31193 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31194 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31195 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
31197 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
31198 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31200 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31203 2017-05-10 Julia Koval <julia.koval@intel.com>
31205 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
31206 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
31207 (_mm256_setr_m128i): New intrinsics.
31209 2017-05-10 Julia Koval <julia.koval@intel.com>
31211 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
31212 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
31213 (_mm_maskz_rcp14_ss): New intrinsics.
31214 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
31215 (__builtin_ia32_rcp14ss_mask): New builtins.
31216 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
31218 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
31220 PR tree-optimization/51513
31221 * tree-cfg.c (gimple_seq_unreachable_p): New function.
31222 (assert_unreachable_fallthru_edge_p): Use it.
31223 (group_case_labels_stmt): Likewise.
31224 * tree-cfg.h: Prototype it.
31225 * stmt.c: Include cfghooks.h and tree-cfg.h.
31226 (emit_case_dispatch_table) <gap_label>: New local variable.
31227 Use it to fill dispatch table gaps.
31228 Test for default_label before updating probabilities.
31229 (expand_case) <default_label>: Remove unneeded initialization.
31230 Test for unreachable default case statement and remove its edge.
31231 Set default_label accordingly.
31232 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
31234 2017-05-10 Carl Love <cel@us.ibm.com>
31236 * config/rs6000/rs6000-c: Add support for built-in functions
31237 vector signed char vec_neg (vector signed char)
31238 vector signed short int vec_neg (vector short int)
31239 vector signed int vec_neg (vector signed int)
31240 vector signed long long vec_neg (vector signed long long)
31241 vector float vec_neg (vector float)
31242 vector double vec_neg (vector double)
31243 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
31245 * config/rs6000/altivec.h: Add define for vec_neg
31246 * doc/extend.texi: Update the built-in documentation for the
31247 new built-in functions.
31249 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31251 PR tree-optimization/77644
31252 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
31254 2017-05-10 Nathan Sidwell <nathan@acm.org>
31256 * dumpfile.h (TDI_lang_all): New.
31257 (TDF_KIND): New. Renumber others
31258 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
31260 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
31262 (get_dump_file_name): Adjust suffix generation.
31263 (dump_enable_all): Use TDF_KIND.
31264 * doc/invoke.texi (-fdump-lang-all): Document.
31266 * dumpfile.h: Tabify.
31268 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
31271 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
31272 Move member access before delete.
31274 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
31276 * tree-inline.c (expand_call_inline): Split block at stmt
31279 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
31282 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
31283 are now unused after splitting mov{sf,sd}_hardfloat.
31284 (f32_lr2): Likewise.
31285 (f32_lm): Likewise.
31286 (f32_lm2): Likewise.
31287 (f32_li): Likewise.
31288 (f32_li2): Likewise.
31289 (f32_lv): Likewise.
31290 (f32_sr): Likewise.
31291 (f32_sr2): Likewise.
31292 (f32_sm): Likewise.
31293 (f32_sm2): Likewise.
31294 (f32_si): Likewise.
31295 (f32_si2): Likewise.
31296 (f32_sv): Likewise.
31297 (f32_dm): Likewise.
31298 (f32_vsx): Likewise.
31299 (f32_av): Likewise.
31300 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
31301 For movsf, order stores so the VSX stores occur before the GPR
31302 store which encourages the register allocator to use a traditional
31303 FPR instead of a GPR. For movsd, order the stores so that the GPR
31304 store comes before the VSX stores to allow the power6 to work.
31305 This is due to the power6 not having a 32-bit integer store
31306 instruction from a FPR.
31307 (movsf_hardfloat): Likewise.
31308 (movsd_hardfloat): Likewise.
31310 2017-05-09 Martin Sebor <msebor@redhat.com>
31312 PR translation/80280
31313 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
31316 PR translation/80280
31317 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
31318 data member added in r247778.
31319 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
31321 2017-05-09 Nathan Sidwell <nathan@acm.org>
31323 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
31325 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
31328 2017-05-09 Marek Polacek <polacek@redhat.com>
31330 * doc/invoke.texi: Fix typo.
31332 2017-05-09 Richard Biener <rguenther@suse.de>
31334 * tree-vrp.c (vrp_val_is_max): Adjust comment.
31335 (vrp_val_is_min): Likewise.
31336 (set_value_range_to_value): Likewise.
31337 (set_value_range_to_nonnegative): Likewise.
31338 (gimple_assign_nonzero_p): Likewise.
31339 (gimple_stmt_nonzero_p): Likewise.
31340 (vrp_int_const_binop): Likewise. Remove unreachable case.
31341 (adjust_range_with_scev): Adjust comments.
31342 (compare_range_with_value): Likewise.
31343 (extract_range_from_phi_node): Likewise.
31344 (test_for_singularity): Likewise.
31346 2017-05-09 Richard Biener <rguenther@suse.de>
31348 * tree-vrp.c (get_single_symbol): Add assert that we don't
31349 get overflowed constants as invariant part.
31350 (compare_values_warnv): Add comment before the TREE_NO_WARNING
31351 checks. Use wi::cmp instead of recursing for integer constants.
31352 (compare_values): Just ignore whether we assumed undefined
31353 overflow instead of failing the compare.
31354 (extract_range_for_var_from_comparison_expr): Add comment before the
31355 TREE_NO_WARNING sets.
31356 (test_for_singularity): Likewise.
31357 (extract_range_from_comparison): Do not disable optimization
31358 when we assumed undefined overflow.
31359 (extract_range_basic): Remove init of unused var.
31361 2017-05-09 Richard Biener <rguenther@suse.de>
31363 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
31364 (extract_range_from_multiplicative_op_1): Adjust.
31365 (extract_range_from_binary_expr_1): Use int_const_binop.
31367 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
31370 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
31371 rs6000_store_data_bypass_p in seven define_bypass directives and
31372 in several comments.
31373 * config/rs6000/rs6000-protos.h: Add prototype for
31374 rs6000_store_data_bypass_p function.
31375 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
31376 function implements slightly different (rs6000-specific) semantics
31377 than store_data_bypass_p, returning false rather than aborting
31378 with assertion error when arguments do not satisfy the
31379 requirements of store data bypass.
31380 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
31381 rs6000_store_data_bypass_p.
31383 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
31385 * config/xtensa/xtensa-protos.h
31386 (xtensa_initial_elimination_offset): New declaration.
31387 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
31388 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
31389 macro definition, add case for FRAME_POINTER_REGNUM when
31390 FRAME_GROWS_DOWNWARD.
31391 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
31392 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
31393 xtensa_initial_elimination_offset.
31395 2017-05-08 Nathan Sidwell <nathan@acm.org>
31397 * doc/invoke.texi: Alphabetize -fdump options.
31399 2017-05-08 Martin Sebor <msebor@redhat.com>
31401 PR translation/80280
31402 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
31404 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
31406 * target.def (compute_frame_layout): New optional target hook.
31407 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
31408 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
31409 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
31411 * reload1.c (verify_initial_elim_offsets): Likewise.
31412 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
31413 (use_simple_return_p): Call arm_compute_frame_layout if needed.
31414 (arm_get_frame_offsets): Split up into this ...
31415 (arm_compute_frame_layout): ... and this function.
31417 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
31419 * config/aarch64/constraints.md (Usa): New constraint.
31420 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
31422 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
31424 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
31425 with_multilib_list after it has been checked.
31427 2017-05-08 Richard Biener <rguenther@suse.de>
31429 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
31430 (bitmap_set_subtract_values): Likewise.
31432 2017-05-08 Richard Biener <rguenther@suse.de>
31434 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
31435 (gimple_assign_nonzero): ... this and remove strict_overflow_p
31437 (gimple_stmt_nonzero_warnv_p): Rename to ...
31438 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
31440 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
31441 (extract_range_basic): Adjust, do not disable propagation on
31442 strict overflow sensitive simplification.
31443 (vrp_visit_cond_stmt): Likewise.
31445 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
31447 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
31448 body size unconditionally.
31450 2017-05-07 Jeff Law <law@redhat.com>
31453 2017-05-06 Jeff Law <law@redhat.com>
31454 PR tree-optimization/78496
31455 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31458 PR tree-optimization/78496
31459 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31460 (simplify_stmt_using_ranges): Call it.
31461 (vrp_dom_walker::before_dom_children): Extract equivalences
31462 from an ASSERT_EXPR with an equality comparison against a
31465 2017-05-06 Jeff Law <law@redhat.com>
31467 PR tree-optimization/78496
31468 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31471 PR tree-optimization/78496
31472 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31473 (simplify_stmt_using_ranges): Call it.
31474 (vrp_dom_walker::before_dom_children): Extract equivalences
31475 from an ASSERT_EXPR with an equality comparison against a
31478 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31480 * lra-constraints.c (lra_copy_reg_equiv): New function.
31481 (split_reg): Use it to copy equivalence information from the
31482 original register to the spill register.
31484 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31486 PR rtl-optimization/75964
31487 * simplify-rtx.c (simplify_const_relational_operation): Remove
31488 invalid handling of comparisons of integer ABS.
31490 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
31492 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
31493 initialize to zero.
31494 (init_regs): Remove declaration.
31495 (function_arg_advance_32): Initialize error_p as boolean variable.
31497 2017-05-05 Nathan Sidwell <nathan@acm.org>
31499 * store-motion.c (remove_reachable_equiv_notes): Reformat long
31500 lines. Use for (;;).
31502 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31504 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
31505 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
31506 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
31507 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
31508 VF=2 that require versioning.
31510 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31512 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
31515 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31517 * diagnostic.h (diagnostic_override_option_index): Convert from
31518 macro to inline function.
31520 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31522 * diagnostic.c (last_module_changed_p): New function.
31523 (set_last_module): New function.
31524 (diagnostic_report_current_module): Convert macro usage to
31525 the above functions.
31526 * diagnostic.h (diagnostic_context::last_module): Strengthen
31527 from const line_map * to const line_map_ordinary *.
31528 (diagnostic_last_module_changed): Delete macro.
31529 (diagnostic_set_last_module): Delete macro.
31531 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31533 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
31534 with diagnostic_report_diagnostic.
31535 (diagnostic_n_impl_richloc): Likewise.
31536 * diagnostic.h (report_diagnostic): Delete macro.
31537 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
31538 with diagnostic_report_diagnostic.
31539 * substring-locations.c (format_warning_va): Likewise.
31541 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31543 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
31544 save/restor of format_spec. Move option-printing code to...
31545 (print_option_information): ...this new function, and
31546 reimplement by simply printing to the pretty_printer,
31547 rather than appending to the format string.
31549 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31551 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
31552 handling logic into...
31553 (update_effective_level_from_pragmas): ...this new function.
31555 2017-05-04 Andrew Waterman <andrew@sifive.com>
31557 * config/riscv/riscv.opt (mstrict-align): New option.
31558 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
31559 (SLOW_UNALIGNED_ACCESS): Define.
31560 (riscv_slow_unaligned_access): Declare.
31561 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
31563 (riscv_slow_unaligned_access): New variable.
31564 (rocket_tune_info): Set slow_unaligned_access to true.
31565 (optimize_size_tune_info): Set slow_unaligned_access to false.
31566 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
31567 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
31568 (riscv_option_override): Set riscv_slow_unaligned_access.
31569 * doc/invoke.texi: Add -mstrict-align to RISC-V.
31571 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
31573 * config/riscv/riscv.md: Unify indentation.
31575 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
31580 * config/rs6000/rs6000.md (u code attribute): Add FIX and
31582 (extendsi<mode>2): Add support for doing sign extension via
31583 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
31584 don't have ISA 3.0 instructions.
31585 (extendsi<mode>2 splitter): Likewise.
31586 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
31587 generate the normal insns since SImode can now go in vector
31588 registers. Disallow the special UNSPECs needed for previous
31589 machines to hide SImode being used. Add new insns
31590 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
31591 (fix_trunc<mode>si2_stfiwx): Likewise.
31592 (fix_trunc<mode>si2_internal): Likewise.
31593 (fixuns_trunc<mode>si2): Likewise.
31594 (fixuns_trunc<mode>si2_stfiwx): Likewise.
31595 (fctiw<u>z_<mode>_smallint): Likewise.
31596 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
31597 of floating point to 32-bit integer from doing a direct move to
31598 the GPR registers to do a store.
31599 (fctiwz_<mode>): Break long line.
31601 2017-05-05 Bin Cheng <bin.cheng@arm.com>
31603 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
31604 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
31605 (addr_list, addr_offset_valid_p): New.
31606 (split_address_groups): Check offset validity with above function.
31607 (gt-tree-ssa-loop-ivopts.h): Include header file.
31609 2017-05-05 Nathan Sidwell <nathan@acm.org>
31611 * config.gcc (arm*-*-*): Add missing 'fi'.
31613 2017-05-05 Steve Ellcey <sellcey@cavium.com>
31615 * doc/invoke.texi (-fopt-info): Explicitly say order of options
31616 included in -fopt-info does not matter.
31617 * doc/optinfo.texi (-fopt-info): Fix description of default
31618 behavour. Explicitly say order of options included in -fopt-info
31621 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
31623 * config.gcc: Allow combinations of aprofile and rmprofile values for
31624 --with-multilib-list.
31625 * config/arm/t-multilib: New file.
31626 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
31627 variables. Remove setting of ISA and floating-point ABI in
31628 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
31629 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
31630 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
31631 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
31633 * config/arm/t-rmprofile: Likewise except for the matches changes.
31634 * doc/install.texi (--with-multilib-list): Document the combination of
31635 aprofile and rmprofile values and warn about pitfalls in doing that.
31637 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31639 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
31640 (movdi_aarch64): Likewise.
31642 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31644 PR tree-optimization/80632
31645 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
31647 (build_arrays): Initialize it for virtual phis.
31648 (fix_phi_nodes): Use it for virtual phis.
31650 PR tree-optimization/80558
31651 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
31652 [x, y] op z into [x op, y op z] for op & or | if conditions
31655 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
31656 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31659 * config/arm/arm.md (use_literal_pool): Remove.
31660 (64-bit immediate split): No longer takes cost into consideration
31661 if arm_disable_literal_pool is enabled.
31662 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
31663 used when arm_disable_literal_pool is enabled.
31664 (arm_max_const_double_inline_cost): Remove use of
31665 arm_disable_literal_pool.
31666 (push_minipool_fix): Add assert.
31667 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
31668 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
31669 (no_literal_pool_sf_immediate): New.
31671 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31673 PR tree-optimization/80613
31674 * tree-ssa-dce.c (propagate_necessity): Remove cases for
31675 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
31677 2017-05-05 Richard Biener <rguenther@suse.de>
31679 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
31681 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
31683 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
31684 of this flag from insn conditions due to removal from r247495.
31686 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31688 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
31690 (arm_early_store_addr_dep_ptr): Likewise.
31691 * config/arm/aarch-common-protos.h
31692 (arm_early_load_addr_dep_ptr): Add prototype.
31693 (arm_early_store_addr_dep_ptr): Likewise.
31694 * config/arm/cortex-a53.md: Add new bypasses.
31696 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31698 * tree.c (next_type_uid): Change type to unsigned.
31699 (type_hash_canon): Decrement back next_type_uid if
31700 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
31701 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
31704 2017-05-04 Martin Sebor <msebor@redhat.com>
31706 * builtins.c: Fix a trivial typo in a comment.
31708 PR middle-end/79234
31709 * builtins.c (check_sizes): Adjust to handle reading past the end.
31710 Avoid printing excessive upper bound of ranges. Use %E to print
31711 tree nodes instead of converting them to %wu.
31712 (expand_builtin_memchr): New function.
31713 (compute_dest_size): Rename...
31714 (compute_objsize): ...to this.
31715 (expand_builtin_memcpy): Adjust.
31716 (expand_builtin_mempcpy): Adjust.
31717 (expand_builtin_strcat): Adjust.
31718 (expand_builtin_strcpy): Adjust.
31719 (check_strncat_sizes): Adjust.
31720 (expand_builtin_strncat): Adjust.
31721 (expand_builtin_strncpy): Adjust and simplify.
31722 (expand_builtin_memset): Adjust.
31723 (expand_builtin_bzero): Adjust.
31724 (expand_builtin_memcmp): Adjust.
31725 (expand_builtin): Handle memcmp.
31726 (maybe_emit_chk_warning): Check strncat just once.
31728 2017-05-04 Martin Sebor <msebor@redhat.com>
31730 PR preprocessor/79214
31731 PR middle-end/79222
31732 PR middle-end/79223
31733 * builtins.c (check_sizes): Add inlining context and issue
31734 warnings even when -Wno-system-headers is set.
31735 (check_strncat_sizes): Same.
31736 (expand_builtin_strncat): Same.
31737 (expand_builtin_memmove): New function.
31738 (expand_builtin_stpncpy): Same.
31739 (expand_builtin): Handle memmove and stpncpy.
31741 2017-05-04 Bin Cheng <bin.cheng@arm.com>
31743 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
31744 which is not used any more.
31746 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31748 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
31750 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31752 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
31753 (cortexa53_tunings): Likewise.
31754 (cortexa57_tunings): Likewise.
31755 (cortexa72_tunings): Likewise.
31756 (cortexa73_tunings): Likewise.
31758 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31760 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
31761 Set loop alignment to 8.
31763 2017-05-04 Martin Sebor <msebor@redhat.com>
31765 PR translation/80280
31766 * builtins.c (expand_builtin_object_size): Add missing quoting to
31767 %D and like directives.
31768 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
31769 (hsa_type_for_tree_type): Same.
31770 (verify_function_arguments): Same.
31771 * symtab.c (symbol_table::change_decl_assembler_name): Same.
31772 * varasm.c (get_section): Same.
31775 2017-05-04 Martin Sebor <msebor@redhat.com>
31777 PR translation/80280
31778 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
31780 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31782 * config/aarch64/aarch64.c (generic_addrcost_table):
31783 Change HI/TI mode setting.
31785 2017-05-04 Martin Jambor <mjambor@suse.cz>
31787 PR tree-optimization/80622
31788 * tree-sra.c (comes_initialized_p): New function.
31789 (build_accesses_from_assign): Only set write lazily when
31790 comes_initialized_p is false.
31791 (analyze_access_subtree): Use comes_initialized_p.
31792 (propagate_subaccesses_across_link): Assert !comes_initialized_p
31793 instead of testing for PARM_DECL.
31795 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31797 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
31798 constraint on operand 0 to allow more general addressing modes.
31799 Adjust output template.
31800 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
31802 * config/aarch64/aarch64-protos.h
31803 (aarch64_address_valid_for_prefetch_p): Declare prototype.
31804 * config/aarch64/constraints.md (Dp): New address constraint.
31805 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
31808 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
31810 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
31811 update use of estimate_ipcp_clone_size_and_time.
31812 (estimate_local_effects): Update use of
31813 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
31814 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
31815 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
31816 Return nonspecialized time.
31818 2017-05-04 Richard Biener <rguenther@suse.de>
31820 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
31821 for the last VUSE which def dominates the PHI. Directly call
31823 (get_continuation_for_phi_1): Remove.
31825 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
31827 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
31828 to explain the use of truncating division. Cap the number of
31829 iterations to the maximum given by nb_iterations_upper_bound,
31832 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31834 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
31835 * configure: Regenerate.
31836 * config.in: Regenerate.
31837 * config/i386/driver-mingw32.c: new file.
31838 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
31839 * config.host: Link driver-mingw32.o on MinGW host.
31840 * doc/install.texi: Document new --enable-mingw-wildcard configure
31843 2017-05-04 Marek Polacek <polacek@redhat.com>
31845 PR tree-optimization/80612
31846 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
31848 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31849 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
31851 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
31852 (movt splitter): Likewise.
31853 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
31854 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
31855 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
31856 block for Thumb-1 with MOVT.
31857 (thumb2_legitimate_address_p): Move code block ...
31858 (can_avoid_literal_pool_for_label_p): ... into this new function.
31859 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
31861 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
31862 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
31863 "M-profile targets with the MOVT instruction".
31865 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31867 * config/arm/arm-builtins.c (arm_init_builtins): Rename
31868 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
31869 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
31871 2017-05-04 Martin Liska <mliska@suse.cz>
31873 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
31874 variable cond_code.
31876 2017-05-04 Richard Biener <rguenther@suse.de>
31878 * tree.c (array_at_struct_end_p): Handle arrays at struct
31879 end with flexarrays more conservatively. Refactor and treat
31880 arrays of arrays or aggregates more strict. Fix
31881 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
31882 * tree.c (array_at_struct_end_p): Adjust prototype.
31883 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
31884 * gimple-fold.c (get_range_strlen): Likewise.
31885 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
31887 2017-05-04 Richard Biener <rguenther@suse.de>
31889 PR tree-optimization/31130
31890 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
31892 (supports_overflow_infinity): Likewise.
31893 (is_negative_overflow_infinity): Likewise.
31894 (is_positive_overflow_infinity): Likewise.
31895 (is_overflow_infinity): Likewise.
31896 (stmt_overflow_infinity): Likewise.
31897 (overflow_infinity_range_p): Likewise.
31898 (usable_range_p): Remove as always returning true.
31899 (make_overflow_infinity): Remove.
31900 (negative_overflow_infinity): Likewise.
31901 (positive_overflow_infinity): Likewise.
31902 (avoid_overflow_infinity): Likewise.
31903 (set_value_range): Adjust accordingly.
31904 (set_value_range_to_nonnegative): Likewise, remove now unused
31905 overflow_infinity arg.
31906 (vrp_operand_equal_p): Adjust.
31907 (update_value_range): Likewise.
31908 (range_int_cst_singleton_p): Likewise.
31909 (operand_less_p): Likewise.
31910 (compare_values_warnv): Likewise.
31911 (extract_range_for_var_from_comparison_expr): Likewise.
31912 (vrp_int_const_binop): Likewise.
31913 (zero_nonzero_bits_from_vr): Likewise.
31914 (extract_range_from_multiplicative_op_1): Likewise.
31915 (extract_range_from_binary_expr_1): Likewise.
31916 (extract_range_from_unary_expr): Likewise.
31917 (extract_range_from_comparison): Likewise.
31918 (extract_range_basic): Likewise.
31919 (adjust_range_with_scev): Likewise.
31920 (compare_ranges): Likewise.
31921 (compare_range_with_value): Likewise.
31922 (dump_value_range): Likewise.
31923 (test_for_singularity): Likewise, remove strict_overflow_p parameter
31925 (simplify_cond_using_ranges): Adjust.
31927 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31929 * brig-builtins.def: Added a builtin for class_f64.
31930 * builtin-types.def: Added a builtin type needed by class_f64.
31932 2017-05-03 Jason Merrill <jason@redhat.com>
31934 * timevar.def: Add TV_CONSTEXPR.
31936 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31938 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
31940 2017-05-03 Martin Jambor <mjambor@suse.cz>
31942 * ipa-prop.c (ipa_update_after_lto_read): Removed.
31943 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
31944 * ipa-cp.c (ipcp_propagate_stage): Do not call
31945 ipa_update_after_lto_read.
31946 * ipa-inline.c (ipa_inline): Likewise.
31948 2017-05-03 Martin Jambor <mjambor@suse.cz>
31950 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
31951 tag. Added a default constructor and a destructor.
31952 (ipa_edge_args_sum_t): New class;
31953 (ipa_edge_args_sum): Declare.
31954 (ipa_edge_args_vector): Remove declaration.
31955 (IPA_EDGE_REF): Use ipa_edge_args_sum.
31956 (ipa_free_edge_args_substructures): Remove declaration.
31957 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
31958 (ipa_edge_args_info_available_for_edge_p): Likewise.
31959 * ipa-prop.c (ipa_edge_args_vector): Removed.
31960 (edge_removal_hook_holder): Likewise.
31961 (edge_duplication_hook_holder): Likewise.
31962 (ipa_edge_args_sum): New variable.
31963 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
31964 ipa_edge_args_vector.
31965 (ipa_free_edge_args_substructures): Likewise.
31966 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
31967 ipa_edge_args_vector.
31968 (ipa_edge_removal_hook): Turned into method
31969 ipa_edge_args_sum_t::remove.
31970 (ipa_edge_duplication_hook): Turned into method
31971 ipa_edge_args_sum_t::duplicate.
31972 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
31973 registering edge hooks.
31974 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
31975 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
31976 ipa_edge_args_sum instead of ipa_edge_args_vector.
31977 * ipa-profile.c (ipa_profile): Likewise.
31979 2017-05-03 Martin Jambor <mjambor@suse.cz>
31981 * symbol-summary.h (function_summary): New method exists.
31982 (function_summary::symtab_removal): Deallocate through release.
31983 (call_summary): New class.
31984 (gt_ggc_mx): New overload.
31985 (gt_pch_nx): Likewise.
31986 (gt_pch_nx): Likewise.
31988 2017-05-03 Jeff Law <law@redhat.com>
31990 PR tree-optimization/78496
31991 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
31992 from simplify_cond_using_ranges. Split off code to walk
31993 backwards through casts into ...
31994 (simplify_cond_using_ranges_2): New function.
31995 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
31996 (execute_vrp): After identifying jump threads, call
31997 simplify_cond_using_ranges_2.
31999 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
32002 * ipa-inline.h (inline_summary): Add ctor.
32003 (create_ggc): Do not use ggc_cleared_alloc.
32005 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
32006 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32008 * gcc.c (handle_braces): Support escaping in switch matching
32010 * doc/invoke.texi (Spec Files): Document it.
32011 Remove superfluous @code markup in items.
32013 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32015 * diagnostic-show-locus.c (struct column_range): New struct.
32016 (get_affected_columns): New function.
32017 (get_printed_columns): New function.
32018 (struct correction): New struct.
32019 (correction::ensure_capacity): New function.
32020 (correction::ensure_terminated): New function.
32021 (struct line_corrections): New struct.
32022 (line_corrections::~line_corrections): New dtor.
32023 (line_corrections::add_hint): New function.
32024 (layout::print_trailing_fixits): Reimplement in terms of the new
32026 (selftest::test_overlapped_fixit_printing): New function.
32027 (selftest::diagnostic_show_locus_c_tests): Call it.
32029 2017-05-03 Nathan Sidwell <nathan@acm.org>
32031 Canonicalize canonical type hashing
32032 * tree.h (type_hash_canon_hash): Declare.
32033 * tree.c (type_hash_list, attribute_hash_list): Move into
32034 type_hash_canon_hash.
32035 (build_type_attribute_qual_variant): Break out hash code calc into
32036 type_hash_canon_hash.
32037 (type_hash_canon_hash): New. Generic type hash computation.
32038 (build_range_type_1, build_array_type_1, build_function_type,
32039 build_method_type_directly, build_offset_type, build_complex_type,
32040 make_vector_type): Call it.
32042 2017-05-03 Richard Biener <rguenther@suse.de>
32044 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
32045 When all DRs have unknown misaligned do not always peel
32046 when there is a store but apply the same costing model as if
32047 there were only loads.
32049 2017-05-03 Richard Biener <rguenther@suse.de>
32052 PR tree-optimization/80492
32053 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32054 compare_base_decls returning dont-know properly.
32056 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
32058 * config/arm/iterators.md (CCSI): New mode iterator.
32059 (arch): New mode attribute.
32060 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
32061 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
32062 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
32063 code iterator for success result mode.
32064 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
32065 the corresponding new insn generators.
32067 2017-05-03 Bin Cheng <bin.cheng@arm.com>
32070 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32071 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32073 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
32075 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
32076 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
32077 (DDR_A): Wrap DDR argument in brackets.
32078 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
32079 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
32080 (DDR_REVERSED_P): Likewise.
32082 2017-05-03 Jakub Jelinek <jakub@redhat.com>
32084 PR tree-optimization/79472
32085 * tree-switch-conversion.c (struct switch_conv_info): Add
32086 contiguous_range and default_case_nonstandard fields.
32087 (collect_switch_conv_info): Compute contiguous_range and
32088 default_case_nonstandard fields, don't clear final_bb if
32089 contiguous_range and only the default case doesn't have the required
32091 (check_all_empty_except_final): Set default_case_nonstandard instead
32092 of failing if contiguous_range and the default case doesn't have empty
32094 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
32095 and only the default case doesn't have the required constants. Skip
32097 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
32098 if default_case_nonstandard.
32099 (build_constructors): Build constant 1 just once. Assert that default
32100 values aren't inserted in between cases if contiguous_range. Skip
32102 (build_arrays): Skip virtual phis.
32103 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
32104 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
32105 Handle virtual phis.
32106 (gen_inbound_check): Handle default_case_nonstandard case.
32107 (process_switch): Adjust check_final_bb caller. Call
32108 gather_default_values with the first non-default case instead of
32109 default case if default_case_nonstandard.
32111 2017-05-02 Nathan Sidwell <nathan@acm.org>
32113 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
32114 check. Fix formatting.
32116 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
32118 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
32119 errors when comparing specialized and unspecialized times.
32121 2017-05-02 David Malcolm <dmalcolm@redhat.com>
32123 * diagnostic-show-locus.c
32124 (layout::should_print_annotation_line_p): Make private.
32125 (layout::print_annotation_line): Make private.
32126 (layout::annotation_line_showed_range_p): Make private.
32127 (layout::show_ruler): Make private.
32128 (layout::print_source_line): Make private. Pass in line and
32129 line_width, rather than calling location_get_source_line. Drop
32131 (layout::print_leading_fixits): New method.
32132 (layout::print_any_fixits): Rename to...
32133 (layout::print_trailing_fixits): ...this, and make private.
32134 Don't print newline fixits.
32135 (diagnostic_show_locus): Move logic for printing one row into...
32136 (layout::print_line): ...this new function. Move the
32137 location_get_source_line call and error-handling from
32138 print_source_line to here. Call print_leading_fixits, and rename
32139 print_any_fixits to print_trailing_fixits.
32140 (selftest::test_fixit_insert_containing_newline): Update now that
32141 newlines are partially supported.
32142 (selftest::test_fixit_insert_containing_newline_2): New test.
32143 (selftest::test_fixit_replace_containing_newline): Update comments.
32144 (selftest::diagnostic_show_locus_c_tests): Call the new test.
32145 * edit-context.c (class added_line): New class.
32146 (class edited_line): Describe newline handling in comment.
32147 (edited_line::actually_edited_p): New method.
32148 (edited_line::print_content): Delete redundant decl.
32149 (edited_line::m_predecessors): New field.
32150 (edited_file::print_content): Call edited_line::print_content.
32151 (edited_file::print_diff): Update to support newlines.
32152 (edited_file::print_diff_hunk): Likewise.
32153 (edited_file::print_run_of_changed_lines): New function.
32154 (edited_file::print_diff_line): Convert to...
32155 (print_diff_line): ...this.
32156 (edited_file::get_effective_line_count): New function.
32157 (edited_line::edited_line): Initialize new field m_predecessors.
32158 (edited_line::~edited_line): Clean up m_predecessors.
32159 (edited_line::apply_fixit): Handle newlines.
32160 (edited_line::get_effective_line_count): New function.
32161 (edited_line::print_content): New function.
32162 (edited_line::print_diff_lines): New function.
32163 (selftest::test_applying_fixits_insert_containing_newline): New
32165 (selftest::test_applying_fixits_replace_containing_newline): New
32167 (selftest::insert_line): New function.
32168 (selftest::test_applying_fixits_multiple_lines): Add example of
32170 (selftest::edit_context_c_tests): Call the new tests.
32172 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32174 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
32175 parameter cand. Update dump information.
32176 (get_computation_cost): Update uses.
32178 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32180 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
32181 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
32182 (get_computation_at, rewrite_use_address): Update use of
32183 get_computation_aff.
32185 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32187 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
32188 (get_computation): Delete.
32189 (get_computation_cost): Implement like get_computation_cost_at.
32190 Use get_computation_at.
32191 (get_computation_cost_at): Delete.
32192 (rewrite_use_nonlinear_expr): Use get_computation_at.
32193 (rewrite_use_compare, remove_unused_ivs): Ditto.
32195 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32197 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
32199 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32201 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
32202 (ivopts_global_cost_for_size): Rename parameter and update uses.
32203 (iv_ca_recount_cost): Update uses.
32204 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
32205 candidates seperately in n_invs and n_cands.
32206 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
32208 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32210 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
32211 (find_inv_vars_cb): New.
32212 (find_depends): Renamed to ...
32213 (find_inv_vars): ... this.
32214 (add_candidate_1, force_var_cost): Call find_inv_vars.
32215 (split_address_cost, determine_group_iv_cost_cond): Ditto.
32217 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32219 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
32220 inv_vars. Add inv_exprs.
32221 (struct iv_cand): Rename depends_on to inv_vars.
32222 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
32223 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
32224 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
32225 (dump_cand): Dump inv_vars.
32226 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
32227 (record_invariant, find_depends, add_candidate_1): Ditto.
32228 (set_group_iv_cost, force_var_cost): Ditto.
32229 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
32230 (get_computation_cost_at, get_computation_cost): Ditto.
32231 (determine_group_iv_cost_generic): Ditto.
32232 (determine_group_iv_cost_address): Ditto.
32233 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
32234 (determine_group_iv_costs): Ditto.
32235 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
32236 (iv_ca_set_remove_invariants): Renamed to ...
32237 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
32238 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
32239 (iv_ca_set_add_invariants): Renamed to ...
32240 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
32241 (iv_ca_set_cp): Use iv_ca_set_add_invs.
32242 (iv_ca_has_deps): Support inv_vars and inv_exprs.
32243 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
32244 (create_new_ivs): Remove useless dump.
32246 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32248 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
32250 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
32251 (iv_ca_set_no_cp, create_new_iv): Ditto.
32253 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32255 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32257 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32259 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
32260 function tree_check2.
32262 2017-05-02 Martin Liska <mliska@suse.cz>
32264 * doc/gcov.texi: Add missing preposition.
32265 * gcov.c (function_info::function_info): Properly fill up
32266 all member variables.
32268 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32270 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
32272 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32274 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
32276 2017-05-02 Martin Liska <mliska@suse.cz>
32279 * lto-streamer-in.c (lto_read_tree_1): Remove
32280 LTO_STREAMER_DEBUG.
32281 * lto-streamer.c (struct tree_hash_entry): Likewise.
32282 (struct tree_entry_hasher): Likewise.
32283 (tree_entry_hasher::hash): Likewise.
32284 (tree_entry_hasher::equal): Likewise.
32285 (lto_streamer_init): Likewise.
32286 (lto_orig_address_map): Likewise.
32287 (lto_orig_address_get): Likewise.
32288 (lto_orig_address_remove): Likewise.
32289 * lto-streamer.h: Likewise.
32290 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
32291 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
32293 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
32295 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
32296 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
32297 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
32298 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
32299 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
32300 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
32301 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
32302 (mm_maskz_sub_ss): New intrinsics.
32303 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32304 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32305 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
32306 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
32307 (__builtin_ia32_subss_mask_round): New builtins.
32308 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32309 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32310 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
32312 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
32313 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32315 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32318 2017-05-02 Martin Jambor <mjambor@suse.cz>
32320 PR tree-optimization/78687
32321 * tree-sra.c (access): New field parent.
32322 (process_subtree_disqualification): New function.
32323 (disqualify_candidate): Call it.
32324 (build_accesses_from_assign): Reset write flag if creating an
32326 (build_access_subtree): Fill in parent field and also prpagate
32327 down grp_write flag.
32328 (create_artificial_child_access): New parameter set_grp_write, set
32329 grp_write to its value.
