1 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
4 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
5 constraints to operand 2.
6 (*ashl<mode>3_mask_1): Ditto.
7 (*<shift_insn><mode>3_mask): Ditto.
8 (*<shift_insn><mode>3_mask_1): Ditto.
9 (*<rotate_insn><mode>3_mask): Ditto.
10 (*<rotate_insn><mode>3_mask_1): Ditto.
12 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
14 * reorg.c (make_return_insns): Reemit each insn with its own location.
16 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
19 * c-family/c-semantics.c (pop_stmt_list): Propagate side
20 effects from single nondebug stmt to container list.
22 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
24 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
25 generation for cases where splatting a value is not useful.
26 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
27 across a vec_duplicate and a paradoxical subreg forming a vector
30 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
32 * combine.c (simplify_set): Do not transform subregs to zero_extends
33 if the destination is not a scalar int mode.
35 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
38 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
39 shared zero if the input has overflowed.
41 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
43 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
44 support for Cortex-A55 and Cortex-A75.
46 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
48 * common/config/arm/arm-common.c (compare_opt_names): Add function
49 comment. Use strcmp instead of manual loop.
51 2017-12-21 Martin Liska <mliska@suse.cz>
54 * gcov-dump.c (dump_gcov_file): Do not read info about
55 support_unexecuted_blocks for gcda files.
57 2017-12-21 Jakub Jelinek <jakub@redhat.com>
59 PR rtl-optimization/82973
60 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
61 (valid_for_const_vector_p): ... this.
62 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
63 (valid_for_const_vector_p): ... this. Adjust function comment.
64 (gen_vec_duplicate): Adjust caller.
65 * optabs.c (expand_vector_broadcast): Likewise.
66 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
67 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
69 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
70 instead of GET_CODE == CONST_FIXED.
71 (simplify_subreg): Use CONST_FIXED_P macro instead of
72 GET_CODE == CONST_FIXED.
75 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
76 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
77 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
78 (ix86_option_override_internal): Adjust for
79 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
80 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
81 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
82 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
83 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
84 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
85 def_builtin2 for CET builtins.
86 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
87 up formatting in IX86_BUILTIN_RDPID code.
88 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
89 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
90 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
91 ix86_isa_flags2 to ix86_isa_flags.
92 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
94 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
95 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
96 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
98 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
99 Or in OPTION_MASK_ISA_AVX512F_SET.
100 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
101 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
102 (ix86_handle_option): Adjust for
103 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
104 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
106 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
107 Alan Hayward <alan.hayward@arm.com>
108 David Sherwood <david.sherwood@arm.com>
110 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
111 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
112 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
113 Track polynomial offsets.
115 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
116 Alan Hayward <alan.hayward@arm.com>
117 David Sherwood <david.sherwood@arm.com>
119 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
120 Treat drb->init as a poly_int. Fail if its misalignment wrt
121 vector_alignment isn't known.
123 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
124 Alan Hayward <alan.hayward@arm.com>
125 David Sherwood <david.sherwood@arm.com>
127 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
129 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
130 (vect_loop_versioning): Take the loop versioning threshold as a
132 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
133 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
134 versioning_threshold.
135 (vect_analyze_loop_2): Compute the loop versioning threshold
136 whenever loop versioning is needed, and store it in the new
137 field rather than combining it with the cost model threshold.
138 (vect_transform_loop): Update call to vect_loop_versioning.
139 Try to combine the loop versioning and cost thresholds here.
141 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
142 Alan Hayward <alan.hayward@arm.com>
143 David Sherwood <david.sherwood@arm.com>
145 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
146 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
147 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
148 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
149 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
150 (group_compare_offset): Update accordingly.
151 (split_small_address_groups_p): Likewise.
152 (record_use): Take addr_offset as a poly_uint64 rather than
153 an unsigned HOST_WIDE_INT.
154 (strip_offset): Return the offset as a poly_uint64 rather than
155 an unsigned HOST_WIDE_INT.
156 (record_group_use, split_address_groups): Track polynomial offsets.
157 (add_iv_candidate_for_use): Likewise.
158 (addr_offset_valid_p): Take the offset as a poly_int64 rather
159 than a HOST_WIDE_INT.
160 (strip_offset_1): Return the offset as a poly_int64 rather than
163 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
164 Alan Hayward <alan.hayward@arm.com>
165 David Sherwood <david.sherwood@arm.com>
167 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
168 rather than a HOST_WIDE_INT.
169 * tree.c (get_binfo_at_offset): Likewise.
171 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
172 Alan Hayward <alan.hayward@arm.com>
173 David Sherwood <david.sherwood@arm.com>
175 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
176 rather than a HOST_WIDE_INT.
177 * tree-sra.c (build_ref_for_offset): Likewise.
179 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
180 Alan Hayward <alan.hayward@arm.com>
181 David Sherwood <david.sherwood@arm.com>
183 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
185 * tree.c (mem_ref_offset): Likewise.
186 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
187 * builtins.c (get_object_alignment_2): Likewise.
188 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
189 * gimple-fold.c (get_base_constructor): Likewise.
190 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
191 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
193 * ipa-polymorphic-call.c
194 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
195 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
196 (get_ancestor_addr_info): Likewise.
197 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
198 * match.pd: Likewise.
199 * tree-data-ref.c (dr_analyze_innermost): Likewise.
200 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
201 * tree-eh.c (tree_could_trap_p): Likewise.
202 * tree-object-size.c (addr_object_size): Likewise.
203 * tree-ssa-address.c (copy_ref_info): Likewise.
204 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
205 (indirect_refs_may_alias_p): Likewise.
206 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
207 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
208 (non_rewritable_mem_ref_base): Likewise.
209 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
210 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
211 * varasm.c (decode_addr_const): Likewise.
213 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
214 Alan Hayward <alan.hayward@arm.com>
215 David Sherwood <david.sherwood@arm.com>
217 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
218 offsets for MEM_REFs.
220 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
221 Alan Hayward <alan.hayward@arm.com>
222 David Sherwood <david.sherwood@arm.com>
224 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
225 as a poly_uint64 rather than an unsigned int.
226 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
228 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
229 Alan Hayward <alan.hayward@arm.com>
230 David Sherwood <david.sherwood@arm.com>
232 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
234 (decode_addr_const): Update accordingly.
236 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
237 Alan Hayward <alan.hayward@arm.com>
238 David Sherwood <david.sherwood@arm.com>
240 * tree.h (bit_field_size, bit_field_offset): New functions.
241 * hsa-gen.c (gen_hsa_addr): Use them.
242 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
243 (simplify_vector_constructor): Likewise.
244 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
245 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
246 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
247 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
250 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
251 Alan Hayward <alan.hayward@arm.com>
252 David Sherwood <david.sherwood@arm.com>
254 * expr.h (emit_group_load, emit_group_load_into_temps)
255 (emit_group_store): Take the size as a poly_int64 rather than an int.
256 * expr.c (emit_group_load_1, emit_group_load): Likewise.
257 (emit_group_load_into_temp, emit_group_store): Likewise.
259 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
260 Alan Hayward <alan.hayward@arm.com>
261 David Sherwood <david.sherwood@arm.com>
263 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
264 an unsigned int to a poly_uint64.
265 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
266 sizes as poly_uint64s rather than unsigned ints.
267 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
270 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
271 Alan Hayward <alan.hayward@arm.com>
272 David Sherwood <david.sherwood@arm.com>
274 * lra-constraints.c (emit_inc): Change inc_amount from an int
277 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
278 Alan Hayward <alan.hayward@arm.com>
279 David Sherwood <david.sherwood@arm.com>
281 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
283 (add_stack_var, stack_var_cmp, partition_stack_vars)
284 (dump_stack_var_partition): Update accordingly.
285 (alloc_stack_frame_space): Take the size as a poly_int64 rather
286 than a HOST_WIDE_INT.
287 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
288 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
289 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
290 rather than a HOST_WIDE_INT.
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 * target.def (return_pops_args): Treat both the input and output
297 sizes as poly_int64s rather than HOST_WIDE_INTS.
298 * targhooks.h (default_return_pops_args): Update accordingly.
299 * targhooks.c (default_return_pops_args): Likewise.
300 * doc/tm.texi: Regenerate.
301 * emit-rtl.h (incoming_args): Change pops_args, size and
302 outgoing_args_size from int to poly_int64_pod.
303 * function.h (expr_status): Change x_pending_stack_adjust and
304 x_stack_pointer_delta from int to poly_int64.
305 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
306 (ARGS_SIZE_RTX): Update accordingly.
307 * calls.c (highest_outgoing_arg_in_use): Change from int to
309 (stack_usage_watermark, stored_args_watermark): New variables.
310 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
311 (emit_call_1): Change the stack_size and rounded_stack_size
312 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
314 (save_fixed_argument_area): Check stack_usage_watermark.
315 (initialize_argument_information): Change old_pending_adj from
316 a HOST_WIDE_INT * to a poly_int64_pod *.
317 (compute_argument_block_size): Return the size as a poly_int64
319 (finalize_must_preallocate): Track polynomial argument sizes.
320 (compute_argument_addresses): Likewise.
321 (internal_arg_pointer_based_exp): Track polynomial offsets.
322 (mem_overlaps_already_clobbered_arg_p): Rename to...
323 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
324 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
325 Check stored_args_used_watermark.
326 (load_register_parameters): Update accordingly.
327 (check_sibcall_argument_overlap_1): Likewise.
328 (combine_pending_stack_adjustment_and_call): Take the unadjusted
329 args size as a poly_int64 rather than an int. Return a bool
330 indicating whether the optimization was possible and return
331 the new adjustment by reference.
332 (check_sibcall_argument_overlap): Track polynomail argument sizes.
333 Update stored_args_watermark.
334 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
335 (expand_call): Likewise. Maintain stack_usage_watermark and
336 stored_args_watermark. Update calls to
337 combine_pending_stack_adjustment_and_call.
338 (emit_library_call_value_1): Handle polynomial argument sizes.
339 Call stack_region_maybe_used_p and mark_stack_region_used.
340 Maintain stack_usage_watermark.
341 (store_one_arg): Likewise. Update call to
342 mem_overlaps_already_clobbered_arg_p.
343 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
345 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
346 * config/microblaze/microblaze.c (microblaze_function_prologue):
348 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
349 TARGET_RETURN_POPS_ARGS interface.
350 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
352 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
353 * config/i386/i386.c (ix86_return_pops_args): Update for new
354 TARGET_RETURN_POPS_ARGS interface.
355 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
356 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
357 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
358 TARGET_RETURN_POPS_ARGS interface.
359 * config/vax/vax.c (vax_return_pops_args): Likewise.
360 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
361 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
362 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
363 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
364 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
365 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
366 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
367 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
368 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
369 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
370 and x_stack_pointer_delta from int to poly_int64.
371 * dojump.c (do_pending_stack_adjust): Update accordingly.
372 * explow.c (allocate_dynamic_stack_space): Handle polynomial
373 stack_pointer_deltas.
374 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
375 (pad_to_arg_alignment): Track polynomial offsets.
376 (assign_parm_find_stack_rtl): Likewise.
377 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
378 * toplev.c (output_stack_usage): Update reference to
379 current_function_pushed_stack_size.
381 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
382 Alan Hayward <alan.hayward@arm.com>
383 David Sherwood <david.sherwood@arm.com>
385 * function.c (in_arg_offset, var_offset, dynamic_offset)
386 (out_arg_offset, cfa_offset): Change from int to poly_int64.
387 (instantiate_new_reg): Return the new offset as a poly_int64_pod
388 rather than a HOST_WIDE_INT.
389 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
390 (instantiate_virtual_regs_in_insn): Likewise.
392 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
393 Alan Hayward <alan.hayward@arm.com>
394 David Sherwood <david.sherwood@arm.com>
396 * rtl.h (get_args_size, add_args_size_note): New functions.
397 (find_args_size_adjust): Return a poly_int64 rather than a
399 (fixup_args_size_notes): Likewise. Make the same change to the
400 end_args_size parameter.
401 * rtlanal.c (get_args_size, add_args_size_note): New functions.
402 * builtins.c (expand_builtin_trap): Use add_args_size_note.
403 * calls.c (emit_call_1): Likewise.
404 * explow.c (adjust_stack_1): Likewise.
405 * cfgcleanup.c (old_insns_match_p): Update use of
406 find_args_size_adjust.
407 * combine.c (distribute_notes): Track polynomial arg sizes.
408 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
409 end_true_args_size, beg_delay_args_size and end_delay_args_size
410 from HOST_WIDE_INT to poly_int64.
411 (add_cfi_args_size): Take the args_size as a poly_int64 rather
412 than a HOST_WIDE_INT.
413 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
414 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
415 polynomial arg sizes.
416 * emit-rtl.c (try_split): Use get_args_size.
417 * recog.c (peep2_attempt): Likewise.
418 * reload1.c (reload_as_needed): Likewise.
419 * expr.c (find_args_size_adjust): Return the adjustment as a
420 poly_int64 rather than a HOST_WIDE_INT.
421 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
422 to a poly_int64 and change the return type in the same way.
423 (emit_single_push_insn): Track polynomial arg sizes.
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 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
430 from HOST_WIDE_INT to poly_int64.
431 * expr.c (push_block, emit_push_insn): Likewise.
433 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
434 Alan Hayward <alan.hayward@arm.com>
435 David Sherwood <david.sherwood@arm.com>
437 * function.h (frame_space): Change start and length from HOST_WIDE_INT
439 (get_frame_size): Return the size as a poly_int64 rather than a
441 (frame_offset_overflow): Take the offset as a poly_int64 rather
442 than a HOST_WIDE_INT.
443 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
444 (assign_stack_temp): Likewise for the size.
445 * function.c (get_frame_size): Return a poly_int64 rather than
447 (frame_offset_overflow): Take the offset as a poly_int64 rather
448 than a HOST_WIDE_INT.
449 (try_fit_stack_local): Take the start, length and size as poly_int64s
450 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
451 rather than a HOST_WIDE_INT.
452 (add_frame_space): Take the start and end as poly_int64s rather than
454 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
455 (assign_stack_temp): Likewise for the size.
456 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
458 (find_temp_slot_from_address): Handle polynomial offsets.
459 (combine_temp_slots): Likewise.
460 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
462 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
463 poly_int64 rather than a HOST_WIDE_INT.
464 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
465 than a HOST_WIDE_INT.
466 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
467 polynomial frame offsets.
468 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
469 as a poly_int64 rather than an int.
470 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
471 * config/v850/v850-protos.h (compute_frame_size): Likewise.
472 * config/v850/v850.c (compute_frame_size): Likewise.
473 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
474 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
475 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
476 * config/pa/pa.c (pa_compute_frame_size): Likewise.
477 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
478 rather than a HOST_WIDE_INT.
479 * explow.c (get_dynamic_stack_base): Likewise.
480 * final.c (final_start_function): Use the constant lower bound
481 of the frame size for -Wframe-larger-than.
482 * ira.c (do_reload): Adjust for new get_frame_size return type.
483 * lra.c (lra): Likewise.
484 * reload1.c (reload): Likewise.
485 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
486 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
487 * rtlanal.c (get_initial_register_offset): Return the offset as
488 a poly_int64 rather than a HOST_WIDE_INT.
490 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
491 Alan Hayward <alan.hayward@arm.com>
492 David Sherwood <david.sherwood@arm.com>
494 * reload1.c (elim_table): Change initial_offset, offset and
495 previous_offset from HOST_WIDE_INT to poly_int64_pod.
496 (offsets_at): Change the target array's element type from
497 HOST_WIDE_INT to poly_int64_pod.
498 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
499 (elimination_costs_in_insn, update_eliminable_offsets)
500 (verify_initial_elim_offsets, set_offsets_for_label)
501 (init_eliminable_invariants): Update after above changes.
503 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
504 Alan Hayward <alan.hayward@arm.com>
505 David Sherwood <david.sherwood@arm.com>
507 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
508 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
509 (decomposition): Change start and end from HOST_WIDE_INT
511 (decompose, immune_p): Update accordingly.
512 (find_inc_amount): Return a poly_int64 rather than an int.
513 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
516 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
517 Alan Hayward <alan.hayward@arm.com>
518 David Sherwood <david.sherwood@arm.com>
520 * tree.h (get_inner_reference): Return the bitsize and bitpos
521 as poly_int64_pods rather than HOST_WIDE_INT.
522 * fold-const.h (ptr_difference_const): Return the pointer difference
523 as a poly_int64_pod rather than a HOST_WIDE_INT.
524 * expr.c (get_inner_reference): Return the bitsize and bitpos
525 as poly_int64_pods rather than HOST_WIDE_INT.
526 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
528 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
529 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
530 (optimize_bit_field_compare): Update call to get_inner_reference.
531 (decode_field_reference): Likewise.
532 (fold_unary_loc): Track polynomial offsets and sizes.
533 (split_address_to_core_and_offset): Return the bitpos as a
534 poly_int64_pod rather than a HOST_WIDE_INT.
535 (ptr_difference_const): Likewise for the pointer difference.
536 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
537 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
538 * dbxout.c (dbxout_expand_expr): Likewise.
539 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
540 (loc_list_from_tree_1, fortran_common): Likewise.
541 * gimple-laddress.c (pass_laddress::execute): Likewise.
542 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
543 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
544 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
545 * tree-affine.c (tree_to_aff_combination): Likewise.
546 (get_inner_reference_aff): Likewise.
547 * tree-data-ref.c (split_constant_offset_1): Likewise.
548 (dr_analyze_innermost): Likewise.
549 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
550 * tree-sra.c (ipa_sra_check_caller): Likewise.
551 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
552 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
553 (instrument_bool_enum_load, instrument_object_size): Likewise.
554 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
555 to get_inner_reference.
556 * hsa-gen.c (gen_hsa_addr): Likewise.
557 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
558 * tsan.c (instrument_expr): Likewise.
559 * match.pd: Update call to ptr_difference_const.
561 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
562 Alan Hayward <alan.hayward@arm.com>
563 David Sherwood <david.sherwood@arm.com>
565 * fold-const.c (fold_comparison): Track sizes and offsets as
566 poly_int64s rather than HOST_WIDE_INTs when folding address
569 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
570 Alan Hayward <alan.hayward@arm.com>
571 David Sherwood <david.sherwood@arm.com>
573 * expr.h (get_bit_range): Return the bitstart and bitend as
574 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
575 as a poly_int64 rather than a HOST_WIDE_INT.
576 * expr.c (get_bit_range): Likewise.
577 (expand_assignment): Update call accordingly.
578 * fold-const.c (optimize_bit_field_compare): Likewise.
580 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
581 Alan Hayward <alan.hayward@arm.com>
582 David Sherwood <david.sherwood@arm.com>
584 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
585 (aarch64_legitimate_address_p): Use it instead of an rtx code,
586 as an optional final parameter.
587 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
588 (aarch64_legitimate_address_p): Likewise.
589 (aarch64_print_address_internal): Take an aarch64_addr_query_type
590 instead of an rtx code.
591 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
592 (aarch64_legitimate_address_hook_p): Likewise.
593 (aarch64_print_ldpstp_address): Likewise.
594 (aarch64_print_operand_address): Likewise.
595 (aarch64_address_cost): Likewise.
596 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
597 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
598 (aarch64_mem_pair_lanes_operand): Likewise.
600 2017-12-20 Richard Biener <rguenther@suse.de>
602 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
603 update_stmt_if_modified.
605 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
607 PR tree-optimization/83491
608 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
609 before walking uses. Improve coding style and comments.
611 2017-12-20 Tom de Vries <tom@codesourcery.com>
613 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
615 2017-12-20 Jakub Jelinek <jakub@redhat.com>
618 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
620 (pass_ipa_free_fn_summary): Add small_p private data member,
621 initialize to false in the ctor.