32330 (propagate_subaccesses_across_link): Also propagate grp_write flag
32332 (propagate_all_subaccesses): Push the closest parent back to work
32333 queue if add_access_to_work_queue returned true.
32335 2017-05-02 Richard Biener <rguenther@suse.de>
32337 * common.opt (fstrict-overflow): Alias negative to fwrapv.
32338 * doc/invoke.texi (fstrict-overflow): Remove all traces of
32339 -fstrict-overflow documentation.
32340 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
32341 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
32342 flag_strict_overflow.
32343 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
32344 * lto-opts.c (lto_write_options): Do not stream it.
32345 * lto-wrapper.c (merge_and_complain): Do not handle it.
32346 * opts.c (default_options_table): Do not set -fstrict-overflow.
32347 (finish_options): Likewise do not clear it when sanitizing.
32348 * simplify-rtx.c (simplify_const_relational_operation): Do not
32349 test flag_strict_overflow.
32351 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32353 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
32354 using enabled attribute.
32355 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
32356 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
32357 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
32358 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
32359 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
32360 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
32361 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
32362 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
32363 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
32364 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
32366 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32368 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
32370 2017-05-02 Richard Biener <rguenther@suse.de>
32372 PR tree-optimization/80591
32374 2017-04-10 Richard Biener <rguenther@suse.de>
32376 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32379 2017-05-02 Richard Biener <rguenther@suse.de>
32381 PR tree-optimization/80549
32382 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
32383 (cleanup_tree_cfg_noloop): Create forwarders to known loop
32384 headers if they do not have a preheader.
32386 2017-05-02 Martin Liska <mliska@suse.cz>
32389 * common.opt: Fix typo.
32390 * doc/invoke.texi: Likewise.
32392 2017-05-01 Jan Beulich <jbeulich@suse.com>
32394 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
32395 swapping, add (x,x,m,x,n) alternative.
32397 2017-05-01 Nathan Sidwell <nathan@acm.org>
32399 * calls.c (combine_pending_stack_adjustment_and_call): Remove
32400 unnecessary unadjusted_alignment check.
32402 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32405 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
32407 * gimplify.c (gimplify_cilk_detach): New function.
32408 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
32409 * tree-core.h: Document EXPR_CILK_SPAWN.
32410 * tree.h (EXPR_CILK_SPAWN): Define.
32412 2017-05-01 David Malcolm <dmalcolm@redhat.com>
32414 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
32415 to use new fixit_hint representation, using the "replace" logic.
32416 (get_line_span_for_fixit_hint): Likewise.
32417 (layout::print_any_fixits): Likewise.
32418 (selftest::test_one_liner_many_fixits): Rename to...
32419 (selftest::test_one_liner_many_fixits_1): ...this, and update
32420 comment and expected output to reflect that the multiple fix-it
32421 hints are now consolidated into one insertion.
32422 (selftest::test_one_liner_many_fixits_2): New test.
32423 (selftest::test_diagnostic_show_locus_one_liner): Update for
32425 (selftest::test_fixit_consolidation): Update for fix-it API
32427 * diagnostic.c (print_parseable_fixits): Likewise.
32428 * edit-context.c (edited_line::m_line_events): Convert from
32429 auto_vec <line_event *> to auto_vec <line_event>.
32430 (class line_event): Convert from abstract base class to a concrete
32431 class, taking over the role of replace_event.
32432 (class insert_event): Delete.
32433 (class replace_event): Rename to class line_event. Convert to
32435 (edit_context::add_fixits): Reimplement.
32436 (edit_context::apply_insert): Delete.
32437 (edit_context::apply_replace): Rename to...
32438 (edit_context::apply_fixit): ...this. Convert to half-open range.
32439 (edited_file::apply_insert): Delete.
32440 (edited_file::apply_replace): Rename to...
32441 (edited_file::apply_fixit): ...this.
32442 (edited_line::~edited_line): Drop deletion of events.
32443 (edited_line::apply_insert): Delete.
32444 (edited_line::apply_replace): Rename to...
32445 (edited_line::apply_fixit): ...this. Convert to half-open range.
32446 Update for change to type of m_line_events.
32447 * edit-context.h (edit_context::apply_insert): Delete.
32448 (edit_context::apply_replace): Rename to...
32449 (edit_context::apply_fixit): ...this.
32451 2017-05-01 Martin Sebor <msebor@redhat.com>
32453 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
32456 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
32459 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
32460 __get_cpuid_max returns 0.
32461 (__get_cpuid_count): Ditto.
32463 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
32465 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
32466 replacement expression is another instance of one of its arguments.
32468 2017-05-01 Jakub Jelinek <jakub@redhat.com>
32471 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
32472 check for stack push/pop autoinc.
32473 * config/i386/i386.c (ix86_agi_dependent): Return false
32474 if the only reason why modified_in_p returned true is that
32475 addr is SP based and set_insn is a push or pop.
32477 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32479 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
32482 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32485 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
32486 (account_size_time): Use two predicates - exec_pred and
32488 (evaluate_conditions_for_known_args): Compute both clause and
32490 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
32491 (inline_summary_t::duplicate): Update.
32492 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
32494 (compute_inline_parameters): Likewise.
32495 (estimate_edge_size_and_time): Update caluclation of time.
32496 (estimate_node_size_and_time): Compute both time and nonspecialized
32498 (estimate_ipcp_clone_size_and_time): Update.
32499 (inline_merge_summary): Update.
32500 (do_estimate_edge_time): Update.
32501 (do_estimate_edge_size): Update.
32502 (do_estimate_edge_hints): Update.
32503 (inline_read_section, inline_write_summary): Stream both new predicates.
32504 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
32506 (compute_inlined_call_time): Cleanup.
32507 (big_speedup_p): Update.
32508 (edge_badness): Update.
32509 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
32510 (size_time_entry): Replace predicate by exec_predicate and
32511 nonconst_predicate.
32512 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
32513 (estimate_edge_time): Return also nonspec_time.
32514 (reset_edge_growth_cache): Update.
32516 2017-04-29 Jakub Jelinek <jakub@redhat.com>
32518 PR rtl-optimization/80491
32519 * ifcvt.c (noce_process_if_block): When looking for x setter
32520 with missing else_bb, don't check only the insn right before
32521 cond_earliest, but look for the last insn that x is modified in
32522 within the same bb.
32524 PR rtl-optimization/80491
32525 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
32527 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
32529 PR tree-optimization/80487
32530 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
32532 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32534 PR tree-optimization/79697
32535 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
32536 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
32537 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
32539 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
32540 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
32542 2017-04-28 Martin Sebor <msebor@redhat.com>
32544 PR tree-optimization/80523
32545 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
32546 (init_target_to_host_charmap, target_to_host, target_strtol10): New
32548 (maybe_warn, format_directive, parse_directive): Use new functions.
32549 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
32551 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
32553 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
32555 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
32557 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
32558 target_header_dir): Set correctly.
32559 * configure: Regenerated.
32560 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32561 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32562 instead of SYSTEM_HEADER_DIR.
32564 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
32566 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
32567 (estimate_local_effects): Likewise.
32568 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
32569 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
32570 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
32571 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
32572 do_estimate_edge_time, estimate_edge_time): Likewise.
32573 * ipa-inline-analysis.c (estimate_node_size_and_time,
32574 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
32575 (estimate_time_after_inlining): Remove.
32577 2017-04-28 Martin Liska <mliska@suse.cz>
32579 * doc/gcov.texi: Enhance documentation of gcov.
32581 2017-04-28 Martin Liska <mliska@suse.cz>
32583 * doc/gcov.texi: Sort options in alphabetic order.
32584 * doc/gcov-dump.texi: Likewise.
32585 * doc/gcov-tool.texi: Likewise.
32586 * gcov.c (print_usage): Likewise.
32587 * gcov-dump.c (print_usage): Likewise.
32588 * gcov-tool.c (print_merge_usage_message): Likewise.
32589 (print_rewrite_usage_message): Likewise.
32590 (print_overlap_usage_message): Likewise.
32592 2017-04-28 Martin Liska <mliska@suse.cz>
32594 PR gcov-profile/53915
32595 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
32597 2017-04-28 Martin Liska <mliska@suse.cz>
32599 PR gcov-profile/79891
32600 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
32601 is marked by compiler as living on a line.
32602 (get_cycles_count): Remove usage of the union.
32603 (output_intermediate_file): Likewise.
32604 (find_source): Fix GNU coding style.
32605 (accumulate_line_counts): Remove old non-all block mode.
32606 (output_lines): Remove usage of the union.
32607 * profile.c (output_location): Include all BBs, even if
32608 belonging to a same line (and file) as a previous BB.
32610 2017-04-28 Martin Liska <mliska@suse.cz>
32612 * gcov.c (process_args): Handle new argument 'w'.
32613 (read_graph_file): Assign ID to BBs.
32614 (output_branch_count): Display BB # if verbose flag is set.
32615 (output_lines): Likewise for arcs.
32616 (print_usage): Add '--verbose' option help.
32617 * doc/gcov.texi: Document --verbose (-w) option.
32619 2017-04-28 Martin Liska <mliska@suse.cz>
32621 * gcov.c (struct block_location_info): New struct.
32622 (process_file): Fill up the new structure.
32623 (read_graph_file): Replace usage of encoding by the newly added
32625 (add_line_counts): Likewise.
32626 (accumulate_line_counts): Remove usage of the union.
32627 (function_info::function_info): New function.
32628 (function_info::~function_info): Likewise.
32629 (process_file): Call delete instead of release_function.
32630 (release_function): Release the function.
32631 (release_structures): Call delete instead of release_function.
32632 (solve_flow_graph): Replace usage of num_blocks.
32633 (find_exception_blocks): Likewise.
32634 (output_lines): Fix GNU coding style.
32636 2017-04-28 Martin Liska <mliska@suse.cz>
32639 * coverage.c (coverage_remove_note_file): New function.
32640 * coverage.h: Declare the function.
32641 * toplev.c (finalize): Clean if an error has been seen.
32643 2017-04-28 Martin Liska <mliska@suse.cz>
32645 PR gcov-profile/80031
32646 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
32647 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
32648 * gcov.c (read_graph_file): Read just number of blocks.
32649 * profile.c (branch_prob): Do not stream 0 flags per a basic
32652 2017-04-28 Martin Liska <mliska@suse.cz>
32654 * gcov-dump.c (tag_*): Add new argument to declarations.
32655 (dump_gcov_file): Likewise.
32656 (tag_blocks): Add and use new argument depth.
32657 (tag_arcs): Likewise.
32658 (tag_lines): Likewise.
32659 (tag_counters): Likewise.
32660 (tag_summary): Likewise.
32661 (dump_working_sets): Use depth to do a proper indentation.
32663 2017-04-28 Jakub Jelinek <jakub@redhat.com>
32666 * cgraph.h (symtab_node::debug_symtab): No longer inline.
32667 * symtab.c (symtab_node::debug_symtab): Move definition here.
32669 2017-04-28 Richard Biener <rguenther@suse.de>
32671 * lto-streamer.h (LTO_major_version): Bump to 7.
32673 2017-04-28 Richard Biener <rguenther@suse.de>
32675 * tree-vrp.c (assert_info): New struct.
32676 (add_assert_info): New helper.
32677 (register_edge_assert_for_2): Refactor to add asserts to a vector
32679 (register_edge_assert_for_1): Likewise.
32680 (register_edge_assert_for): Likewise.
32681 (finish_register_edge_assert_for): New helper actually registering
32682 asserts where live on edge.
32683 (find_conditional_asserts): Adjust.
32684 (find_switch_asserts): Likewise.
32685 (evrp_dom_walker::try_find_new_range): Generalize.
32686 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
32688 2017-04-27 Marek Polacek <polacek@redhat.com>
32691 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
32692 arg10 and arg11 to itype.
32694 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
32696 * doc/extend.texi (Object Size Checking): Improve grammar.
32698 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
32701 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
32702 that the logic for permitting reciprocal estimates matches that
32705 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32708 * tree.c (type_cache_hasher::equal): Only compare
32709 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
32710 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
32711 non-aggregate element types.
32712 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
32713 about the flag on ARRAY_TYPEs in the comment, formatting fix.
32715 2017-04-27 Richard Biener <rguenther@suse.de>
32717 PR middle-end/80533
32718 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
32719 stripping ARRAY_REFs from MEM_EXPR make sure we're not
32720 keeping a reference to a trailing array.
32722 2017-04-27 Richard Biener <rguenther@suse.de>
32724 PR middle-end/80539
32725 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
32726 being in loop-closed SSA form conservatively.
32727 (chrec_fold_multiply_poly_poly): Likewise.
32729 2017-04-27 Tamar Christina <tamar.christina@arm.com>
32731 PR middle-end/79665
32732 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
32733 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
32735 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32738 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
32739 (aarch64_function_arg_alignment): Return unsigned int again, but still
32740 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
32741 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32742 Don't emit -Wpsabi note.
32743 (aarch64_function_arg_boundary): Likewise.
32744 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32747 2017-04-26 Nathan Sidwell <nathan@acm.org>
32749 * tree.h (crc32_unsigned_n): Declare.
32750 (crc32_unsigned, crc32_unsigned): Make inline.
32751 * tree.c (crc32_unsigned_bits): Replace with ...
32752 (crc32_unsigned_n): ... this.
32753 (crc32_unsigned, crc32_byte): Remove.
32754 (crc32_string): Remove unnecessary braces.
32756 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32758 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
32759 * ipa-inline-analysis.c (MAX_TIME): Remove.
32760 (account_size_time): Use sreal for time.
32761 (dump_inline_summary): Update.
32762 (estimate_function_body_sizes): Update.
32763 (estimate_edge_size_and_time): Update.
32764 (estimate_calls_size_and_time): Update.
32765 (estimate_node_size_and_time): Update.
32766 (inline_merge_summary): Update.
32767 (inline_update_overall_summary): Update.
32768 (estimate_time_after_inlining): Update.
32769 (inline_read_section): Update.
32770 (inline_write_summary): Update.
32771 * ipa-inline.c (compute_uninlined_call_time): Update.
32772 (compute_inlined_call_time): Update.
32773 (recursive_inlining): Update.
32774 (inline_small_functions): Update.
32775 (dump_overall_stats): Update.
32776 * ipa-inline.h: Include sreal.h.
32777 (size_time_entry): Turn time to sreal.
32778 (inline_summary): Turn self_time nad time to sreal.
32780 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32782 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
32784 (sreal::stream_out, sreal::stream_in): New.
32785 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
32787 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32789 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
32792 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
32795 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
32797 (dimode_scalar_chain::compute_convert_gain): Ditto.
32798 (dimode_scalar_chain::make_vector_copies): Ditto.
32799 (dimode_scalar_chain::convert_reg): Ditto.
32800 (dimode_scalar_chain::convert_insn): Ditto.
32801 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
32802 (VI248_AVX512BW_1): New mode iterator.
32803 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
32804 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
32807 2017-04-25 Martin Sebor <msebor@redhat.com>
32809 PR tree-optimization/80497
32810 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
32811 constants are representable in HOST_WIDE_INT.
32812 (parse_directive): Ditto.
32814 2017-04-25 Martin Sebor <msebor@redhat.com>
32817 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
32818 (new_zero_array): Adjust signature.
32819 (dom_info::dom_init): Used unsigned rather that size_t.
32820 (dom_info::dom_info): Same.
32822 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32823 Jakub Jelinek <jakub@redhat.com>
32826 * config/arm/arm.c: Include gimple.h.
32827 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32828 returns negative, increment ncrn only if it returned positive.
32829 (arm_needs_doubleword_align): Return int instead of bool,
32830 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
32831 members, but if there is any such non-FIELD_DECL
32832 > PARM_BOUNDARY aligned decl, return -1 instead of false.
32833 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32834 returns negative, increment nregs only if it returned positive.
32835 (arm_setup_incoming_varargs): Likewise.
32836 (arm_function_arg_boundary): Emit -Wpsabi note if
32837 arm_needs_doubleword_align returns negative, return
32838 DOUBLEWORD_ALIGNMENT only if it returned positive.
32840 2017-04-25 Marek Polacek <polacek@redhat.com>
32843 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32844 first argument to type.
32846 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
32849 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
32850 type checks to test for compatibility instead of equality.
32852 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32853 Jakub Jelinek <jakub@redhat.com>
32856 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
32858 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
32859 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
32860 the alignment computation, but return their maximum in warn_alignment.
32861 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32862 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
32864 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
32865 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32868 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32870 * config/arc/simdext.md (dmpyh): Fix typo.
32872 2017-04-25 Richard Biener <rguenther@suse.de>
32874 PR tree-optimization/80492
32875 * alias.c (compare_base_decls): Handle registers with asm
32876 specification conservatively.
32877 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32878 compare_base_decls returning dont-know properly.
32880 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32882 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
32883 (legitimate_offset_address_p): New function.
32884 (arc_legitimate_address_p): Use above function.
32886 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32888 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
32890 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32892 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
32893 ACCH registers whenever they are available.
32895 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32897 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
32898 double regs fix when not used.
32900 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32902 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
32904 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
32905 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
32907 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32909 * config/arc/arc.c (arc_output_addsi): Check for h-register class
32910 when emitting short ADD instructions.
32912 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32914 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
32916 (cmpsi_cc_c_insn): Likewise.
32917 (cbranchsi4_scratch): Compute proper instruction length using
32918 compact_hreg_operand.
32919 * config/arc/predicates.md (compact_hreg_operand): New predicate.
32921 2017-04-25 Richard Biener <rguenther@suse.de>
32923 PR middle-end/80509
32924 * passes.c (pass_manager::pass_manager): Initialize
32925 m_name_to_pass_map.
32927 2017-04-25 Richard Biener <rguenther@suse.de>
32929 PR tree-optimization/79201
32930 * tree-ssa-sink.c (statement_sink_location): Handle calls.
32932 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32935 * config/s390/vector.md: Split MEM->GPR vector moves for
32936 non-s_operand addresses.
32938 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32941 * config/s390/predicates.md (reload_const_wide_int_operand): New
32943 * config/s390/s390.md ("movti"): Remove d/P alternative.
32944 ("movti_bigconst"): New pattern definition.
32946 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32949 * s390-protos.h (s390_expand_cs_hqi): Removed.
32950 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
32951 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
32952 modes as well as CCZ1mode and CCZmode.
32953 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
32954 signature of s390_emit_compare_and_swap.
32955 (s390_expand_cs_hqi): Likewise, make static.
32956 (s390_expand_cs_tdsi): Generate an explicit compare before trying
32957 compare-and-swap, in some cases.
32958 (s390_expand_cs): Wrapper function.
32959 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
32961 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
32962 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
32963 patterns for small and large integers. Forbid symref memory operands.
32964 Move expander to s390.c. Require cc register.
32965 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
32966 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
32967 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
32968 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
32969 symref memory operands. Remove CC mode and call s390_match_ccmode
32971 ("atomic_exchange<mode>"): Allow and implement all integer modes.
32973 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32975 * config/s390/s390.md (define_peephole2): New peephole to help
32976 combining the load-and-test pattern with volatile memory.
32978 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32980 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
32981 with CCZmode for TARGET_Z196.
32983 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32985 PR rtl-optimization/80501
32986 * combine.c (make_compound_operation_int): Set subreg_code to SET
32987 even for AND with mask of the sign bit of mode.
32989 PR rtl-optimization/80500
32990 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
32991 sum's initial value.
32993 2017-04-25 Julian Brown <julian@codesourcery.com>
32994 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32996 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
32998 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
33000 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
33002 2017-04-25 Julian Brown <julian@codesourcery.com>
33003 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33005 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
33006 (thunderx2t99_sha): New Reservation.
33008 2017-04-25 Julian Brown <julian@codesourcery.com>
33009 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33011 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
33012 type for 1-element load.
33014 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
33016 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
33018 2017-04-24 Martin Jambor <mjambor@suse.cz>
33020 PR tree-optimization/80293
33021 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
33022 char arrays not totally scalarizable if it is false.
33023 (analyze_all_variable_accesses): Pass correct value in the new
33024 parameter. Add a statistics counter.
33026 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
33028 PR middle-end/79931
33029 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
33031 2017-04-24 Richard Biener <rguenther@suse.de>
33033 PR tree-optimization/80494
33034 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
33035 out for complex types.
33037 2017-04-24 Richard Biener <rguenther@suse.de>
33039 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
33040 * tree-ssa-sccvn.c (print_scc): Print SCC size.
33041 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
33042 (DFS): Adjust and never fail.
33043 (sccvn_dom_walker::fail): Remove.
33044 (sccvn_dom_walker::before_dom_children): Adjust.
33045 (run_scc_vn): Likewise and never fail.
33046 * tree-ssa-pre.c (pass_pre::execute): Adjust.
33047 (pass_fre::execute): Likewise.
33049 2017-04-24 Richard Biener <rguenther@suse.de>
33051 PR tree-optimization/79725
33052 * tree-ssa-sink.c (statement_sink_location): Return whether
33053 failure reason was zero uses. Move that check later.
33054 (sink_code_in_bb): Deal with zero uses by removing the stmt
33057 2017-04-24 Richard Biener <rguenther@suse.de>
33060 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
33061 pointer-based references.
33063 2017-04-24 Richard Biener <rguenther@suse.de>
33066 * pass_manager.h (pass_manager::operator new): Remove.
33067 (pass_manager::operator delete): Likewise.
33068 * passes.c (pass_manager::operator new): Remove.
33069 (pass_manager::operator delete): Likewise.
33070 (pass_manager::pass_manager): Zero individual pass members.
33072 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
33075 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
33076 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
33077 Check "XEXP (src, 1)" operand here.
33078 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
33079 Check "XEXP (src, 1)" operand here.
33080 (dimode_scalar_chain::make_vector_copies): Detect count register
33081 of a shift instruction. Zero extend count register from QImode
33082 to DImode to satisfy vector shift pattern count operand predicate.
33083 Substitute vector shift count operand with a DImode copy.
33084 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
33087 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
33089 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
33090 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33091 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
33092 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33093 (UNSPEC_NOREX_MEM): Remove definition.
33095 2017-04-21 Richard Biener <rguenther@suse.de>
33097 PR tree-optimization/79547
33098 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
33099 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
33100 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
33101 without any constraints.
33103 2017-04-21 Richard Biener <rguenther@suse.de>
33105 PR tree-optimization/78847
33106 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
33108 2017-04-21 Richard Biener <rguenther@suse.de>
33110 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
33111 (build_distinct_type_copy): Likewise.
33112 (build_variant_type_copy): Likewise.
33113 * tree.c (build_qualified_type): Pass down mem-stat info.
33114 (build_distinct_type_copy): Likewise.
33115 (build_variant_type_copy): Likewise.
33117 2017-04-21 Richard Biener <rguenther@suse.de>
33119 PR tree-optimization/80237
33120 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
33122 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
33123 for a simplified result.
33125 2016-04-21 Richard Biener <rguenther@suse.de>
33127 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
33128 sth as strict as a simple_iv but a chrec without symbols and an
33129 operand defined in the loop we are peeling (and not some subloop).
33130 (propagate_constants_for_unrolling): Propagate all constants.
33132 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33135 * config/i386/i386.c (print_reg): Remove assert for disalowed
33136 regno values, call output_operand_lossage instead.
33138 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33141 * config/i386/constraints.md (Yc): New register constraint.
33142 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
33143 Use Yc constraint for alternative 2 of operand 0. Remove
33144 preferred_for_speed attribute.
33146 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33148 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
33149 lastprivate clauses in SIMT case.
33151 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
33153 * doc/invoke.texi (-Wextra-semi): Document new warning option.
33155 2017-04-20 Richard Biener <rguenther@suse.de>
33157 PR tree-optimization/57796
33158 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
33159 as N scalar stores.
33160 (vect_model_load_cost): Cost gathers as N scalar loads.
33162 2017-04-20 Richard Biener <rguenther@suse.de>
33164 * ggc-page.c (ggc_allocated_p): Rename to ...
33165 (safe_lookup_page_table_entry): ... this and return the lookup
33167 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
33169 2017-04-20 Richard Biener <rguenther@suse.de>
33171 PR tree-optimization/80453
33172 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
33173 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
33174 from the conditions.
33175 (vn_phi_eq): Pass them down.
33176 (vn_phi_lookup): Record them.
33177 (vn_phi_insert): Likewise.
33179 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
33181 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
33182 uninitialized variable warning to avoid buffer overrun.
33184 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33187 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
33188 is suppressed for '{ 0 }' in C.
33190 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33192 * BASE-VER: Set to 8.0.0.
33194 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
33196 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
33197 priority .init_array and .fini_array section with SECTION_NOTYPE
33200 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33202 PR middle-end/80423
33203 * tree.h (build_array_type): Add typeless_storage default argument.
33204 * tree.c (type_cache_hasher::equal): Also compare
33205 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
33206 (build_array_type): Add typeless_storage argument, set
33207 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
33209 (build_nonshared_array_type): Adjust build_array_type_1 caller.
33210 (build_array_type): Likewise. Add typeless_storage argument.
33212 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33213 Jakub Jelinek <jakub@redhat.com>
33215 PR tree-optimization/80426
33216 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
33217 operation on symbolic operands, also compute the overflow for the
33218 invariant part when the operation degenerates into a negation.
33220 2017-04-19 Jakub Jelinek <jakub@redhat.com>
33223 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
33224 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
33227 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
33229 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
33232 * config/avr/avr.c (tree.h): Include it.
33233 (cgraph.h): Include it.
33234 (avr_encode_section_info): Don't warn for uninitialized progmem
33235 variable if it's just an alias.
33237 2017-04-19 Richard Biener <rguenther@suse.de>
33240 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
33241 when needed by AutoPGO.
33243 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
33246 * doc/lto.texi: Remove an extra 'that'.
33248 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
33250 PR rtl-optimization/80429
33251 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
33252 are only used in debug insns.
33254 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33255 Vladimir Makarov <vmakarov@redhat.com>
33257 * config/sparc/predicates.md (input_operand): Add comment. Return
33258 true for any memory operand when LRA is in progress.
33259 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
33261 2017-04-18 Jeff Law <law@redhat.com>
33264 * mips.md ({return,simple_return}_internal): Do not overwrite
33267 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33269 PR tree-optimization/80443
33270 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
33271 instead of adding 1, subtract -1 and similarly instead of subtracting
33274 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
33276 PR rtl-optimization/80357
33277 * haifa-sched.c (tmp_bitmap): New variable.
33278 (model_recompute): Handle duplicate use records.
33279 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
33280 (free_global_sched_pressure_data): Free it.
33282 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
33285 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33286 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33287 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33288 instead of SYSTEM_HEADER_DIR.
33290 2017-04-18 Jeff Law <law@redhat.com>
33292 PR middle-end/80422
33293 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
33294 predecessors after walking up the insn chain.
33296 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33299 * dwarf2out.c (modified_type_die): Try harder not to emit internal
33300 sizetype type into debug info.
33302 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33305 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
33306 unneeded test for TARGET_UPPER_REGS_SF.
33307 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
33309 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33312 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
33313 instead of gsi_after_labels.
33315 2017-04-18 Jeff Law <law@redhat.com>
33317 * regcprop.c (maybe_mode_change): Avoid creating copies of the
33321 2017-04-13 Jeff Law <law@redhat.com>
33322 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33323 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33325 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
33328 * config/avr/avr.c (intl.h): Include it.
33329 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
33331 2017-04-18 Martin Liska <mliska@suse.cz>
33333 PR gcov-profile/78783
33334 * gcov-tool.c (gcov_output_files): Validate that destination
33335 file is either removed by the tool or by a user.
33337 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33338 Guy Benyei <guybe@mellanox.com>
33340 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
33341 block, and do not negate it, the stored id is already negative.
33343 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33345 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
33347 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33350 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
33351 masks of options that should be turned off if the VSX vector
33352 options are turned off.
33353 (OTHER_P8_VECTOR_MASKS): Likewise.
33354 (OTHER_VSX_VECTOR_MASKS): Likewise.
33355 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
33356 rs6000_disable_incompatible_switches to validate no type switches
33358 (rs6000_incompatible_switch): New function to disallow turning on
33359 other vector options if -mno-vsx, -mno-power8-vector, or
33360 -mno-power9-vector are specified.
33362 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33364 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
33366 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33368 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
33369 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
33370 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
33371 (ARG_POINTER_CFA_OFFSET): Likewise.
33373 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33375 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
33376 conditions to take advantage of various optimizations.
33378 2017-04-13 Jeff Law <law@redhat.com>
33380 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33381 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33382 (zero_extendsidi2_dext): Likewise.
33384 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33387 * fold-const.c (fold_ternary_loc): Revert
33388 use op0 instead of fold_convert_loc (loc, type, arg0) part of
33391 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
33393 PR rtl-optimization/80343
33394 * lra-remat.c (update_scratch_ops): Assign original hard reg to
33395 new scratch pseudo.
33397 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
33400 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
33401 to ubsan_encode_value.
33403 2017-04-13 Jeff Law <law@redhat.com>
33405 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
33406 appearing in DEBUG_INSNs.
33408 2017-04-13 Martin Liska <mliska@suse.cz>
33410 PR gcov-profile/80413
33411 * gcov-io.c (gcov_write_string): Copy to buffer just when
33412 allocated size is greater than zero.
33414 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33417 * dwarf2out.c (decls_for_scope): Ignore declarations of
33418 current_function_decl in BLOCK_NONLOCALIZED_VARS.
33420 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
33423 * ipa-visibility.c (non_local_p): Fix typos.
33424 (localize_node): When localizing symbol in same comdat group,
33425 dissolve the group only when we know external symbols are going
33427 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
33429 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33431 PR tree-optimization/79390
33432 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
33433 order does not result in usable sequence, retry with reversed operand
33439 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
33440 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
33441 op0 instead of fold_convert_loc (loc, type, arg0).
33443 2017-04-12 Jeff Law <law@redhat.com>
33445 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
33446 has a delay slot in the generated code.
33448 * config/cris/cris.md (cris_preferred_reload_class): Return
33449 GENNONACR_REGS rather than GENERAL_REGS.
33451 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33454 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
33455 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
33456 signedness of the result type.
33458 2017-04-12 Richard Biener <rguenther@suse.de>
33459 Jeff Law <law@redhat.com>
33461 PR tree-optimization/80359
33462 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
33463 trim stores to TARGET_MEM_REFs.
33465 2017-04-12 Richard Biener <rguenther@suse.de>
33467 PR tree-optimization/79390
33468 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
33469 threading case even more.
33471 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
33474 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
33475 for quad_address_p for TImode, instead of just not indexed_address.
33477 2017-04-12 Richard Biener <rguenther@suse.de>
33478 Bernd Edlinger <bernd.edlinger@hotmail.de>
33480 PR middle-end/79671
33481 * alias.c (component_uses_parent_alias_set_from): Handle
33482 TYPE_TYPELESS_STORAGE.
33483 (get_alias_set): Likewise.
33484 * tree-core.h (tree_type_common): Add typeless_storage flag.
33485 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
33486 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
33487 for types containing members with TYPE_TYPELESS_STORAGE.
33488 (place_field): Likewise.
33489 (layout_type): Likewise for ARRAY_TYPE.
33490 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
33491 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
33492 TYPE_TYPELESS_STORAGE.
33493 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
33495 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33498 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33499 first argument to type.
33501 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33505 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
33506 CONST0_RTX (mode) rather than const0_rtx where appropriate.
33507 (rs6000_expand_binop_builtin): Likewise.
33508 (rs6000_expand_ternop_builtin): Likewise; also add missing
33509 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
33510 vshasigma built-ins.
33511 * doc/extend.texi: Document that vec_xxpermdi's third argument
33512 must be a constant.
33514 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
33516 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
33517 Use shift_const cost parameter when calculating gain of STV shifts.
33519 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
33521 PR rtl-optimization/70478
33522 * lra-constraints.c (process_alt_operands): Check memory for
33523 disfavoring memory insn operand.
33525 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33527 PR middle-end/80100
33528 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
33529 left shift in unsigned HOST_WIDE_INT type.
33531 PR rtl-optimization/80385
33532 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
33533 (not (neg X)) into (plus X -1) for complex or non-integral modes.
33536 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
33537 if they have any depend clauses.
33539 2017-04-11 Martin Liska <mliska@suse.cz>
33542 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
33543 * ipa-split.c (split_function): Create a local comdat symbol
33544 if caller is in a comdat group.
33546 2017-04-11 Martin Liska <mliska@suse.cz>
33549 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
33552 2017-04-11 Martin Sebor <msebor@redhat.com>
33554 PR middle-end/80364
33555 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
33556 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
33558 (directive::set_width, directive::set_precision, format_character):
33560 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
33563 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
33566 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
33567 conflict, set target->arch_name instead of target->cpu_name.
33569 2017-04-11 Richard Biener <rguenther@suse.de>
33571 PR tree-optimization/80374
33572 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
33573 build_zero_cst, remove fold_convertible_p check again.
33575 2017-04-11 Martin Liska <mliska@suse.cz>
33578 * ubsan.c (instrument_object_size): Do not instrument register
33581 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33584 * config/i386/i386-builtin-types.def
33585 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
33586 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
33587 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
33588 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
33589 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
33590 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
33591 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
33592 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
33593 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
33594 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
33595 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
33596 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
33597 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
33598 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
33599 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
33600 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
33601 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
33602 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
33603 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
33604 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
33605 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
33606 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
33607 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
33608 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
33609 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
33610 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
33611 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
33612 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
33613 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
33614 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
33615 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
33616 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
33617 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
33618 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
33619 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
33620 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
33621 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
33622 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
33623 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
33624 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
33625 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
33626 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
33627 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
33628 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
33629 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
33630 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
33632 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
33633 flag to second_arg_count, handle 4 argument function type _COUNT
33634 aliases, handle second_arg_count on second argument rather than last.
33636 2017-04-10 Jeff Law <law@redhat.com>
33638 PR tree-optimization/80374
33639 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
33640 record anything if we can not convert integer_zero_node to the
33643 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33646 * config/rs6000/rs6000.c (rs6000_option_override_internal):
33647 Enhance special handling given to the TARGET_P9_MINMAX option in
33648 relation to certain other options.
33650 2017-04-10 Bin Cheng <bin.cheng@arm.com>
33652 PR tree-optimization/80153
33653 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
33654 remove POINTER_PLUS_EXPR's base part directly, rather than through
33657 2017-04-10 Richard Biener <rguenther@suse.de>
33658 Bin Cheng <bin.cheng@arm.com>
33660 PR tree-optimization/80153
33661 * tree-affine.c (aff_combination_to_tree): Get base pointer from
33662 the first element of pointer type aff_tree. Build result expr in
33664 (add_elt_to_tree): Convert to type unconditionally. Remove other
33665 fold_convert calls.
33666 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
33667 (rewrite_use_nonlinear_expr): Check invariant using iv information.
33669 2017-04-10 Richard Biener <rguenther@suse.de>
33671 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33674 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
33676 PR rtl-optimization/70478
33677 * lra-constraints.c (curr_small_class_check): New.
33678 (update_and_check_small_class_inputs): New.
33679 (process_alt_operands): Update curr_small_class_check. Disfavor
33680 alternative insn memory operands. Check available regs for small
33683 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33686 * config/mips/mips.opt (-mvirt): Update description.
33687 * doc/invoke.texi (-mvirt): Likewise.
33689 2017-04-10 Richard Biener <rguenther@suse.de>
33691 PR middle-end/80362
33692 * fold-const.c (fold_binary_loc): Look at unstripped ops when
33693 looking for NEGATE_EXPR in -A / -B to A / B folding.
33695 2017-04-10 Martin Liska <mliska@suse.cz>
33697 PR gcov-profile/80224
33698 * gcov.c (print_usage): Fix usage string.
33699 (get_gcov_intermediate_filename): Remove.
33700 (output_gcov_file): Use both for normal and intermediate format.
33701 (generate_results): Do not initialize special file for
33702 intermediate format.
33704 2017-04-10 Richard Biener <rguenther@suse.de>
33706 PR tree-optimization/80304
33707 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
33710 2017-04-10 Nathan Sidwell <nathan@acm.org>
33713 * config/rs6000/rs6000.c (rs6000_vector_type): New.
33714 (rs6000_init_builtins): Use it.
33716 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33718 * config/arm/arm.md (<mrc>): Add mode to SET source.
33719 (<mrrc>): Likewise.
33721 2017-04-10 Richard Biener <rguenther@suse.de>
33723 PR middle-end/80344
33724 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
33726 2017-04-10 Jakub Jelinek <jakub@redhat.com>
33729 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
33730 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
33731 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
33732 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
33733 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
33734 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
33735 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
33736 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
33737 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
33738 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
33739 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
33740 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
33741 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
33742 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
33743 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
33744 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
33745 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
33746 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
33747 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
33748 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
33749 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
33750 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
33751 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
33753 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
33755 PR rtl-optimization/70478
33756 * lra-constraints.c: Reverse the last patch.
33758 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
33760 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
33761 Add comment for WCHAR_T.
33763 2017-04-08 Martin Liska <mliska@suse.cz>
33766 2017-04-07 Martin Liska <mliska@suse.cz>
33769 * ipa-split.c (split_function): Add function part to a same comdat
33772 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33775 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
33777 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
33779 * rs6000/rs6000.c (vec_load_pendulum): Rename...
33780 (vec_pairing): ...to this.
33781 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
33782 (rs6000_sched_init): Adjust for name change.
33783 (struct rs6000_sched_context): Likewise.
33784 (rs6000_init_sched_context): Likewise.
33785 (rs6000_set_sched_context): Likewise.
33787 2017-04-07 Jakub Jelinek <jakub@redhat.com>
33793 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
33795 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
33796 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
33797 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
33799 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
33801 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
33803 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33805 PR rtl-optimization/70703
33806 * ira-color.c (update_conflict_hard_regno_costs): Use
33807 int64_t instead of HOST_WIDE_INT.
33809 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33811 PR rtl-optimization/70478
33812 * lra-constraints.c (process_alt_operands): Disfavor alternative
33813 insn memory operands.
33815 2017-04-07 Jeff Law <law@redhat.com>
33817 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
33818 CALL and NOTE_INSN_CALL_ARG_LOCATION.
33820 2017-04-07 Martin Liska <mliska@suse.cz>
33823 * config/aarch64/aarch64.c (aarch64_process_target_attr):
33824 Show error message instead of an ICE.
33826 2017-04-07 Martin Liska <mliska@suse.cz>
33829 * ipa-split.c (split_function): Add function part to a same comdat
33832 2017-04-07 Richard Biener <rguenther@suse.de>
33834 PR middle-end/80341
33835 * tree.c (get_unwidened): Also handle ! for_type case for
33837 * convert.c (do_narrow): Split out from ...
33838 (convert_to_integer_1): ... here. Do not pass final truncation
33839 type to get_unwidened for TRUNC_DIV_EXPR.
33841 2017-04-07 Richard Biener <rguenther@suse.de>
33843 * tree-affine.c (wide_int_ext_for_comb): Take type rather
33845 (aff_combination_const): Adjust.
33846 (aff_combination_scale): Likewise.
33847 (aff_combination_add_elt): Likewise.
33848 (aff_combination_add_cst): Likewise.
33849 (aff_combination_convert): Likewise.
33850 (add_elt_to_tree): Likewise. Remove unused argument.
33851 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
33853 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
33855 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
33857 * config/arm/arm.c (arm_default_short_enums): Use
33858 ARM_DEFAULT_SHORT_ENUMS.
33859 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
33861 2017-04-06 Jakub Jelinek <jakub@redhat.com>
33864 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
33865 members with redundant out-of-class redeclaration.
33867 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33870 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
33871 * config/i386/i386.md (*zero_extendsidi2):
33872 Add (?*x,*x) and (?*v,*v) alternatives.
33874 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33877 * config/i386/i386.c (ix86_expand_builtin)
33878 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
33879 mode from insn data. Convert operands to insn operand mode.
33880 Copy operands that don't satisfy insn predicate to a register.
33882 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
33884 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
33887 2017-04-06 Richard Biener <rguenther@suse.de>
33889 PR tree-optimization/80334
33890 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
33891 preserve alignment of accesses.
33893 2017-04-06 Richard Biener <rguenther@suse.de>
33895 PR tree-optimization/80262
33896 * tree-sra.c (build_ref_for_offset): Preserve address-space
33898 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
33899 Drop useless address-space information on MEM_REF offsets.
33901 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
33903 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
33905 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33907 PR rtl-optimization/70703
33908 * ira-color.c (update_conflict_hard_regno_costs): Use
33909 HOST_WIDE_INT instead of long.
33911 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
33914 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
33915 not defined for x86_64 target. Add -mmmx target option when __SSE2__
33917 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
33918 for x86_64 target. Handle -m3dnowa option.
33920 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33922 PR rtl-optimization/70703
33923 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
33924 (update_conflict_hard_regno_costs): Use long instead of unsigned
33925 arithmetic for cost calculation.
33927 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33928 Bernd Edlinger <bernd.edlinger@hotmail.de>
33931 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
33934 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
33937 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
33938 ptr_mode with Pmode throughout.
33939 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
33940 into probe_stack_range and use DImode.
33942 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
33945 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
33946 call_eh_return is true.
33948 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33950 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
33951 Initialize last_match_fntype_index.
33953 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33956 * tree-nvr.c: Include internal-fn.h.
33957 (pass_return_slot::execute): Ignore internal calls without
33960 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33961 Richard Biener <rguenther@suse.de>
33964 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
33965 captures used multiple times, except for the last use.
33966 * generic-match-head.c: Include gimplify.h.
33968 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33970 PR tree-optimization/79390
33971 * target.h (struct noce_if_info): Declare.
33972 * targhooks.h (default_noce_conversion_profitable_p): Declare.
33973 * target.def (noce_conversion_profitable_p): New target hook.
33974 * ifcvt.h (struct noce_if_info): New type, moved from ...
33975 * ifcvt.c (struct noce_if_info): ... here.
33976 (noce_conversion_profitable_p): Renamed to ...
33977 (default_noce_conversion_profitable_p): ... this. No longer
33979 (noce_try_store_flag_constants, noce_try_addcc,
33980 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
33981 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
33982 instead of noce_conversion_profitable_p.
33983 * config/i386/i386.c: Include ifcvt.h.
33984 (ix86_option_override_internal): Don't override
33985 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
33986 (ix86_noce_conversion_profitable_p): New function.
33987 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
33988 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
33989 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
33990 * doc/tm.texi: Regenerated.
33992 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33994 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
33997 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
34000 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
34001 instructions for small multiply cores.
34003 2017-04-04 Jeff Law <law@redhat.com>
34005 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
34007 (mips_expand_vec_perm_const): Initialize elements in orig_perm
34008 that are not set by the loop over the elements.
34010 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34013 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
34014 int mode, convert_modes it to mode as unsigned, otherwise use
34015 lowpart_subreg to mode rather than SImode.
34016 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
34017 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
34018 Use DImode instead of SImode for the shift count operand.
34019 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
34022 2017-04-04 Richard Biener <rguenther@suse.de>
34024 PR middle-end/80281
34025 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
34026 arithmetic done for the negate or the plus. Simplify.
34027 (A - (-B) -> A + B): Likewise.
34028 * fold-const.c (split_tree): Make sure to not negate pointers.
34030 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
34032 PR rtl-optimization/60818
34033 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
34034 a compare of comparisons with the thing compared if this results
34035 in a different machine mode.
34037 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
34039 * alias.c (base_alias_check): Fix typo in comment.
34040 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
34041 * cgraphunit.c (symbol_table::compile): Likewise.
34042 * collect2.c (maybe_run_lto_and_relink): Likewise.
34043 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
34044 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
34045 * config/avr/avr.c (avr_map_op_t): Likewise.
34046 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
34047 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
34048 * config/epiphany/epiphany.md (movcc): Likewise.
34049 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
34050 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
34052 * config/mips/mips.c (mips_save_restore_reg): Likewise.
34053 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
34054 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
34055 * config/sh/sh.c (sh_rtx_costs): Likewise.
34056 * fold-const.c (fold_truth_andor): Likewise.
34057 * genautomata.c (collapse_flag): Likewise.
34058 * gengtype.h (struct type::u::s): Likewise.
34059 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
34060 * input.c (FORMAT_AMOUNT): Likewise.
34061 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
34062 (known_aggs_to_agg_replacement_list): Likewise.
34063 * ipa-inline-analysis.c: Likewise.
34064 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
34065 * ipa-polymorphic-call.c
34066 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
34067 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
34068 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
34070 * modulo-sched.c (apply_reg_moves): Likewise.
34071 * omp-expand.c (build_omp_regions_1): Likewise.
34072 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
34073 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
34074 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
34075 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
34076 * value-prof.c: Likewise.
34077 * var-tracking.c (val_reset): Likewise.
34079 2017-04-03 Richard Biener <rguenther@suse.de>
34081 PR tree-optimization/80275
34082 * fold-const.c (split_address_to_core_and_offset): Handle
34085 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
34087 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
34088 descriptors is at least equal to that of functions.
34090 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34092 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
34094 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34097 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
34098 (mov<IMOD4:mode>): New expander.
34099 (*mov<IMOD4:mode>_internal): New insn and split pattern.
34101 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
34103 PR rtl-optimization/79405
34104 * fwprop.c (propagations_left): New variable.
34105 (forward_propagate_into): Decrement it.
34106 (fwprop_init): Initialize it.
34107 (fw_prop): If the variable has reached zero, stop propagating.
34108 (fwprop_addr): Ditto.
34110 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34113 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
34114 a FUNCTION_DECL, pass it as decl instead of origin to
34117 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
34119 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
34122 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
34125 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
34126 TARGET_VSX_SMALL_INTEGER.
34128 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34130 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
34131 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
34133 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34135 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
34136 extraction from odd-numbered MSA register.
34138 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34140 PR middle-end/80173
34141 * expmed.c (store_bit_field_1): Don't attempt to create
34142 a word subreg out of hard registers wider than word if they
34143 have HARD_REGNO_NREGS of 1 for their mode.
34145 PR middle-end/80163
34146 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
34147 conversions to integer types wider than word and pointer.
34150 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
34151 (rtx_equal_for_cselib_p): Pass 0 to it.
34152 * cselib.c (cselib_hasher::equal): Likewise.
34153 (rtx_equal_for_cselib_1): Add depth argument. If depth
34154 is 128, don't look up VALUE locs and punt. Increment
34155 depth in recursive calls when walking VALUE locs.
34157 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
34159 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
34160 (make_gcov_file_name): Use the canonical path name for generating
34162 (read_line): Fix handling of files with ascii null bytes.
34164 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
34166 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
34167 to initialise a vector register instead
34168 of using a const_int.
34170 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34172 PR translation/80189
34173 * gimplify.c (omp_default_clause): Use %qs instead of %s in
34174 diagnostic messages.
34176 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
34179 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
34180 (dfp_diex_<mode>): Update mode of operand 1.
34181 * doc/extend.texi (dxex, dxexq): Document change to return type.
34182 (diex, diexq): Document change to argument type.
34184 2017-03-30 Martin Jambor <mjambor@suse.cz>
34187 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
34188 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
34189 it reflects the signature changes performed at the callee side.
34190 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
34191 to cgraph_build_function_type_skip_args.
34192 (build_function_decl_skip_args): Adjust call to the above function.
34194 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34197 * config/i386/sse.md
34198 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
34199 register as dest whenever it is a MEM not rtx_equal_p to the
34200 corresponding dup operand, and when forcing into reg move the
34201 reg into the memory afterwards.
34202 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
34203 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
34204 for the force_reg mode.
34205 (avx512vl_vextractf128<mode>): Use register as dest either
34206 always when a MEM, or when it is a MEM not rtx_equal_p to the
34207 corresponding dup operand, or even not when it is a CONST_VECTOR
34208 depending on the mode and lo vs. hi.
34209 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
34211 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
34212 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
34213 Likewise. Require that operands[2] is even.
34214 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
34215 Remove extraneous parens. Require that operands[2] is a multiple
34217 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
34218 operands[0] is a MEM if <mask_applied>, the predicates/constraints
34219 disallow memory then.
34221 2017-03-30 Richard Biener <rguenther@suse.de>
34223 PR tree-optimization/77498
34224 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
34225 to non-constants over backedges.
34227 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
34229 PR rtl-optimization/80233
34230 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
34231 as last_combined_insn. Do not test for BARRIER_P separately.
34233 2017-03-29 Andreas Schwab <schwab@suse.de>
34236 * calls.c (prepare_call_address): Convert funexp to Pmode before
34237 copying to temp reg.
34239 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34241 PR tree-optimization/80158
34242 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
34243 Handle possible future case of more than one alternate
34245 (replace_rhs_if_not_dup): Likewise.
34246 (replace_one_candidate): Likewise.
34248 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
34250 PR rtl-optimization/80193
34251 * ira.c (ira): Do not check allocation for LRA.
34253 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
34255 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
34256 (nvptx_output_simt_exit): Declare.
34257 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
34258 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
34259 (init_softstack_frame): Move initialization of crtl->is_leaf to...
34260 (nvptx_declare_function_name): ...here. Emit declaration of local
34261 memory space buffer for omp_simt_enter insn.
34262 (nvptx_output_unisimt_switch): New.
34263 (nvptx_output_softstack_switch): New.
34264 (nvptx_output_simt_enter): New.
34265 (nvptx_output_simt_exit): New.
34266 * config/nvptx/nvptx.h (struct machine_function): New fields
34267 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
34268 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
34269 (UNSPECV_SIMT_EXIT): Ditto.
34270 (omp_simt_enter_insn): New insn.
34271 (omp_simt_enter): New expansion.
34272 (omp_simt_exit): New insn.
34273 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
34275 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
34276 (expand_GOMP_SIMT_ENTER_ALLOC): New.
34277 (expand_GOMP_SIMT_EXIT): New.
34278 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
34279 (GOMP_SIMT_ENTER_ALLOC): Ditto.
34280 (GOMP_SIMT_EXIT): Ditto.
34281 * target-insns.def (omp_simt_enter): New insn.
34282 (omp_simt_exit): Ditto.
34283 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
34285 (lower_rec_simd_input_clauses): Implement SIMT privatization.
34286 (lower_rec_input_clauses): Likewise.
34287 (lower_lastprivate_clauses): Handle SIMT privatization.
34289 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
34290 (ompdevlow_adjust_simt_enter): New.
34291 (find_simtpriv_var_op): New.
34292 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
34293 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
34295 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
34296 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
34297 (copy_decl_for_dup_finish): Ditto.
34299 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
34301 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
34304 * config/i386/i386.c (ix86_option_override_internal): Always
34305 allow -mpreferred-stack-boundary=3 for 64-bit targets.
34307 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34309 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
34311 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34313 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
34314 mark new edge's irreducible flag accordign to it.
34315 (vect_do_peeling): Check loop preheader edge's irreducible flag
34316 and pass it to function slpeel_add_loop_guard.
34318 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
34320 PR tree-optimization/80218
34321 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34322 Update block frequencies and counts.
34324 2017-03-28 Richard Biener <rguenther@suse.de>
34326 PR tree-optimization/78644
34327 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
34328 of a simplification result we may not use it at all.
34330 2017-03-28 Richard Biener <rguenther@suse.de>
34333 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
34334 without arguments, generate default definition of a SSA name.
34336 2017-03-28 Richard Biener <rguenther@suse.de>
34338 PR middle-end/80222
34339 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
34340 TYPE_REF_CAN_ALIAS_ALL references.
34341 * fold-const.c (fold_indirect_ref_1): Likewise.
34343 2017-03-28 Martin Liska <mliska@suse.cz>
34346 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
34347 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
34349 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34350 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
34352 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
34353 (EXTRA_SPECS): Define.
34354 (SUBTARGET_EXTRA_SPECS): Likewise.
34355 (SUBTARGET_CPP_SPEC): Likewise.
34356 * config/arc/elf.h (EXTRA_SPECS): Renamed to
34357 SUBTARGET_EXTRA_SPECS.
34358 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
34360 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34362 * config/arc/simdext.md (vst64_insn): Update pattern.
34363 (vld32wh_insn): Likewise.
34364 (vld32wl_insn): Likewise.
34365 (vld64_insn): Likewise.
34366 (vld32_insn): Likewise.
34368 2017-03-28 Marek Polacek <polacek@redhat.com>
34371 * fold-const.c (fold_comparison): Use protected_set_expr_location
34372 instead of SET_EXPR_LOCATION.
34374 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
34376 * tree.c (add_expr): Avoid name lookup warning.
34378 2017-03-27 Jeff Law <law@redhat.com>
34380 PR tree-optimization/80216
34381 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
34382 function name. Limit recursion depth.
34383 (record_temporary_equivalences): Corresponding changes.
34385 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
34387 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
34390 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34393 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
34395 * cfgcleanup.c (reg_note_cfa_p): New array.
34396 (insns_have_identical_cfa_notes): New function.
34397 (old_insns_match_p): Don't cross-jump in between /f
34398 and non-/f instructions. If both i1 and i2 are frame related,
34399 verify all CFA notes, their order and content.
34401 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
34404 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
34405 HImode and SImode with zero extend to DImode to one insn.
34406 (bswap<mode>2_extenddi): Likewise.
34407 (bswapsi2_extenddi): Likewise.
34408 (bswaphi2_extendsi): Likewise.
34409 (bswaphi2): Combine bswap HImode and SImode into one insn.
34410 Separate memory insns from swapping register.
34411 (bswapsi2): Likewise.
34412 (bswap<mode>2): Likewise.
34413 (bswaphi2_internal): Delete, no longer used.
34414 (bswapsi2_internal): Likewise.
34415 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
34416 store, and gpr<-gpr swap insns.
34417 (bswap<mode>2_store): Likewise.
34418 (bswaphi2_reg): Register only splitter, combine with the splitter.
34419 (bswaphi2 splitter): Likewise.
34420 (bswapsi2_reg): Likewise.
34421 (bswapsi2 splitter): Likewise.
34422 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
34423 the insns into load, store, and register/register insns.
34424 (bswapdi2_ldbrx): Likewise.
34425 (bswapdi2_load): Likewise.
34426 (bswapdi2_store): Likewise.
34427 (bswapdi2_reg): Likewise.
34429 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
34431 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
34432 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
34434 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
34437 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
34439 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34440 special handling for target option conflicts between dform
34441 options (-mpower9-dform, -mpower9-dform-vector,
34442 -mpower9-dform-scalar) and -mno-direct-move.
34444 2017-03-27 Richard Biener <rguenther@suse.de>
34446 PR tree-optimization/80181
34447 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
34449 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34451 * config/arc/predicates.md (move_double_src_operand): Replace the
34452 call to move_double_src_operand with a call to address_operand.
34454 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34456 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
34457 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
34458 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
34460 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34462 * config/arc/predicates.md (long_immediate_loadstore_operand):
34463 Consider scaled addresses cases.
34465 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34467 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
34468 restored when in interrupt.
34469 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
34470 doesn't have delay slot.
34472 2017-03-27 Richard Biener <rguenther@suse.de>
34475 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
34476 inlined thunk clones.
34478 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34481 * asan.c (instrument_derefs): Copy over last operand from
34482 original COMPONENT_REF to the new COMPONENT_REF with
34483 DECL_BIT_FIELD_REPRESENTATIVE.
34484 * ubsan.c (instrument_object_size): Likewise.
34486 2017-03-27 Richard Biener <rguenther@suse.de>
34488 PR tree-optimization/80170
34489 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
34490 sure DR/SCEV didnt fold in constants we do not see when looking
34491 at the reference base alignment.
34493 2017-03-27 Richard Biener <rguenther@suse.de>
34495 PR middle-end/80171
34496 * gimple-fold.c (fold_ctor_reference): Properly guard against
34497 NULL return value from canonicalize_constructor_val.
34499 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
34502 * config/i386/i386.c (ix86_expand_builtin)
34503 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
34504 flags reg setting and flags reg using instructions.
34505 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
34506 clobbering instructions to zero extend op2.
34508 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
34510 * doc/install.texi (Configuration) <--with-aix-soname>:
34511 Update link to AIX ld.
34513 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
34515 PR rtl-optimization/80160
34516 PR rtl-optimization/80159
34517 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
34518 reg_alternate_class into account.
34520 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
34523 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
34524 to consider in curr_insn_transform.
34526 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34528 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
34529 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
34530 and emit_mode_inner.
34532 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34534 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
34535 argument to the overloaded builtin variants. Use the new flag to
34536 deprecate certain builtin variants.
34537 * config/s390/s390-builtin-types.def: Add new builtin types.
34538 * config/s390/s390-builtins.h: Support new flags field for
34539 overloaded builtins.
34540 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
34541 (s390_macro_to_expand): Enable vector float data type.
34542 (s390_cpu_cpp_builtins_internal): Indicate support of the new
34543 builtins by incrementing the __VEC__ version number.
34544 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
34546 (s390_resolve_overloaded_builtin): Emit error messages depending
34547 on the builtin flags.
34548 * config/s390/s390.c (s390_expand_builtin): Support additional
34549 flags argument. Change error message to match the messages
34550 emitted in s390-c.c.
34551 * config/s390/s390.md: New UNSPEC_* constants.
34552 (op_type): Add new instruction types.
34553 * config/s390/vecintrin.h: Add new builtins and test data class
34555 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
34556 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
34557 (VEC_INEXACT, VEC_NOINEXACT): New constants.
34558 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
34559 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
34560 ("vec_mergel<mode>"): V_HW -> VEC_HW.
34562 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
34563 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
34564 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
34565 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
34567 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
34568 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
34569 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
34570 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
34572 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
34573 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
34574 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
34575 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
34576 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
34577 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
34578 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
34580 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
34581 ("vec_scatter_element<V_HW_4:mode>_DI")
34582 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
34583 ("vec_fpint<mode>", "vflls")
34584 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
34585 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
34586 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
34587 ("*vec_cmphe<mode>_cc"): ... these.
34589 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
34590 mode constant instead of magic value.
34592 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34594 * config/s390/s390.c (s390_expand_vec_compare): Support other
34595 vector floating point modes than just V2DF.
34596 (s390_expand_vcond): Likewise.
34597 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
34598 (s390_cannot_change_mode_class): Prevent mode changes between TF
34599 and V1TF in vector registers.
34600 * config/s390/s390.md (DF, SF): New mode attributes.
34601 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
34602 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
34603 SFmode support for VRs.
34604 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
34606 (VFT, VF_HW): New mode iterators.
34607 (vw, sdx): New mode attributes.
34608 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
34609 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
34610 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
34611 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
34612 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
34613 also the new vector floating point modes. Renaming to ...
34615 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
34616 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
34617 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
34618 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
34619 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
34620 ("vec_unordered<mode>"): ... these.
34622 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
34623 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
34624 ("*vec_extendv2df"): New insn definitions.
34626 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34628 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
34629 ("mulditi3_2", "*muldi3_sign"): New patterns.
34630 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
34631 rename the pattern definition.
34633 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34635 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
34637 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
34639 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34641 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
34642 instruction if possible.
34643 * config/s390/vector.md (vec_halfnumelts): New mode
34645 ("*vec_vllezlf<mode>"): New pattern.
34647 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34649 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
34650 ("popcountv4si2", "popcountv2di2"): Rename to ...
34651 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
34652 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
34654 ("popcount<mode>2_vxe"): New pattern.
34656 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34658 * common/config/s390/s390-common.c (processor_flags_table): Add
34660 * config.gcc: Add arch12.
34661 * config/s390/driver-native.c (s390_host_detect_local_cpu):
34662 Default to arch12 for unknown CPU model numbers.
34663 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
34664 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
34665 PROCESSOR_max sanity check.
34666 * config/s390/s390-opts.h (enum processor_type): Add
34668 * config/s390/s390.c (processor_table): Add arch12.
34669 (s390_expand_builtin): Add check for B_VXE flag.
34670 (s390_issue_rate): Add PROCESSOR_ARCH12.
34671 (s390_get_sched_attrmask): Likewise.
34672 (s390_get_unit_mask): Likewise.
34673 (s390_sched_score): Enable z13 scheduling for arch12.
34674 (s390_sched_reorder): Likewise.
34675 (s390_sched_variable_issue): Likewise.
34676 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
34678 (s390_tune_attr): Use z13 scheduling also for arch12.
34679 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
34680 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
34681 (TARGET_VXE_P): New macros.
34682 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
34683 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
34684 * config/s390/s390.opt: Add arch12 as processor_type.
34686 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34688 * config/s390/s390.md
34689 ("fixuns_truncdddi2", "fixuns_trunctddi2")
34690 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
34691 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
34693 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
34694 Rename expanders to ...
34696 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
34697 ("fixuns_truncdddi2_emu"): ... these.
34699 ("fixuns_trunc<mode>si2_emu"): New expander.
34701 ("*fixuns_truncdfdi2_z13"): Rename to ...
34702 ("*fixuns_truncdfdi2_vx"): ... this.
34704 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34706 * config/s390/2964.md: Remove the single element vector compare
34707 instructions which are no longer used.
34708 * config/s390/s390.c (s390_select_ccmode): Remove handling of
34710 (s390_canonicalize_comparison): Remove handling of DFmode
34712 (s390_expand_vec_compare_scalar): Remove function.
34713 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
34714 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
34716 ("*cmp<mode>_ccs"): Add wfcdb instruction.
34718 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34720 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
34722 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
34723 will anyway by matched by mov<mode>_64dfp.
34725 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34727 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
34728 vlef/vstef. Add missing operand to vleif.
34730 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34732 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
34733 pair for all vector types with 64 bit elements.
34734 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
34735 * config/s390/vector.md (V_HW_64): ... here.
34736 (V_128_NOSINGLE): New mode iterator.
34737 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
34738 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
34739 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
34740 ("*vec_load_pairv2di"): Change to ...
34741 ("*vec_load_pair<mode>"): ... this one.
34743 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34745 * config/s390/constraints.md: Add comments.
34746 (jKK): Reject element sizes > 8 bytes.
34747 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
34749 * config/s390/s390.md: Add the s_operand checks formerly in
34750 s390_split_ok_p to various splitters where they are still
34752 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
34753 for 128 bit vectors. Plus two splitters.
34755 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34757 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
34760 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34763 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
34764 error if the boundary argument is not constant.
34766 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34768 PR rtl-optimization/80112
34769 * loop-doloop.c (doloop_condition_get): Don't check condition
34770 if cmp isn't SET with IF_THEN_ELSE src.
34772 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34774 PR tree-optimization/80158
34775 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
34776 replacing a candidate statement, also replace it for the
34777 candidate's alternate interpretation.
34778 (replace_rhs_if_not_dup): Likewise.
34779 (replace_one_candidate): Likewise.
34781 2017-03-24 Richard Biener <rguenther@suse.de>
34783 PR tree-optimization/80167
34784 * graphite-isl-ast-to-gimple.c
34785 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
34787 (translate_isl_ast_to_gimple::get_rename): Likewise.
34789 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34791 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
34792 handling of certain combinations of target options, including the
34793 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
34794 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
34796 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34799 * config/arm/arm.md (*load_multiple): Add reload_completed to
34800 matching condition.
34802 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34803 Richard Biener <rguenth@suse.de>
34805 PR tree-optimization/79908
34806 PR tree-optimization/80136
34807 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34808 been cast away, gimplify_and_add suffices.
34810 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
34812 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
34814 2017-03-23 Richard Biener <rguenther@suse.de>
34816 PR tree-optimization/80032
34817 * gimplify.c (gimple_push_cleanup): Forced unconditional
34818 cleanups still have to go to the conditional_cleanups
34821 2017-03-22 Jakub Jelinek <jakub@redhat.com>
34823 PR tree-optimization/80072
34824 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
34826 (next_operand_entry_id): Change type to unsigned int.
34827 (sort_by_operand_rank): Make sure to return the right return value
34828 even if unsigned fields are bigger than INT_MAX.
34829 (struct oecount): Change cnt and id type to unsigned int.
34830 (oecount_hasher::equal): Formatting fix.
34831 (oecount_cmp): Make sure to return the right return value
34832 even if unsigned fields are bigger than INT_MAX.
34833 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
34836 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
34837 TREE_READONLY on result if writing it more than once.
34840 * doc/invoke.texi (-fsanitize=thread): Document that with
34841 -fnon-call-exceptions atomics are not able to throw
34845 * tsan.c: Include tree-eh.h.
34846 (instrument_builtin_call): Call maybe_clean_eh_stmt or
34847 maybe_clean_or_replace_eh_stmt where needed.
34848 (instrument_memory_accesses): Add cfg_changed argument.
34849 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
34850 if it returned true.
34851 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
34853 PR rtl-optimization/63191
34854 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
34855 wrapper function, moved the whole old content into ...
34856 (ix86_delegitimize_address_1): ... this. New inline function.
34857 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
34858 true as last argument instead of ix86_delegitimize_address.
34860 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34862 * config/aarch64/aarch64.c (generic_branch_cost): Copy
34863 cortexa57_branch_cost.
34865 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34867 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
34869 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34872 * doc/md.texi (Constraints): Document wA constraint.
34873 * config/rs6000/constraints.md (wA): New.
34874 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
34875 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
34876 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
34877 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
34879 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
34882 * gimplify.c (is_oacc_declared): New function.
34883 (oacc_default_clause): Use it to set default flags for acc declared
34884 variables inside parallel regions.
34885 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
34886 declared variables.
34887 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
34888 declare attribute to any decl as necessary.
34890 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
34893 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
34894 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
34895 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
34896 (arm_arch_lpae): This.
34897 * config/arm/arm.c (arm_arch7ve): Rename into ...
34898 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
34899 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
34902 2017-03-22 Martin Liska <mliska@suse.cz>
34905 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
34906 error message instead of an ICE.
34908 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34910 * doc/extend.texi (6.11 Additional Floating Types): Revise.
34912 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
34914 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34916 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34919 2017-03-21 Martin Sebor <msebor@redhat.com>
34921 * doc/extend.texi: Use "cannot" instead of "can't."
34922 * doc/hostconfig.texi: Same.
34923 * doc/install.texi: Same.
34924 * doc/invoke.texi: Same.
34925 * doc/loop.texi: Same.
34926 * doc/md.texi: Same.
34927 * doc/objc.texi: Same.
34928 * doc/rtl.texi: Same.
34929 * doc/tm.texi: Same.
34930 * doc/tm.texi.in: Same.
34931 * doc/trouble.texi: Same.
34933 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
34936 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
34937 (collect_checksum_attributes): Set it.
34938 (die_checksum_ordered): Use it.
34940 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34942 PR tree-optimization/79908
34943 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
34944 change: For a VA_ARG whose LHS has been cast away, use
34945 force_gimple_operand to construct the side effects.
34947 2017-03-21 David Malcolm <dmalcolm@redhat.com>
34949 PR translation/80001
34950 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
34951 more amenable to translation.
34952 (oacc_loop_auto_partitions): Likewise.
34954 2017-03-21 Marek Polacek <polacek@redhat.com>
34955 Martin Sebor <msebor@redhat.com>
34957 PR tree-optimization/80109
34958 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
34959 on INTEGRAL_TYPE_P.
34961 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34962 Segher Boessenkool <segher@kernel.crashing.org>
34965 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
34966 check reg_used_between_p between insn and one of succ or succ2
34967 depending on if succ is artificial insn not inserted into insn
34970 2017-03-21 Martin Liska <mliska@suse.cz>
34972 PR gcov-profile/80081
34973 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
34974 * doc/gcc.texi: Include gcov-dump stuff.
34975 * doc/gcov-dump.texi: New file.
34977 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
34979 PR rtl-optimization/79150
34980 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
34981 conditional jump, if the jump is the last insn of the loop.
34983 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34984 Richard Biener <rguenth@suse.de>
34986 PR tree-optimization/79908
34987 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34988 been cast away, use force_gimple_operand to construct the side
34991 2017-03-21 Martin Liska <mliska@suse.cz>
34993 PR libfortran/79956
34994 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
34997 2017-03-21 Brad Spengler <spender@grsecurity.net>
35000 * plugin.c (htab_hash_plugin): New function.
35001 (add_new_plugin): Use it and adjust.
35002 (parse_plugin_arg_opt): Adjust.
35003 (init_one_plugin): Likewise.
35005 2017-03-21 Richard Biener <rguenther@suse.de>
35007 PR tree-optimization/80032
35008 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
35009 if set force the cleanup to happen unconditionally.
35010 (gimplify_target_expr): Push inserted clobbers with force_uncond
35011 to avoid them being removed by control-dependent DCE.
35013 2017-03-21 Richard Biener <rguenther@suse.de>
35015 PR tree-optimization/80122
35016 * tree-inline.c (copy_bb): Do not expans va-arg packs or
35017 va_arg_pack_len when the inlined call stmt requires pack
35019 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
35021 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35024 * tsan.c (instrument_builtin_call): If the memory model argument
35025 is not a constant, assume it is valid.
35028 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
35031 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35033 PR rtl-optimization/79910
35034 * combine.c (can_combine_p): Do not allow combining an I0 or I1
35035 if its dest is used by an insn before I2 (other than the combined
35036 insns themselves, which are properly handled already).
35038 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35041 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35043 * combine.c (record_used_regs): New static function.
35044 (try_combine): Handle situations where there is an additional
35045 instruction between I2 and I3 which needs to have a LOG_LINK
35049 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35051 * combine.c (try_combine): Delete redundant i1 test. Call
35052 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35054 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35057 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
35058 alternatives 13/14.
35060 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35062 PR tree-optimization/80054
35063 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
35064 the optimization if a PHI or any of its arguments is not dominated
35065 by the candidate's basis. Use gphi* rather than gimple* as
35067 (replace_profitable_candidates): Clean up a gimple* variable that
35068 should be a gphi* variable.
35070 2017-03-20 Martin Sebor <msebor@redhat.com>
35073 * doc/extend.texi (attribute constructor): Document present limitation.
35075 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
35078 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
35079 __POWER9_VECTOR__ #ifdef control, change template definition to
35080 use Power9-specific built-in function.
35081 (vec_any_eq): Likewise.
35082 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
35083 to control outcomes from this test.
35084 (vector_ae_<mode>p): For VEC_F modes, likewise.
35086 2017-03-20 Ian Lance Taylor <iant@google.com>
35088 * config/i386/i386.c (ix86_function_regparm): Save an extra
35089 register for -fsplit-stack with DECL_STATIC_CHAIN.
35091 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35094 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
35095 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
35097 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35099 * config/riscv/riscv.c (riscv_print_operand): Use "fence
35101 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
35104 2017-03-20 Marek Polacek <polacek@redhat.com>
35107 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
35109 2017-03-20 Richard Biener <rguenther@suse.de>
35111 PR tree-optimization/80113
35112 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
35113 allocate extra SSA name for PHI def.
35114 (add_close_phis_to_outer_loops): Likewise.