622 (pass_ipa_free_fn_summary::clone,
623 pass_ipa_free_fn_summary::set_pass_param,
624 pass_ipa_free_fn_summary::gate): New methods.
625 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
626 | TODO_dump_symtab if small_p.
627 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
628 entry and add another instance of the pass with false parm after
630 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
631 ipa_free_fn_summary here.
633 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
635 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
637 2017-12-20 Martin Sebor <msebor@redhat.com>
640 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
641 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
642 (expand_builtin_strncmp): Same.
644 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
647 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
648 allowed debug stmts before labels.
649 (expand_gimple_basic_block): Likewise.
650 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
651 * gimple-iterator.h (gsi_after_labels): Likewise.
652 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
653 rename reused variable, and simplify using gsi_move_before.
654 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
655 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
656 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
657 (gimple_verify_flow_info, gimple_block_label): Likewise.
658 (make_blocks): Move debug markers after adjacent labels.
659 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
660 allowed debug insns outside blocks.
661 * df-scan.c (df_insn_delete): Likewise.
662 * lra-constraints.c (update_ebb_live_info): Likewise.
663 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
664 (vt_initialize, delete_vta_debug_insns): Likewise.
665 (reemit_marker_as_note): Drop BB parm. Adjust callers.
667 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
668 Alan Hayward <alan.hayward@arm.com>
669 David Sherwood <david.sherwood@arm.com>
671 * poly-int-types.h (round_down_to_byte_boundary): New macro.
672 (round_up_to_byte_boundary): Likewise.
673 * expr.h (get_bit_range): Add temporary shim.
674 * gimple-ssa-store-merging.c (store_operand_info): Change the
675 bitsize, bitpos, bitregion_start and bitregion_end fields from
676 unsigned HOST_WIDE_INT to poly_uint64.
677 (merged_store_group): Likewise load_align_base.
678 (compatible_load_p, compatible_load_p): Update accordingly.
679 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
680 (split_group, imm_store_chain_info::output_merged_store): Likewise.
681 (mem_valid_for_store_merging): Return the bitsize, bitpos,
682 bitregion_start and bitregion_end as poly_uint64s rather than
683 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
684 (handled_load): Take the bitsize, bitpos,
685 bitregion_start and bitregion_end as poly_uint64s rather than
686 unsigned HOST_WIDE_INTs.
687 (pass_store_merging::process_store): Update call to
688 mem_valid_for_store_merging.
690 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
691 Alan Hayward <alan.hayward@arm.com>
692 David Sherwood <david.sherwood@arm.com>
694 * builtins.c (get_object_alignment_2): Track polynomial offsets
695 and sizes. Update the alignment handling.
697 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
698 Alan Hayward <alan.hayward@arm.com>
699 David Sherwood <david.sherwood@arm.com>
701 * tree.h (get_inner_reference): Add a version that returns the
702 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
703 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
704 the case in which bitpos is not associated with the first byte.
706 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
707 Alan Hayward <alan.hayward@arm.com>
708 David Sherwood <david.sherwood@arm.com>
710 * tree-affine.h (get_inner_reference_aff): Return the size as a
712 * tree-affine.c (get_inner_reference_aff): Likewise.
713 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
714 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
716 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
717 Alan Hayward <alan.hayward@arm.com>
718 David Sherwood <david.sherwood@arm.com>
720 * fold-const.c (pointer_may_wrap_p): Take the offset as a
721 HOST_WIDE_INT rather than a poly_int64.
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 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
728 HOST_WIDE_INT to poly_int64_pod.
729 (perform_symbolic_merge): Update accordingly.
730 (bswap_replace): Likewise.
732 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
733 Alan Hayward <alan.hayward@arm.com>
734 David Sherwood <david.sherwood@arm.com>
736 * tree-affine.h (aff_tree::offset): Change from widest_int
738 (wide_int_ext_for_comb): Delete.
739 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
740 constants as poly_widest_int rather than widest_int.
741 (aff_combination_constant_multiple_p): Return the multiplier
742 as a poly_widest_int.
743 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
745 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
746 version static and add an overload for poly_widest_int.
747 (aff_combination_const, aff_combination_add_cst)
748 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
749 the constants as poly_widest_int rather than widest_int.
750 (tree_to_aff_combination): Generalize INTEGER_CST case to
752 (aff_combination_to_tree): Track offsets as poly_widest_ints.
753 (aff_combination_add_product, aff_combination_mult): Handle
755 (aff_combination_constant_multiple_p): Return the multiplier
756 as a poly_widest_int.
757 * tree-predcom.c (determine_offset): Return the offset as a
759 (split_data_refs_to_components, suitable_component_p): Update
761 (valid_initializer_p): Update call to
762 aff_combination_constant_multiple_p.
763 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
764 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
765 as a poly_int64 rather than a HOST_WIDE_INT.
766 (get_address_cost): Handle polynomial offsets.
767 (iv_elimination_compare_lt): Likewise.
768 (rewrite_use_nonlinear_expr): Likewise.
770 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
771 Alan Hayward <alan.hayward@arm.com>
772 David Sherwood <david.sherwood@arm.com>
774 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
775 as a poly_int64_pod rather than a HOST_WIDE_INT.
776 (get_addr_base_and_unit_offset): Likewise.
777 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
778 (get_addr_base_and_unit_offset): Likewise.
779 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
780 to poly_int64 in example.
781 * fold-const.c (fold_binary_loc): Update call to
782 get_addr_base_and_unit_offset.
783 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
784 (maybe_canonicalize_mem_ref_addr): Likewise.
785 (gimple_fold_stmt_to_constant_1): Likewise.
786 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
788 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
789 * match.pd: Likewise.
790 * omp-low.c (lower_omp_target): Likewise.
791 * tree-sra.c (build_ref_for_offset): Likewise.
792 (build_debug_ref_for_model): Likewise.
793 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
794 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
795 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
796 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
797 (constant_pointer_difference): Likewise.
798 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
799 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
800 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
801 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
802 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
803 (set_ssa_val_to): Likewise.
804 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
805 (maybe_diag_stxncpy_trunc): Likewise.
806 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
807 * tree.c (build_simple_mem_ref_loc): Likewise.
808 (array_at_struct_end_p): Likewise.
810 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
811 Alan Hayward <alan.hayward@arm.com>
812 David Sherwood <david.sherwood@arm.com>
814 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
815 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
816 (get_ref_base_and_extent_hwi): Declare.
817 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
818 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
819 (get_ref_base_and_extent_hwi): New function.
820 * cfgexpand.c (expand_debug_expr): Update call to
821 get_ref_base_and_extent.
822 * dwarf2out.c (add_var_loc_to_decl): Likewise.
823 * gimple-fold.c (get_base_constructor): Return the offset as a
824 poly_int64_pod rather than a HOST_WIDE_INT.
825 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
826 * ipa-polymorphic-call.c
827 (ipa_polymorphic_call_context::set_by_invariant)
828 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
829 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
830 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
831 rather than get_ref_base_and_extent.
832 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
833 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
834 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
836 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
837 call to get_ref_base_and_extent.
838 * tree-sra.c (create_access, get_access_for_expr): Likewise.
839 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
840 (stmt_kills_ref_p): Likewise.
841 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
842 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
844 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
845 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
846 when calling native_encode_expr.
847 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
848 call to get_ref_base_and_extent.
849 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
850 get_ref_base_and_extent.
851 * var-tracking.c (track_expr_p): Likewise.
853 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
854 Alan Hayward <alan.hayward@arm.com>
855 David Sherwood <david.sherwood@arm.com>
857 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
858 HOST_WIDE_INT to poly_int64_pod.
859 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
860 polynomail parameter offsets.
862 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
863 Alan Hayward <alan.hayward@arm.com>
864 David Sherwood <david.sherwood@arm.com>
866 * gengtype.c (main): Handle poly_int64_pod.
867 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
868 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
869 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
870 from HOST_WIDE_INT to poly_int64_pod.
871 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
872 (copy_cfa): New function.
873 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
874 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
875 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
876 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
877 (initial_return_save): Treat offsets as poly_ints.
878 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
879 offset is nonconstant.
880 (reg_save): Take the offset as a poly_int64. Fall back to
881 DW_CFA_expression for nonconstant offsets.
882 (queue_reg_save): Take the offset as a poly_int64.
883 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
885 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
886 Alan Hayward <alan.hayward@arm.com>
887 David Sherwood <david.sherwood@arm.com>
889 * rtl.h (operand_subword, operand_subword_force): Take the offset
890 as a poly_uint64 an unsigned int.
891 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
893 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
894 Alan Hayward <alan.hayward@arm.com>
895 David Sherwood <david.sherwood@arm.com>
897 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
898 'p' format code. Use INT_LIST rather than SUBREG as the example of
899 a code with an XINT and an XEXP. Remove the implication that
900 accessing an rtx field using XINT is expected to work.
901 * rtl.def (SUBREG): Change format from "ei" to "ep".
902 * rtl.h (rtunion::rt_subreg): New field.
903 (XCSUBREG): New macro.
904 (SUBREG_BYTE): Use it.
905 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
906 Update constructor accordingly.
907 (subreg_shape::operator ==): Update accordingly.
908 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
909 than an unsigned int.
910 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
911 a poly_uint64 rather than an unsigned int.
912 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
913 than an unsigned int.
914 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
915 (subreg_size_highpart_offset): Return a poly_uint64 rather than
916 an unsigned int. Take the sizes as poly_uint64s.
917 (subreg_offset_from_lsb): Return a poly_uint64 rather than
918 an unsigned int. Take the shift as a poly_uint64 rather than
920 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
921 as a poly_uint64 rather than an unsigned int.
922 (simplify_subreg_regno): Likewise.
923 (byte_lowpart_offset): Return the memory offset as a poly_int64
925 (subreg_memory_offset): Likewise. Take the subreg offset as a
926 poly_uint64 rather than an unsigned int.
927 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
928 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
929 poly_uint64 rather than an unsigned int.
930 * rtl.c (rtx_format): Describe 'p' in comment.