35115 (add_close_phis_to_merge_points): Likewise.
35116 (copy_loop_close_phi_args): Likewise.
35117 (copy_cond_phi_nodes): Likewise.
35119 2017-03-20 Martin Liska <mliska@suse.cz>
35121 PR middle-end/79753
35122 * tree-chkp.c (chkp_build_returned_bound): Do not build
35123 returned bounds for a LHS that's not a BOUNDED_P type.
35125 2017-03-20 Martin Liska <mliska@suse.cz>
35129 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
35130 COMPLEX_CST and VECTOR_CST.
35132 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35135 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
35136 target operand. A new splitter adds the clobber statement in case
35137 the target operand is dead anyway.
35139 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
35141 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
35142 to age-old versions of binutils and glibc.
35144 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
35146 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
35148 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35150 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
35152 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35154 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
35155 requirement for binutils 2.13.
35157 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35159 * combine.c (try_combine): Delete redundant i1 test. Call
35160 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35162 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
35164 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
35165 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
35167 <riscv64-*-elf>: Re-arrange section
35168 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
35169 <riscv32-*-linux>: Likewise.
35170 <riscv64-*-elf>: Likewise
35171 <riscv64-*-linux>: Likewise.
35173 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
35176 * aarch64.opt(verbose-cost-dump): Fix typo.
35178 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
35181 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
35182 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
35184 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35186 * reload.c (find_reloads): When reloading a nonoffsettable address,
35187 use RELOAD_OTHER for it and its address reloads.
35189 PR rtl-optimization/79910
35190 * combine.c (record_used_regs): New static function.
35191 (try_combine): Handle situations where there is an additional
35192 instruction between I2 and I3 which needs to have a LOG_LINK
35195 2017-03-17 Jeff Law <law@redhat.com>
35197 PR tree-optimization/71437
35198 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
35199 conditional in the hash table first.
35200 (vrp_dom_walker::before_dom_children): Extract condition from
35201 ASSERT_EXPR. Record condition, its inverion and any implied
35202 conditions as well.
35204 2017-03-17 Marek Polacek <polacek@redhat.com>
35205 Markus Trippelsdorf <markus@trippelsdorf.de>
35207 PR tree-optimization/80079
35208 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
35211 2017-03-17 Richard Biener <rguenther@suse.de>
35213 PR middle-end/80075
35214 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
35215 Properly verify the LHS before the RHS possibly claims to be
35217 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
35220 2017-03-17 Martin Jambor <mjambor@suse.cz>
35222 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
35223 (List of -O2 options): Likewise.
35224 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
35227 2017-03-17 Tom de Vries <tom@codesourcery.com>
35229 * gcov-dump.c (print_usage): Print bug_report_url.
35231 2017-03-17 Richard Biener <rguenther@suse.de>
35233 PR middle-end/80050
35234 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
35235 (parser::peek): Likewise.
35237 2017-03-17 Richard Biener <rguenther@suse.de>
35239 PR tree-optimization/80048
35240 * sese.c (free_sese_info): Properly release rename_map and
35241 copied_bb_map elements.
35243 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
35245 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
35246 Add linked-list forward and backlinks. Insert on
35247 construction, remove on destruction.
35248 (class pass_store_merging): Add m_stores_head field.
35249 (pass_store_merging::terminate_and_process_all_chains):
35250 Iterate over m_stores_head list.
35251 (pass_store_merging::terminate_all_aliasing_chains):
35253 (pass_store_merging::execute): Check for debug stmts first.
35254 Push new chains onto the m_stores_head stack.
35256 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
35259 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
35260 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
35261 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
35263 2017-03-16 Jeff Law <law@redhat.com>
35265 PR tree-optimization/71437
35266 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
35267 member function. Implementation moved into after_dom_children
35268 member function and into the threader's thread_outgoing_edges
35270 (dom_opt_dom_walker::after_dom_children): Simplify by moving
35271 some code into new thread_outgoing_edges.
35272 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
35273 definition. Simplify marker handling (do it here). Assume we always
35274 have the available expression and the const/copies tables.
35275 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
35277 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
35278 * tree-vrp.c (equiv_stack): No longer file scoped.
35279 (vrp_dom_walker): New class.
35280 (vrp_dom_walker::before_dom_children): New member function.
35281 (vrp_dom_walker::after_dom_children): Likewise.
35282 (identify_jump_threads): Setup domwalker. Use it rather than
35283 walking edges in a random order by hand. Simplify setup/finalization.
35284 (finalize_jump_threads): Remove.
35285 (vrp_finalize): Do not call identify_jump_threads here.
35286 (execute_vrp): Do it here instead and call thread_through_all_blocks
35289 PR tree-optimization/71437
35290 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
35292 (simplify_stmt_for_jump_threading): Add basic_block argument. All
35294 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
35295 (dom_opt_dom_walker::thread_across_edge): Remove
35296 handle_dominating_asserts argument. All callers changed.
35297 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
35298 changes. Remove calls to lhs_of_dominating_assert. Other
35299 uses of handle_dominating_asserts turn into unconditional code
35300 (simplify_control_stmt_condition_1): Likewise.
35301 (simplify_control_stmt_condition): Likewise.
35302 (thread_through_normal_block, thread_across_edge): Likewise.
35303 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
35304 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
35305 object if it is not an SSA_NAME.
35306 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
35307 before calling into the VRP specific simplifiers.
35308 (identify_jump_threads): Remove handle_dominating_asserts
35311 2017-03-16 Jakub Jelinek <jakub@redhat.com>
35314 * tree-diagnostic.c (default_tree_printer): No longer static.
35315 * tree-diagnostic.h (default_tree_printer): New prototype.
35317 2017-03-16 Tamar Christina <tamar.christina@arm.com>
35319 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
35320 Change ins into fmov.
35322 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35324 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
35325 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
35326 Use h_con constraint for operand 1.
35327 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
35328 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
35330 2017-03-15 Jeff Law <law@redhat.com>
35332 PR tree-optimization/71437
35333 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
35334 (record_temporary_equivalences): Use it.
35336 PR tree-optimization/71437
35337 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
35338 tree-ssa-scopedtables.
35339 (lookup_avail_expr, build_and_record_new_cond): Likewise.
35340 (record_conditions, record_cond, vuse_eq): Likewise.
35341 (record_edge_info): Adjust to API tweak of record_conditions.
35342 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
35343 (record_temporary_equivalences, optimize_stmt): Likewise.
35344 (eliminate_redundant_computations): Likewise.
35345 (record_equivalences_from_stmt): Likewise.
35346 * tree-ssa-scopedtables.c: Include options.h and params.h.
35347 (vuse_eq): New function, moved from tree-ssa-dom.c
35348 (build_and_record_new_cond): Likewise.
35349 (record_conditions): Likewise. Accept vector of conditions rather
35350 than edge_equivalence structure for first argument.
35351 for the first argument.
35352 (avail_exprs_stack::lookup_avail_expr): New member function, moved
35353 from tree-ssa-dom.c.
35354 (avail_exprs_stack::record_cond): Likewise.
35355 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
35356 from tree-ssa-dom.c.
35357 (avail_exprs_stack): Add new member functions lookup_avail_expr
35359 (record_conditions): Declare.
35361 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
35364 * lra-constraints.c (process_alt_operands): Increase reject for
35365 reloading an input/output operand.
35367 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
35370 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
35371 insns to convert from signed/unsigned char/short to IEEE 128-bit
35373 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
35375 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
35378 * config/i386/i386.c (ix86_vector_duplicate_value): Create
35379 subreg of inner mode for values already in registers.
35381 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
35383 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
35384 iteration reg is used after the loop.
35386 2017-03-14 Martin Sebor <msebor@redhat.com>
35388 PR tree-optimization/79800
35389 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
35390 precision in negative-positive range.
35391 (format_floating): Call non-const overload with adjusted precision.
35393 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
35396 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
35399 2017-03-14 Martin Sebor <msebor@redhat.com>
35401 PR middle-end/80020
35402 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
35403 * builtins.def (aligned_alloc): Use it.
35406 * Makefile.in (GTFILES): Add calls.c.
35407 * calls.c: Include "gt-calls.h".
35409 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
35411 PR rtl-optimization/79728
35412 * regs.h (struct target_regs): New field
35413 x_contains_allocatable_regs_of_mode.
35414 (contains_allocatable_regs_of_mode): New macro.
35415 * reginfo.c (init_reg_sets_1): Initialize it, and change
35416 contains_reg_of_mode so it includes global regs as well.
35417 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
35418 rather than contains_regs_of_mode.
35420 2017-03-14 Martin Liska <mliska@suse.cz>
35422 * doc/invoke.texi: Document options that can't be combined with
35423 -fcheck-pointer-bounds.
35425 2017-03-14 Martin Liska <mliska@suse.cz>
35427 PR middle-end/79831
35428 * doc/invoke.texi (-Wchkp): Document the option.
35430 2017-03-14 Martin Liska <mliska@suse.cz>
35432 * Makefile.in: Install gcov-dump.
35434 2017-03-14 Martin Liska <mliska@suse.cz>
35436 * multiple_target.c (expand_target_clones): Bail out for
35437 an invalid attribute.
35439 2017-03-14 Richard Biener <rguenther@suse.de>
35441 * alias.c (struct alias_set_entry): Pack properly.
35442 * cfgloop.h (struct loop): Likewise.
35443 * cse.c (struct set): Likewise.
35444 * ipa-utils.c (struct searchc_env): Likewise.
35445 * loop-invariant.c (struct invariant): Likewise.
35446 * lra-remat.c (struct cand): Likewise.
35447 * recog.c (struct change_t): Likewise.
35448 * rtl.h (struct address_info): Likewise.
35449 * symbol-summary.h (function_summary): Likewise.
35450 * tree-loop-distribution.c (struct partition): Likewise.
35451 * tree-object-size.c (struct object_size_info): Likewise.
35452 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
35453 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
35454 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
35455 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
35456 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
35457 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
35458 (struct _stmt_vec_info): Likewise.
35460 2017-03-14 Martin Liska <mliska@suse.cz>
35463 * multiple_target.c (create_dispatcher_calls): Check that
35464 a target can create a function dispatcher.
35466 2017-03-14 Martin Liska <mliska@suse.cz>
35469 * multiple_target.c (expand_target_clones): Drop local.local
35470 flag for default implementation.
35472 2017-03-14 Richard Biener <rguenther@suse.de>
35474 PR tree-optimization/80030
35475 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
35477 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
35479 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
35480 gcc_fallthrough() instead of __attribute__((fallthrough));
35482 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35484 * doc/gcc.texi: Remove "up" link to (DIR).
35485 * doc/gccint.texi: Ditto.
35487 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35489 * doc/install.texi (Specific) <avr>: Remove reference to
35492 2017-03-13 Jeff Law <law@redhat.com>
35494 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
35495 attribute rather than comments.
35497 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
35498 match_scratch operand is highest.
35500 2017-03-13 Martin Liska <mliska@suse.cz>
35502 PR middle-end/78339
35503 * ipa-pure-const.c (warn_function_noreturn): If the declarations
35504 is a CHKP clone, use original declaration.
35506 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35508 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
35509 (arc_conditional_register_usage): Use a different allocation order
35510 when optimizing for size.
35511 * common/config/arc/arc-common.c (arc_option_optimization_table):
35512 Section anchors default on when optimizing for size.
35514 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35516 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
35518 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35520 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
35521 * config/arc/arc.md (cpu_facility): Add cd variant.
35522 (*movqi_insn): Add code density variant.
35523 (*movhi_insn): Likewise.
35524 (*movqi_insn): Likewise.
35525 (*addsi3_mixed): Likewise.
35526 (subsi3_insn): Likewise.
35528 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35530 * config/arc/arc.md (movsi_cond_exec): Update constraint.
35532 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35534 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
35535 expressions with MINUS and UNARY ops.
35537 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35540 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
35542 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
35543 between vec_select and vector argument.
35544 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
35545 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
35546 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
35547 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
35548 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
35549 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
35551 2017-03-13 Richard Biener <rguenther@suse.de>
35554 * params.def (vect-max-peeling-for-alignment): Fix typo.
35556 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35558 * doc/install.texi (Specific) <mips-*-*>: Remove description of
35559 issue that only occurred with binutils below 2.18.
35561 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35563 * doc/install.texi (Specific) <cris-axis-elf>: No longer
35564 refer to binutils 2.11/2.12 minimum.
35566 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35568 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
35569 ftp.kernel.org and simplify binutils requirement.
35571 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
35573 * doc/invoke.texi (Warning Options): Fix spelling of link-time
35575 (Optimize Options): Ditto. Also remove redundancy.
35577 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35579 PR translation/79848
35580 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
35582 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
35583 to G_ to avoid double translation.
35585 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35587 PR translation/79923
35588 * auto-profile.c (get_combined_location): Convert leading
35589 character of diagnostics to lower case and remove trailing period.
35590 (read_profile): Likewise for various diagnostics.
35591 * config/arm/arm.c (arm_option_override): Remove trailing period
35592 from various diagnostics.
35593 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
35594 (msp430_expand_delay_cycles): Likewise.
35596 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35599 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
35600 full command-line argument, rather than just "str".
35601 (aarch64_validate_march): Likewise.
35602 (aarch64_validate_mtune): Likewise.
35604 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
35606 PR rtl-optimization/78911
35607 * lra-assigns.c (must_not_spill_p): New function.
35608 (spill_for): Use it.
35610 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35612 PR tree-optimization/79981
35613 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
35614 ATOMIC_COMPARE_EXCHANGE ifn result.
35615 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
35616 IFN_ATOMIC_COMPARE_EXCHANGE.
35618 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35621 * opts.c (parse_sanitizer_options): Add missing question mark to
35622 "did you mean" message.
35624 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35626 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
35628 (VMULEUH_UNS): Likewise.
35629 (VMULOUB_UNS): Likewise.
35630 (VMULOUH_UNS): Likewise.
35631 * config/rs6000/rs6000.c (builtin_function_type): Remove
35632 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
35634 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35637 * read-rtl-function.c (function_reader::read_rtx_operand): Update
35638 x with result of extra_parsing_for_operand_code_0.
35639 (function_reader::extra_parsing_for_operand_code_0): Convert
35640 return type from void to rtx, returning x. When reading
35641 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
35642 larger size containing struct block_symbol.
35644 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
35646 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
35647 -mfloat128-hardware without -m64.
35649 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
35652 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
35653 entries to the case statement that marks unsigned arguments to
35654 overloaded functions.
35656 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
35658 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
35659 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
35661 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
35664 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
35665 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
35667 2017-03-10 Martin Liska <mliska@suse.cz>
35671 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
35672 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
35675 2017-03-10 Olivier Hainque <hainque@adacore.com>
35677 * tree-switch-conversion (array_value_type): Start by resetting
35678 candidate type to it's main variant.
35680 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35682 PR rtl-optimization/79909
35683 * combine.c (try_combine): Use simplify_replace_rtx on individual
35684 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
35685 of the whole CALL_INSN_FUNCTION_USAGE.
35687 PR tree-optimization/79972
35688 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
35689 get_range_info on SSA_NAMEs. Formatting fixes.
35691 2017-03-10 Richard Biener <rguenther@suse.de>
35692 Jakub Jelinek <jakub@redhat.com>
35694 PR tree-optimization/77975
35695 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
35696 edge to be constant.
35697 (get_val_for): For constant x return it. Formatting fix.
35698 (loop_niter_by_eval): Avoid pointless looping if the next iteration
35699 would use the same bases as the current one.
35701 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35703 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
35704 instead of vec_select for V1TImode.
35705 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
35707 (VSX_LE_128): Add V1TI to this mode iterator.
35708 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
35709 (*vsx_le_perm_store_<mode>): Likewise.
35710 (pre-reload splitter for VSX stores): Likewise.
35711 (post-reload splitter for VSX stores): Likewise.
35712 (*vsx_xxpermdi2_le_<mode>): Likewise.
35713 (*vsx_lxvd2x2_le_<mode>): Likewise.
35714 (*vsx_stxvd2x2_le_<mode>): Likewise.
35716 2017-03-09 Michael Eager <eager@eagercon.com>
35718 Correct failures with --enable-checking=yes,rtl.
35720 * config/microblaze/microblaze.c (microblaze_expand_shift):
35721 Replace GET_CODE test with CONST_INT_P and INTVAL test with
35722 test for const0_rtx.
35723 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
35724 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
35726 2017-03-09 Richard Biener <rguenther@suse.de>
35728 PR tree-optimization/79977
35729 * graphite-scop-detection.c (scop_detection::merge_sese):
35730 Handle the case of extra exits to blocks dominating the entry.
35732 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
35734 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
35737 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
35739 PR rtl-optimization/79949
35740 * lra-constraints.c (process_alt_operands): Check memory when
35741 trying to predict a cycle. Print about the overall increase.
35743 2017-03-09 Richard Biener <rguenther@suse.de>
35745 PR middle-end/79971
35746 * gimple-expr.c (useless_type_conversion_p): Preserve
35747 TYPE_SATURATING for fixed-point types.
35749 2017-03-09 Richard Biener <rguenther@suse.de>
35752 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
35753 alignment of BLKmode params.
35755 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35758 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
35759 (VALL_NO_V2Q): Likewise.
35761 * config/aarch64/aarch64-simd.md
35762 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
35764 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
35765 VALL_NO_V2Q mode iterator.
35766 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
35768 2017-03-09 Martin Liska <mliska@suse.cz>
35770 PR tree-optimization/79631
35771 * tree-chkp-opt.c (chkp_is_constant_addr): Call
35772 tree_int_cst_sign_bit just for INTEGER constants.
35774 2017-03-09 Martin Liska <mliska@suse.cz>
35778 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
35781 2017-03-09 Marek Polacek <polacek@redhat.com>
35784 * tree.c (inchash::add_expr): Handle TREE_VEC.
35786 2017-03-09 Martin Liska <mliska@suse.cz>
35789 (chkp_narrow_size_and_offset): New function.
35790 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
35791 (void chkp_parse_bit_field_ref): New function.
35792 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
35793 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
35795 2017-03-09 Martin Liska <mliska@suse.cz>
35798 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
35799 (chkp_find_bounds_1): Remove gcc_unreachable.
35801 2017-03-09 Jakub Jelinek <jakub@redhat.com>
35804 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
35805 BUILT_IN_SYNC*, determine the access type from the size suffix and
35806 always build a MEM_REF with that type. Handle forgotten
35807 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
35810 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
35811 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
35812 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
35813 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
35814 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
35815 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
35816 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
35817 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
35818 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
35819 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
35820 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
35821 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
35822 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
35823 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
35824 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
35825 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
35826 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
35827 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
35828 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
35829 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
35830 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
35831 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
35832 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
35833 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
35834 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
35835 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
35836 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
35837 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
35838 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
35839 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
35840 definitions outside of __OPTIMIZE__ guarded section.
35843 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
35844 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
35845 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
35846 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
35849 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35851 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
35852 ("vfenez<mode>"): Add missing constraints.
35854 2017-03-08 Martin Sebor <msebor@redhat.com>
35857 * config/nds32/nds32.c (nds32_option_override):
35858 Fix misspelled diagnostic.
35860 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35863 * gimplify.c (gimplify_omp_for): Replace index var in outer
35864 taskloop statement with an artificial variable and add
35865 OMP_CLAUSE_PRIVATE clause for it.
35867 2017-03-08 Richard Biener <rguenther@suse.de>
35869 PR tree-optimization/79955
35870 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
35871 for accesses that are completely outside of the variable.
35873 2017-03-08 Andrew Haley <aph@redhat.com>
35875 PR tree-optimization/79943
35876 * tree-ssa-loop-split.c (compute_new_first_bound): When
35877 calculating the new upper bound, (END-BEG) should be added, not
35880 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35882 * config/avr/avr.md (setmemhi): Make sure match_dup
35883 operand number comes before match_scratch.
35885 2017-03-08 Richard Biener <rguenther@suse.de>
35887 PR tree-optimization/79920
35888 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
35889 with ncopies == 1 to ...
35890 (vect_transform_slp_perm_load): ... here. Properly compute
35891 all element loads by iterating VF times over the group. Do
35892 not handle ncopies (computed in a broken way) in
35893 vect_create_mask_and_perm.
35895 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35898 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
35899 is a uniform vector, use uniform_vector_p return value instead of
35900 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
35902 2017-03-07 Marek Polacek <polacek@redhat.com>
35904 PR middle-end/79809
35905 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
35906 (alloca_call_type): Likewise.
35908 2017-03-07 Martin Liska <mliska@suse.cz>
35910 * gcov.c (process_args): Put comment to correct location.
35912 2017-03-07 Martin Liska <mliska@suse.cz>
35914 PR middle-end/68270
35915 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
35916 Use array_at_struct_end_p instead of DECL_CHAIN (field).
35917 (chkp_narrow_bounds_for_field): Likewise.
35918 (chkp_parse_array_and_component_ref): Pass one more argument to
35921 2017-03-07 Richard Biener <rguenther@suse.de>
35923 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
35926 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
35928 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
35929 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
35931 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35934 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
35935 to end of description.
35936 (PARAM_MAX_STORES_TO_MERGE): Likewise.
35938 2017-03-07 Jakub Jelinek <jakub@redhat.com>
35940 PR rtl-optimization/79901
35941 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
35943 (*avx512f_<code><mode>3<mask_name>): ... this.
35944 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
35945 iterator instead of VI8_AVX2_AVX512BW.
35947 PR rtl-optimization/79901
35948 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
35949 min/max expander, expand it using expand_vec_cond_expr.
35952 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
35955 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35958 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
35959 to void * for PCH reasons.
35960 * dwarf2out.c (output_loc_operands, output_die): Cast
35961 v.val_vec.array to unsigned char *.
35963 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
35966 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
35969 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
35971 PR rtl-optimization/79571
35972 * lra-constraints.c (process_alt_operands): Calculate static
35973 reject and subtract it from overall when only addresses will be
35976 2017-03-06 Julia Koval <julia.koval@intel.com>
35979 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
35980 incoming stack boundary to 128 for 64-bit targets.
35982 2017-03-06 Richard Biener <rguenther@suse.de>
35984 PR tree-optimization/79894
35985 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
35986 to NULL after folding it.
35988 2017-03-06 Richard Biener <rguenther@suse.de>
35990 PR tree-optimization/79824
35991 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
35992 check disabling peeling for gaps.
35994 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
35996 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
35997 attributes): Document gettimeofday.
35999 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36001 * config/s390/s390.c (s390_option_override_internal): Set
36002 PARAM_MIN_VECT_LOOP_BOUND
36004 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36006 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
36007 * config/s390/s390.md: Likewise.
36009 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36012 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
36013 (<avx2_avx512>_perm<mode>): Rename to ...
36014 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
36016 (<avx512>_perm<mode>_mask): Rename to ...
36017 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
36019 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
36020 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
36021 instead of VI8F_256_512.
36022 (avx512f_perm<mode>): New define_expand.
36023 (avx512f_perm<mode>_mask): Likewise.
36024 (avx512f_perm<mode>_1<mask_name>): New define_insn.
36025 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
36027 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36029 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
36030 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
36032 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
36034 2017-03-06 Martin Liska <mliska@suse.cz>
36037 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
36038 when having a SSA NAME w/o VAR_DECL assigned to it.
36040 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36042 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
36043 msa_dpsub_<su>_d): Fix MODE for vec_select.
36045 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36047 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
36049 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
36051 2017-03-06 Richard Biener <rguenther@suse.de>
36053 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
36054 * plugin.c (register_plugin_info): Likewise.
36055 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
36057 2017-03-05 Jakub Jelinek <jakub@redhat.com>
36059 * config/i386/sse.md (sse_storehps, sse_storelps,
36060 avx_<castmode><avxsizesuffix>_<castmode>,
36061 avx512f_<castmode><avxsizesuffix>_<castmode>,
36062 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
36063 in condition that at least one operand is not a MEM.
36065 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36067 PR middle-end/79805
36068 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
36069 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
36071 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
36072 gimple_call_nothrow_p flag based on whether original builtin can throw.
36073 If it can, emit following stmts on the fallthrough edge.
36074 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
36075 don't create new bb if inserting just debug stmts on the edge, try to
36076 insert them on the fallthru bb or just reset debug stmts.
36078 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
36081 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
36082 restore recog_data (including the operand rtxes inside it) around
36083 the call to get_insn_template.
36085 2017-03-03 Martin Sebor <msebor@redhat.com>
36087 PR tree-optimization/79699
36088 * context.c (context::~context): Free MPFR caches to avoid
36089 a memory leak on program exit.
36091 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36093 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
36094 Use wide_int::ulow () instead of .elt (0).
36096 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36098 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
36099 (*pushxf): Limit oF constraint to 32bit targets and add oC
36100 constraint for 64bit targets.
36101 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
36102 (*pushdf): Change rmF constraint to rmC.
36104 2017-03-03 Martin Liska <mliska@suse.cz>
36106 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
36107 Remove unused variable.
36109 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36112 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
36113 is a memory operand, increase num_memory.
36114 (ix86_expand_args_builtin): Likewise.
36116 2017-03-03 Jan Hubicka <jh@suse.cz>
36119 * ipa-devirt.c (maybe_record_node): Properly handle
36120 __cxa_pure_virtual visibility.
36122 2017-03-03 Martin Liska <mliska@suse.cz>
36124 PR tree-optimization/79803
36125 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
36127 (pass_loop_prefetch::execute): Disabled optimization if an
36128 assumption about L1 cache size is not met.
36130 2017-03-03 Martin Liska <mliska@suse.cz>
36132 PR rtl-optimization/79574
36133 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
36134 (hash_scan_set): Likewise.
36135 (dump_hash_table): Likewise.
36136 (hoist_code): Likewise.
36138 2017-03-03 Richard Biener <rguenther@suse.de>
36140 * fixed-value.c (fixed_from_string): Restore use of elt (1)
36141 in place of uhigh ().
36142 (fixed_convert_from_real): Likewise.
36144 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36147 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
36149 2017-03-03 Richard Biener <rguenther@suse.de>
36151 PR middle-end/79818
36152 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
36153 TYPE_OVERFLOW_UNDEFINED check.
36155 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36157 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
36159 (vector_ae_<mode>_p): Likewise.
36160 (vector_nez_<mode>_p): Likewise.
36161 (vector_ne_v2di_p): Likewise.
36162 (vector_ae_v2di_p): Likewise.
36163 (vector_ne_<mode>_p): Likewise.
36164 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
36166 (vsx_tsqrt<mode>2_fe): Likewise.
36168 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
36171 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
36173 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36175 PR rtl-optimization/79780
36176 * cprop.c (one_cprop_pass): When second and further conditional trap
36177 in a single basic block is turned into an unconditional trap, turn it
36178 into a deleted note to avoid RTL verification failures.
36180 2017-03-02 Richard Biener <rguenther@suse.de>
36182 * fold-const.c (const_binop): Use ulow () instead of elt (0).
36184 2017-03-02 Richard Biener <rguenther@suse.de>
36186 PR tree-optimization/79345
36188 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
36189 param and abort the walk, returning -1 if it is hit.
36190 (walk_aliased_vdefs): Take a limit param and pass it on.
36191 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
36192 defaulting to 0 and return a signed int.
36193 * tree-ssa-uninit.c (struct check_defs_data): New struct.
36194 (check_defs): New helper.
36195 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
36196 about uninitialized memory.
36197 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
36198 bogus uninitialized warning.
36199 (fixed_convert_from_real): Likewise.
36201 2017-03-02 Bin Cheng <bin.cheng@arm.com>
36203 PR tree-optimization/66768
36204 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
36205 iv_use if base object can't be determined.
36207 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36209 PR tree-optimization/79345
36210 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
36211 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
36212 (get_pattern_stats): Initialize it.
36213 * genemit.c (gen_expand): Verify match_scratch numbers come after
36214 match_operand/match_dup numbers.
36215 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
36216 match_scratch numbers.
36217 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
36219 * config/s390/s390.md (trunctdsd2): Likewise.
36221 2017-03-02 Richard Biener <rguenther@suse.de>
36223 * wide-int.h (wide_int_storage::operator=): Implement in terms
36226 2017-03-02 Richard Biener <rguenther@suse.de>
36228 PR tree-optimization/79777
36229 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
36230 the to insert expression to sth existing.
36232 2017-03-01 Martin Sebor <msebor@redhat.com>
36234 PR middle-end/79692
36235 * gimple-ssa-sprintf.c
36236 (directive::known_width_and_precision): New function.
36237 (format_integer): Use it.
36238 (get_mpfr_format_length): Consider the full range of precision
36239 when computing %g output with the # flag. Set the likely byte
36240 count to 3 rather than 1 when precision is indeterminate.
36241 (format_floating): Correct the lower bound of precision.
36243 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36245 * doc/invoke.texi: Document default code model for 64-bit Linux.
36247 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36250 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
36251 udiv rather than div since input pattern is unsigned.
36253 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
36255 * config/i386/i386.c (print_reg): Warn for values of
36256 unsupported size in integer register.
36258 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
36261 * config/rs6000/predicates.md (current_file_function_operand): Do
36262 not allow self calls to be local if the function is replaceable.
36264 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
36267 * config/rs6000/altivec.h (vec_ctz and others): Change the
36268 preprocessor macro that controls conditional compilation from
36269 _ARCH_PWR9 to __POWER9_VECTOR__.
36270 (vec_all_ne): Change parameterization of __altivec_scalar_pred
36271 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
36272 control (instead of _ARCH_PWR9 control) so that template
36273 definition uses power9-specific function.
36274 (vec_any_eq): Likewise.
36275 (vec_all_ne): Change macro definition to use a power9-specific
36276 expansion under #ifdef __POWER9_VECTOR__ control (instead of
36277 _ARCH_PWR9 control).
36278 (vec_any_eq) Likewise.
36279 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
36280 expansion for CMPNEF to remove support for xvcmpnesp instruction.
36281 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
36282 support for xvcmpnedp instruction.
36283 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
36284 macro expansion so that Power9 implementation of vec_all_ne does
36285 not use the AltiVec predicate framework.
36286 (VCMPNEH_P): Likewise.
36287 (VCMPNEW_P): Likewise.
36288 (VCMPNED_P): Likewise.
36289 (VCMPNEFP_P): Likewise.
36290 (VCMPNEDP_P): Likewise.
36291 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
36292 implementation of vec_any_eq to not use AltiVec predicate
36294 (VCMPAEH_P): Likewise.
36295 (VCMPAEW_P): Likewise.
36296 (VCMPAED_P): Likewise.
36297 (VCMPAEFP_P): Likewise.
36298 (VCMPAEDP_P): Likewise.
36299 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
36300 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
36301 not use the AltiVec predicate framework.
36302 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
36303 of vec_any_eq to not use AltiVec predicate framework.
36304 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36305 support for predefined __POWER9_VECTOR__ macro to indicate that
36306 Power9 instruction selection is enabled.
36307 (altivec_overloaded_builtins): Remove extraneous
36308 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
36309 function argument types RS6000_BTI_bool_V16QI and
36310 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
36311 entry for overloaded function argument types RS6000_BTI_bool_V4SI
36312 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
36313 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
36314 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
36315 Power9 for implementations of vec_cmpne. Change the signature for
36316 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
36317 (representing vec_all_ne) to remove the previously described first
36318 argument of type RS6000_BTI_INTSI, as this was an artifact of
36319 reliance on the AltiVec predicate framework, which is no longer
36320 used in the implementation of these functions. Add
36321 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
36322 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
36323 since, unlike the AltiVec predicate framework implementation, we
36324 do not share function descriptors between vec_alle and vec_anyeq.
36325 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
36326 set of modes that receive special treatment even when
36327 TARGET_P9_VECTOR is true. The special treatment emits code that
36328 does not depend on Power9 instructions.
36329 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
36330 define_expand to not rely on AltiVec predicate framework.
36331 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36333 (vector_ne_v2di_p): Change this define_expand to not rely on
36334 AltiVec predicate framework.
36335 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
36337 (vector_ne_<mode>_p): Change this define_expand to not rely on
36338 AltiVec predicate framework.
36339 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36341 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
36342 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
36343 define_insn pattern.
36344 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
36345 define_insn pattern because the xvcmpne<VSs>. instruction is not
36347 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
36348 instruction is not supported.
36350 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36352 * config/nvptx/nvptx.c: Include intl.h.
36354 2017-03-01 Martin Jambor <mjambor@suse.cz>
36357 * ipa-prop.h (ipa_bits): Removed field known.
36358 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
36359 to pointers. Adjusted their comments to warn about their sharing.
36360 (ipcp_transformation_summary): Change bits to a vector of pointers.
36361 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
36362 (ipa_get_ipa_bits_for_value): Declare.
36363 * tree-vrp.h (value_range): Mark as GTY((for_user)).
36364 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
36365 (ipa_bits_hash_table): Likewise.
36366 (ipa_vr_ggc_hash_traits): Likewise.
36367 (ipa_vr_hash_table): Likewise.
36368 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
36369 being pointers and vr_known being removed.
36370 (ipa_set_jf_unknown): Likewise.
36371 (ipa_get_ipa_bits_for_value): New function.
36372 (ipa_set_jfunc_bits): Likewise.
36373 (ipa_get_value_range): New overloaded functions.
36374 (ipa_set_jfunc_vr): Likewise.
36375 (ipa_compute_jump_functions_for_edge): Use the above functions to
36376 construct bits and vr parts of jump functions.
36377 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
36378 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36380 (ipcp_grow_transformations_if_necessary): Also allocate
36381 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36383 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
36384 them. Fix too long lines.
36385 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
36386 vr_known being removed.
36387 (ipa_read_jump_function): Use new setter functions to construct bits
36388 and vr parts of jump functions or set them to NULL.
36389 (write_ipcp_transformation_info): Adjust for bits being pointers.
36390 (read_ipcp_transformation_info): Likewise.
36391 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
36393 Include gt-ipa-prop.h.
36394 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
36396 (ipcp_store_bits_results): Likewise.
36397 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
36398 Do not write to existing jump functions but use a temporary instead.
36400 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36403 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
36404 attempt to use its first operand as BIT_FIELD_REF base.
36406 2017-03-01 Richard Biener <rguenther@suse.de>
36408 PR middle-end/79721
36409 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
36410 interpolating formula in wrapping arithmetic.
36411 (chrec_apply): Convert chrec_evaluate return value to wanted type.
36413 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36415 PR tree-optimization/79734
36416 * tree-vect-generic.c (expand_vector_condition): Optimize
36417 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
36418 Handle VEC_COND_EXPR where comparison has different inner width from
36419 type's inner width.
36421 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
36423 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
36424 markup, and similar issues. Remove @opindex entries for things
36425 that aren't options. Add missing -mmpy-option entries.
36427 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36429 PR tree-optimization/79737
36430 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
36431 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
36432 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
36433 instead of byte_size. Formatting fix.
36434 (shift_bytes_in_array_right): Formatting fix.
36436 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
36439 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
36440 condition on optimize for the leaf function test.
36442 2017-02-28 Martin Liska <mliska@suse.cz>
36445 * read-rtl-function.c (function_reader::handle_unknown_directive):
36446 Bail out when one uses -flto.
36448 2017-02-28 Martin Liska <mliska@suse.cz>
36450 * common.opt: Replace space with tabular for options of <number>
36452 * config/i386/i386.opt: Show <number> value for
36453 -mlarge-data-threshold.
36454 * opts.c (print_filtered_help): Do not display number in hexadecimal
36457 2017-02-28 Martin Liska <mliska@suse.cz>
36459 * common.opt: Fix --help=option -Q for options which are of
36462 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
36464 * config/i386/i386.c (print_reg): Error out for values
36465 of 8-bit size in invalid integer register.
36467 2017-02-28 Martin Sebor <msebor@redhat.com>
36469 PR tree-optimization/79691
36470 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
36472 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36475 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
36476 gcc_unreachable with output_operand_lossage.
36478 2017-02-28 Richard Biener <rguenther@suse.de>
36480 PR tree-optimization/79740
36481 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
36483 (visit_nary_op): Insert the nary into the hashtable if we
36484 pattern-matched sth.
36485 * tree-ssa-pre.c (eliminate_insert): Robustify.
36487 2017-02-28 Richard Biener <rguenther@suse.de>
36489 PR middle-end/79731
36490 * fold-const.c (decode_field_reference): Reject out-of-bound
36493 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36495 * config/i386/i386.c: Include intl.h.
36496 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
36497 instead of just cond ? "..." : "...".
36498 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
36499 * coverage.c (read_counts_file): Likewise.
36500 * omp-offload.c: Include intl.h.
36501 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
36502 of just cond ? "..." : "...".
36503 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
36504 of just cond ? "..." : "...".
36506 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
36509 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
36511 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
36512 'tune for' CPU name.
36513 * config/arm/arm-cpu-data.h: Regenerated.
36515 2017-02-28 Richard Biener <rguenther@suse.de>
36517 PR tree-optimization/79732
36518 * tree-inline.c (expand_call_inline): Do not shadow var.
36520 2017-02-28 Richard Biener <rguenther@suse.de>
36522 PR tree-optimization/79723
36523 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
36524 address-space properly.
36526 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
36528 * doc/optinfo.texi (Optimization groups): Fix option used for
36530 * doc/invoke.texi (-fopt-info): Document "omp".
36531 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
36532 (OPTGROUP_ALL): Add OPTGROUP_OMP.
36533 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
36534 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
36535 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
36537 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
36539 * dumpfile.c (optgroup_options): Instead of "openmp", associate
36540 OPTGROUP_OMP with "omp".
36542 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
36545 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
36546 for arithmetic shift of unsigned V2DI.
36548 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
36550 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
36551 arc/linux.h headers.
36552 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
36553 (LINK_SPEC): Likewise.
36554 (ARC_TLS_EXTRA_START_SPEC): Likewise.
36555 (EXTRA_SPECS): Likewise.
36556 (STARTFILE_SPEC): Likewise.
36557 (ENDFILE_SPEC): Likewise.
36558 (LIB_SPEC): Likewise.
36559 (TARGET_SDATA_DEFAULT): Likewise.
36560 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
36561 (MULTILIB_DEFAULTS): Likewise.
36562 (DWARF2_UNWIND_INFO): Likewise.
36563 * config/arc/big.h: New file.
36564 * config/arc/elf.h: Likewise.
36565 * config/arc/linux.h: Likewise.
36566 * config/arc/t-uClibc: Remove.
36568 2017-02-27 Bin Cheng <bin.cheng@arm.com>
36570 PR tree-optimization/77536
36571 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
36572 (tree_transform_and_unroll_loop): Use above function to compute the
36573 estimated niter of unrolled loop and use it when scaling profile.
36574 Also use count info rather than frequency if it's non-zero.
36575 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
36576 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
36577 (vect_transform_loop): Call above function.
36579 2017-02-27 Richard Biener <rguenther@suse.de>
36581 PR tree-optimization/45397
36582 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
36583 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
36584 (visit_nary_op): Add pattern matching for CSEing sign-changed
36585 or truncated operations with wider ones.
36587 2017-02-27 Richard Biener <rguenther@suse.de>
36589 PR tree-optimization/79690
36590 * tree-vect-stmts.c (vectorizable_store): Use vector type
36591 built from the DR with address-space.
36593 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
36595 * doc/invoke.texi (Optimize Options): Refine the description
36596 of asan-use-after-return.
36598 2017-02-25 Alan Modra <amodra@gmail.com>
36600 PR rtl-optimization/79584
36601 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
36602 base, not ad->base_term, the reg within base. Remove assertion
36603 that ad->base == ad->base_term. Replace gen_int_mode using
36604 bogus mode with const0_rtx.
36606 2017-02-25 Jakub Jelinek <jakub@redhat.com>
36608 PR middle-end/79396
36609 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
36610 FMA_EXPR like tcc_binary or tcc_unary.
36612 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
36615 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
36617 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
36618 (output_loc_operands): Handle DW_OP_call_ref and
36619 DW_OP_GNU_variable_value.
36620 (struct variable_value_struct): New type.
36621 (struct variable_value_hasher): Likewise.
36622 (variable_value_hash): New variable.
36623 (string_types): Remove.
36624 (copy_loc_descr): New function.
36625 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
36626 (prepend_loc_descr_to_each): New function.
36627 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
36628 instead of add_loc_descr_to_each if the first argument is single
36629 location list and the second has multiple.
36630 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
36631 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
36632 when looking for variable value which doesn't have other location info.
36633 (loc_list_from_tree): Formatting fix.
36634 (gen_array_type_die): Simplify DW_AT_string_length handling.
36635 (adjust_string_types): Remove.
36636 (gen_subprogram_die): Don't call adjust_string_types nor test/set
36637 string_types. Call resolve_variable_values.
36638 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
36639 (resolve_addr_in_expr): Likewise. Add A argument.
36640 (copy_deref_exprloc): Remove deref argument. Adjust for the
36641 original expression being DW_OP_GNU_variable_value with optionally
36642 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
36643 optionally after it.
36644 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
36645 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
36646 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
36647 (variable_value_hasher::hash, variable_value_hasher::equal): New
36649 (resolve_variable_value_in_expr, resolve_variable_value,
36650 resolve_variable_values, note_variable_value_in_expr,
36651 note_variable_value): New functions.
36652 (dwarf2out_early_finish): Call note_variable_value on all toplevel
36655 2017-02-24 Jakub Jelinek <jakub@redhat.com>
36658 * opts.h (handle_generated_option): Add GENERATED_P argument.
36659 * opts-common.c (handle_option): Adjust function comment.
36660 (handle_generated_option): Add GENERATED_P argument, pass it to
36662 (control_warning_option): Pass false to handle_generated_option
36664 * opts.c (maybe_default_option): Pass true to handle_generated_option
36666 * optc-gen.awk: Likewise.
36668 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36670 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
36671 a REG, look at the REG it is a SUBREG of.
36672 (splitter for cmpeqsi_t): Ditto.
36674 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36676 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
36677 the special USEs with the pattern of the insn, not the insn itself.
36679 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
36682 * doc/invoke.texi: Document -mload-store-pairs.
36684 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36685 Sandra Loosemore <sandra@codesourcery.com>
36687 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
36688 argument isn't a CONST_INT.
36689 (nios2_alternate_compare_const): Assert op is a CONST_INT.
36690 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
36691 (nios2_validate_compare): Bypass alternate compare logic if *op2
36692 is not a CONST_INT.
36693 (ldstwm_operation_p): Return false if first_base is not a REG or
36694 if first_offset is not a CONST_INT.
36696 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36698 * config/cris/cris.md: Use correct operand in a define_peephole2.
36700 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36702 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
36704 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36706 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
36707 this_insn if it is an INSN or JUMP_INSN.
36708 (force_offsettable): Look at base, not at addr.
36709 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
36710 on things that aren't necessarily CONST_INTs.
36712 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
36714 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
36715 -mfpmath=sse is the default also for x86-32 targets with SSE2
36716 instruction set when @option{-ffast-math} is enabled
36718 2017-02-24 Jeff Law <law@redhat.com>
36720 PR rtl-optimizatoin/79286
36721 * ira.c (update_equiv_regs): Drop may_trap_p exception to
36724 2017-02-24 Richard Biener <rguenther@suse.de>
36726 PR tree-optimization/79389
36727 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
36730 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
36732 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
36733 function comment to reflect reality.
36734 (loop_exits_before_overflow): Fix typo in function description.
36736 2017-02-24 Richard Biener <rguenther@suse.de>
36738 PR tree-optimization/79389
36739 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
36740 properly that a threading opportunity exists. Detect conditional
36741 copy/constant propagation opportunities.
36743 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
36745 * config/visium/visium.md (type): Add trap.
36746 (b): New mode attribute.
36747 (*btst): Rename into...
36748 (*btst<mode>): ...this and adjust.
36749 (*cbranchsi4_btst_insn): Rename into...
36750 (*cbranch<mode>4_btst_insn): ...this and adjust.
36751 (trap): New define_insn.
36753 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36755 PR tree-optimization/79389
36756 * ifcvt.c (struct noce_if_info): Add rev_cond field.
36757 (noce_reversed_cond_code): New function.
36758 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
36759 reversed_comparison_code. Formatting fix.
36760 (noce_try_store_flag): Test rev_cond != NULL in addition to
36761 reversed_comparison_code.
36762 (noce_try_store_flag_constants): Likewise.
36763 (noce_try_store_flag_mask): Likewise.
36764 (noce_try_addcc): Use rev_cond if non-NULL instead of
36765 reversed_comparison_code.
36766 (noce_try_cmove_arith): Likewise. Formatting fixes.
36767 (noce_try_minmax, noce_try_abs): Clear rev_cond.
36768 (noce_find_if_block): Initialize rev_cond.
36769 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
36770 instead of false as last argument never attempt to reverse it
36773 2017-02-23 Bin Cheng <bin.cheng@arm.com>
36775 PR tree-optimization/79663
36776 * tree-predcom.c (combine_chains): Process refs in reverse order
36777 only for ZERO length chains, and add explaining comment.
36779 2017-02-23 Jeff Law <law@redhat.com>
36781 PR tree-optimization/79578
36782 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
36783 in call to operand_equal_p.
36785 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
36788 * config/i386/cpuid.h: Fix another undefined behavior.
36790 2017-02-23 Richard Biener <rguenther@suse.de>
36792 PR tree-optimization/79683
36793 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
36794 vector types for data-refs.
36796 2017-02-23 Martin Liska <mliska@suse.cz>
36798 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
36800 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36802 PR middle-end/79665
36803 * internal-fn.c (get_range_pos_neg): Moved to ...
36804 * tree.c (get_range_pos_neg): ... here. No longer static.
36805 * tree.h (get_range_pos_neg): New prototype.
36806 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
36807 are known to be in between 0 and signed maximum inclusive, try to
36808 expand both unsigned and signed divmod and use the cheaper one from
36811 2017-02-22 Jeff Law <law@redhat.com>
36813 PR tree-optimization/79578
36814 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
36815 to compare base operands.
36817 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
36820 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
36821 gpc_reg_operand instead of fpr_reg_operand.
36823 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
36825 * config/mips/mips.c (mips_return_in_memory): Force FP
36826 vector types to be returned in memory for o32 ABI.
36828 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36830 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
36831 instead of DW_TAG_member for static data member declarations and don't
36832 set no_linkage_name for static inline data members.
36833 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
36836 2017-02-22 Martin Liska <mliska@suse.cz>
36838 * doc/invoke.texi: Replace inequality signs with square brackets
36841 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
36844 * lra-constraints.c (simplify_operand_subreg): Handle
36845 WORD_REGISTER_OPERATIONS targets.
36847 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36850 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
36851 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
36852 elimination by swapping fld*.
36854 2017-02-22 Richard Biener <rguenther@suse.de>
36856 PR tree-optimization/79673
36857 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
36858 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
36859 irrelevant address-space qualifiers and avoiding a
36860 ADDR_SPACE_CONVERT_EXPR from fold_convert.
36862 2017-02-22 Richard Biener <rguenther@suse.de>
36864 PR tree-optimization/79666
36865 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
36866 to not symbolically negate if that may introduce undefined
36869 2017-02-22 Martin Liska <mliska@suse.cz>
36872 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
36873 * data-streamer-out.c (streamer_write_gcov_count_stream):
36875 * value-prof.c (stream_out_histogram_value): Make assert more
36876 precise based on type of counter.
36878 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
36881 * config/i386/i386.md (standard_x87sse_constant_load splitter):
36882 Use nonimmediate_operand instead of memory_operand for operand 1.
36883 (float-extend standard_x87sse_constant_load splitter): Ditto.
36885 2017-02-21 Jeff Law <law@redhat.com>
36887 PR tree-optimization/79621
36888 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
36889 blocks with edges to themselves.
36891 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36894 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
36895 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
36896 Use gimple_call_builtin_p.
36899 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
36900 on temporarily removed DEBUG_INSNs.
36902 PR tree-optimization/79649
36903 * tree-loop-distribution.c (classify_partition): Give up on
36904 non-generic address space loads/stores.
36906 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
36908 * doc/loop.texi (Loop manipulation): Remove nonexistent
36909 tree_ssa_loop_version from the documentation.
36910 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
36912 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36915 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
36916 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
36917 * config/rs6000/rs6000.c: Include except.h.
36918 (rs6000_expand_split_stack_prologue): Call
36919 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
36921 2017-02-21 Martin Jambor <mjambor@suse.cz>
36924 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
36925 have been analyzed.
36927 2017-02-21 Martin Jambor <mjambor@suse.cz>
36929 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
36930 for backward compatibility only.
36931 * doc/invoke.texi (Option Summary): Remove all references to
36932 -fipa-cp-alignment.
36934 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
36938 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36940 * lra-constraints.c (curr_insn_transform): Handle
36941 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36943 2017-02-21 Martin Liska <mliska@suse.cz>
36945 * config/i386/i386.opt: Replace -masm-dialect with -masm.
36947 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
36949 PR translation/79638
36950 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
36952 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
36955 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
36956 (arm_function_ok_for_sibcall): Return false for an indirect call by
36957 descriptor if all the argument registers are used.
36958 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
36959 alignment of the function.
36961 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36963 PR tree-optimization/61441
36964 * simplify-rtx.c (simplify_const_unary_operation): For
36965 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
36966 the sNaN unmodified.
36968 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
36970 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
36971 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
36972 instead of SYSTEM_HEADER_DIR.
36974 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
36975 Martin Liška <mliska@suse.cz>
36977 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
36978 Fix typos and grammar, use active voice, and clarify.
36980 2017-02-20 Marek Polacek <polacek@redhat.com>
36982 PR middle-end/79537
36983 * gimplify.c (gimplify_expr): Handle unused *&&L;.
36986 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
36988 2017-02-20 Jakub Jelinek <jakub@redhat.com>
36991 * config/i386/i386.c (ix86_expand_builtin): Handle
36992 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
36993 ix86_builtins_isa[fcode].isa as a requirement of those
36994 flags and any other flag in the bitmask.
36995 (ix86_init_mmx_sse_builtins): Use 0 instead of
36996 ~OPTION_MASK_ISA_64BIT as mask.
36997 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
36998 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
36999 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
37000 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
37002 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37005 * lra-constraints.c (split_reg): Check requested split mode
37006 is supported by the register.
37008 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37010 * lra-constraints.c (simplify_operand_subreg): Remove early
37013 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37016 * lra-constraints.c (curr_insn_transform): Tighten condition
37017 for converting SUBREG reloads from OP_OUT to OP_INOUT.
37019 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37022 * lra-constraints.c (curr_insn_transform): Handle
37023 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37025 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
37028 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
37030 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
37032 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
37035 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
37038 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37040 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
37043 2017-02-18 Jakub Jelinek <jakub@redhat.com>
37045 * final.c (last_columnnum, override_columnnum): New variables.
37046 (final_start_function): Set last_columnnum, pass it to begin_prologue
37047 hook and pass 0 to dwarf2out_begin_prologue.
37048 (final_scan_insn): Update override_columnnum. Pass last_columnnum
37049 to source_line debug hook.
37050 (notice_source_line): Compute last_columnnum and for debug_column_info
37051 return true on column changes.
37052 * debug.h (struct gcc_debug_hooks): Add column argument to
37053 source_line and begin_prologue hooks.
37054 (debug_nothing_int_charstar_int_bool): Remove prototype.
37055 (debug_nothing_int_int_charstar,
37056 debug_nothing_int_int_charstar_int_bool): New prototypes.
37057 (dwarf2out_begin_prologue): Add column argument.
37058 * debug.c (do_nothing_debug_hooks): Adjust source_line and
37059 begin_prologue hooks.
37060 (debug_nothing_int_charstar_int_bool): Remove.
37061 (debug_nothing_int_int_charstar,
37062 debug_nothing_int_int_charstar_int_bool): New functions.
37063 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
37064 through to dwarf2out_source_line.
37065 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
37066 (dwarf2out_source_line): Add column argument, emit it if requested.
37067 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
37069 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37070 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37071 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
37072 through to dwarf2out_begin_prologue.
37073 (vmsdbgout_source_line): Add column argument, pass it through to
37074 dwarf2out_source_line.
37075 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
37076 dbxout_source_line caller.
37077 (dbxout_source_line): Add column argument.
37079 * common.opt (gno-column-info, gcolumn-info): New options.
37080 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
37081 (check_die): Also test for multiple DW_AT_decl_column attributes.
37082 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
37083 DW_AT_decl_column if requested.
37084 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
37086 (gen_variable_die): Likewise.
37087 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
37088 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
37091 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
37092 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
37093 (ix86_handle_option): Handle OPT_m3dnowa.
37094 * doc/invoke.texi (-m3dnowa): Document.
37095 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
37096 -m3dnowa instead of -m3dnow -march=athlon.
37099 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
37100 instead of gcc_assert for K, r and R code checks. Formatting fixes.
37102 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37105 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
37106 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
37107 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
37108 generator for vsx_xxpermdi_<mode>_be.
37109 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
37110 force big-endian semantics.
37111 (vsx_xxpermdi_<mode>_be): New define_expand with same
37112 implementation as previous version of vsx_xxpermdi_<mode>.
37114 2017-02-17 Jakub Jelinek <jakub@redhat.com>
37116 PR tree-optimization/79327
37117 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
37118 variable, its initialization and use.
37120 2017-02-17 Julia Koval <julia.koval@intel.com>
37122 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
37123 (OPTION_MASK_ISA_PKU_UNSET): New.
37124 (ix86_handle_option): Handle -mrdpid.
37125 * config/i386/cpuid.h (bit_RDPID): New.
37126 * config/i386/driver-i386.c (host_detect_local_cpu):
37127 Detect RDPID feature.
37128 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
37129 * config/i386/i386-c.c (ix86_target_macros_internal):
37131 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
37132 (ix86_valid_target_attribute_inner_p): Add "rdpid".
37133 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
37134 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
37135 * config/i386/i386.md (define_insn "rdpid"): New.
37136 * config/i386/i386.opt Add -mrdpid.
37137 * config/i386/immintrin.h (_rdpid_u32): New.
37139 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
37141 PR rtl-optimization/79541
37142 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
37143 instead of transforming it into USE.
37145 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
37147 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
37148 If HONOR_SNANS (SFmode) force the input to a register.
37149 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
37150 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
37151 an frsp or similar insn.
37153 2017-02-17 Martin Liska <mliska@suse.cz>
37155 PR rtl-optimization/79577
37156 * params.def (selsched-max-sched-times): Increase minimum to 1.
37158 2017-02-17 Martin Liska <mliska@suse.cz>
37160 PR rtl-optimization/79574
37161 * gcse.c (want_to_gcse_p): Prevent integer overflow.
37163 2017-02-17 Martin Liska <mliska@suse.cz>
37165 PR tree-optimization/79529
37166 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
37167 ssa_defined_default_def_p to handle cases which are implicitly
37169 * tree-ssa.c (ssa_defined_default_def_p): New function.
37170 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
37171 which are implicitly defined.
37172 * tree-ssa.h (ssa_defined_default_def_p): Declare.
37174 2017-02-17 Richard Biener <rguenther@suse.de>
37176 PR middle-end/79576
37177 * params.def (max-ssa-name-query-depth): Limit to 10.
37179 2017-02-17 Richard Biener <rguenther@suse.de>
37181 PR tree-optimization/79552
37182 * tree-ssa-structalias.c (visit_loadstore): Properly verify
37185 2017-02-17 Richard Biener <rguenther@suse.de>
37188 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
37190 2017-02-17 Marek Polacek <polacek@redhat.com>
37192 PR middle-end/79536
37193 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
37194 (fold_negate_expr): New wrapper.
37196 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
37198 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
37199 Correct terminology and de-emphasize pre-standard behavior.
37201 2017-02-16 Alan Modra <amodra@gmail.com>
37203 PR rtl-optimization/79286
37204 * ira.c (def_dominates_uses): New function.
37205 (update_equiv_regs): Don't create an equivalence for insns that
37206 may trap where the register def does not dominate the use.
37208 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
37210 PR rtl-optimization/78127
37211 * lra.c (lra): Call lra_eliminate before finish the loop after
37214 2017-02-16 Richard Biener <rguenther@suse.de>
37216 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
37218 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
37219 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
37220 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
37221 (isl_val_int_from_wi): New function.
37222 (extract_affine_gmp): Rename to ...
37223 (extract_affine_wi): ... this, take a widest_int.
37224 (extract_affine_int): Just wrap extract_affine_wi.
37225 (add_param_constraints): Use isl_val_int_from_wi.
37226 (add_loop_constraints): Likewise, and extract_affine_wi.
37228 2017-02-15 Jeff Law <law@redhat.com>
37230 PR middle-end/79521
37231 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
37232 ira_init_register_move_cost_if_necessary.
37234 2017-02-15 Martin Sebor <msebor@redhat.com>
37236 PR middle-end/32003
37237 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
37238 removed in a prior commit.
37240 2017-02-15 Bin Cheng <bin.cheng@arm.com>
37242 PR tree-optimization/79347
37243 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
37244 counters during peeling.
37246 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
37248 * Makefile.in (site.exp): Remove "set ISLVER".
37250 2017-02-15 Jakub Jelinek <jakub@redhat.com>
37253 * real.c (real_from_integer): Call real_convert even for decimal.
37255 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
37258 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
37260 2017-02-14 Andrew Pinski <apinski@cavium.com>
37262 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
37263 cores and change the partno/implementer to be correct.
37264 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
37265 the 'B" as the implementer.
37266 * config/aarch64/aarch64-tune.md: Regenerate.
37268 2017-02-14 Carl Love <cel@us.ibm.com>
37270 * config/rs6000/rs6000.c: Add case statement entry to make the
37271 xvcvuxdsp built-in argument unsigned.
37272 * config/rs6000/vsx.md: Fix the source and return operand types so they
37273 match the instruction definitions from the ISA document. Fix typo
37274 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
37277 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
37280 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
37281 member early_clobber_alts.
37282 * lra-lives.c (reg_early_clobber_p): New.
37283 (process_bb_lives): Use it.
37284 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
37285 (debug_operand_data): Initialize early_clobber_alts.
37286 (setup_operand_alternative): Set up early_clobber_alts.
37287 (collect_non_operand_hard_regs): Ditto. Pass early clobber
37288 alternatives to new_insn_reg.
37289 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
37291 (lra_update_insn_regno_info): Pass the new arg.
37293 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37295 PR middle-end/79505
37296 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
37297 (new_oacc_loop_raw): Don't clear already cleared fields.
37300 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
37301 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
37302 _mm512_prefetch_i64gather_ps): New inline functions and macros.
37304 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
37307 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
37309 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
37312 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
37313 the extra instruction to the right place to store 128-bit constant
37316 2017-02-14 Martin Sebor <msebor@redhat.com>
37318 PR middle-end/79448
37319 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
37320 warning for strings of unknown length.
37322 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
37324 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
37326 2017-02-14 Jeff Law <law@redhat.com>
37329 * ira-costs.c (scan_one_insn): Initialize register move costs
37330 for pseudos seen in USE/CLOBBER insns.
37332 PR tree-optimization/79095
37333 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
37334 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
37335 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
37336 if the operands are known to be not equal, then the resulting range
37338 (intersect_ranges): If the new range is ~[0,0] and the old range is
37339 wide, then prefer ~[0,0].
37340 * tree-vrp.c (overflow_comparison_p_1): New function.
37341 (overflow_comparison_p): New function.
37342 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
37343 if NAME is used in an overflow test.
37344 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
37345 overflow check that can be expressed as an equality test, then adjust
37346 ops to be that equality test.
37348 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37350 * config/s390/s390-builtin-types.def: Remove flags argument.
37351 * config/s390/s390.c (s390_init_builtins): Likewise.
37353 2017-02-14 Martin Liska <mliska@suse.cz>
37355 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
37356 vector. Fix trailing white spaces.
37358 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
37360 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
37363 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37365 PR rtl-optimization/68664
37366 * config/arm/arm.c (arm_sched_can_speculate_insn):
37367 New function. Declare prototype.
37368 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37370 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37372 PR rtl-optimization/68664
37373 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
37375 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37377 2017-02-14 Amit Pawar <amit.pawar@amd.com>
37379 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
37380 max skip bytes for function, loop and jump.
37382 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37384 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
37385 ABS_EXPR for gimple dump.
37387 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37390 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
37392 PR tree-optimization/79408
37393 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
37394 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
37395 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
37396 also if rhs1 is INTEGER_CST.
37398 2017-02-14 Richard Biener <rguenther@suse.de>
37400 PR middle-end/79432
37401 * tree-into-ssa.c (insert_phi_nodes): When the function can
37402 have abnormal edges rewrite SSA names with broken use-def
37403 dominance out of SSA and register them for PHI insertion.
37405 2017-02-13 Martin Sebor <msebor@redhat.com>
37407 PR middle-end/79496
37408 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
37409 clearing info.nowrite flag when snprintf size argument is a range.
37411 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37413 * cprop.c (cprop_jump): Add missing space in string literal.
37414 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
37415 (get_constraint_for_component_ref): Likewise.
37416 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
37417 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
37418 * lra-constraints.c (process_alt_operands): Likewise.
37419 * ipa-inline.c (inline_small_functions): Likewise.
37420 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
37421 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
37422 * trans-mem.c (diagnose_tm_1_op): Likewise.
37423 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
37424 (grid_parallel_clauses_gridifiable): Likewise.
37426 * config/nvptx/mkoffload.c (process): Add space in between
37429 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
37430 "MOD4_SSE_REGS" and "ALL_REGS".
37432 * spellcheck.c (test_data): Add , in between "foo" and "food".
37434 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37437 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
37438 boundary crossing check and subsequent code generation agree.
37440 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37442 * config/aarch64/aarch64.c (has_memory_op): Delete.
37443 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
37446 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37448 PR rtl-optimization/79388
37449 PR rtl-optimization/79450
37450 * combine.c (distribute_notes): When removing TEM_INSN for which
37451 corresponding dest has last value recorded, invalidate that last
37454 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37456 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
37457 of explicit '@'. Add missing assembly comment marker on branch costs
37460 2017-02-13 Nathan Sidwell <nathan@acm.org>
37462 * gengtype-lex.l (<in_struct>): Add '/'.
37464 2017-02-13 Martin Liska <mliska@suse.cz>
37467 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
37469 2017-02-13 Richard Biener <rguenther@suse.de>
37471 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
37473 * configure: Re-generate.
37474 * config.in: Likewise.
37475 * graphite-dependences.c: Simplify as if
37476 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
37477 * graphite-isl-ast-to-gimple.c: Likewise.
37478 * graphite-optimize-isl.c: Likewise.
37479 * graphite-poly.c: Likewise.
37480 * graphite-sese-to-poly.c: Likewise.
37481 * graphite.h: Likewise.
37482 * toplev.c: Include isl/version.h and use isl_version () for
37483 printing the ISL version.
37484 * doc/install.texi: Update ISL requirement.
37486 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37488 * doc/standards.texi (Standards): Update reference to
37491 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37493 * doc/extend.texi (Named Address Spaces): sourceware.org now
37495 * doc/install.texi (Binaries): Ditto.
37498 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37500 * doc/cpp.texi: Replace "stringify"/"stringification" with C
37501 standard terminology "stringize"/"stringizing" throughout.
37502 * doc/cppinternals.texi: Likewise.
37504 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37506 * doc/extend.texi: Fix some spelling mistakes and typos.
37507 * doc/invoke.texi: Likewise.
37509 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37512 * params.def (inline-min-speedup) Change from 10 to 8.
37514 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37516 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
37519 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37522 * ipa-inline-analysis.c (get_minimal_bb): New function.
37523 (record_modified): Use it.
37524 (remap_edge_change_prob): Handle also ancestor functions.
37526 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
37528 * doc/contrib.texi (Contributors): Remove broken link into
37529 the Mauve CVS repository.
37531 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37533 PR middle-end/79454
37534 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
37535 result computation whenever lhs doesn't have vector mode, not
37536 just when it has BLKmode.
37538 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37540 * doc/makefile.texi (profiledbootstrap): Refer to the
37541 installation instructions only in textual form.
37543 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37546 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
37548 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37550 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
37551 (Specific): Update mingw-w64 reference.
37553 (Specific): Remove broken link to Renesas RX processor.
37555 2017-02-10 Richard Biener <rguenther@suse.de>
37557 * toplev.c (process_options): Do not mention obsolete graphite
37558 options when printing sorry message about missing graphite support.
37559 Mention -floop-nest-optimize.
37561 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
37563 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
37564 (vtst_p16): Likewise.
37565 (vtstq_p8): Likewise.
37566 (vtstq_p16): Likewise.
37568 (vtstq_p64): Likewise.
37569 * config/arm/arm_neon.h (vgetq_lane_p64): New.
37570 (vset_lane_p64): New.
37571 (vsetq_lane_p64): New.
37573 2017-02-10 Jakub Jelinek <jakub@redhat.com>
37575 PR tree-optimization/79411
37576 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
37577 stmt operands are SSA_NAMEs used in abnormal phis.
37578 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
37581 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37584 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
37587 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37589 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
37591 2017-02-09 Jakub Jelinek <jakub@redhat.com>
37593 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
37597 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
37598 not arbitrary TREE_CONSTANT.
37601 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
37602 "omp declare target link" attribute unless is_global_var.
37603 * omp-offload.c (find_link_var_op): Likewise.
37605 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
37606 Chung-Lin Tang <cltang@codesourcery.com>
37608 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
37610 (gimplify_adjust_omp_clauses): Don't delete TILE.
37611 (gimplify_omp_for): Deal with TILE.
37612 * internal-fn.c (expand_GOACC_TILE): New function.
37613 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
37615 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
37616 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
37618 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
37619 avoid DIV for outermost collapse var.
37620 (expand_oacc_for): Insert tile element loop as needed. Adjust.
37621 Remove out of date comments, fix whitespace.
37622 * omp-general.c (omp_extract_for_data): Deal with tiling.
37623 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
37624 adjust OLF_DIM_BASE value.
37625 (struct omp_for_data): Add tiling field.
37626 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
37627 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
37628 for auto loops. Remove default auto determining, moved to
37629 oacc_loop_fixed_partitions.
37630 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
37631 stmts, add e_mask field.
37632 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
37633 (oacc_thread_numbers): Use oacc_dim_call.
37634 (oacc_xform_tile): New.
37635 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
37636 (finish_oacc_loop): Adjust for ifns vector.
37637 (oacc_loop_discover_walk): Append loop abstraction sites to list,
37638 add case for GOACC_TILE fns.
37639 (oacc_loop_xform_loop): Delete.
37640 (oacc_loop_process): Iterate over call list directly, and add
37641 handling for GOACC_TILE fns.
37642 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
37644 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
37645 vector partitioning to outer loops. Assign 2 partitions to loops
37646 when available. Add TILE handling.
37647 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
37648 (execite_oacc_device_lower): Process GOACC_TILE fns,
37649 ignore unknown specs.
37650 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
37651 * tree.c (omp_clause_num_ops): Adjust TILE ops.
37652 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
37654 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
37656 * configure.ac (ACX_BUGURL): Update.
37657 * configure: Regenerate.
37659 2017-02-09 Richard Biener <rguenther@suse.de>
37661 PR tree-optimization/69823
37662 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
37663 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
37665 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37667 * config/arc/arc-c.def: Add __NPS400__ definition.
37668 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
37669 (TARGET_NPS400): Define.
37671 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37673 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
37675 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
37676 pointer, arch_info.
37677 (arc_cpu_types): Fill the arch_info field with a pointer into the
37678 arc_arch_types table.
37679 (arc_selected_cpu): Declare.
37680 * config/arc/arc.c (arc_selected_cpu): Make global.
37681 (arc_selected_arch): Delete.
37682 (arc_base_cpu): Delete.
37683 (arc_override_options): Remove references to deleted variables,
37684 update access to arch information.
37685 (ARC_OPT): Update access to arch information.
37686 (ARC_OPTX): Likewise.
37687 * config/arc/arc.h (arc_base_cpu): Remove declaration.
37688 (TARGET_ARC600): Update access to arch information.
37689 (TARGET_ARC601): Likewise.
37690 (TARGET_ARC700): Likewise.
37691 (TARGET_EM): Likewise.
37692 (TARGET_HS): Likewise.
37693 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
37696 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
37699 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
37700 condition/operands for integer GE/LE/GEU/LEU operations.
37702 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
37704 PR translation/79397
37705 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
37708 2017-02-08 Martin Jambor <mjambor@suse.cz>
37711 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
37712 whether allocation happened.
37713 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
37714 nothing was allocated.
37716 2017-02-08 Jakub Jelinek <jakub@redhat.com>
37718 PR tree-optimization/79408
37719 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
37720 constant, but SSA_NAME with a known integer range, use the minimum
37721 of that range instead of op1 to determine if modulo can be replaced
37722 with its first operand.
37724 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37726 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
37728 2017-02-08 Richard Biener <rguenther@suse.de>
37730 PR tree-optimization/71824
37731 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37732 Check all loops contained in the merged region.
37734 2017-02-07 Andrew Pinski <apinski@cavium.com>
37736 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
37738 2017-02-07 Andrew Pinski <apinski@cavium.com>
37740 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
37741 (thunderxt88): Likewise.
37742 (thunderxt81): Disable LSE and change v8.1 to v8.
37743 (thunderxt83): Likewise.
37745 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37746 Richard Biener <rguenther@suse.de>
37748 PR middle-end/79399
37749 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
37750 type from int to size_t.
37751 * ira-costs.c (struct_costs_size): Change type from int to size_t.
37753 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37755 PR rtl-optimization/79386
37756 * cprop.c (bypass_conditional_jumps): Initialize
37757 bypass_last_basic_block already before splitting bbs after
37758 unconditional traps...
37759 (bypass_conditional_jumps): ... rather than here.
37762 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
37763 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
37764 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
37765 fix -masm=intel patterns.
37767 2017-02-07 Richard Biener <rguenther@suse.de>
37769 PR tree-optimization/79256
37770 PR middle-end/79278
37771 * builtins.c (get_object_alignment_2): Use min_align_of_type
37772 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
37773 and ADJUST_FIELD_ALIGN.
37775 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
37777 * doc/tm.texi: Regenerate.
37778 * stor-layout.c (layout_decl): Adjust.
37779 (update_alignment_for_field): Likewise.
37780 (place_field): Likewise.
37781 (min_align_of_type): Likewise.
37782 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
37783 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
37784 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
37785 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
37786 * config/frv/frv.c (frv_adjust_field_align): Likewise.
37787 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
37788 * config/i386/i386.c (x86_field_alignment): Likewise.
37789 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
37790 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
37791 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
37792 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
37793 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
37794 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
37798 2017-01-30 Richard Biener <rguenther@suse.de>
37800 PR tree-optimization/79256
37801 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37802 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37805 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
37807 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
37808 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
37809 builtins to SImode and emit a zero-extend, if necessary.