931 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
932 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
933 offset as a poly_uint64 rather than an unsigned int.
934 (byte_lowpart_offset): Return the memory offset as a poly_int64
936 (subreg_memory_offset): Likewise. Take the subreg offset as a
937 poly_uint64 rather than an unsigned int.
938 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
939 mode sizes as poly_uint64s rather than unsigned ints. Return a
940 poly_uint64 rather than an unsigned int.
941 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
942 (copy_insn_1): Handle 'p'.
943 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
944 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
945 an unsigned int. Return the shift in the same way.
946 (subreg_lsb): Return the shift as a poly_uint64 rather than an
948 (subreg_size_offset_from_lsb): Take the sizes and shift as
949 poly_uint64s rather than unsigned ints. Return the offset as
951 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
952 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
954 * rtlhash.c (add_rtx): Handle 'p'.
955 * genemit.c (gen_exp): Likewise.
956 * gengenrtl.c (type_from_format, gendef): Likewise.
957 * gensupport.c (subst_pattern_match, get_alternatives_number)
958 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
959 (subst_dup): Likewise.
960 * gengtype.c (adjust_field_rtx_def): Likewise.
961 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
962 (match_pattern_2): Likewise.
963 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
964 (rtx_test::subreg_field): New function.
965 (operator ==, safe_to_hoist_p, transition_parameter_type)
966 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
967 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
968 * genpeep.c (match_rtx): Likewise.
969 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
970 (rtx_writer::print_rtx_operand): Handle 'p'.
971 (print_value): Handle SUBREG.
972 * read-rtl.c (apply_int_iterator): Likewise.
973 (rtx_reader::read_rtx_operand): Handle 'p'.
974 * alias.c (rtx_equal_for_memref_p): Likewise.
975 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
976 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
978 * calls.c (expand_call): Likewise.
979 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
980 (make_extraction, gen_lowpart_for_combine): Likewise.
981 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
983 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
984 rather than an unsigned int. Treat subreg offsets as poly_ints.
985 (exp_equiv_p): Handle 'p'.
986 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
987 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
988 * dse.c (find_shift_sequence): Likewise.
989 * dwarf2out.c (rtl_for_decl_location): Likewise.
990 * expmed.c (extract_low_bits): Likewise.
991 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
992 (expand_expr_real_2): Likewise.
993 * final.c (alter_subreg): Likewise.
994 (leaf_renumber_regs_insn): Handle 'p'.
995 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
996 Treat subreg offsets as poly_ints.
997 * fwprop.c (forward_propagate_and_simplify): Likewise.
998 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
999 * ira.c (get_subreg_tracking_sizes): Likewise.
1000 * ira-conflicts.c (go_through_subreg): Likewise.
1001 * ira-lives.c (process_single_reg_class_operands): Likewise.
1002 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
1003 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
1004 as a poly_uint64 rather than an unsigned int.
1005 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
1006 subreg offsets as poly_ints.
1007 * lra-constraints.c (operands_match_p): Handle 'p'.
1008 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
1009 * lra-spills.c (assign_mem_slot): Likewise.
1010 * postreload.c (move2add_valid_value_p): Likewise.
1011 * recog.c (general_operand, indirect_operand): Likewise.
1012 * regcprop.c (copy_value, maybe_mode_change): Likewise.
1013 (copyprop_hardreg_forward_1): Likewise.
1014 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
1015 (record_subregs_of_mode): Likewise.
1016 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
1017 * reload.c (operands_match_p): Handle 'p'.
1018 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
1019 * reload1.c (alter_reg, choose_reload_regs): Likewise.
1020 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
1021 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
1022 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
1023 (simplify_const_poly_int_tests<N>::run): Likewise.
1024 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
1025 a poly_uint64 rather than an unsigned int.
1026 * valtrack.c (debug_lowpart_subreg): Likewise.
1027 * var-tracking.c (var_lowpart): Likewise.
1028 (loc_cmp): Handle 'p'.
1030 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1031 Alan Hayward <alan.hayward@arm.com>
1032 David Sherwood <david.sherwood@arm.com>
1034 * ira.c (get_subreg_tracking_sizes): New function.
1035 (init_live_subregs): Take an integer size rather than a register.
1036 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
1037 to init_live_subregs.
1039 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1040 Alan Hayward <alan.hayward@arm.com>
1041 David Sherwood <david.sherwood@arm.com>
1043 * expr.c (store_constructor_field): Change bitsize from a
1044 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
1045 HOST_WIDE_INT to a poly_int64.
1046 (store_constructor): Change size from a HOST_WIDE_INT to
1048 (store_field): Likewise bitsize and bitpos.
1050 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1051 Alan Hayward <alan.hayward@arm.com>
1052 David Sherwood <david.sherwood@arm.com>
1054 * expmed.h (store_bit_field): Change bitregion_start and
1055 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
1056 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
1057 (store_bit_field_1, store_integral_bit_field, store_bit_field)
1058 (store_fixed_bit_field, store_split_bit_field): Likewise.
1059 * expr.c (store_constructor_field, store_field): Likewise.
1060 (optimize_bitfield_assignment_op): Likewise. Make the same change
1061 to bitsize and bitpos.
1062 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
1063 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
1064 same change in the constructor arguments.
1065 (get_best_mode): Change bitregion_start and bitregion_end from
1066 unsigned HOST_WIDE_INT to poly_uint64.
1067 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
1068 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
1070 (bit_field_mode_iterator::next_mode): Update for new types
1071 of m_bitregion_start and m_bitregion_end.
1072 (get_best_mode): Change bitregion_start and bitregion_end from
1073 unsigned HOST_WIDE_INT to poly_uint64.
1075 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1076 Alan Hayward <alan.hayward@arm.com>
1077 David Sherwood <david.sherwood@arm.com>
1079 * rtl.h (simplify_gen_subreg): Add a temporary overload that
1080 accepts poly_uint64 offsets.
1081 * expmed.h (extract_bit_field): Take bitsize and bitnum as
1082 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1083 * expmed.c (lowpart_bit_field_p): Likewise.
1084 (extract_bit_field_as_subreg): New function, split out from...
1085 (extract_bit_field_1): ...here. Take bitsize and bitnum as
1086 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
1087 extractions, check that BITSIZE matches the size of the extracted
1088 value and that BITNUM is an exact multiple of that size.
1089 If all else fails, try forcing the value into memory if
1090 BITNUM is variable, and adjusting the address so that the
1091 offset is constant. Split the part that can only handle constant
1092 bitsize and bitnum out into...
1093 (extract_integral_bit_field): ...this new function.
1094 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
1095 rather than unsigned HOST_WIDE_INTs.
1097 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1098 Alan Hayward <alan.hayward@arm.com>
1099 David Sherwood <david.sherwood@arm.com>
1101 * expmed.h (store_bit_field): Take bitsize and bitnum as
1102 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1103 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
1104 that returns the byte size.
1105 (store_bit_field_1): Take bitsize and bitnum as
1106 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
1107 to simple_mem_bitfield_p. Split the part that can only handle
1108 constant bitsize and bitnum out into...
1109 (store_integral_bit_field): ...this new function.
1110 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
1111 than unsigned HOST_WIDE_INTs.
1112 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
1114 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1115 Alan Hayward <alan.hayward@arm.com>
1116 David Sherwood <david.sherwood@arm.com>
1118 * lra-int.h (lra_reg): Change offset from int to poly_int64.
1119 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
1121 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1122 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
1123 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
1124 offset as a poly_int64 rather than an int.
1125 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
1126 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1127 * lra-constraints.c (equiv_address_substitution): Track offsets
1129 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
1130 (curr_insn_transform): Handle the new form of sp_offset.
1131 * lra-eliminations.c (lra_elim_table): Change previous_offset
1132 and offset from HOST_WIDE_INT to poly_int64.
1133 (print_elim_table, update_reg_eliminate): Update accordingly.
1134 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
1135 (get_elimination): Update accordingly.
1136 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
1137 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1138 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
1139 poly_int64 offsets generally.
1140 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
1141 (mark_not_eliminable, init_elimination): Update accordingly.
1142 (remove_reg_equal_offset_note): Return a bool and pass the new
1143 offset back by pointer as a poly_int64.
1144 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
1145 rather than a HOST_WIDE_INT.
1146 (do_remat): Track offsets poly_int64s.
1147 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
1149 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1150 Alan Hayward <alan.hayward@arm.com>
1151 David Sherwood <david.sherwood@arm.com>
1153 * rtl.h (mem_attrs): Add a default constructor. Change size and
1154 offset from HOST_WIDE_INT to poly_int64.
1155 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
1156 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
1157 (widen_memory_access): Take the sizes and offsets as poly_int64s
1158 rather than HOST_WIDE_INTs.
1159 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
1160 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
1162 (adjust_offset_for_component_ref): Change the offset from a
1163 HOST_WIDE_INT to a poly_int64.
1164 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
1165 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
1166 * dce.c (find_call_stack_args): Likewise.
1167 * dse.c (record_store): Likewise.
1168 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
1169 * print-rtl.c (rtx_writer::print_rtx): Likewise.
1170 * read-rtl-function.c (test_loading_mem): Likewise.
1171 * rtlanal.c (may_trap_p_1): Likewise.
1172 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1173 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
1174 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
1175 (mem_attrs::mem_attrs): New function.
1176 (set_mem_attributes_minus_bitpos): Change bitpos from a
1177 HOST_WIDE_INT to poly_int64.
1178 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
1179 (clear_mem_offset, clear_mem_size, change_address)
1180 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
1181 initialize mem_attrs.
1182 (set_mem_offset, set_mem_size, adjust_address_1)
1183 (adjust_automodify_address_1, offset_address, widen_memory_access):
1184 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
1186 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1187 Alan Hayward <alan.hayward@arm.com>
1188 David Sherwood <david.sherwood@arm.com>
1190 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
1191 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
1192 rather than 0 to represent an unknown size. Assert that the size
1193 is known when the mode isn't BLKmode.
1194 (may_trap_p_1): Use -1 for unknown sizes.