37811 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37813 * docs/invoke.texi (RISC-V Options): Alphabetize.
37815 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37817 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
37820 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37822 * config/riscv/riscv.c: New file.
37823 * common/config/riscv/riscv-common.c: Likewise.
37824 * config.gcc: Likewise.
37825 * config/riscv/constraints.md: Likewise.
37826 * config/riscv/elf.h: Likewise.
37827 * config/riscv/generic.md: Likewise.
37828 * config/riscv/linux.h: Likewise.
37829 * config/riscv/multilib-generator: Likewise.
37830 * config/riscv/peephole.md: Likewise.
37831 * config/riscv/pic.md: Likewise.
37832 * config/riscv/predicates.md: Likewise.
37833 * config/riscv/riscv-builtins.c: Likewise.
37834 * config/riscv/riscv-c.c: Likewise.
37835 * config/riscv/riscv-ftypes.def: Likewise.
37836 * config/riscv/riscv-modes.def: Likewise.
37837 * config/riscv/riscv-opts.h: Likewise.
37838 * config/riscv/riscv-protos.h: Likewise.
37839 * config/riscv/riscv.h: Likewise.
37840 * config/riscv/riscv.md: Likewise.
37841 * config/riscv/riscv.opt: Likewise.
37842 * config/riscv/sync.md: Likewise.
37843 * config/riscv/t-elf-multilib: Likewise.
37844 * config/riscv/t-linux: Likewise.
37845 * config/riscv/t-linux-multilib: Likewise.
37846 * config/riscv/t-riscv: Likewise.
37847 * configure.ac: Likewise.
37848 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
37849 Waterman as RISC-V maintainers.
37850 * doc/install.texi: Add RISC-V entries.
37851 * doc/invoke.texi: Add RISC-V options section.
37852 * doc/md.texi: Add RISC-V constraints section.
37853 * configure: Regenerated.
37855 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
37858 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
37859 false values to be constant vectors with all 0 or all 1 bits set.
37860 (vcondu<mode><mode>): Likewise.
37861 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
37863 (fpmask_comparison_operator): Update comment.
37864 (vecint_comparison_operator): New predicate.
37865 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
37866 vector conditionals when the true and false values are constant
37867 vectors with all 0 bits or all 1 bits set.
37869 2017-02-06 Martin Sebor <msebor@redhat.com>
37871 PR tree-optimization/79376
37872 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
37874 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
37876 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
37877 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
37878 to simplify split condition.
37880 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37882 * omp-expand.c (oxpand_omp_atomic_fetch_op,
37883 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
37886 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
37888 PR rtl-optimization/68664
37889 * target.def (can_speculate_insn): New hook.
37890 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
37891 * doc/tm.texi: Regenerate.
37892 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
37893 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
37894 (rs6000_sched_can_speculate_insn): New function.
37896 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37898 PR tree-optimization/79284
37899 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
37900 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
37901 vectorizable_mask_load_store, vectorizable_operation,
37902 vect_is_simple_cond, get_same_sized_vectype): Use it instead
37903 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
37904 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
37905 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
37906 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
37907 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
37908 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
37909 is_gimple_assign (stmt). Replace another such test with
37910 is_gimple_assign (stmt).
37912 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37915 * config/avr/avr.c (rtl-iter.h): Include it.
37916 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
37917 (avr_legitimate_combined_insn): ...and implementation.
37919 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
37921 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
37922 * config/s390/s390.c (s390_const_operand_ok)
37923 (s390_canonicalize_comparison, s390_extract_part)
37924 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
37925 (s390_contiguous_bitmask_p, s390_rtx_costs)
37926 (legitimize_pic_address): Likewise.
37927 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
37928 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
37929 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
37930 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
37931 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
37933 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37935 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
37936 REGNO($0) == REGNO($1).
37938 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37940 * config/s390/linux.h(SIZE_TYPE): Add comment.
37942 2017-02-06 Julian Brown <julian@codesourcery.com>
37943 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37944 Virendra Pathak <virendra.pathak@broadcom.com>
37946 * config/aarch64/aarch64-cores.def: Change the scheduler
37948 * config/aarch64/aarch64.md: Include thunderx2t99.md.
37949 * config/aarch64/thunderx2t99.md: New file.
37951 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37953 * doc/standards.texi (Go Language): Update link to language
37956 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37958 * tree-eh.c (lower_resx): Sanitize profile.
37959 (cleanup_empty_eh_move_lp): Likewise.
37961 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37964 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
37966 * cfgloopmanip.h (loop_version): Update prototype.
37967 * modulo-sched.c (sms_schedule): Update call of loop_version.
37968 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
37969 * tree-parloops.c (gen_parallel_loop): Likewise.
37970 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
37971 * tree-ssa-loop-split.c (split_loop): Likewise.
37972 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
37973 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
37975 2017-02-05 Martin Liska <mliska@suse.cz>
37978 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
37980 (print_operand_address): Initialize a struct to zero.
37982 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37984 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
37985 garbage collector only in textual form.
37987 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37989 * doc/extend.texi (x86 specific memory model extensions for
37990 transactional memory): Simplify a phrase.
37992 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
37995 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
37996 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
37997 (atomic_storedi_1): Likewise.
37999 2017-02-04 Jakub Jelinek <jakub@redhat.com>
38001 PR tree-optimization/79338
38002 * tree-parloops.c (gather_scalar_reductions): Don't call
38003 vect_analyze_loop_form for loop->inner before destroying loop's
38006 2017-02-03 Martin Sebor <msebor@redhat.com>
38008 PR tree-optimization/79327
38009 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
38010 when precision has resulted in leading zeros.
38011 (format_integer): Adjust the likely counter to assume an unknown
38012 argument that may be zero is non-zero.
38014 2017-02-03 Jason Merrill <jason@redhat.com>
38017 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
38018 avoid copying non-taken branch.
38020 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38022 PR tree-optimization/79340
38023 * tree-vect-loop.c (vectorizable_reduction): Release
38024 vec_defs elements after safe_splicing them into other vectors.
38027 PR tree-optimization/79327
38028 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
38029 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
38031 (format_integer): Use wide_int_to_tree instead of build_int_cst
38032 + to_?hwi. If argmin is NULL, just set argmin and argmax to
38033 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
38034 of shortest and longest sequence.
38036 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
38038 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
38039 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
38041 2017-02-03 Walter Lee <walt@tilera.com>
38044 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
38045 after initial stackframe link reg save.
38046 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
38048 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38051 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
38052 wu for stxssp alternative.
38054 2017-02-03 Martin Sebor <msebor@redhat.com>
38056 PR tree-optimization/79352
38057 * gimple-fold.c (get_range_strlen): Add argument.
38058 (get_range_strlen): Change return type to bool.
38059 (get_maxval_strlen): Pass in a dummy argument.
38060 * gimple-fold.h (get_range_strlen): Change return type to bool.
38061 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
38062 * tree.h (array_at_struct_end_p): Add argument.
38063 * tree.c (array_at_struct_end_p): Handle it.
38065 2017-02-03 Martin Liska <mliska@suse.cz>
38068 * multiple_target.c (create_dispatcher_calls): Redirect edge
38069 from a caller of a dispatcher.
38070 (expand_target_clones): Make the clones local.
38071 (ipa_target_clone): Do both target clones and resolvers.
38072 (ipa_dispatcher_calls): Remove the pass.
38073 (pass_dispatcher_calls::gate): Likewise.
38074 (make_pass_dispatcher_calls): Likewise.
38075 * passes.def (pass_target_clone): Put as very first IPA early
38078 2017-02-03 Martin Liska <mliska@suse.cz>
38080 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
38081 in case of a function with ifunc attribute.
38083 2017-02-03 Martin Liska <mliska@suse.cz>
38085 * cgraph.c (cgraph_node::dump): Dump function version info.
38086 * symtab.c (symtab_node::dump_base): Add missing new line.
38088 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38090 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
38091 (ifcombine_ifandif): Use it.
38093 2017-02-03 Martin Liska <mliska@suse.cz>
38095 * doc/invoke.texi: Document default value for
38096 use-after-scope-direct-emission-threshold.
38098 2017-02-03 Martin Liska <mliska@suse.cz>
38100 PR tree-optimization/79339
38101 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
38102 (format_floating): Likewise.
38104 2017-02-03 Martin Liska <mliska@suse.cz>
38107 * ipa-prop.c (ipa_node_params_t::insert): Remove current
38109 (ipa_node_params_t::remove): Likewise.
38110 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
38111 initialization from removed ipa_node_params_t::insert.
38112 (ipa_node_params::~ipa_node_params): Move from removed
38113 ipa_node_params_t::release.
38114 * symbol-summary.h (symbol_summary::m_released): New member.
38115 Do not release a summary twice. Do not allow to call finalizer
38116 for types of a summary that live in GGC memory.
38118 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38120 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
38123 2017-02-02 Martin Sebor <msebor@redhat.com>
38125 PR middle-end/79275
38126 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
38127 (format_string): Tighten up the range of output for non-constant
38128 strings and correct the expected range for wide non-constant strings.
38130 2017-02-02 Martin Sebor <msebor@redhat.com>
38132 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
38134 PR middle-end/32003
38135 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
38137 (-fdump-tree-@var): Add to index and document how to come up
38138 with pass-specific option and dump file names.
38139 (-fdump-passes): Clarify where to look for output.
38141 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38143 PR middle-end/77445
38144 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
38145 statistics of the analyzed path; allow threading for speed when
38146 any of BBs along the path are optimized for speed.
38148 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
38150 PR middle-end/78468
38151 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
38152 settings of the virtual registers.
38155 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
38157 * explow.c (get_dynamic_stack_size): Take known alignment of stack
38158 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
38161 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38163 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
38164 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
38166 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38168 * config/s390/s390.md: Add missing comments with the expanded
38170 * config/s390/vector.md: Likewise.
38171 * config/s390/vx-builtins.md: Likewise.
38173 2017-02-02 Jakub Jelinek <jakub@redhat.com>
38176 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
38177 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
38178 conditions on a single line.
38180 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38182 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38183 __S390_VX__ to __VX__.
38185 2017-02-01 Andrew Pinski <apinski@cavium.com>
38187 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
38188 stmt_info to record_stmt_cost.
38189 (vect_get_known_peeling_cost): Pass stmt_info if known to
38191 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
38192 cpu_vector_cost field into
38193 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
38194 field into vec_int_stmt_cost and vec_fp_stmt_cost.
38195 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
38196 splitting of scalar_stmt_cost and vec_stmt_cost.
38197 (thunderx_vector_cost): Likewise.
38198 (cortexa57_vector_cost): LIkewise.
38199 (exynosm1_vector_cost): Likewise.
38200 (xgene1_vector_cost): Likewise.
38201 (thunderx2t99_vector_cost): Improve after the splitting of the two
38203 (aarch64_builtin_vectorization_cost): Update for the splitting of
38204 scalar_stmt_cost and vec_stmt_cost.
38206 2017-02-01 Torvald Riegel <triegel@redhat.com>
38207 Richard Henderson <rth@redhat.com>
38209 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
38210 conditional on existance of a fast atomic load.
38211 * optabs-query.c (can_atomic_load_p): New function.
38212 * optabs-query.h (can_atomic_load_p): Declare it.
38213 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
38214 no fast atomic load is available for the particular size of access.
38215 (expand_atomic_compare_and_swap): Likewise.
38216 (expand_atomic_load): Likewise.
38217 (expand_atomic_store): Likewise.
38218 (expand_atomic_fetch_op): Likewise.
38219 * testsuite/lib/target-supports.exp
38220 (check_effective_target_sync_int_128): Remove x86 because it provides
38221 no fast atomic load.
38222 (check_effective_target_sync_int_128_runtime): Likewise.
38224 2017-02-01 Richard Biener <rguenther@suse.de>
38226 * graphite.c: Include tree-vectorizer.h for find_loop_location.
38227 (graphite_transform_loops): Provide opt-info for optimized nests.
38228 * tree-parloop.c (parallelize_loops): Provide opt-info for
38229 parallelized loops.
38231 2017-02-01 Richard Biener <rguenther@suse.de>
38233 PR middle-end/79315
38234 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
38235 was not set before.
38237 2017-02-01 Richard Biener <rguenther@suse.de>
38239 PR tree-optimization/71824
38240 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38241 Verify the loops are valid in the merged SESE region.
38242 (scop_detection::can_represent_loop_1): Check analyzing the
38243 evolution of the number of iterations in the region succeeds.
38245 2017-01-31 Ian Lance Taylor <iant@golang.org>
38247 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
38248 REG_ARGS_SIZE note to 32-bit push insns and call insn.
38250 2017-01-31 David Malcolm <dmalcolm@redhat.com>
38252 PR preprocessor/79210
38253 * input.c (get_substring_ranges_for_loc): Replace line_width
38254 assertion with error-handling.
38256 2017-01-31 Richard Biener <rguenther@suse.de>
38258 PR tree-optimization/77318
38259 * graphite-sese-to-poly.c (extract_affine): Fix assert.
38260 (create_pw_aff_from_tree): Take loop parameter.
38261 (add_condition_to_pbb): Pass loop of the condition to
38262 create_pw_aff_from_tree.
38264 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38266 * config/s390/s390.c (s390_asan_shadow_offset): New function.
38267 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
38269 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
38273 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
38275 (convert_int_to_float128): Likewise.
38276 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
38277 (convert_int_to_float128): Likewise.
38278 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
38279 (UNSPEC_IEEE128_CONVERT): Likewise.
38280 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
38281 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
38282 Use local variables for IBM extended format.
38283 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
38284 (fix_trunc<mode>si2_fprs): Likewise.
38285 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
38286 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
38287 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
38288 to know that we can now have integers of all sizes in vector
38290 (fix<uns>_<mode>di2_hw): Likewise.
38291 (float<uns>_<mode>si2_hw): Likewise.
38292 (fix_<mode>si2_hw): Likewise.
38293 (fixuns_<mode>si2_hw): Likewise.
38294 (float<uns>_<mode>di2_hw): Likewise.
38295 (float_<mode>di2_hw): Likewise.
38296 (float_<mode>si2_hw): Likewise.
38297 (floatuns_<mode>di2_hw): Likewise.
38298 (floatuns_<mode>si2_hw): Likewise.
38299 (xscvqp<su>wz_<mode>): Delete, no longer used.
38300 (xscvqp<su>dz_<mode>): Likewise.
38301 (xscv<su>dqp_<mode>): Likewise.
38302 (ieee128_mfvsrd_64bit): Likewise.
38303 (ieee128_mfvsrd_32bit): Likewise.
38304 (ieee128_mfvsrwz): Likewise.
38305 (ieee128_mtvsrw): Likewise.
38306 (ieee128_mtvsrd_64bit): Likewise.
38307 (ieee128_mtvsrd_32bit): Likewise.
38309 2017-01-31 Martin Liska <mliska@suse.cz>
38312 * ipa-prop.c (ipa_free_all_node_params): Call release method
38313 instead of ~sumbol_summary to not to trigger double times
38316 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
38318 PR tree-optimization/71691
38319 * bitmap.h (class auto_bitmap): New.
38320 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
38321 is_maybe_undefined instead of ssa_undefined_value_p.
38323 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38325 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38326 __S390_ARCH_LEVEL__ to __ARCH__.
38328 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38330 PR tree-optimization/79267
38331 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
38332 if should_remove_lhs_p is true.
38334 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
38337 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
38338 (add_alignment_attribute): New.
38339 (base_type_die): Add alignment attribute.
38340 (subrange_type_die): Likewise.
38341 (modified_type_die): Likewise.
38342 (gen_array_type_die): Likewise.
38343 (gen_descr_array_type_die: Likewise.
38344 (gen_enumeration_type_die): Likewise.
38345 (gen_subprogram_die): Likewise.
38346 (gen_variable_die): Likewise.
38347 (gen_field_die): Likewise.
38348 (gen_ptr_to_mbr_type_die): Likewise.
38349 (gen_struct_or_union_type_die): Likewise.
38350 (gen_subroutine_type_die): Likewise.
38351 (gen_typedef_die): Likewise.
38352 (base_type_cmp): Compare alignment attribute.
38354 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38357 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
38358 (setb_unsigned) New pattern for setb with CCUNS.
38359 * config/rs6000/rs6000.c (expand_block_compare): Use a different
38360 subfc./subfe sequence to avoid overflow problems. Generate a
38361 shorter sequence with cmpld/setb for power9.
38362 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
38363 for generating subfc. instruction.
38364 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
38365 now uses this instruction.
38367 2017-01-30 Ian Lance Taylor <iant@google.com>
38370 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
38371 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
38373 2017-01-30 Martin Sebor <msebor@redhat.com>
38375 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
38376 Move constant to the right of a relational operator.
38377 (get_mpfr_format_length, format_character, format_string): Ditto.
38378 (should_warn_p, maybe_warn): Same.
38380 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
38382 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
38385 * asan.c (get_translation_unit_decl): Remove function.
38386 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
38388 2017-01-30 Martin Liska <mliska@suse.cz>
38390 PR gcov-profile/79259
38391 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
38392 -fprofile-generate.
38394 2017-01-30 Martin Liska <mliska@suse.cz>
38397 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
38398 Initialize variables with NULL value.
38400 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
38403 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
38405 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
38407 2017-01-30 Richard Biener <rguenther@suse.de>
38409 PR tree-optimization/79276
38410 * tree-vrp.c (process_assert_insertions): Properly adjust common
38411 when removing a duplicate.
38413 2017-01-30 Richard Biener <rguenther@suse.de>
38415 PR tree-optimization/79256
38416 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38417 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38419 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
38421 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
38424 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
38425 ("*r<noxa>sbg_<mode>_sll_bitmask")
38426 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
38427 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
38428 Use contiguous_bitmask_nowrap_operand.
38430 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38433 * config/rs6000/altivec.h (vec_xl): Revise #define.
38434 (vec_xst): Likewise.
38436 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
38438 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
38440 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
38442 PR rtl-optimization/79194
38443 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
38444 traps before call to bypass_conditional_jumps.
38446 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38448 PR tree-optimization/71374
38449 * lra-constraints.c (check_conflict_input_operands): New.
38450 (match_reload): Use it.
38452 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38455 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
38456 account to calculate conflict_set.
38458 2017-01-27 Bin Cheng <bin.cheng@arm.com>
38460 PR rtl-optimization/78559
38461 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
38462 other_insn in combine.
38464 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
38466 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
38467 uint16_type_node for BT_UINT16.
38469 2017-01-27 David Malcolm <dmalcolm@redhat.com>
38471 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
38472 "RTL Tests" to menu.
38473 (GIMPLE Tests): New node.
38474 (RTL Tests): New node.
38476 2017-01-27 Richard Biener <rguenther@suse.de>
38478 PR tree-optimization/79245
38479 * tree-loop-distribution.c (distribute_loop): Apply cost
38480 modeling also to detected patterns.
38482 2017-01-27 Richard Biener <rguenther@suse.de>
38484 PR tree-optimization/71433
38485 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
38486 (compare_assert_loc): New function.
38487 (process_assert_insertions): Sort and optimize assert locations
38488 to remove duplicates and push down identical assertions on
38489 edges to their destination block.
38491 2017-01-27 Richard Biener <rguenther@suse.de>
38493 PR tree-optimization/79244
38494 * tree-vrp.c (remove_range_assertions): Forcefully propagate
38495 out SSA names even if abnormal.
38497 2017-01-27 Jakub Jelinek <jakub@redhat.com>
38499 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
38500 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
38501 instead of MPFR_RNDN.
38503 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
38506 * arm.c (arm_option_override): Don't call build_target_option_node
38507 until after doing all option overrides.
38508 (arm_valid_target_attribute_tree): Likewise.
38510 2017-01-27 Martin Liska <mliska@suse.cz>
38512 * doc/invoke.texi (-fprofile-arcs): Document profiling support
38513 for {cd}tors and C++ {cd}tors.
38515 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
38517 * config/s390/s390.md ("*setmem_long_and")
38518 ("*setmem_long_and_31z"): Use zero_extend instead of and.
38520 2017-01-26 Martin Sebor <msebor@redhat.com>
38522 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
38525 2017-01-26 Martin Sebor <msebor@redhat.com>
38527 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
38528 HAVE_DFmode before using XFmode or DFmode.
38529 (parse_directive): Avoid using the z length modifier to avoid
38530 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
38532 PR middle-end/78703
38533 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
38534 to accept adjustment as an array.
38535 (get_int_range): New function.
38536 (struct directive): Make width and prec arrays.
38537 (directive::set_width, directive::set_precision): Call get_int_range.
38538 (format_integer, format_floating): Handle width and precision ranges.
38539 (format_string, parse_directive): Same.
38541 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38544 * dwarf2out.c (generate_skeleton_bottom_up): For children with
38545 comdat_type_p set, just clone them, but keep the children in the
38549 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
38550 which have direct callers with -fvar-tracking-assignments enabled
38552 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
38553 inside of type units.
38555 2017-01-26 Martin Sebor <msebor@redhat.com>
38557 PR middle-end/78703
38558 * gimple-ssa-sprintf.c (struct result_range): Add likely and
38560 (struct format_result): Replace number_chars, number_chars_min,
38561 and number_chars_max with a single member of struct result_range.
38563 (format_result::operator+=): Adjust.
38564 (struct fmtresult): Remove bounded. Handle likely and unlikely
38566 (fmtresult::adjust_for_width_or_precision): New function.
38567 (fmtresult:type_max_digits): New function.
38568 (bytes_remaining): Handle likely and unlikely counters.
38569 (min_bytes_remaining): Remove.
38570 (format_percent): Simplify.
38571 (format_integer, format_floating): Set likely and unlikely counters.
38572 (get_string_length, format_character, format_string): Same.
38573 (format_plain, should_warn_p): New function.
38574 (maybe_warn): Call should_warn_p. Update diagnostic messages
38575 and handle those for all directives, including plain strings.
38576 (format_directive): Handle likely and unlikely counters.
38577 Remove unnecessary quoting from diagnostics. Add an informational
38579 (add_bytes): Remove.
38580 (pass_sprintf_length::compute_format_length): Simplify.
38581 (try_substitute_return_value): Handle likely and unlikely counters.
38583 2017-01-26 Carl Love <cel@us.ibm.com>
38585 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
38586 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
38588 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
38591 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
38592 endianess for subregs into account.
38593 * lra-constraints.c (lra_constraints): Do risky transformations
38594 always on the first iteration.
38595 * lra-lives.c (check_pseudos_live_through_calls): Add arg
38596 last_call_used_reg_set.
38597 (process_bb_lives): Define and use last_call_used_reg_set.
38598 * lra.c (lra): Always continue after lra_constraints on the first
38601 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
38603 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
38605 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
38607 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38609 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
38610 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
38611 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
38612 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
38613 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
38614 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
38615 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
38616 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
38617 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
38619 2017-01-26 Marek Polacek <polacek@redhat.com>
38622 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
38623 for the third operand.
38625 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38627 PR middle-end/79236
38628 * omp-low.c (struct omp_context): Add simt_stmt field.
38629 (scan_omp_for): Return omp_context *.
38630 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
38631 context to the _simt_ SIMD stmt.
38632 (lower_omp_for): For combined SIMD with sibling _simt_
38633 SIMD, make sure to use the same decls in _looptemp_
38634 clauses as in the sibling.
38636 2017-01-26 David Sherwood <david.sherwood@arm.com>
38638 PR middle-end/79212
38639 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
38642 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38645 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
38646 emit fld b; fld a; if possible.
38648 * brig-builtins.def: Update copyright years.
38649 * config/arm/arm_acle_builtins.def: Update copyright years.
38651 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
38654 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
38655 constraint instead of o for the stxsd instruction.
38657 2017-01-25 Carl Love <cel@us.ibm.com>
38659 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
38660 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
38662 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
38664 * doc/invoke.texi (C++ Dialect Options): Fix typo.
38666 2017-01-25 Richard Biener <rguenther@suse.de>
38668 PR tree-optimization/69264
38669 * target.def (vector_alignment_reachable): Improve documentation.
38670 * doc/tm.texi: Regenerate.
38671 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
38673 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
38674 earlier changes with respect to TYPE_USER_ALIGN.
38675 (vector_alignment_reachable_p): Likewise. Improve dumping.
38677 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38680 * config/arm/arm.md (xordi3): Force constant operand into a register
38683 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38685 * doc/invoke.texi (-fstore-merging): Correct default optimization
38686 levels at which it is enabled.
38687 (-O): Move -fstore-merging from list to...
38690 2017-01-25 Richard Biener <rguenther@suse.de>
38693 * omp-expand.c: Include debug.h.
38694 (expand_omp_taskreg): Make sure to generate early debug before
38695 outlining anything from a function.
38696 (expand_omp_target): Likewise.
38697 (grid_expand_target_grid_body): Likewise.
38699 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
38702 * asan.c (get_translation_unit_decl): New function.
38703 (asan_add_global): Extract modules file name from globals
38704 TRANSLATION_UNIT_DECL name.
38706 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
38709 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
38710 for long calls with APCS frame and VFP.
38712 2017-01-24 David Malcolm <dmalcolm@redhat.com>
38714 * cfg.c (original_copy_tables_initialized_p): New function.
38715 * cfg.h (original_copy_tables_initialized_p): New decl.
38716 * cfgrtl.c (relink_block_chain): Guard the call to
38717 free_original_copy_tables with a call to
38718 original_copy_tables_initialized_p.
38719 * cgraph.h (symtab_node::native_rtl_p): New decl.
38720 * cgraphunit.c (symtab_node::native_rtl_p): New function.
38721 (symtab_node::needed_p): Don't assert for early assembly output
38722 for __RTL functions.
38723 (cgraph_node::finalize_function): Set "force_output" for __RTL
38725 (cgraph_node::analyze): Bail out early for __RTL functions.
38726 (analyze_functions): Update assertion to support __RTL functions.
38727 (cgraph_node::expand): Bail out early for __RTL functions.
38728 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
38730 * function.h (struct function): Update comment for field
38732 * gimple-expr.c: Include "tree-pass.h".
38733 (gimple_has_body_p): Return false for __RTL functions.
38734 * Makefile.in (OBJS): Add run-rtl-passes.o.
38735 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
38737 (gcc::pass_manager::get_clean_slate): New accessor.
38738 * passes.c: Include "insn-addr.h".
38739 (should_skip_pass_p): Add logging. Update logic for running
38740 "expand" to be compatible with both __GIMPLE and __RTL. Guard
38741 property-provider override so it is only done for gimple passes.
38743 (skip_pass): New function.
38744 (execute_one_pass): Call skip_pass when skipping passes.
38745 * read-md.c (md_reader::read_char): Support filtering
38746 the input to a subset of line numbers.
38747 (md_reader::md_reader): Initialize fields
38748 m_first_line and m_last_line.
38749 (md_reader::read_file_fragment): New function.
38750 * read-md.h (md_reader::read_file_fragment): New decl.
38751 (md_reader::m_first_line): New field.
38752 (md_reader::m_last_line): New field.
38753 * read-rtl-function.c (function_reader::create_function): Only
38754 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
38755 curr_properties. Set DECL_INITIAL to a dummy block.
38756 (read_rtl_function_body_from_file_range): New function.
38757 * read-rtl-function.h (read_rtl_function_body_from_file_range):
38759 * run-rtl-passes.c: New file.
38760 * run-rtl-passes.h: New file.
38762 2017-01-24 Jeff Law <law@redhat.com>
38764 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
38767 2017-01-24 Bin Cheng <bin.cheng@arm.com>
38769 PR tree-optimization/79159
38770 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
38771 (record_nonwrapping_iv): Improve boundary using above function if no
38772 value range information.
38774 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
38775 Martin Jambor <mjambor@suse.cz>
38777 * brig-builtins.def: New file.
38778 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
38779 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
38780 (DEF_HSAIL_SAT_BUILTIN): Likewise.
38781 (DEF_HSAIL_INTR_BUILTIN): Likewise.
38782 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
38783 * builtin-types.def (BT_INT8): New.
38784 (BT_INT16): Likewise.
38785 (BT_UINT8): Likewise.
38786 (BT_UINT16): Likewise.
38787 (BT_FN_ULONG): Likewise.
38788 (BT_FN_UINT_INT): Likewise.
38789 (BT_FN_UINT_ULONG): Likewise.
38790 (BT_FN_UINT_LONG): Likewise.
38791 (BT_FN_UINT_PTR): Likewise.
38792 (BT_FN_ULONG_PTR): Likewise.
38793 (BT_FN_INT8_FLOAT): Likewise.
38794 (BT_FN_INT16_FLOAT): Likewise.
38795 (BT_FN_UINT32_FLOAT): Likewise.
38796 (BT_FN_UINT16_FLOAT): Likewise.
38797 (BT_FN_UINT8_FLOAT): Likewise.
38798 (BT_FN_UINT64_FLOAT): Likewise.
38799 (BT_FN_UINT16_UINT32): Likewise.
38800 (BT_FN_UINT32_UINT16): Likewise.
38801 (BT_FN_UINT16_UINT16_UINT16): Likewise.
38802 (BT_FN_INT_PTR_INT): Likewise.
38803 (BT_FN_UINT_PTR_UINT): Likewise.
38804 (BT_FN_LONG_PTR_LONG): Likewise.
38805 (BT_FN_ULONG_PTR_ULONG): Likewise.
38806 (BT_FN_VOID_UINT64_UINT64): Likewise.
38807 (BT_FN_UINT8_UINT8_UINT8): Likewise.
38808 (BT_FN_INT8_INT8_INT8): Likewise.
38809 (BT_FN_INT16_INT16_INT16): Likewise.
38810 (BT_FN_INT_INT_INT): Likewise.
38811 (BT_FN_UINT_FLOAT_UINT): Likewise.
38812 (BT_FN_FLOAT_UINT_UINT): Likewise.
38813 (BT_FN_ULONG_UINT_UINT): Likewise.
38814 (BT_FN_ULONG_UINT_PTR): Likewise.
38815 (BT_FN_ULONG_ULONG_ULONG): Likewise.
38816 (BT_FN_UINT_UINT_UINT): Likewise.
38817 (BT_FN_VOID_UINT_PTR): Likewise.
38818 (BT_FN_UINT_UINT_PTR: Likewise.
38819 (BT_FN_UINT32_UINT64_PTR): Likewise.
38820 (BT_FN_INT_INT_UINT_UINT): Likewise.
38821 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
38822 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
38823 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
38824 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
38825 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
38826 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
38827 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
38828 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
38829 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
38830 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
38831 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
38832 * doc/frontends.texi: List BRIG FE.
38833 * doc/install.texi (Testing): Add BRIG tesring requirements.
38834 * doc/invoke.texi (Overall Options): Mention BRIG.
38835 * doc/standards.texi (Standards): Doucment BRIG HSA version.
38837 2017-01-24 Richard Biener <rguenther@suse.de>
38839 PR translation/79208
38840 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
38842 2017-01-24 Martin Jambor <mjambor@suse.cz>
38845 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
38846 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
38847 and known_contexts.
38849 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
38851 PR middle-end/79123
38852 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
38853 casts from signed to unsigned really don't have a range.
38855 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
38857 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
38858 GMP_RNDx for compatiblity.
38860 2017-01-24 Martin Liska <mliska@suse.cz>
38863 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
38864 that would prevent us to call alloca with -1 as argument.
38866 2017-01-24 Jakub Jelinek <jakub@redhat.com>
38868 * dwarf2out.c (output_compilation_unit_header, output_file_names):
38869 Avoid -Wformat-security warning.
38871 2017-01-23 Andrew Pinski <apinski@cavium.com>
38873 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
38876 2017-01-23 Martin Sebor <msebor@redhat.com>
38878 PR middle-end/78703
38879 * gimple-ssa-sprintf.c (warn_level): New global.
38880 (format_integer): Use it here and throughout the rest of the file.
38881 Use the same switch to compute sign as base.
38882 (maybe_warn): New function.
38883 (format_directive): Factor out warnings into maybe_warn.
38884 Add debugging output. Use warn_level.
38885 (add_bytes): Use warn_level.
38886 (pass_sprintf_length::compute_format_length): Add debugging output.
38887 (try_substitute_return_value): Same.
38888 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
38890 PR middle-end/78703
38891 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
38892 (struct fmtresult, format_integer, format_floating): Adjust.
38893 (fmtresult::fmtresult): Set max correctly in two argument ctor.
38894 (get_string_length, format_string,format_directive): Same.
38895 (pass_sprintf_length::compute_format_length): Same.
38896 (try_substitute_return_value): Simplify slightly.
38898 PR middle-end/78703
38899 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
38900 (fmtresult::operator+=): Outlined.
38901 (struct fmtresult): Add ctors.
38902 (struct conversion_spec): Rename...
38903 (struct directive): ...to this. Add and remove data members.
38904 (directive::set_width, directive::set_precision): New functions.
38905 (format_percent): Use fmtresult ctor.
38906 (get_width_and_precision): Remove.
38907 (format_integer): Make naming changes. Avoid computing width and
38909 (format_floating): Same. Adjust indentation.
38910 (format_character, format_none): New functions.
38911 (format_string): Moved character handling to format_character.
38912 (format_directive): Remove arguments, change return type.
38913 (parse_directive): New function.
38914 (pass_sprintf_length::compute_format_length): Move directive
38915 parsing to parse_directive.
38917 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38919 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
38920 (assign_assembler_name_if_needed): ... this.
38921 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
38922 (assign_assembler_name_if_needed): ... this.
38923 (free_lang_data_in_cgraph): Adjust callers.
38924 * cgraphunit.c (cgraph_node::analyze): Likewise.
38925 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
38928 2017-01-23 Richard Biener <rguenther@suse.de>
38930 PR tree-optimization/79088
38931 PR tree-optimization/79188
38932 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
38933 resetting loop bounds after last path deletion. Reset loop
38934 bounds of the target loop, make code match the comments.
38935 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
38936 Make sure loops need no fixups.
38938 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
38940 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
38941 exponent support with double type for first argument.
38942 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
38943 type returned by __builtin_vec_extract_sig,
38944 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
38945 functions from "vector int" to "vector unsigned int" or from
38946 "vector long long int" to "vector unsigned long long int".
38947 Changed type returned by __builtin_vec_extract_exp,
38948 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
38949 functions from "vector int" to "vector unsigned int" or from
38950 "vector long long int" to "vector unsigned long long int".
38951 Changed return type of __builtin_vec_test_data_class,
38952 __builtin_vec_test_data_class_sp, and
38953 __builtin_vec_test_data_class_dp from "vector int" to
38954 "vector bool int" or from "vector long long int" to "vector bool
38955 long long int" and changed second argument type from "unsigned
38956 int" to "int". Added new overloaded function forms "vector float
38957 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
38958 "vector float __builtin_vec_insert_exp_sp (vector float, vector
38959 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
38960 double, vector unsigned long long int)" and "vector double
38961 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
38962 long int)". Changed return type of
38963 __builtin_scalar_test_data_class and
38964 __builtin_scalar_test_data_class_sp and
38965 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
38966 int" and changed second argument from "unsigned int" to "int".
38967 Changed type returned by __builtin_scalar_test_neg,
38968 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
38969 from "int" to "bool int". Added new overloaded function form
38970 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
38971 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
38972 exponent double-precision with floating point first argument.
38973 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
38974 documentation of scalar_test_data_class, scalar_test_neg,
38975 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
38976 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
38977 vec_test_data_class built-in functions to reflect refinements in
38978 their type signatures.
38980 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
38982 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
38984 (aarch64_elf_asm_destructor): Likewise.
38986 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
38988 PR rtl-optimization/78634
38989 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
38990 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
38991 * ifcvt.c (noce_try_cmove): Add missing cost check.
38993 PR rtl-optimization/71724
38994 * combine.c (if_then_else_cond): Look for situations where it is
38995 beneficial to undo the work of one of the recursive calls.
38997 2017-01-23 Bin Cheng <bin.cheng@arm.com>
38999 PR tree-optimization/70754
39000 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
39001 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
39002 combined stmt before it if not NULL.
39003 (combine_chains): Process refs reversely and compute dominance point
39006 2017-01-23 Martin Liska <mliska@suse.cz>
39008 PR tree-optimization/79196
39009 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
39010 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
39012 (strlen_optimize_stmt): Call the renamed function.
39014 2017-01-23 Michael Matz <matz@suse.de>
39016 PR tree-optimization/78384
39017 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
39019 2017-01-23 Richard Biener <rguenther@suse.de>
39021 PR tree-optimization/79186
39022 * tree-vrp.c (register_new_assert_for): Make sure we've seen
39023 both incoming edges before moving an assert.
39025 2017-01-23 Martin Jambor <mjambor@suse.cz>
39027 * ipa-prop.c (load_from_param_1): Removed.
39028 (load_from_unmodified_param): Bits from load_from_param_1 put back
39030 (load_from_param): Removed.
39031 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
39032 with stmt. Reverted back to use of load_from_unmodified_param.
39034 2017-01-23 Martin Jambor <mjambor@suse.cz>
39037 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
39038 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
39039 field a pointer to garbage collected vector, mark lattices and
39040 ipcp_orig_node with GTY((skip)).
39041 (ipa_get_param_count): Adjust to descriptors being a pointer.
39042 (ipa_get_param): Likewise.
39043 (ipa_get_type): Likewise.
39044 (ipa_get_param_move_cost): Likewise.
39045 (ipa_set_param_used): Likewise.
39046 (ipa_get_controlled_uses): Likewise.
39047 (ipa_set_controlled_uses): Likewise.
39048 (ipa_is_param_used): Likewise.
39049 (ipa_node_params_t): Move into garbage collector. New methods insert
39051 (ipa_node_params_sum): Annotate wth GTY(()).
39052 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
39054 (ipa_load_from_parm_agg): Adjust declaration.
39055 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
39056 * ipa-profile.c (ipa_profile): Likewise.
39057 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
39058 (ipa_populate_param_decls): Make descriptors parameter garbage
39060 (ipa_dump_param): Adjust to descriptors being a pointer.
39061 (ipa_alloc_node_params): Likewise.
39062 (ipa_initialize_node_params): Likewise.
39063 (load_from_param_1): Make descriptors parameter garbage collected.
39064 (load_from_unmodified_param): Likewise.
39065 (load_from_param): Likewise.
39066 (ipa_load_from_parm_agg): Likewise.
39067 (ipa_node_params::~ipa_node_params): Removed.
39068 (ipa_free_all_node_params): Remove call to delete operator.
39069 (ipa_node_params_t::insert): New.
39070 (ipa_node_params_t::remove): Likewise.
39071 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
39072 copy known_csts and known_contexts vectors.
39073 (ipa_read_node_info): Adjust to descriptors being a pointer.
39074 (ipcp_modif_dom_walker): Make m_descriptors field garbage
39076 (ipcp_transform_function): Make descriptors variable garbage
39079 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
39081 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
39082 * config/i386/avx512dqintrin.h: Ditto.
39083 * config/i386/avx512fintrin.h: Ditto.
39084 * config/i386/i386.c: Handle new builtins.
39085 * config/i386/i386-builtin.def: Add new builtins.
39086 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
39087 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
39089 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39090 Martin Liska <mliska@suse.cz>
39092 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
39093 * asan.c (asan_expand_poison_ifn): Support stores and use
39094 appropriate ASAN report function.
39095 * internal-fn.c (expand_ASAN_POISON_USE): New function.
39096 * internal-fn.def (ASAN_POISON_USE): Declare.
39097 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
39098 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
39099 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
39100 ASAN_POISON calls w/o LHS.
39101 * tree-ssa.c (execute_update_addresses_taken): Create clobber
39102 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
39103 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
39104 * gimplify.c (asan_poison_variables): Add attribute
39105 use_after_scope_memory to variables that really needs to live
39107 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
39108 having the attribute.
39110 2017-01-23 Martin Liska <mliska@suse.cz>
39112 * asan.c (create_asan_shadow_var): New function.
39113 (asan_expand_poison_ifn): Likewise.
39114 * asan.h (asan_expand_poison_ifn): New declaration.
39115 * internal-fn.c (expand_ASAN_POISON): Likewise.
39116 * internal-fn.def (ASAN_POISON): New builtin.
39117 * sanopt.c (pass_sanopt::execute): Expand
39118 asan_expand_poison_ifn.
39119 * tree-inline.c (copy_decl_for_dup_finish): Make function
39121 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
39122 * tree-ssa.c (is_asan_mark_p): New function.
39123 (execute_update_addresses_taken): Rewrite local variables
39124 (identified just by use-after-scope as addressable) into SSA.
39126 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39128 * doc/install.texi (Specific): opensource.apple.com uses https
39129 now. Remove trailing slash.
39131 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39133 * README.Portability: Remove note on an Irix compatibility issue.
39135 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
39137 * gcov.c (INCLUDE_ALGORITHM): Define.
39138 (INCLUDE_VECTOR): Define.
39139 No longer include <vector> and <algorithm> directly.
39141 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
39143 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
39145 * doc/invoke.texi (Code Gen Options): Ditto.
39147 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
39150 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
39152 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
39154 rtl-optimization/79125
39155 * cprop.c (local_cprop_pass): Handle cases where we make an
39156 unconditional trap.
39158 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
39162 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
39163 read from, for big endian.
39165 2017-01-20 Jiong Wang <jiong.wang@arm.com>
39167 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
39168 register pauth builtins for LP64 only.
39170 2017-01-20 Marek Polacek <polacek@redhat.com>
39173 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
39176 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39178 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
39180 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
39181 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
39182 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
39184 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39187 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
39188 in big-endian mode when they are not a single duplicated value.
39190 2017-01-20 Richard Biener <rguenther@suse.de>
39192 * BASE-VER: Bump to 7.0.1.
39194 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39196 * omp-low.c (omplow_simd_context): New struct. Use it...
39197 (lower_rec_simd_input_clauses): ...here and...
39198 (lower_rec_input_clauses): ...here to hold common data. Adjust all
39199 references to idx, lane, max_vf, is_simt.
39201 2017-01-20 Graham Markall <graham.markall@embecosm.com>
39203 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
39206 2017-01-20 Martin Jambor <mjambor@suse.cz>
39208 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
39209 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
39211 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
39212 (GTFILES): Rename hsa.c to hsa-common.c.
39213 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
39214 * hsa-dump.c: Likewise.
39215 * hsa-gen.c: Likewise.
39216 * hsa-regalloc.c: Likewise.
39217 * ipa-hsa.c: Likewise.
39218 * omp-expand.c: Likewise.
39219 * omp-low.c: Likewise.
39220 * toplev.c: Likewise.
39222 2017-01-20 Marek Polacek <polacek@redhat.com>
39225 * doc/invoke.texi: Document -Wduplicated-branches.
39226 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
39227 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
39228 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
39229 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
39230 return 0 only when not OEP_LEXICOGRAPHIC.
39231 (fold_build_cleanup_point_expr): Use the expression
39232 location when building CLEANUP_POINT_EXPR.
39233 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
39234 * tree.c (add_expr): Handle error_mark_node.
39236 2017-01-20 Martin Liska <mliska@suse.cz>
39239 * tree-profile.c (init_ic_make_global_vars): Do not call
39241 (gimple_init_gcov_profiler): Likewise.
39243 2017-01-20 Martin Liska <mliska@suse.cz>
39246 * cgraph.h (maybe_create_reference): Remove argument and
39248 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
39250 * ipa-cp.c (create_specialized_node): Likewise.
39251 * symtab.c (symtab_node::maybe_create_reference): Handle
39252 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
39254 2017-01-20 Martin Liska <mliska@suse.cz>
39256 * read-rtl-function.c (function_reader::create_function): Use
39257 build_decl instread of build_decl_stat.
39259 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
39261 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
39262 * config/i386/avx512dqintrin.h: Ditto.
39263 * config/i386/avx512fintrin.h: Ditto.
39264 * config/i386/i386-builtin-types.def: Add new types.
39265 * config/i386/i386.c: Handle new types.
39266 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
39267 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
39268 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
39269 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
39270 (__builtin_ia32_kshiftridi): New.
39271 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
39273 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
39277 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
39278 define to rs6000_init_stack_protect_guard.
39279 (rs6000_init_stack_protect_guard): New function.
39281 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39282 Yunqiang Su <yunqiang.su@imgtec.com>
39284 * config.gcc (supported_defaults): Add madd4.
39285 (with_madd4): Add validation.
39286 (all_defaults): Add madd4.
39287 * config/mips/mips.opt (mmadd4): New option.
39288 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39290 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39292 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
39293 (ISA_HAS_FUSED_MADD4): Likewise.
39294 * doc/invoke.texi (-mmadd4): Document the new option.
39295 * doc/install.texi (--with-madd4): Document the new option.
39297 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39299 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
39300 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
39301 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
39302 (aarch64_init_pauth_hint_builtins): New.
39303 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
39304 (aarch64_expand_builtin): Expand new builtins.
39306 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39308 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
39309 * combine-stack-adj.c (no_unhandled_cfa): Handle
39310 REG_CFA_TOGGLE_RA_MANGLE.
39311 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
39312 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
39313 info for return address signing.
39314 (aarch64_expand_epilogue): Likewise.
39316 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39318 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
39319 * config/aarch64/aarch64-protos.h
39320 (aarch64_return_address_signing_enabled): New declaration.
39321 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
39323 (aarch64_expand_prologue): Sign return address before it's pushed onto
39325 (aarch64_expand_epilogue): Authenticate return address fetched from
39327 (aarch64_override_options): Sanity check for ILP32 and ISA level.
39328 (aarch64_attributes): New function attributes for "sign-return-address".
39329 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
39330 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
39331 ("*do_return"): Generate combined instructions according to key index.
39332 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
39333 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
39335 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
39336 * config/aarch64/aarch64.opt (msign-return-address=): New.
39337 * doc/extend.texi (AArch64 Function Attributes): Documents
39338 "sign-return-address=".
39339 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
39341 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39343 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
39344 overall option summary.
39346 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39348 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
39349 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
39350 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
39351 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
39353 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
39355 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
39356 -mpower9-minmax by default for -mcpu=power9.
39357 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
39358 128-bit floating point.
39360 2017-01-20 Alan Modra <amodra@gmail.com>
39362 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
39363 optimizing for size.
39365 2017-01-20 Alan Modra <amodra@gmail.com>
39368 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
39369 for strcmp and strncmp from corresponding builtin decl.
39371 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39373 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
39374 instead of i386/rtems-64.h.
39375 * config/i386/rtems-64.h: Remove.
39377 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39381 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
39383 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
39385 2017-01-19 Tamar Christina <tamar.christina@arm.com>
39387 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
39388 Change int to HOST_WIDE_INT.
39389 * config/aarch64/aarch64-protos.h
39390 (aarch64_simd_gen_const_vector_dup): Likewise.
39391 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
39393 2017-01-19 David Malcolm <dmalcolm@redhat.com>
39395 * langhooks-def.h (lhd_type_for_size): New decl.
39396 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
39397 * langhooks.c (lhd_type_for_size): New function, taken from
39400 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
39402 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
39403 define_bypass for CR latency.
39404 (power9-cracked-alu): Update bypass latency and remove power9-branch.
39405 (power9-alu2): Add define_bypass for CR latency.
39407 (power9-mul): Update insn latency.
39408 (power9-mul-compare): Update insn latency, bypass latency and remove
39411 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39413 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
39415 * config/aarch64/aarch64.md
39416 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
39417 aarch64_nopcrelative_literal_loads.
39418 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
39420 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
39422 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
39423 TARGET_LOONGSON_3A.
39424 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
39426 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
39429 * config.gcc (supported_defaults): Add lxc1-sxc1.
39430 (with_lxc1_sxc1): Add validation.
39431 (all_defaults): Add lxc1-sxc1.
39432 * config/mips/mips.opt (mlxc1-sxc1): New option.
39433 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39435 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39436 __mips_no_lxc1_sxc1.
39437 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
39438 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
39439 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
39441 2017-01-19 Richard Biener <rguenther@suse.de>
39443 PR tree-optimization/72488
39444 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
39445 sure to restore SSA info.
39446 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
39448 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
39450 PR rtl-optimization/79121
39451 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
39452 of the inner type when shifting an extended value.
39454 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39457 * symtab.c (symtab_node::equal_address_to): Fix comparing of
39458 interposable aliases.
39460 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
39463 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
39464 Use the evmergelohi instruction.
39465 (mov_si<mode>_e500_subreg4_2_le): Likewise.
39466 (mov_sitf_e500_subreg8_2_be): Likewise.
39467 (mov_sitf_e500_subreg12_2_le): Likewise.
39468 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
39469 (mov_si<mode>_e500_subreg4_2_be): Likewise.
39470 (mov_sitf_e500_subreg8_2_le): Likewise.
39471 (mov_sitf_e500_subreg12_2_be): Likewise.
39473 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39475 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
39476 attribute from vecsimple to vecperm.
39477 (altivec_vbpermq2): Likewise.
39479 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39482 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
39484 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39485 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
39486 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
39487 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
39488 case where N arg is SIZE_MAX.
39489 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
39490 (cmpstrsi): Add pattern.
39492 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
39494 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
39495 __builtin_vec_revb builtins.
39496 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
39497 built-in functions to support generation of the ISA 3.0 XXBR<x>
39498 vector byte reverse instructions.
39499 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
39500 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
39501 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
39502 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
39503 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
39504 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
39505 (P9V_BUILTIN_VEC_REVB): Likewise.
39506 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
39507 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
39508 (p9_xxbrq_v16qi): Likewise.
39509 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
39510 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
39511 (p9_xxbrh_v8hi): Likewise.
39512 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
39513 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
39514 vec_revb built-in functions.
39516 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
39518 PR rtl-optimization/78952
39519 * config/i386/i386.md (any_extract): New code iterator.
39520 (*insvqi_2): Use any_extract for source operand.
39521 (*insvqi_3): Use any_shiftrt for source operand.
39523 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
39525 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
39527 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
39529 2017-01-18 Matthias Klose <doko@ubuntu.com>
39531 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
39533 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39535 * config/rs6000/altivec.h (vec_bperm): Change #define.
39536 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
39537 (altivec_vbpermq2): New define_insn.
39538 (altivec_vbpermd): Likewise.
39539 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
39540 function interface.
39541 (VBPERMD): Likewise.
39542 (VBPERM): New polymorphic function interface.
39543 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
39544 Add entries for P9V_BUILTIN_VEC_VBPERM.
39545 * doc/extend.texi: Add interfaces for vec_bperm.
39547 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39549 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
39550 first letter of error messages.
39551 (s390_resolve_overloaded_builtin): Likewise.
39552 * config/s390/s390.c (s390_expand_builtin): Likewise.
39553 (s390_invalid_arg_for_unprototyped_fn): Likewise.
39554 (s390_valid_target_attribute_inner_p): Likewise.
39555 * config/s390/s390.md ("tabort"): Likewise.
39557 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
39559 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
39560 (ISA_AVOID_DIV_HILO): New macro.
39561 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
39562 (ISA_HAS_DDIV): Likewise.
39564 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39566 * doc/invoke.texi (fabi-version): Correct number of occurrences.
39568 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39570 * doc/invoke.texi (fabi-version): Spelling fix.
39572 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39575 * doc/invoke.texi (fabi-version): Mention mangling fix for
39578 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39581 * doc/invoke.texi (fabi-version): Document discriminator mangling.
39583 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
39586 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
39587 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
39589 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
39591 (stack_protect_test): Ditto.
39592 * config/rs6000/rs6000.opt (mstack-protector-guard=,
39593 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
39595 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
39596 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
39597 -mstack-protector-guard-offset=.
39598 (RS/6000 and PowerPC Options): Ditto.
39600 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39602 * config/i386/i386.h (MASK_CLASS_P): New define.
39603 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
39604 there are no registers from different register sets also when
39605 mask registers are used. Update function comment.
39606 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
39607 to (*k/*r) and (*k/*km) alternatives.
39609 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
39611 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
39612 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
39613 (EH_RETURN_HANDLER_RTX): New define.
39614 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
39615 Force frame pointer in EH return functions.
39616 (aarch64_expand_epilogue): Add barrier for eh_return.
39617 (aarch64_final_eh_return_addr): Remove.
39618 (aarch64_eh_return_handler_rtx): New function.
39619 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
39621 (aarch64_eh_return_handler_rtx): New prototype.
39623 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39625 * config/rs6000/altivec.h (vec_rlmi): New #define.
39626 (vec_vrlnm): Likewise.
39627 (vec_rlnm): Likewise.
39628 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
39629 (UNSPEC_VRLNM): Likewise.
39630 (VIlong): New mode iterator.
39631 (altivec_vrl<VI_char>mi): New define_insn.
39632 (altivec_vrl<VI_char>nm): Likewise.
39633 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
39635 (VRLDNM): Likewise.
39636 (RLNM): New polymorphic function entry.
39637 (VRLWMI): New monomorphic function entry.
39638 (VRLDMI): Likewise.
39639 (RLMI): New polymorphic function entry.
39640 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
39641 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
39642 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
39645 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39648 * dwarf2out.c (field_byte_offset): Restore the
39649 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
39650 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
39651 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
39654 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39657 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
39659 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39662 * dwarf2out.c (add_data_member_location_attribute): For constant
39663 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
39664 instead of DW_AT_data_member_location, DW_AT_bit_offset and
39665 DW_AT_byte_size attributes.
39667 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39669 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
39670 after forcing to constant memory when the code model is medium.
39672 2017-01-17 Julia Koval <julia.koval@intel.com>
39675 * config/i386/avx512fintrin.h
39676 (_mm512_i32gather_ps): Change __addr type to void const*.
39677 (_mm512_mask_i32gather_ps): Ditto.
39678 (_mm512_i32gather_pd): Ditto.
39679 (_mm512_mask_i32gather_pd): Ditto.
39680 (_mm512_i64gather_ps): Ditto.
39681 (_mm512_mask_i64gather_ps): Ditto.
39682 (_mm512_i64gather_pd): Ditto.
39683 (_mm512_mask_i64gather_pd): Ditto.
39684 (_mm512_i32gather_epi32): Ditto.
39685 (_mm512_mask_i32gather_epi32): Ditto.
39686 (_mm512_i32gather_epi64): Ditto.
39687 (_mm512_mask_i32gather_epi64): Ditto.
39688 (_mm512_i64gather_epi32): Ditto.
39689 (_mm512_mask_i64gather_epi32): Ditto.
39690 (_mm512_i64gather_epi64): Ditto.
39691 (_mm512_mask_i64gather_epi64): Ditto.
39692 (_mm512_i32scatter_ps): Change __addr type to void*.
39693 (_mm512_mask_i32scatter_ps): Ditto.
39694 (_mm512_i32scatter_pd): Ditto.
39695 (_mm512_mask_i32scatter_pd): Ditto.
39696 (_mm512_i64scatter_ps): Ditto.
39697 (_mm512_mask_i64scatter_ps): Ditto.
39698 (_mm512_i64scatter_pd): Ditto.
39699 (_mm512_mask_i64scatter_pd): Ditto.
39700 (_mm512_i32scatter_epi32): Ditto.
39701 (_mm512_mask_i32scatter_epi32): Ditto.
39702 (_mm512_i32scatter_epi64): Ditto.
39703 (_mm512_mask_i32scatter_epi64): Ditto.
39704 (_mm512_i64scatter_epi32): Ditto.
39705 (_mm512_mask_i64scatter_epi32): Ditto.
39706 (_mm512_i64scatter_epi64): Ditto.
39707 (_mm512_mask_i64scatter_epi64): Ditto.
39708 * config/i386/avx512pfintrin.h
39709 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
39710 (_mm512_mask_prefetch_i32gather_ps): Ditto.
39711 (_mm512_mask_prefetch_i64gather_pd): Ditto.
39712 (_mm512_mask_prefetch_i64gather_ps): Ditto.
39713 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
39714 (_mm512_prefetch_i32scatter_ps): Ditto.
39715 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
39716 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
39717 (_mm512_prefetch_i64scatter_pd): Ditto.
39718 (_mm512_prefetch_i64scatter_ps): Ditto.
39719 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
39720 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
39721 * config/i386/avx512vlintrin.h
39722 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
39723 (_mm_mmask_i32gather_ps): Ditto.
39724 (_mm256_mmask_i32gather_pd): Ditto.
39725 (_mm_mmask_i32gather_pd): Ditto.
39726 (_mm256_mmask_i64gather_ps): Ditto.
39727 (_mm_mmask_i64gather_ps): Ditto.
39728 (_mm256_mmask_i64gather_pd): Ditto.
39729 (_mm_mmask_i64gather_pd): Ditto.
39730 (_mm256_mmask_i32gather_epi32): Ditto.
39731 (_mm_mmask_i32gather_epi32): Ditto.
39732 (_mm256_mmask_i32gather_epi64): Ditto.
39733 (_mm_mmask_i32gather_epi64): Ditto.
39734 (_mm256_mmask_i64gather_epi32): Ditto.
39735 (_mm_mmask_i64gather_epi32): Ditto.
39736 (_mm256_mmask_i64gather_epi64): Ditto.
39737 (_mm_mmask_i64gather_epi64): Ditto.
39738 (_mm256_i32scatter_ps): Change __addr type to void*.
39739 (_mm256_mask_i32scatter_ps): Ditto.
39740 (_mm_i32scatter_ps): Ditto.
39741 (_mm_mask_i32scatter_ps): Ditto.
39742 (_mm256_i32scatter_pd): Ditto.
39743 (_mm256_mask_i32scatter_pd): Ditto.
39744 (_mm_i32scatter_pd): Ditto.
39745 (_mm_mask_i32scatter_pd): Ditto.
39746 (_mm256_i64scatter_ps): Ditto.
39747 (_mm256_mask_i64scatter_ps): Ditto.
39748 (_mm_i64scatter_ps): Ditto.
39749 (_mm_mask_i64scatter_ps): Ditto.
39750 (_mm256_i64scatter_pd): Ditto.
39751 (_mm256_mask_i64scatter_pd): Ditto.
39752 (_mm_i64scatter_pd): Ditto.
39753 (_mm_mask_i64scatter_pd): Ditto.
39754 (_mm256_i32scatter_epi32): Ditto.
39755 (_mm256_mask_i32scatter_epi32): Ditto.
39756 (_mm_i32scatter_epi32): Ditto.
39757 (_mm_mask_i32scatter_epi32): Ditto.
39758 (_mm256_i32scatter_epi64): Ditto.
39759 (_mm256_mask_i32scatter_epi64): Ditto.
39760 (_mm_i32scatter_epi64): Ditto.
39761 (_mm_mask_i32scatter_epi64): Ditto.
39762 (_mm256_i64scatter_epi32): Ditto.
39763 (_mm256_mask_i64scatter_epi32): Ditto.
39764 (_mm_i64scatter_epi32): Ditto.
39765 (_mm_mask_i64scatter_epi32): Ditto.
39766 (_mm256_i64scatter_epi64): Ditto.
39767 (_mm256_mask_i64scatter_epi64): Ditto.
39768 (_mm_i64scatter_epi64): Ditto.
39769 (_mm_mask_i64scatter_epi64): Ditto.
39770 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
39771 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
39772 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
39773 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
39774 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
39775 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
39776 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
39777 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
39778 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
39779 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
39780 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
39781 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
39782 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
39783 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
39784 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
39785 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
39786 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
39787 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
39788 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
39789 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
39790 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
39791 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
39792 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
39793 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
39794 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
39795 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
39796 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
39797 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
39798 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
39799 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
39800 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
39801 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
39802 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
39803 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
39804 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
39805 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
39806 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
39807 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
39808 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
39809 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
39810 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
39811 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
39812 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
39813 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
39814 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
39815 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
39816 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
39817 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
39818 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
39819 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
39820 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
39821 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
39822 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
39823 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
39824 definitions accordingly.
39826 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
39827 Kuan-Lin Chen <kuanlinchentw@gmail.com>
39830 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
39833 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
39836 * ira-conflicts.c (ira_build_conflicts): Update total conflict
39837 hard regs for inner regno.
39839 2017-01-17 Martin Liska <mliska@suse.cz>
39842 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
39843 assumption and add comment.
39845 2017-01-17 Nathan Sidwell <nathan@acm.org>
39847 * ipa-visibility.c (localize_node): New function, broken out of ...
39848 (function_and_variable_visibility): ... here. Call it.
39850 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39852 PR middle-end/77445
39853 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
39854 correctly set frequency of oudgoing edge.
39855 (duplicate_thread_path): Fix profile updating.
39857 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39860 * configure.ac: Add GCC_BASE_VER.
39861 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
39862 version from BASE-VER file.
39863 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
39864 (gcc.o): Depend on $(BASEVER).
39865 * common.opt (dumpfullversion): New option.
39866 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
39867 * doc/invoke.texi: Document -dumpfullversion.
39868 * doc/install.texi: Document --with-gcc-major-version-only.
39869 * configure: Regenerated.
39871 2017-01-17 Richard Biener <rguenther@suse.de>
39873 PR tree-optimization/71433
39874 * tree-vrp.c (register_new_assert_for): Merge same asserts
39875 on all incoming edges.
39876 (process_assert_insertions_for): Handle insertions at the
39879 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
39881 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
39882 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
39884 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
39887 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
39890 2017-01-17 Alan Modra <amodra@gmail.com>
39893 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
39894 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
39895 symbolic stack limit when pic.
39897 2017-01-16 Martin Sebor <msebor@redhat.com>
39899 PR tree-optimization/78608
39900 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
39902 2017-01-16 Jeff Law <law@redhat.com>
39905 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
39906 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
39907 for several include directories that may be relative to sysroot.
39908 * config/i386/x-mingw32 (gplus_includedir): Define.
39909 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
39910 (native_system_includedir): Likewise.
39911 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
39912 override if TARGET_SYSTEM_ROOT is defined.
39913 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
39915 PR tree-optimization/79090
39916 PR tree-optimization/33562
39917 PR tree-optimization/61912
39918 PR tree-optimization/77485
39919 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
39920 and computed trims into the dump file.
39922 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39924 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
39926 2017-01-16 Jakub Jelinek <jakub@redhat.com>
39929 * gimplify.c (gimplify_init_constructor): If want_value and
39930 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
39934 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
39935 sequence. Formatting fixes.
39936 (doloop_optimize): Formatting fixes.
39939 * gcc.c (debug_level_greater_than_spec_func): New function.
39940 (static_spec_functions): Add debug-level-gt spec function.
39941 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
39943 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39944 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39945 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
39946 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
39947 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
39948 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
39950 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
39952 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
39953 QImode fixups to general and mask registers only.
39955 2017-01-16 Carl Love <cel@us.ibm.com>
39957 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
39958 for built-in functions
39959 vector signed char vec_nabs (vector signed char)
39960 vector signed short vec_nabs (vector signed short)
39961 vector signed int vec_nabs (vector signed int)
39962 vector signed long long vec_nabs (vector signed long long)
39963 vector float vec_nabs (vector float)
39964 vector double vec_nabs (vector double)
39965 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
39967 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
39968 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
39969 * doc/extend.texi: Update the documentation file for the new built-in
39972 2017-01-16 Martin Sebor <msebor@redhat.com>
39974 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
39977 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39979 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
39980 UNSPEC_VSX__XXSPLTD to require special splat handling.
39982 2017-01-16 David Malcolm <dmalcolm@redhat.com>
39985 * system.h: Poison strndup.
39987 2017-01-16 Alan Modra <amodra@gmail.com>
39990 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
39993 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
39995 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
39997 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
39999 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
40000 call recog here. Assert that INSN_CODE (insn) is non-negative.
40002 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40005 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
40007 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
40008 in the currently scheduled RTL fragment.
40010 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40012 PR rtl-optimization/78751
40013 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
40016 2017-01-14 Jeff Law <law@redhat.com>
40018 PR tree-optimization/79090
40019 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
40020 variable length stores.
40021 (compute_trims): Delete dead assignment to *trim_tail.
40022 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
40025 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
40027 PR rtl-optimization/78626
40028 PR rtl-optimization/78727
40029 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
40030 of a block, and split such blocks after everything else is finished.
40032 2017-01-14 Alan Modra <amodra@gmail.com>
40035 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
40036 target legitimate_combined_insn.
40037 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
40038 (rs6000_legitimate_combined_insn): New function.
40039 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
40041 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
40042 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
40043 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
40045 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
40047 * doc/frontends.texi (G++ and GCC): Remove references to Java.
40049 2017-01-13 Jeff Law <law@redhat.com>
40051 PR tree-optimization/33562
40052 PR tree-optimization/61912
40053 PR tree-optimization/77485
40054 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
40056 (delete_dead_assignment): Likewise.
40057 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
40058 statement to delete_dead_call and delete_dead_assignment.
40060 2017-01-13 David Malcolm <dmalcolm@redhat.com>
40063 * substring-locations.c (format_warning_va): Strengthen case 1 so
40064 that both endpoints of the substring must be within the format
40065 range for just the substring to be printed.
40067 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
40069 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
40070 * config/i386/i386.c (ix86_target_string): Add missing options
40071 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
40072 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
40073 flags_other and ix86_target_other to flags2_other. Display unknown
40075 (ix86_valid_target_attribute_inner_p): Add missing options and
40076 reorder options by implied ISAs, as in ix86_target_string.
40078 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40080 * hash-table.h (hash_table::too_empty_p): New function.
40081 (hash_table::expand): Use it.
40082 (hash_table::traverse): Likewise.
40083 (hash_table::empty_slot): Use sizeof (value_type) instead of
40084 sizeof (PTR) to convert bytes to elements. Shrink the table
40085 if the current size is excessive for the current number of
40088 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40090 * ira-costs.c (record_reg_classes): Break from the inner loop
40091 early once alt_fail is known to be true. Update outer loop
40092 handling accordingly.
40094 2017-01-13 Jeff Law <law@redhat.com>
40096 * tree-ssa-dse.c (decrement_count): New function.
40097 (increment_start_addr, maybe_trim_memstar_call): Likewise.
40098 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
40099 when we know the partially dead statement is a mem* function.
40101 PR tree-optimization/61912
40102 PR tree-optimization/77485
40103 * tree-ssa-dse.c: Include expr.h.
40104 (maybe_trim_constructor_store): New function.
40105 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
40107 PR tree-optimization/33562
40108 PR tree-optimization/61912
40109 PR tree-optimization/77485
40110 * doc/invoke.texi: Document new dse-max-object-size param.
40111 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
40112 * tree-ssa-dse.c: Include params.h.
40113 (dse_store_status): New enum.
40114 (initialize_ao_ref_for_dse): New, partially extracted from
40116 (valid_ao_ref_for_dse, normalize_ref): New.
40117 (setup_live_bytes_from_ref, compute_trims): Likewise.
40118 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
40119 (maybe_trim_partially_dead_store): Likewise.
40120 (maybe_trim_complex_store): Likewise.
40121 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
40122 Track what bytes live from the original store. Return tri-state
40123 for dead, partially dead or live.
40124 (dse_dom_walker): Add constructor, destructor and new private members.
40125 (delete_dead_call, delete_dead_assignment): New extracted from
40127 (dse_optimize_stmt): Make a member of dse_dom_walker.
40128 Use initialize_ao_ref_for_dse.
40130 PR tree-optimization/33562
40131 PR tree-optimization/61912
40132 PR tree-optimization/77485
40133 * sbitmap.h (bitmap_count_bits): Prototype.
40134 (bitmap_clear_range, bitmap_set_range): Likewise.
40135 * sbitmap.c (bitmap_clear_range): New function.
40136 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
40138 2017-01-13 Martin Liska <mliska@suse.cz>
40141 * function.c (set_cfun): Add new argument force.
40142 * function.h (set_cfun): Likewise.
40143 * ipa-inline-transform.c (inline_call): Use the function when
40144 strict alising from is dropped for function we inline to.
40146 2017-01-13 Richard Biener <rguenther@suse.de>
40148 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
40149 for dumping GIMPLE INTEGER_CSTs.
40151 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
40153 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
40154 to 201112L since C++17.
40156 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
40159 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
40160 if -fsanitize=kernel-address is present.
40162 2017-01-13 Richard Biener <rguenther@suse.de>
40164 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
40165 as _Literal ( type ) number in case usual suffixes do not
40166 preserve all information.
40168 2017-01-13 Richard Biener <rguenther@suse.de>
40170 PR tree-optimization/77283
40171 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
40172 and ssa-iterators.h.
40173 (is_feasible_trace): Implement a cost model based on joiner
40176 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
40179 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
40180 char or short to __float128/_Float128 directly.
40182 2017-01-12 Martin Sebor <msebor@redhat.com>
40184 to -Wformat-overflow.
40185 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
40186 (min_bytes_remaining): Same.
40187 (get_string_length): Same.
40188 (format_string): Same.
40189 (format_directive): Same.
40191 (pass_sprintf_length::handle_gimple_call): Same.
40193 2017-01-12 Jakub Jelinek <jakub@redhat.com>
40195 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
40196 info.nowrite calls with no lhs that can't throw. Return bool
40197 whether gsi_remove has been called or not.
40198 (pass_sprintf_length::handle_gimple_call): Return bool whether
40199 try_substitute_return_value called gsi_remove. Formatting fix.
40200 (pass_sprintf_length::execute): Don't use gsi_remove if
40201 handle_gimple_call returned true.
40204 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
40205 be removed due to side-effects, don't remove following barrier nor
40206 turn the successor edge into fallthru edge.
40208 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40211 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
40212 element-reversing loads and stores as not swappable.
40214 2017-01-12 Nathan Sidwell <nathan@acm.org>
40215 Nicolai Stange <nicstange@gmail.com>
40217 * combine.c (try_combine): Don't ignore result of overlap checking
40218 loop. Combine overlap & asm check into single loop.
40220 2017-01-12 Richard Biener <rguenther@suse.de>
40222 * tree-pretty-print.c (dump_generic_node): Provide -gimple
40223 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
40225 2017-01-12 Richard Biener <rguenther@suse.de>
40227 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
40228 and TS_TARGET_OPTION directly derive from TS_BASE.
40229 * tree-core.h (tree_optimization_option): Derive from tree_base.
40230 (tree_target_option): Likewise.
40232 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
40234 * config/i386/i386.c (memory_address_length): Increase len
40235 only when rip_relative_addr_p returns false.
40237 2017-01-11 Julia Koval <julia.koval@intel.com>
40239 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
40240 (OPTION_MASK_ISA_SGX_SET): New.
40241 (ix86_handle_option): Handle OPT_msgx.
40242 * config.gcc: Added sgxintrin.h.
40243 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
40244 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
40245 * config/i386/i386.c (ix86_target_string): Add -msgx.
40247 (ix86_option_override_internal): Handle new options.