1195 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
1197 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1198 Alan Hayward <alan.hayward@arm.com>
1199 David Sherwood <david.sherwood@arm.com>
1201 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
1202 to poly_int64. Update commentary for positions_needed.large.
1203 (read_info_type): Change offset and width from HOST_WIDE_INT
1205 (set_usage_bits): Likewise.
1206 (canon_address): Return the offset as a poly_int64 rather than
1207 a HOST_WIDE_INT. Use strip_offset_and_add.
1208 (set_all_positions_unneeded, any_positions_needed_p): Use
1209 positions_needed.large to track stores with non-constant widths.
1210 (all_positions_needed_p): Likewise. Take the offset and width
1211 as poly_int64s rather than ints. Assert that rhs is nonnull.
1212 (record_store): Cope with non-constant offsets and widths.
1213 Nullify the rhs of an earlier store if we can't tell which bytes
1215 (find_shift_sequence): Take the access_size and shift as poly_int64s
1217 (get_stored_val): Take the read_offset and read_width as poly_int64s
1218 rather than HOST_WIDE_INTs.
1219 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
1220 non-constant offsets and widths.
1222 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1223 Alan Hayward <alan.hayward@arm.com>
1224 David Sherwood <david.sherwood@arm.com>
1226 * inchash.h (inchash::hash::add_poly_int): New function.
1227 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
1228 Use poly_int64 rather than HOST_WIDE_INT.
1229 (ao_ref::max_size_known_p): New function.
1230 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
1231 rather than HOST_WIDE_INT.
1232 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
1233 to temporaries until its interface is adjusted to match.
1234 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
1235 (aliasing_component_refs_p, decl_refs_may_alias_p)
1236 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
1237 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
1238 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
1240 * alias.c (ao_ref_from_mem): Likewise.
1241 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1242 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
1243 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
1244 (maybe_trim_complex_store, maybe_trim_constructor_store)
1245 (live_bytes_read, dse_classify_store): Likewise.
1246 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
1247 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
1248 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
1249 (vn_reference_lookup_3): Likewise.
1250 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
1252 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1253 Alan Hayward <alan.hayward@arm.com>
1254 David Sherwood <david.sherwood@arm.com>
1256 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
1257 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
1258 instead of ranges_overlap_p.
1260 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1261 Alan Hayward <alan.hayward@arm.com>
1262 David Sherwood <david.sherwood@arm.com>
1264 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
1265 sizes as poly_int64s rather than HOST_WIDE_INTs.
1267 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1268 Alan Hayward <alan.hayward@arm.com>
1269 David Sherwood <david.sherwood@arm.com>
1271 * gimple-fold.h (fold_ctor_reference): Take the offset and size
1272 as poly_uint64 rather than unsigned HOST_WIDE_INT.
1273 * gimple-fold.c (fold_ctor_reference): Likewise.
1275 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1276 Alan Hayward <alan.hayward@arm.com>
1277 David Sherwood <david.sherwood@arm.com>
1279 * target.def (dwarf_poly_indeterminate_value): New hook.
1280 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
1281 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
1282 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
1283 * doc/tm.texi: Regenerate.
1284 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
1285 offset as a poly_int64.
1286 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
1287 offset as a poly_int64.
1288 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
1289 Take the offset as a poly_int64.
1290 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
1291 (frame_pointer_fb_offset): Change to a poly_int64.
1292 (int_loc_descriptor): Take the offset as a poly_int64. Use
1293 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
1294 (based_loc_descr): Take the offset as a poly_int64.
1295 Use strip_offset_and_add to handle (plus X (const)).
1296 Use new_reg_loc_descr instead of an open-coded version of the
1297 previous implementation.
1298 (mem_loc_descriptor): Handle CONST_POLY_INT.
1299 (compute_frame_pointer_to_fb_displacement): Take the offset as a
1300 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
1302 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1303 Alan Hayward <alan.hayward@arm.com>
1304 David Sherwood <david.sherwood@arm.com>
1306 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
1307 (gen_rtx_REG_offset): Take the offset as a poly_int64.
1308 * inchash.h (inchash::hash::add_poly_hwi): New function.
1309 * gengtype.c (main): Register poly_int64.
1310 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
1311 offset as a poly_int.
1312 (reg_attr_hasher::equal): Use must_eq to compare offsets.
1313 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
1314 offset as a poly_int64.
1315 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
1316 * print-rtl.c (print_poly_int): New function.
1317 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
1319 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
1321 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
1322 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
1323 (vt_get_decl_and_offset): Return the offset as a poly_int64.
1324 Enforce track_offset_p for parts of a PARALLEL.
1325 (vt_add_function_parameter): Use const_offset for the final
1326 offset to track. Use get_tracked_reg_offset for the parts
1329 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1330 Alan Hayward <alan.hayward@arm.com>
1331 David Sherwood <david.sherwood@arm.com>
1333 * target.def (truly_noop_truncation): Take poly_uint64s instead of
1334 unsigned ints. Change default to hook_bool_puint64_puint64_true.
1335 * doc/tm.texi: Regenerate.
1336 * hooks.h (hook_bool_uint_uint_true): Delete.
1337 (hook_bool_puint64_puint64_true): Declare.
1338 * hooks.c (hook_bool_uint_uint_true): Delete.
1339 (hook_bool_puint64_puint64_true): New function.
1340 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
1341 instead of unsigned ints.
1342 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
1343 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
1345 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1346 Alan Hayward <alan.hayward@arm.com>
1347 David Sherwood <david.sherwood@arm.com>
1349 * optabs.h (expand_operand): Add an int_value field.
1350 (create_expand_operand): Add an int_value parameter and use it
1351 to initialize the new expand_operand field.
1352 (create_integer_operand): Replace with a declaration of a function
1353 that accepts poly_int64s. Move the implementation to...
1354 * optabs.c (create_integer_operand): ...here.
1355 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
1356 the mode preserves the value of int_value, instead of calling
1357 const_int_operand on the rtx. Use gen_int_mode to generate
1360 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1361 Alan Hayward <alan.hayward@arm.com>
1362 David Sherwood <david.sherwood@arm.com>
1364 * dumpfile.h (dump_dec): Declare.
1365 * dumpfile.c (dump_dec): New function.
1366 * pretty-print.h (pp_wide_integer): Turn into a function and
1367 declare a poly_int version.
1368 * pretty-print.c (pp_wide_integer): New function for poly_ints.
1370 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1371 Alan Hayward <alan.hayward@arm.com>
1372 David Sherwood <david.sherwood@arm.com>
1374 * doc/generic.texi (POLY_INT_CST): Document.
1375 * tree.def (POLY_INT_CST): New tree code.
1376 * treestruct.def (TS_POLY_INT_CST): New tree layout.
1377 * tree-core.h (tree_poly_int_cst): New struct.
1378 (tree_node): Add a poly_int_cst field.
1379 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
1380 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
1381 instead of a wide_int_ref.
1382 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
1384 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1385 instead of an unsigned HOST_WIDE_INT.
1386 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
1387 (ptrdiff_tree_p): Declare.
1388 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
1389 extern inline implementations if the target doesn't use POLY_INT_CST.
1390 (poly_int_tree_p): New function.
1391 (wi::unextended_tree): New class.
1392 (wi::int_traits <unextended_tree>): New override.
1393 (wi::extended_tree): Add a default constructor.
1394 (wi::extended_tree::get_tree): New function.
1395 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
1396 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
1397 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
1398 (wi::tree_to_poly_wide_ref): New typedefs.
1399 (wi::ints_for): Provide overloads for extended_tree and
1401 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
1402 (wi::to_wide): New functions.
1403 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
1404 * tree.c (poly_int_cst_hasher): New struct.
1405 (poly_int_cst_hash_table): New variable.
1406 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
1407 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
1409 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
1410 (init_ttree): Initialize poly_int_cst_hash_table.
1411 (build_int_cst, build_int_cst_type, build_invariant_address): Take
1412 a poly_int64 instead of a HOST_WIDE_INT.
1413 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1414 instead of an unsigned HOST_WIDE_INT.
1415 (wide_int_to_tree): Rename to...
1416 (wide_int_to_tree_1): ...this.
1417 (build_new_poly_int_cst, build_poly_int_cst): New functions.
1418 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
1419 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
1420 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
1421 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
1422 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
1424 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
1425 (streamer_read_tree_body): Likewise.
1426 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
1427 (streamer_write_tree_body): Likewise.
1428 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
1429 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
1430 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
1431 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
1432 * gimple-expr.h (is_gimple_constant): Likewise.
1433 * gimplify.c (maybe_with_size_expr): Likewise.
1434 * print-tree.c (print_node): Likewise.
1435 * tree-data-ref.c (data_ref_compare_tree): Likewise.
1436 * tree-pretty-print.c (dump_generic_node): Likewise.
1437 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1438 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
1439 * tree-vrp.c (compare_values_warnv): Likewise.
1440 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
1441 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
1442 (force_expr_to_var_cost): Likewise.
1443 * tree-ssa-loop.c (for_each_index): Likewise.
1444 * fold-const.h (build_invariant_address, size_int_kind): Take a
1445 poly_int64 instead of a HOST_WIDE_INT.
1446 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
1447 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
1449 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
1450 (int_const_binop_2): New function, split out from...
1451 (int_const_binop_1): ...here. Handle POLY_INT_CST.
1452 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
1453 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
1454 * gimple-ssa-strength-reduction.c (slsr_process_add)
1455 (slsr_process_mul): Check for INTEGER_CSTs before using them
1457 * stor-layout.c (bits_from_bytes): New function.
1458 (bit_from_pos): Use it.
1459 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
1460 by BITS_PER_UNIT to get the TYPE_SIZE.
1461 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
1462 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
1464 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1465 Alan Hayward <alan.hayward@arm.com>
1466 David Sherwood <david.sherwood@arm.com>
1468 * doc/rtl.texi (const_poly_int): Document. Also document the
1469 rtl sharing behavior.
1470 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
1471 * rtl.h (const_poly_int_def): New struct.
1472 (rtx_def::u): Add a cpi field.
1473 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
1474 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
1475 (wi::rtx_to_poly_wide_ref): New typedef
1476 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
1477 (poly_int_rtx_p): New functions.
1478 (trunc_int_for_mode): Declare a poly_int64 version.
1479 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1480 (immed_wide_int_const): Take a poly_wide_int_ref rather than
1482 (strip_offset): Declare.
1483 (strip_offset_and_add): New function.
1484 * rtl.def (CONST_POLY_INT): New rtx code.
1485 * rtl.c (rtx_size): Handle CONST_POLY_INT.
1486 (shared_const_p): Use poly_int_rtx_p.
1487 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
1489 (gen_int_shift_amount): Likewise.
1490 * emit-rtl.c (const_poly_int_hasher): New class.
1491 (const_poly_int_htab): New variable.
1492 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
1493 (const_poly_int_hasher::hash): New function.
1494 (const_poly_int_hasher::equal): Likewise.
1495 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
1496 (immed_wide_int_const): Rename to...
1497 (immed_wide_int_const_1): ...this and make static.
1498 (immed_wide_int_const): New function, taking a poly_wide_int_ref
1499 instead of a wide_int_ref.
1500 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
1501 (gen_lowpart_common): Handle CONST_POLY_INT.
1502 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
1503 * cselib.c (cselib_hash_rtx): Likewise.
1504 * dwarf2out.c (const_ok_for_output_1): Likewise.
1505 * expr.c (convert_modes): Likewise.
1506 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
1507 * rtlhash.c (add_rtx): Likewise.
1508 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
1509 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1510 Handle existing CONST_POLY_INT rtxes.
1511 * expmed.h (expand_shift): Take a poly_int64 instead of a
1513 * expmed.c (expand_shift): Likewise.
1514 * rtlanal.c (strip_offset): New function.
1515 (commutative_operand_precedence): Give CONST_POLY_INT the same
1516 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
1518 * rtl-tests.c (const_poly_int_tests): New struct.
1519 (rtl_tests_c_tests): Use it.
1520 * simplify-rtx.c (simplify_const_unary_operation): Handle
1522 (simplify_const_binary_operation): Likewise.
1523 (simplify_binary_operation_1): Fold additions of symbolic constants
1524 and CONST_POLY_INTs.
1525 (simplify_subreg): Handle extensions and truncations of
1527 (simplify_const_poly_int_tests): New struct.
1528 (simplify_rtx_c_tests): Use it.
1529 * wide-int.h (storage_ref): Add default constructor.
1530 (wide_int_ref_storage): Likewise.
1531 (trailing_wide_ints): Use GTY((user)).
1532 (trailing_wide_ints::operator[]): Add a const version.
1533 (trailing_wide_ints::get_precision): New function.
1534 (trailing_wide_ints::extra_size): Likewise.
1536 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1537 Alan Hayward <alan.hayward@arm.com>
1538 David Sherwood <david.sherwood@arm.com>
1540 * emit-rtl.h (gen_int_shift_amount): Declare.
1541 * emit-rtl.c (gen_int_shift_amount): New function.
1542 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
1544 * calls.c (shift_return_value): Likewise.
1545 * cse.c (fold_rtx): Likewise.
1546 * dse.c (find_shift_sequence): Likewise.
1547 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
1548 (expand_shift, expand_smod_pow2): Likewise.
1549 * lower-subreg.c (shift_cost): Likewise.
1550 * optabs.c (expand_superword_shift, expand_doubleword_mult)
1551 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
1552 (expand_vec_perm_var): Likewise.
1553 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1554 (simplify_binary_operation_1): Likewise.
1555 * combine.c (try_combine, find_split_point, force_int_to_mode)
1556 (simplify_shift_const_1, simplify_shift_const): Likewise.
1557 (change_zero_ext): Likewise. Use simplify_gen_binary.
1559 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1561 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
1563 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1565 * doc/invoke.texi (ARM Options): Document accepted extension options
1566 for -march=armv8.3-a.
1568 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
1571 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
1572 or --with-float=softfp, set the default CPU to arm10e.
1574 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
1576 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
1577 * config/visium/predicates.md (const_shift_operand): Likewise.
1578 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
1579 (visium_legitimize_reload_address): Likewise.
1581 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1584 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
1586 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1589 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
1591 2017-12-20 Julia Koval <julia.koval@intel.com>
1593 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
1594 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
1595 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
1596 * config.gcc: Include vpclmulqdqintrin.h.
1597 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
1598 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
1600 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
1601 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
1602 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
1603 * config/i386/i386.c (isa2_opts): Add -mcx16.
1604 (isa_opts): Add -mpclmulqdq, remove -mcx16.
1605 (ix86_option_override_internal): Move mcx16 to flags2.
1606 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
1607 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
1608 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
1609 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
1610 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
1611 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
1612 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
1613 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
1614 * doc/invoke.texi: Add -mvpclmulqdq.
1616 2017-12-20 Tom de Vries <tom@codesourcery.com>
1619 * config/i386/i386.c (ix86_static_chain): Move
1620 DECL_STATIC_CHAIN test ...
1621 * calls.c (rtx_for_static_chain): ... here. New function.
1622 * calls.h (rtx_for_static_chain): Declare.
1623 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
1624 instead of targetm.calls.static_chain.
1625 * df-scan.c (df_get_entry_block_def_set): Same.
1627 2017-12-19 Tom de Vries <tom@codesourcery.com>
1629 PR tree-optimization/83493
1630 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
1633 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1635 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
1636 inform with hardcoded english plural handling.
1638 2017-12-18 Jeff Law <law@redhat.com>
1640 PR tree-optimization/83477
1641 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
1642 a non-virtual PHI, always push a new range.
1644 2017-12-19 Martin Sebor <msebor@redhat.com>
1647 * builtins.c (compute_objsize): Handle non-constant offsets.
1649 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1651 PR tree-optimization/83444
1652 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
1653 character load case, if get_stridx on MEM_REF's operand doesn't
1654 look usable, retry with get_addr_stridx.
1656 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
1659 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
1660 (variable_tracking_main_1): Keep markers even when VTA fails.
1663 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
1664 even if there are markers before it.
1665 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
1667 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1669 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
1670 typical order conditions.
1671 * sel-sched.c (extract_new_fences_from): Likewise.
1672 * config/visium/constraints.md (J, K, L): Likewise.
1673 * config/visium/predicates.md (const_shift_operand): Likewise.
1674 * config/visium/visium.c (visium_legitimize_address,
1675 visium_legitimize_reload_address): Likewise.
1676 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
1677 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
1678 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
1679 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
1680 SET_DUMP_DETAIL): Likewise.
1681 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
1682 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
1683 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
1684 avr_set_core_architecture, avr_set_current_function,
1685 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
1686 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
1687 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
1688 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
1689 avr_map_decompose, avr_fold_builtin): Likewise.
1690 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
1691 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
1692 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
1693 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
1694 * config/m32c/m32c.c (m32c_conditional_register_usage,
1695 m32c_address_cost): Likewise.
1696 * config/m32c/predicates.md (shiftcount_operand,
1697 longshiftcount_operand): Likewise.
1698 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
1699 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
1700 can_use_cdx_ldstw): Likewise.
1701 * config/nios2/nios2.h (CDX_REG_P): Likewise.
1702 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
1704 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
1705 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
1706 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
1707 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
1708 * config/vax/vax.c (adjacent_operands_p): Likewise.
1709 * config/ft32/constraints.md (L, b, KA): Likewise.
1710 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
1712 * cfgexpand.c (expand_stack_alignment): Likewise.
1713 * gcse.c (insert_expr_in_table): Likewise.
1714 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
1715 * cgraphunit.c (cgraph_node::expand): Likewise.
1716 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
1717 * emit-rtl.c (add_insn): Likewise.
1718 * input.c (dump_location_info): Likewise.
1719 * passes.c (NEXT_PASS): Likewise.
1720 * read-rtl-function.c (parse_note_insn_name,
1721 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
1723 * sched-rgn.c (sched_rgn_init): Likewise.
1724 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
1725 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
1726 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
1728 * explow.c (eliminate_constant_term): Likewise.
1729 * final.c (leaf_renumber_regs_insn): Likewise.
1730 * cfgrtl.c (print_rtl_with_bb): Likewise.
1731 * genhooks.c (emit_init_macros): Likewise.
1732 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
1733 * tree-data-ref.c (conflict_fn): Likewise.
1734 * selftest.c (assert_streq): Likewise.
1735 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
1736 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
1737 fold_binary_loc, multiple_of_p): Likewise.
1738 * reload.c (push_reload, find_equiv_reg): Likewise.
1739 * et-forest.c (et_nca, et_below): Likewise.
1740 * dbxout.c (dbxout_symbol_location): Likewise.
1741 * reorg.c (relax_delay_slots): Likewise.
1742 * dojump.c (do_compare_rtx_and_jump): Likewise.
1743 * gengtype-parse.c (type): Likewise.
1744 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
1745 simplify_const_relational_operation): Likewise.
1746 * reload1.c (do_output_reload): Likewise.
1747 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
1748 * gengtype.c (type_for_name): Likewise.
1749 * gimple-ssa-sprintf.c (format_directive): Likewise.
1751 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1754 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
1755 accessing it. Adjust comment.
1757 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1760 * predict.c (zero_one_minusone): New function.
1761 (apply_return_prediction): Avoid return prediction for functions
1762 returning only -1, 0 and 1 values, unless they only return -1 and 0
1765 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1767 * config/arc/arc.c (legitimate_scaled_address_p): Clean
1768 fall-through warning.
1769 (arc_compute_frame_size): Remove unused variables.
1770 (arc_print_operand): Fix fprintif format.
1771 (arc_can_follow_jump): Clean fall-through warning.
1773 2017-12-19 Marek Polacek <polacek@redhat.com>
1776 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
1779 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1781 * config/arc/arc.c (overriderregs): New variable.
1782 (arc_override_options): Track fixed/call saved/call options.