40248 (ix86_valid_target_attribute_inner_p): Add sgx.
40249 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
40250 * config/i386/i386.opt: Add msgx.
40251 * config/i386/sgxintrin.h: New file.
40252 * config/i386/x86intrin.h: Add sgxintrin.h.
40254 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40257 * fold-const.c (maybe_nonzero_address): Return 1 for function
40259 (tree_single_nonzero_warnv_p): Don't handle function local objects
40263 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
40266 2017-01-11 David Malcolm <dmalcolm@redhat.com>
40269 * opts.c: Include "spellcheck.h"
40270 (struct string_fragment): New struct.
40271 (struct edit_distance_traits<const string_fragment &>): New
40273 (get_closest_sanitizer_option): New function.
40274 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
40276 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40278 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
40280 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
40281 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
40282 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
40283 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
40284 for initial die_offset if dwarf_split_debug_info.
40285 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
40286 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
40288 (output_skeleton_debug_sections): Formatting fix. Use
40289 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
40290 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
40292 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
40294 * config/arm/cortex-a53.md: Add bypasses for
40295 cortex_a53_r2f_cvt.
40296 (cortex_a53_r2f): Only use for transfers.
40297 (cortex_a53_f2r): Likewise.
40298 (cortex_a53_r2f_cvt): Add reservation for conversions.
40299 (cortex_a53_f2r_cvt): Likewise.
40301 2017-01-11 Tamar Christina <tamar.christina@arm.com>
40303 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
40304 to all inlined functions, change static to extern.
40306 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
40309 * config/arm/arm.c (legitimize_pic_address): Handle reference to
40311 (arm_assemble_integer): Likewise.
40313 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
40315 * config.gcc: Use new awk script to check CPU, FPU and architecture
40316 parameters for --with-... options.
40317 * config/arm/parsecpu.awk: New file
40318 * config/arm/arm-cpus.in: New file.
40319 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
40321 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
40323 * config/arm/t-arm: Update dependency rules.
40324 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
40325 of processing .def files.
40326 * config/arm/genopt.sh: Deleted.
40327 * config/arm/gentune.sh: Deleted.
40328 * config/arm/arm-cores.def: Deleted.
40329 * config/arm/arm-arches.def: Deleted.
40330 * config/arm/arm-fpus.def: Deleted.
40331 * config/arm/arm-tune.md: Regenerated.
40332 * config/arm/arm-tables.opt: Regenerated.
40333 * config/arm/arm-cpu.h: New generated file.
40334 * config/arm/arm-cpu-data.h: New generated file.
40335 * config/arm/arm-cpu-cdata.h: New generated file.
40337 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
40340 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
40342 (input_varpool_node): Unpack dynamically_initialized bit.
40344 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
40346 PR rtl-optimization/79032
40347 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
40348 the alignment of the adjusted memory reference against that of MODE,
40349 instead of the alignment of the original memory reference.
40351 2017-01-11 Martin Jambor <mjambor@suse.cz>
40353 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
40355 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
40356 decorated functions.
40358 2017-01-11 Richard Biener <rguenther@suse.de>
40360 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
40361 set range/nonnull info for PHI results. Do not set it on
40362 stmts marked for removal.
40364 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
40366 * expr.c (store_field): In the bitfield case, fetch the return value
40367 from the registers before applying a single big-endian adjustment.
40368 Always do a final load for a BLKmode value not larger than a word.
40370 2017-01-10 David Malcolm <dmalcolm@redhat.com>
40373 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40374 that we correctly handle column numbers greater than
40375 LINE_MAP_MAX_COLUMN_NUMBER.
40377 2017-01-10 Martin Sebor <msebor@redhat.com>
40379 PR middle-end/78245
40380 * gimple-ssa-sprintf.c (get_destination_size): Call
40381 {init,fini}object_sizes.
40382 * tree-object-size.c (addr_object_size): Adjust.
40383 (pass_through_call): Adjust.
40384 (pass_object_sizes::execute): Adjust.
40385 * tree-object-size.h (fini_object_sizes): Declare.
40387 2017-01-10 Martin Sebor <msebor@redhat.com>
40389 PR tree-optimization/78775
40390 * builtins.c (get_size_range): Move...
40391 * calls.c: ...to here.
40392 (alloc_max_size): Accept zero argument.
40393 (operand_signed_p): Remove.
40394 (maybe_warn_alloc_args_overflow): Call get_size_range.
40395 * calls.h (get_size_range): Declare.
40397 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
40399 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
40400 from TI's devices.csv file as of September 2016.
40401 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
40403 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
40405 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
40406 * doc/invoke.texi: Likewise.
40407 * doc/md.texi: Likewise.
40408 * doc/objc.texi: Likewise.
40410 2017-01-10 Joshua Conner <joshconner@google.com>
40412 * config/arm/fuchsia-elf.h: New file.
40413 * config/fuchsia.h: New file.
40414 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
40415 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
40417 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
40419 2016-01-10 Richard Biener <rguenther@suse.de>
40421 PR tree-optimization/79034
40422 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
40423 Propagate out degenerate PHIs in the joiner.
40425 2017-01-10 Martin Liska <mliska@suse.cz>
40427 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
40428 (sort_congruence_classes_by_decl_uid): Likewise.
40429 (sort_congruence_class_groups_by_decl_uid): Likewise.
40430 (sem_item_optimizer::merge_classes): Sort class, groups in these
40431 classes and members in the groups by DECL_UID of declarations.
40432 This would make merge operations stable.
40434 2017-01-10 Martin Liska <mliska@suse.cz>
40436 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
40437 usage of m_classes_vec.
40438 (sem_item_optimizer::~sem_item_optimizer): Likewise.
40439 (sem_item_optimizer::get_group_by_hash): Likewise.
40440 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
40441 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
40442 (sem_item_optimizer::verify_classes): Likewise.
40443 (sem_item_optimizer::process_cong_reduction): Likewise.
40444 (sem_item_optimizer::dump_cong_classes): Likewise.
40445 (sem_item_optimizer::merge_classes): Likewise.
40446 * ipa-icf.h (congruence_class_hash): Rename from
40447 congruence_class_group_hash. Remove declaration of m_classes_vec.
40449 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
40451 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
40452 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
40453 * config.gcc: Add avx512vpopcntdqintrin.h.
40454 * config/i386/avx512vpopcntdqintrin.h: New.
40455 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
40456 * config/i386/i386-builtin-types.def: Add new types.
40457 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
40458 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
40459 __builtin_ia32_vpopcountq_v8di_mask): New.
40460 * config/i386/i386-c.c (ix86_target_macros_internal): Define
40461 __AVX512VPOPCNTDQ__.
40462 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
40463 (PTA_AVX512VPOPCNTDQ): Define.
40464 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
40465 TARGET_AVX512VPOPCNTDQ_P): Define.
40466 * config/i386/i386.opt: Add mavx512vpopcntdq.
40467 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
40468 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
40470 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40472 PR middle-end/77484
40473 * predict.def (PRED_CALL): Set to 67.
40475 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
40477 * expr.c (store_field): In the bitfield case, if the value comes from
40478 a function call and is of an aggregate type returned in registers, do
40479 not modify the field mode; extract the value in all cases if the mode
40480 is BLKmode and the size is not larger than a word.
40482 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
40485 * config/i386/cpuid.h: Fix undefined behavior.
40487 2017-01-04 Jeff Law <law@redhat.com>
40489 PR tree-optimization/79007
40490 PR tree-optimization/67955
40491 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
40492 conservative for pt.null when flag_non_call_exceptions is on.
40494 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40496 PR translation/79019
40497 PR translation/79020
40498 * params.def (PARAM_INLINE_MIN_SPEEDUP,
40499 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
40500 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
40502 * config/avr/avr.opt (maccumulate-args): Likewise.
40503 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
40504 * common.opt (freport-bug): Likewise.
40505 * cif-code.def (CIF_FINAL_ERROR): Likewise.
40506 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
40507 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
40508 translatable string.
40509 * config/i386/i386.c (function_value_32): Likewise.
40510 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
40511 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
40513 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
40514 * common/config/msp430/msp430-common.c (msp430_handle_option):
40516 * symtab.c (symtab_node::verify_base): Likewise.
40517 * opts.c (set_debug_level): Likewise.
40518 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
40519 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
40520 missing whitespace to translatable strings.
40521 * config/avr/avr.md (bswapsi2): Fix typo in comment.
40522 * config/sh/superh.h: Likewise.
40523 * config/i386/xopintrin.h: Likewise.
40524 * config/i386/znver1.md: Likewise.
40525 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
40526 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
40527 * double-int.h (struct double_int): Likewise.
40528 * double-int.c (div_and_round_double): Likewise.
40529 * wide-int.cc: Likewise.
40530 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
40531 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
40532 * cfgcleanup.c (crossjumps_occured): Renamed to ...
40533 (crossjumps_occurred): ... this.
40534 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
40537 PR tree-optimization/78899
40538 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
40539 returning bool return struct loop *, NULL for failure and the new
40541 (versionable_outer_loop_p): Don't version outer loop if it has
40542 dont_vectorized bit set.
40543 (tree_if_conversion): When versioning outer loop, ensure
40544 tree_if_conversion is performed also on the inner loop of the
40545 non-vectorizable outer loop copy.
40546 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
40547 LOOP_VECTORIZED in inner loop of the scalar outer loop and
40548 prevent vectorization of it.
40549 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
40550 the outer loop vectorization of the non-scalar version is attempted
40551 before vectorization of the inner loop in scalar version. If
40552 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
40553 vectorization of its inner loop.
40554 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
40555 has 2 inner loops, rename also on edges from bb whose single pred
40556 is outer_loop->header. Fix typo in function comment.
40558 2017-01-09 Martin Sebor <msebor@redhat.com>
40561 * asan.c (asan_emit_stack_protection): Increase local buffer size
40562 to avoid snprintf truncation warning.
40564 2017-01-09 Andrew Pinski <apinski@cavium.com>
40566 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
40567 to reference thunderx2t99 for the tuning structure
40568 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
40570 (thunderx2t99_extra_costs): This.
40571 * config/aarch64/aarch64-tune.md: Regenerate.
40572 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
40573 (vulcan_addrcost_table): This.
40574 (vulcan_regmove_cost): Rename to ...
40575 (thunderx2t99_regmove_cost): This.
40576 (vulcan_vector_cost): Rename to ...
40577 (thunderx2t99_vector_cost): this.
40578 (vulcan_branch_cost): Rename to ...
40579 (thunderx2t99_branch_cost): This.
40580 (vulcan_tunings): Rename to ...
40581 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
40582 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
40584 2017-01-09 Martin Jambor <mjambor@suse.cz>
40588 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
40589 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
40590 (propagate_vr_accross_jump_function): Use the above function for all
40591 value range computations for pass-through jump functions and type
40592 converasion from explicit value range values.
40593 (ipcp_propagate_stage): Do not attempt to deduce types of formal
40594 parameters from TYPE_ARG_TYPES.
40595 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
40596 (ipa_write_node_info): Stream type of the actual argument.
40597 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
40599 2017-01-09 Martin Liska <mliska@suse.cz>
40602 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
40603 (lookup_compiler): Do not show error message with have_E.
40605 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40607 PR tree-optimization/78938
40608 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
40609 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
40610 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
40611 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
40614 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40616 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
40619 2017-01-09 Richard Biener <rguenther@suse.de>
40621 PR tree-optimization/78997
40622 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
40623 name condition properly.
40625 2017-01-09 Richard Biener <rguenther@suse.de>
40628 * dwarf2out.c (is_cxx): New overload with context.
40629 (is_naming_typedef_decl): Use it.
40631 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40633 * invoke.texi (Option Summary): Correct spacing in option lists
40634 and add line breaks to fix over-long lines.
40636 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40638 PR middle-end/17660
40640 * extend.texi (Common Variable Attributes): Add xref to GCC
40641 Internals manual to explain mode attribute keywords.
40643 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40646 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
40647 and Preprocessor Options.
40648 (Options for Linking): Document -pthread here....
40649 (RS/6000 and PowerPC Options): ...not here.
40650 (Solaris 2 Options): ...or here.
40651 * doc/cppopts.texi: Document -pthread.
40653 2017-01-08 Martin Sebor <msebor@redhat.com>
40655 PR middle-end/77708
40656 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
40657 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
40658 New member functions.
40659 (format_directive): Used them.
40661 (pass_sprintf_length::handle_gimple_call): Same.
40662 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
40663 to avoid truncation for any argument.
40664 (extract_affine_mul): Same.
40665 * tree.c (get_file_function_name): Same.
40667 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40669 PR middle-end/77484
40670 * predict.def (PRED_INDIR_CALL): Set to 86.
40672 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40674 PR preprocessor/54124
40675 * doc/cppopts.texi: Reformat -d subtable to list the full name
40676 of the options. Add cross-reference to the docs for the general
40677 compiler -d options.
40678 * doc/invoke.texi (Developer Options): Add cross-reference to the
40679 preprocessor-specific -d option documentation.
40681 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40683 PR preprocessor/13498
40684 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
40685 redudant material, and reflect new command-line options.
40686 (System Headers): Likewise.
40688 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40690 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
40691 -isystem, and -idirafter. Copy-edit.
40692 * doc/cppopts.texi: Copy-edit. Remove contradiction about
40693 default for -ftrack-macro-expansion. Delete obsolete and
40694 badly-formatted implementation details about -fdebug-cpp output.
40695 * doc/cppwarnopts.texi: Copy-edit.
40697 2017-01-07 David Malcolm <dmalcolm@redhat.com>
40700 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40701 that the transition from a max line width >= 1<<10 to narrower
40702 lines works correctly.
40704 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
40706 * doc/options.texi (PerFunction): New.
40707 * opt-functions.awk (switch_flags): Map both Optimization and
40708 PerFunction to CL_OPTIMIZATION.
40709 * opth-gen.awk: Test for PerFunction flag along with
40711 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
40712 it only when the latter is present. Skip those that don't in
40713 the hash function generator.
40714 * common.opt (fvar-tracking): Mark as PerFunction instead of
40716 (fvar-tracking-assignments): Likewise.
40717 (fvar-tracking-assignments-toggle): Likewise.
40718 (fvar-tracking-uninit): Likewise.
40720 2017-01-07 Jakub Jelinek <jakub@redhat.com>
40722 PR translation/79018
40723 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
40726 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
40729 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
40730 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
40731 TARGET_LONG_JUMP_TABLE_OFFSETS.
40732 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
40733 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
40734 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
40735 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
40736 * config/m68k/m68k.md (tablejump expander): Likewise.
40737 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
40738 TARGET_LONG_JUMP_TABLE_OFFSETS.
40739 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
40740 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
40742 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
40743 David Holsgrove <david.holsgrove@xilinx.com>
40745 * common/config/microblaze/microblaze-common.c
40746 (TARGET_EXCEPT_UNWIND_INFO): Remove.
40747 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
40749 * config/microblaze/microblaze.c (microblaze_must_save_register)
40750 (microblaze_expand_epilogue, microblaze_return_addr): Handle
40752 (microblaze_eh_return): New function.
40753 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
40754 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
40755 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
40756 * config/microblaze/microblaze.md (eh_return): New pattern.
40758 2017-01-06 Jakub Jelinek <jakub@redhat.com>
40760 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
40761 GCC_DIAGNOSTIC_STRINGIFY): Define.
40763 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
40765 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40767 * config/arm/arm.md (<mcrr>): New.
40769 * config/arm/arm.c (arm_arch5te): New.
40770 (arm_option_override): Set arm_arch5te.
40771 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
40773 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
40774 (arm_mcrr_qualifiers): ... this. New.
40775 (MRRC_QUALIFIERS): Define to...
40776 (arm_mrrc_qualifiers): ... this. New.
40777 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
40779 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
40780 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
40781 (MRRCI, mrrc, MRRC): New.
40782 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
40783 VUNSPEC_MRRC2): New.
40785 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40787 * config/arm/arm.md (<mcr>): New.
40789 * config/arm/arm.c (arm_coproc_builtin_available): Add
40790 support for mcr, mrc, mcr2 and mrc2.
40791 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
40792 (arm_mcr_qualifiers): ... this. New.
40793 (MRC_QUALIFIERS): Define to ...
40794 (arm_mrc_qualifiers): ... this. New.
40795 (MCR_QUALIFIERS): Define to ...
40796 (arm_mcr_qualifiers): ... this. New.
40797 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
40799 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
40800 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
40801 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
40802 VUNSPEC_MRC2): New.
40804 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40806 * config/arm/arm.md (*ldc): New.
40810 * config/arm/arm.c (arm_coproc_builtin_available): Add
40811 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
40812 (arm_coproc_ldc_stc_legitimate_address): New.
40813 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
40814 'qualifier_const_pointer'.
40815 (LDC_QUALIFIERS): Define to...
40816 (arm_ldc_qualifiers): ... this. New.
40817 (STC_QUALIFIERS): Define to...
40818 (arm_stc_qualifiers): ... this. New.
40819 * config/arm/arm-protos.h
40820 (arm_coproc_ldc_stc_legitimate_address): New.
40821 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
40822 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
40823 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
40824 stc2, stcl, stc2l): New.
40825 * config/arm/constraints.md (Uz): New.
40826 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
40827 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
40828 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
40829 VUNSPEC_STC2L): New.
40831 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40833 * config/arm/arm.md (<cdp>): New.
40834 * config/arm/arm.c (neon_const_bounds): Rename this ...
40835 (arm_const_bounds): ... this.
40836 (arm_coproc_builtin_available): New.
40837 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
40838 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
40839 (CDP_QUALIFIERS): Define to...
40840 (arm_cdp_qualifiers): ... this. New.
40842 (arm_expand_builtin_args): Add case for 6 arguments.
40843 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
40844 (arm_const_bounds): ... this.
40845 (arm_coproc_builtin_available): New.
40846 * config/arm/arm_acle.h (__arm_cdp): New.
40848 * config/arm/arm_acle_builtins.def (cdp): New.
40850 * config/arm/iterators.md (CDPI,CDP,cdp): New.
40851 * config/arm/neon.md: Rename all 'neon_const_bounds' to
40852 'arm_const_bounds'.
40853 * config/arm/types.md (coproc): New.
40854 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
40855 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
40856 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
40857 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
40859 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40861 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
40862 (UBINOP_QUALIFIERS): New.
40864 (acle_builtin_data): New. Change comment.
40865 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
40866 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
40867 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
40868 arm_acle_builtins.def.
40869 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
40870 (arm_init_acle_builtins): New.
40871 (CRC32_BUILTIN): Remove.
40872 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
40873 crc32cb, crc32ch and crc32cw.
40874 (arm_init_crc32_builtins): Remove.
40875 (arm_init_builtins): Use arm_init_acle_builtins rather
40876 than arm_init_crc32_builtins.
40877 (arm_expand_acle_builtin): New.
40878 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
40879 * config/arm/arm_acle_builtins.def: New.
40881 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40883 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
40884 (arm_builtin_datum): ... this.
40885 (arm_init_neon_builtin): Rename to ...
40886 (arm_init_builtin): ... this. Add a new parameters PREFIX
40887 and USE_SIG_IN_NAME.
40888 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
40889 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
40890 'arm_builtin_datum'.
40891 (arm_init_vfp_builtins): Likewise.
40892 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
40893 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
40894 (arm_expand_neon_args): Rename to ...
40895 (arm_expand_builtin_args): ... this. Rename builtin_arg
40896 enum values and differentiate between ARG_BUILTIN_MEMORY
40897 and ARG_BUILTIN_NEON_MEMORY.
40898 (arm_expand_neon_builtin_1): Rename to ...
40899 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
40900 values, arm_expand_builtin_args and add bool parameter NEON.
40901 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
40902 (arm_expand_vfp_builtin): Likewise.
40903 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
40905 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40907 PR middle-end/77484
40908 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
40909 * predict.c (tree_estimate_probability_bb): Reverse direction of
40910 polymorphic call predictor.
40912 2017-01-06 David Malcolm <dmalcolm@redhat.com>
40914 * passes.c (execute_one_pass): Split out pass-skipping logic into...
40915 (determine_pass_name_match): ...this new function and...
40916 (should_skip_pass_p): ...this new function.
40918 2017-01-06 Nathan Sidwell <nathan@acm.org>
40920 * ipa-visibility.c (function_and_variable_visibility): Reformat
40921 comments and long lines. Remove extrneous if.
40922 * symtab.c (symtab_node::make_decl_local): Fix code format.
40923 (symtab_node::set_section_for_node): Fix comment typo.
40925 2017-01-06 Martin Liska <mliska@suse.cz>
40928 * lra-constraints.c: Rename invariant to lra_invariant.
40929 * predict.c (set_even_probabilities): Initialize e to NULL.
40931 2017-01-05 Martin Sebor <msebor@redhat.com>
40933 PR tree-optimization/78910
40934 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
40935 (format_integer): Correct off-by-one error in the handling
40936 of precision with negative numbers in signed conversions..
40938 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
40940 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
40942 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40944 PR tree-optimization/71016
40945 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
40946 factor_out_conditional_conversion. Formatting fix.
40947 (factor_out_conditional_conversion): Add cond_stmt argument.
40948 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
40949 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
40952 2017-01-05 David Malcolm <dmalcolm@redhat.com>
40954 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
40955 read-rtl-function.o, and selftest-rtl.o.
40956 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
40957 (selftest::aarch64_test_loading_full_dump): New function.
40958 (selftest::aarch64_run_selftests): New function.
40959 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
40960 selftest::aarch64_run_selftests.
40961 * config/i386/i386.c
40962 (selftest::ix86_test_loading_dump_fragment_1): New function.
40963 (selftest::ix86_test_loading_call_insn): New function.
40964 (selftest::ix86_test_loading_full_dump): New function.
40965 (selftest::ix86_test_loading_unspec): New function.
40966 (selftest::ix86_run_selftests): Call the new functions.
40967 * emit-rtl.c (maybe_set_max_label_num): New function.
40968 * emit-rtl.h (maybe_set_max_label_num): New decl.
40969 * function.c (instantiate_decls): Guard call to
40970 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
40971 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
40973 * gensupport.c (gen_reader::gen_reader): Pass "false"
40974 for new "compact" param of rtx_reader.
40975 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
40976 rather than an empty string for NULL strings.
40977 * read-md.c: Potentially include config.h rather than bconfig.h.
40978 Wrap include of errors.h with #ifdef GENERATOR_FILE.
40979 (have_error): New global, copied from errors.c.
40980 (md_reader::read_name): Rename to...
40981 (md_reader::read_name_1): ...this, adding "out_loc" param,
40982 and converting "missing name or number" to returning false, rather
40984 (md_reader::read_name): Reimplement in terms of read_name_1.
40985 (md_reader::read_name_or_nil): New function.
40986 (md_reader::read_string): Handle "(nil)" by returning NULL.
40987 (md_reader::md_reader): Add new param "compact".
40988 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
40989 (md_reader::read_file): New method.
40990 * read-md.h (md_reader::md_reader): Add new param "compact".
40991 (md_reader::read_file): New method.
40992 (md_reader::is_compact): New accessor.
40993 (md_reader::read_name): Convert return type from void to file_location.
40994 (md_reader::read_name_or_nil): New decl.
40995 (md_reader::read_name_1): New decl.
40996 (md_reader::m_compact): New field.
40997 (noop_reader::noop_reader): Pass "false" for new "compact" param
40999 (rtx_reader::rtx_reader): Add new "compact" param.
41000 (rtx_reader::read_rtx_operand): Make virtual and convert return
41001 type from void to rtx.
41002 (rtx_reader::read_until): New decl.
41003 (rtx_reader::handle_any_trailing_information): New virtual function.
41004 (rtx_reader::postprocess): New virtual function.
41005 (rtx_reader::finalize_string): New virtual function.
41006 (rtx_reader::m_in_call_function_usage): New field.
41007 (rtx_reader::m_reuse_rtx_by_id): New field.
41008 * read-rtl-function.c: New file.
41009 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
41010 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
41011 (selftest::verify_three_block_rtl_cfg): New decl.
41012 * read-rtl-function.h: New file.
41013 * read-rtl.c: Potentially include config.h rather than bconfig.h.
41014 For host, include function.h, memmodel.h, and emit-rtl.h.
41015 (one_time_initialization): New function.
41016 (struct compact_insn_name): New struct.
41017 (compact_insn_names): New array.
41018 (find_code): Handle insn codes in compact dumps.
41019 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
41020 (bind_subst_iter_and_attr): Likewise.
41021 (add_condition_to_string): Likewise.
41022 (add_condition_to_rtx): Likewise.
41023 (apply_attribute_uses): Likewise.
41024 (add_current_iterators): Likewise.
41025 (apply_iterators): Likewise.
41026 (initialize_iterators): Guard usage of apply_subst_iterator with
41027 #ifdef GENERATOR_FILE.
41028 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
41029 (md_reader::read_mapping): Likewise.
41030 (add_define_attr_for_define_subst): Likewise.
41031 (add_define_subst_attr): Likewise.
41032 (read_subst_mapping): Likewise.
41033 (check_code_iterator): Likewise.
41034 (rtx_reader::read_rtx): Likewise. Move one-time initialization
41036 (one_time_initialization): New function.
41037 (rtx_reader::read_until): New method.
41038 (read_flags): New function.
41039 (parse_reg_note_name): New function.
41040 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
41041 Handle reuse_rtx ids.
41042 Wrap iterator lookup within #ifdef GENERATOR_FILE.
41043 Add parsing support for RTL dumps, mirroring the special-cases in
41044 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
41045 values, and calling handle_any_trailing_information.
41046 (rtx_reader::read_rtx_operand): Convert return type from void
41047 to rtx, returning return_rtx. Handle case 'e'. Call
41048 finalize_string on XSTR and XTMPL fields.
41049 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
41050 "(nil)" values were omitted. Call the postprocess vfunc on the
41052 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
41053 class ctor. Initialize m_in_call_function_usage. Call
41054 one_time_initialization.
41055 * rtl-tests.c (selftest::test_uncond_jump): Call
41056 set_new_first_and_last_insn.
41057 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
41058 * selftest-rtl.c: New file.
41059 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
41060 (selftest::get_insn_by_uid): New decl.
41061 * selftest-run-tests.c (selftest::run_tests): Call
41062 read_rtl_function_c_tests.
41063 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
41064 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
41067 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
41069 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
41070 operands in a special way. Assert that pos+len <= mode precision.
41072 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41074 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
41075 3 argument Alias with unlimited for the negative form.
41076 (fno-vect-cost-model): Removed.
41078 2017-01-05 Martin Liska <mliska@suse.cz>
41080 * hsa-gen.c (gen_hsa_divmod): New function.
41081 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
41083 2017-01-05 Martin Liska <mliska@suse.cz>
41086 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
41089 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41091 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
41092 small constant length operands.
41094 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41096 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
41097 between loop iterations.
41099 2017-01-05 Martin Liska <mliska@suse.cz>
41102 * gimplify.c (gimplify_decl_expr): Compare to
41103 asan_poisoned_variables instread of checking flags.
41104 (gimplify_target_expr): Likewise.
41105 (gimplify_expr): Likewise.
41106 (gimplify_function_tree): Conditionally initialize
41107 asan_poisoned_variables.
41109 2017-01-04 Jeff Law <law@redhat.com>
41111 PR tree-optimizatin/78812
41112 * rtl.h (contains_mem_rtx_p): Prototype.
41113 * ifcvt.c (containts_mem_rtx_p): Move from here to...
41114 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
41115 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
41116 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
41117 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
41119 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41121 * input.c (assert_char_at_range): Default-initialize actual_range.
41123 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41125 * df-scan.c (df_ref_create_structure): Make regno unsigned,
41126 to match the caller.
41128 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41130 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
41131 insns after final jump in test to emit dummy move.
41133 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41135 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
41136 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
41138 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41140 * multiple_target.c (create_dispatcher_calls): Init e_next.
41141 * tree-ssa-loop-split.c (split_loop): Init border.
41142 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
41145 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
41150 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
41151 (altivec_register_operand): Do not return true if the operand
41152 contains a SUBREG mixing SImode and SFmode.
41153 (vsx_register_operand): Likewise.
41154 (vsx_reg_sfsubreg_ok): New predicate.
41155 (vfloat_operand): Do not return true if the operand contains a
41156 SUBREG mixing SImode and SFmode.
41157 (vint_operand): Likewise.
41158 (vlogical_operand): Likewise.
41159 (gpc_reg_operand): Likewise.
41160 (int_reg_operand): Likewise.
41161 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
41162 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
41163 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
41165 (rs6000_emit_move_si_sf_subreg): New helper function.
41166 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
41167 fixup SUBREGs involving SImode and SFmode.
41168 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
41169 numbers for the new peephole2 optimization.
41170 (peephole2 for SFmode unions): New peephole2 to optimize cases in
41171 the GLIBC math library that do AND/IOR/XOR operations on single
41172 precision floating point.
41173 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
41174 target macros to say whether we need to avoid SUBREGs mixing
41176 (TARGET_ALLOW_SF_SUBREG): Likewise.
41177 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
41178 (UNSPEC_SI_FROM_SF): Likewise.
41179 (iorxor): Change spacing.
41180 (and_ior_xor): New iterator for AND, IOR, and XOR.
41181 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
41182 (movdi_from_sf_zero_ext): Likewise.
41183 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
41184 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
41185 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
41186 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
41187 (fms<mode>4): Likewise.
41188 (fnma<mode>4): Likewise.
41189 (fnms<mode>4): Likewise.
41190 (nfma<mode>4): Likewise.
41191 (nfms<mode>4): Likewise.
41193 2017-01-04 Marek Polacek <polacek@redhat.com>
41196 * doc/invoke.texi: Document -Wpointer-compare.
41198 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41200 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
41203 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
41204 descriptions for -gdwarf-5 and emit them as uleb128 instead of
41207 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
41210 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
41211 documentation of the powerpc_popcntb_ok attribute.
41212 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
41213 code to issue warning messages if a requested CPU configuration is
41214 not supported by the binary (assembler and loader) toolchain.
41215 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
41216 made to define a built-in function that has been disabled.
41217 (paired_init_builtins): Add assertion to prevent ICE if attempt is
41218 made to define a built-in function that has been disabled.
41219 (altivec_init_builtins): Add comment explaining why definition
41220 of the DST built-in functions is not preceded by an assertion
41221 check. Add assertions to prevent ICE if attempts are made to
41222 define an altivec predicate or an abs* built-in function that has
41224 (htm_init_builtins): Add comment explaining why definition of the
41225 htm built-in functions is not preceded by an assertion check.
41227 2017-01-04 Jeff Law <law@redhat.com>
41229 PR tree-optimizatin/67955
41230 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
41231 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
41232 the points-to solution does not include pt_null. Use DECL_PT_UID
41235 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
41237 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
41238 Use gen_int_mode instead of gen_lopwart for const_int operands.
41240 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41242 PR tree-optimization/71563
41243 * match.pd: Simplify X << Y into X if Y is known to be 0 or
41244 out of range value - has low bits known to be zero.
41246 2017-01-04 Alan Modra <amodra@gmail.com>
41248 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
41249 * configure: Regenerate.
41250 * config.in: Regenerate.
41252 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41255 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
41256 a substring of the message, but strcmp with the whole message. Ifdef
41257 ENABLE_NLS, translate the message first using dgettext.
41259 2017-01-03 Jeff Law <law@redhat.com>
41261 PR tree-optimizatin/78856
41262 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
41263 (mark_threaded_blocks): Remove code to truncate thread paths that
41264 cross multiple loop headers. Instead invalidate the cached loop
41265 iteration information and handle case of a thread path walking
41266 into an irreducible region.
41268 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
41271 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
41272 assertions. Add support for doing the signbit if the IEEE 128-bit
41273 floating point value is in a GPR.
41274 * config/rs6000/rs6000.md (Fsignbit): Delete.
41275 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
41276 Update the length attribute if the value is in a GPR.
41277 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
41278 the sign or zero extension instruction, since the value is always 0/1.
41279 (signbit<mode>2_dm2): Delete using <Fsignbit>.
41282 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
41283 extracting SImode to a GPR register so that we can generate a
41284 store, limit the vector to be in a traditional Altivec register
41285 for the vextuwrx instruction.
41287 2017-01-03 Ian Lance Taylor <iant@google.com>
41289 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
41291 2017-01-03 Martin Sebor <msebor@redhat.com>
41293 PR tree-optimization/78696
41294 * gimple-ssa-sprintf.c (format_floating): Correct handling of
41295 precision. Use MPFR for %f for greater fidelity. Correct handling
41297 (pass_sprintf_length::compute_format_length): Set width and precision
41298 specified by asrerisk to void_node for vararg functions.
41299 (try_substitute_return_value): Adjust dump output.
41301 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
41303 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
41305 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
41307 * doc/invoke.texi (SPARC options): Document -mlra as the default.
41308 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
41309 -mlra/-mno-lra was passed to the compiler.
41311 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
41313 PR rtl-optimization/65618
41314 * emit-rtl.c (try_split): Move initialization of "before" and
41315 "after" to just before the call to emit_insn_after_setloc.
41317 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
41319 * doc/md.texi (Standard Names): Remove reference to Java frontend.
41321 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
41323 * dwarf2out.c (gen_enumeration_type_die): When
41324 -gno-strict-dwarf, add a DW_AT_encoding attribute.
41326 2017-01-03 Jakub Jelinek <jakub@redhat.com>
41328 PR tree-optimization/78965
41329 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
41330 Change first argument from const call_info & to call_info &. For %n
41331 set info.nowrite to false.
41333 PR middle-end/78901
41334 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
41335 possibly throwing calls.
41337 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
41338 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
41339 and fns handling, rather than in a separate case SSA_NAME.
41341 2017-01-02 Jeff Law <law@redhat.com>
41343 * config/darwin-driver.c (darwin_driver_init): Const-correctness
41344 fixes for first_period and second_period variables.
41346 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
41349 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
41350 (*insvqi_1): New insn pattern.
41351 (*insvqi_1_mem_rex64): Ditto.
41352 (*insvqi_2): Ditto.
41353 (*insvqi_3): Rename from *insvqi.
41355 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
41357 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
41359 * doc/cfg.texi (Edges): Remove reference to Java.
41360 (Maintaining the CFG): Ditto.
41362 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41364 PR middle-end/77674
41365 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
41366 transparent aliases.
41368 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41370 PR middle-end/77484
41371 * predict.def (PRED_CALL): Update hitrate.
41372 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
41373 * predict.c (tree_estimate_probability_bb): Split CALL predictor
41374 into direct/indirect/polymorphic variants.
41376 2017-01-01 Jakub Jelinek <jakub@redhat.com>
41378 Update copyright years.
41380 * gcc.c (process_command): Update copyright notice dates.
41381 * gcov-dump.c (print_version): Ditto.
41382 * gcov.c (print_version): Ditto.
41383 * gcov-tool.c (print_version): Ditto.
41384 * gengtype.c (create_file): Ditto.
41385 * doc/cpp.texi: Bump @copying's copyright year.
41386 * doc/cppinternals.texi: Ditto.
41387 * doc/gcc.texi: Ditto.
41388 * doc/gccint.texi: Ditto.
41389 * doc/gcov.texi: Ditto.
41390 * doc/install.texi: Ditto.
41391 * doc/invoke.texi: Ditto.
41393 Copyright (C) 2017 Free Software Foundation, Inc.
41395 Copying and distribution of this file, with or without modification,
41396 are permitted in any medium without royalty provided the copyright
41397 notice and this notice are preserved.