1783 (arc_conditional_register_usage): Check against overrideregs
1784 variable whenever we change register properties.
1786 2017-12-19 Nathan Sidwell <nathan@acm.org>
1788 * opts.c (finish_options): Don't prefix dump_base_name if it
1789 already contains directories.
1791 2017-12-19 Martin Liska <mliska@suse.cz>
1793 PR rtl-optimization/82675
1794 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
1795 more element in sbitmap.
1797 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
1799 * gimplify.c (gimplify_expr): Use error_operand_p.
1801 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
1804 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
1807 2017-12-19 Daniel Cederman <cederman@gaisler.com>
1809 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
1810 the jump is to a label.
1812 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1814 PR tree-optimization/83444
1815 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
1820 * ipa-inline.c (flatten_remove_node_hook): New function.
1821 (ipa_inline): Keep only nodes with flatten attribute at the end of
1822 the array in the order from ipa_reverse_postorder, only walk that
1823 portion of array for flattening, if there is more than one such
1824 node, temporarily register a removal hook and ignore removed nodes.
1826 PR tree-optimization/80631
1827 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
1828 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
1829 IFN_REDUC_MAX or IFN_REDUC_MIN.
1831 2017-12-18 Jeff Law <law@redhat.com>
1833 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
1834 record symbolic equivalences from backedges in the CFG.
1837 2017-11-19 Jeff Law <law@redhat.com>
1839 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1840 of degenerates resulting from ignoring an edge.
1842 2017-12-18 Martin Sebor <msebor@redhat.com>
1845 PR tree-optimization/78450
1846 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
1847 (handle_builtin_strlen): Call it.
1849 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
1851 PR rtl-optimization/83424
1852 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
1854 2017-12-18 Marek Polacek <polacek@redhat.com>
1857 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1858 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
1861 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1863 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
1864 (umaddsidi4, umaddsidi_split): Likewise.
1866 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1868 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
1871 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1874 * config/s390/s390.c (s390_option_override): Avoid strncpy.
1876 2017-12-18 Richard Biener <rguenther@suse.de>
1878 PR tree-optimization/81877
1879 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
1880 (outermost_indep_loop): Adjust.
1881 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
1882 (can_sm_ref_p): Adjust.
1884 2017-12-18 Richard Biener <rguenther@suse.de>
1887 * tree.c (array_at_struct_end_p): Return true if the underlying
1888 object has space for at least one element in excess of what
1889 the array domain specifies.
1891 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1893 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
1894 example to avoid overfull hbox.
1895 * doc/invoke.texi (Option Summary): Add missing @gol.
1896 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
1899 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1900 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1902 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
1903 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
1904 (AArch64 Options): Document -mverbose-cost-dump.
1905 (ARM Options): Likewise, plus -mflip-thumb.
1907 2017-12-17 Martin Sebor <msebor@redhat.com>
1910 * gimple-ssa-warn-restrict.c
1911 (builtin_memref::offset_out_of_bounds): Correct the handling of
1914 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1916 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
1918 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
1920 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
1922 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
1924 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
1927 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
1929 * doc/invoke.texi: Fix some typos.
1931 2017-12-16 Martin Sebor <msebor@redhat.com>
1933 PR tree-optimization/78918
1934 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
1935 * builtins.c (check_sizes): Rename...
1936 (check_access): ...to this. Rename function arguments for clarity.
1937 (check_memop_sizes): Adjust names.
1938 (expand_builtin_memchr, expand_builtin_memcpy): Same.
1939 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
1940 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
1941 (check_strncat_sizes, expand_builtin_strncat): Same.
1942 (expand_builtin_strncpy, expand_builtin_memset): Same.
1943 (expand_builtin_bzero, expand_builtin_memcmp): Same.
1944 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
1945 (maybe_emit_sprintf_chk_warning): Same.
1946 (expand_builtin_strcpy): Adjust.
1947 (expand_builtin_stpcpy): Same.
1948 (expand_builtin_with_bounds): Detect out-of-bounds accesses
1949 in pointer-checking forms of memcpy, memmove, and mempcpy.
1950 (gcall_to_tree_minimal, max_object_size): Define new functions.
1951 * builtins.h (max_object_size): Declare.
1952 * calls.c (alloc_max_size): Call max_object_size instead of
1953 hardcoding ssizetype limit.
1954 (get_size_range): Handle new argument.
1955 * calls.h (get_size_range): Add a new argument.
1956 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
1957 * doc/invoke.texi (-Wrestrict): Adjust, add example.
1958 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
1960 (gimple_fold_builtin_memory_chk): Same.
1961 (gimple_fold_builtin_stxcpy_chk): New function.
1962 * gimple-ssa-warn-restrict.c: New source.
1963 * gimple-ssa-warn-restrict.h: New header.
1964 * gimple.c (gimple_build_call_from_tree): Propagate location.
1965 * passes.def (pass_warn_restrict): Add new pass.
1966 * tree-pass.h (make_pass_warn_restrict): Declare.
1967 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
1969 (handle_builtin_strcat): Same.
1970 (strlen_optimize_stmt): Rename...
1971 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
1972 stpncpy, strncpy, and their checking forms.
1974 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
1976 PR rtl-optimization/82849
1977 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
1978 and get_max_loop_iterations_int.
1980 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1981 Alan Hayward <alan.hayward@arm.com>
1982 David Sherwood <david.sherwood@arm.com>
1984 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
1985 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
1986 size as a poly_uint64.
1987 (mode_for_vector, mode_for_int_vector): Take the number of vector
1988 elements as a poly_uint64.
1989 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
1990 size as a poly_uint64.
1991 (mode_for_vector, mode_for_int_vector): Take the number of vector
1992 elements as a poly_uint64.
1994 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1995 Alan Hayward <alan.hayward@arm.com>
1996 David Sherwood <david.sherwood@arm.com>
1998 * machmode.h (MACRO_MODE): New macro.
1999 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
2000 * targhooks.c (default_libcall_value, default_secondary_reload)
2001 (default_memory_move_cost, default_register_move_cost)
2002 (default_class_max_nregs): Likewise.
2004 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2005 Alan Hayward <alan.hayward@arm.com>
2006 David Sherwood <david.sherwood@arm.com>
2008 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
2009 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
2011 * genautomata.c (main): Likewise.
2012 * genconditions.c (write_header): Likewise.
2013 * genemit.c (main): Likewise.
2014 * genextract.c (print_header): Likewise.
2015 * genopinit.c (main): Likewise.
2016 * genoutput.c (output_prologue): Likewise.
2017 * genpeep.c (main): Likewise.
2018 * genpreds.c (write_insn_preds_c): Likewise.
2019 * genrecog.c (writer_header): Likewise.
2020 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
2021 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
2022 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
2023 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
2024 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
2025 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
2026 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
2027 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
2028 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
2029 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
2030 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
2031 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
2032 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
2033 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
2034 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
2035 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
2036 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
2037 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
2038 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
2039 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
2040 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
2041 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
2042 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
2043 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
2044 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
2045 * config/darwin.c (IN_TARGET_CODE): Likewise.
2046 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
2047 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
2048 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
2049 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
2050 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
2051 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
2052 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
2053 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
2054 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
2055 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
2056 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
2057 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
2058 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
2059 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
2060 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
2061 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
2062 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
2063 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
2064 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
2065 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
2066 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
2067 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
2068 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
2069 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
2070 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
2071 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
2072 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
2073 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
2074 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
2075 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
2076 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
2077 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
2078 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
2079 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
2080 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
2081 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
2082 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
2083 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
2084 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
2085 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
2086 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
2087 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
2088 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
2089 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
2090 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
2091 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
2092 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
2093 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
2094 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
2095 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
2096 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
2097 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
2098 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
2099 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
2100 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
2101 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
2102 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
2103 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
2104 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
2105 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
2106 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2107 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
2108 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
2109 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
2110 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
2111 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
2112 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
2113 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
2114 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
2115 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
2116 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
2117 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2118 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
2119 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
2120 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
2121 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
2122 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
2123 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
2124 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
2125 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
2126 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
2127 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
2128 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
2129 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
2130 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
2131 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
2132 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
2133 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
2134 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
2135 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
2136 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
2137 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
2138 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
2139 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
2140 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
2141 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
2142 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
2143 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
2144 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
2145 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
2146 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
2147 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
2148 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
2149 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
2150 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
2151 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
2153 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2154 Alan Hayward <alan.hayward@arm.com>
2155 David Sherwood <david.sherwood@arm.com>
2157 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
2160 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2161 Alan Hayward <alan.hayward@arm.com>
2162 David Sherwood <david.sherwood@arm.com>
2164 * doc/generic.texi (VEC_SERIES_EXPR): Document.
2165 * doc/md.texi (vec_series@var{m}): Document.
2166 * tree.def (VEC_SERIES_EXPR): New tree code.
2167 * tree.h (build_vec_series): Declare.
2168 * tree.c (build_vec_series): New function.
2169 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
2170 * tree-pretty-print.c (dump_generic_node): Likewise.
2171 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
2172 * tree-inline.c (estimate_operator_cost): Likewise.
2173 * expr.c (expand_expr_real_2): Likewise.
2174 * optabs-tree.c (optab_for_tree_code): Likewise.
2175 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
2176 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
2177 * expmed.c (make_tree): Handle VEC_SERIES.
2178 * optabs.def (vec_series_optab): New optab.
2179 * optabs.h (expand_vec_series_expr): Declare.
2180 * optabs.c (expand_vec_series_expr): New function.
2181 * tree-vect-generic.c (expand_vector_operations_1): Check that
2182 the operands also have vector type.
2184 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2185 Alan Hayward <alan.hawyard@arm.com>
2186 David Sherwood <david.sherwood@arm.com>
2188 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
2189 (VEC_COND_EXPR): Add missing @tindex.
2190 * doc/md.texi (vec_duplicate@var{m}): Document.
2191 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
2192 * tree.c (build_vector_from_val): Add stubbed-out handling of
2193 variable-length vectors, using VEC_DUPLICATE_EXPR.
2194 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
2195 * cfgexpand.c (expand_debug_expr): Likewise.
2196 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2197 * tree-inline.c (estimate_operator_cost): Likewise.
2198 * tree-pretty-print.c (dump_generic_node): Likewise.
2199 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
2200 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
2201 (test_vec_duplicate_folding): New function.
2202 (fold_const_c_tests): Call it.
2203 * optabs.def (vec_duplicate_optab): New optab.
2204 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
2205 * optabs.h (expand_vector_broadcast): Declare.
2206 * optabs.c (expand_vector_broadcast): Make non-static. Try using
2207 vec_duplicate_optab.
2208 * expr.c (store_constructor): Try using vec_duplicate_optab for
2210 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
2212 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
2215 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
2216 div/mod latencies a bit.
2218 2017-12-15 Jeff Law <law@redhat.com>
2220 PR tree-optimization/36550
2221 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
2222 (mark_threaded_blocks): Rewrite code to avoid block copying when
2223 optimizing for size. Don't pessimize blocks which will be
2224 copied, but all the statements will be dead.
2226 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
2228 PR tree-optimization/81165
2229 * tree-ssa-threadupdate.c (uses_in_bb): New.
2230 (estimate_threading_killed_stmts): New.
2231 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
2232 * tree-ssa-threadedge.c
2233 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
2236 2017-12-15 Jeff Law <law@redhat.com>
2238 PR tree-optimization/83410
2239 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
2240 threads when parallelizing loops.
2242 2017-12-15 Jakub Jelinek <jakub@redhat.com>
2244 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
2246 * config/alpha/alpha.c (vms_attribute_table): Swap
2247 affects_type_identity and handler fields, adjust comments.
2248 * config/mips/mips.c (mips_attribute_table): Likewise.
2249 * config/visium/visium.c (visium_attribute_table): Likewise.
2250 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
2251 * config/microblaze/microblaze.c (microblaze_attribute_table):
2253 * config/spu/spu.c (spu_attribute_table): Likewise.
2254 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
2255 * config/arc/arc.c (arc_attribute_table): Likewise.
2256 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
2257 * config/v850/v850.c (v850_handle_interrupt_attribute,
2258 v850_handle_data_area_attribute): Formatting fixes.
2259 (v850_attribute_table): Swap affects_type_identity and handler
2260 fields, adjust comments.
2261 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
2262 * config/arm/arm.c (arm_attribute_table): Likewise.
2263 * config/avr/avr.c (avr_attribute_table): Likewise.
2264 * config/s390/s390.c (s390_attribute_table): Likewise.
2265 * config/sh/sh.c (sh_attribute_table): Likewise.
2266 * config/i386/i386.c (ix86_handle_cconv_attribute,
2267 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
2268 (ix86_attribute_table): Swap affects_type_identity and handler
2269 fields, adjust comments.
2270 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2271 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
2272 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
2273 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
2274 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
2275 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2276 * config/rx/rx.c (rx_attribute_table): Likewise.
2277 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
2278 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
2279 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
2280 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
2281 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2282 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
2283 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
2284 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
2285 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2286 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2287 * doc/plugins.texi (user_attr): Likewise. Add NULL for
2289 * attribs.c (empty_attribute_table): Swap affects_type_identity and
2291 (register_scoped_attributes, decl_attributes): Formatting fixes.
2293 PR tree-optimization/83269
2294 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
2295 subtraction in arg0's type if type is signed and arg0 is unsigned.
2299 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
2300 convert? on @0 instead of convert. Check type of @1, not @0.
2301 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
2302 convert? on @0 instead of convert. Check type of @1, not @0.
2303 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
2304 only optimize if either both @1 and @2 types are narrower
2305 precision, or both are wider or equal precision, and in the former
2306 case only if both have undefined overflow.
2308 2017-12-15 Richard Biener <rguenther@suse.de>
2311 * internal-fn.def (IFN_NOP): Add.
2312 * internal-fn.c (expand_NOP): Do nothing.
2313 * lto-streamer-in.c (input_function): Instead of removing
2314 sanitizer calls replace them with IFN_NOP calls.
2316 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
2317 Alan Hayward <alan.hayward@arm.com>
2318 David Sherwood <david.sherwood@arm.com>
2320 * dse.c (store_info, read_info_type): Replace begin and end with
2322 (print_range): New function.
2323 (set_all_positions_unneeded, any_positions_needed_p)
2324 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
2326 (record_store): Likewise. Optimize the case in which all positions
2328 (get_stored_val): Replace read_begin and read_end with read_offset
2330 (replace_read): Update call accordingly.
2332 2017-12-15 Bin Cheng <bin.cheng@arm.com>
2334 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
2335 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
2336 (loop_cand::loop_cand): Initialize above members.
2337 (loop_cand::supported_operations): Delete.
2338 (loop_cand::can_interchange_p): Inline above function.
2339 (loop_cand::classify_simple_reduction): Record number of constant
2340 initialized simple reductions.
2341 (should_interchange_loops): New parameters. Check stmt cost of loops
2343 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
2344 Update call to should_interchange_loops.
2345 (should_interchange_loop_nest): Update call to
2346 should_interchange_loops.
2348 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
2351 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
2352 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
2353 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
2355 2017-12-15 Julia Koval <julia.koval@intel.com>
2357 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
2358 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
2360 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
2361 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
2362 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
2364 2017-12-15 Julia Koval <julia.koval@intel.com>
2366 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
2367 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
2368 * config/i386/sse.md (vaesenc_<mode>): New pattern.
2369 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
2370 _mm_aesenc_epi128): New intrinsics.
2372 2017-12-15 Julia Koval <julia.koval@intel.com>
2374 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
2375 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
2377 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
2378 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
2379 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
2381 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2383 * gimple-ssa-strength-reduction.c (analyze_increments):
2384 Distinguish replacement costs for constant strides from those for
2387 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2389 * var-tracking.c (variable_tracking_main_1): Formatting fix.
2391 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2393 * doc/invoke.texi: Document -Wcast-function-type.
2394 * recog.h (stored_funcptr): Change signature.
2395 * tree-dump.c (dump_node): Avoid warning.
2396 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
2398 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
2401 * gimple-fold.c (get_range_strlen): Add the handling of non-member
2404 2017-12-14 David Malcolm <dmalcolm@redhat.com>
2406 PR tree-optimization/83312
2407 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
2408 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
2409 "val" param, and to cope with arbitrary basic blocks.
2410 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
2411 handle NULL_TREE for "val", dropping "bb" param.
2412 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
2413 drop "bb" param. Handle NULL_TREE for "val".
2414 (find_case_label_for_value): Make "switch_stmt" param const.
2415 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
2417 (vrp_prop::check_all_array_refs): Reimplement as...
2418 (check_array_bounds_dom_walker::before_dom_children): ...this new
2419 vfunc. Replace linear search through BB block list, excluding
2420 those with non-executable in-edges via dominator walk.
2422 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2424 * config/arm/arm.opt (mverbose-cost-dump): New option.
2425 * config/arm/arm.c (arm_rtx_costs): Use it.
2427 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
2430 * reload1.c (emit_input_reload_insns): Skip debug markers.
2432 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
2434 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
2438 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
2439 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
2440 * config/sh/sh.c: Skip debug insns besides notes.
2441 * config/sh/sh.md: Likewise.
2442 * config/sh/sh_treg_combine.cc: Likewise.
2443 * config/sh/sync.md: Likewise.
2445 2017-12-14 Tom de Vries <tom@codesourcery.com>
2447 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2448 Add item for weak_undefined.
2450 2017-12-14 Richard Biener <rguenther@suse.de>
2452 PR tree-optimization/67842
2453 * sese.h (bb_in_region): Remove #if 0'ed code.
2455 2017-12-14 Richard Biener <rguenther@suse.de>
2457 PR tree-optimization/83326
2458 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2459 may_be_zero parameter and handle it by not marking the first
2460 peeled copy as not exiting the loop.
2461 (try_peel_loop): Likewise.
2462 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
2463 to handle the case of constant or zero iterations and perform
2464 loop header copying on-the-fly.
2466 2017-12-14 Richard Biener <rguenther@suse.de>
2468 PR tree-optimization/83418
2469 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
2470 Instead of asserting we don't get unfolded comparisons deal with them.
2472 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2475 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
2476 basic blocks. Assert debug bind insns don't appear outside of bbs,
2477 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
2480 PR tree-optimization/83198
2481 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
2482 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
2483 value if arg is a REAL_CST with incompatible type.
2485 2017-12-14 Sudakshina Das <sudi.das@arm.com>
2486 Bin Cheng <bin.cheng@arm.com>
2489 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
2491 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
2494 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2496 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
2497 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
2499 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
2500 Alan Hayward <alan.hayward@arm.com>
2501 David Sherwood <david.sherwood@arm.com>
2503 * poly-int.h: New file.
2504 * poly-int-types.h: Likewise.
2505 * coretypes.h: Include them.
2506 (POLY_INT_CONVERSION): Define.
2507 * target.def (estimated_poly_value): New hook.
2508 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
2509 * doc/tm.texi: Regenerate.
2510 * doc/poly-int.texi: New file.
2511 * doc/gccint.texi: Include it.
2512 * doc/rtl.texi: Describe restrictions on subreg modes.
2513 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
2514 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
2515 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
2516 * targhooks.h (default_estimated_poly_value): Declare.
2517 * targhooks.c (default_estimated_poly_value): New function.
2518 * target.h (estimated_poly_value): Likewise.
2519 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
2520 (wi::unary_traits): Delete.
2521 (wi::binary_traits::signed_shift_result_type): Define for
2522 offset_int << HOST_WIDE_INT, etc.
2523 (generic_wide_int::operator <<=): Define for all types and use
2524 wi::lshift instead of <<.
2525 (wi::hwi_with_prec): Add a default constructor.
2526 (wi::ints_for): New class.
2527 (operator <<): Define for all wide-int types.
2528 (operator /): New function.
2529 (operator %): Likewise.
2530 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
2531 (ASSERT_MAYBE_NE_AT): New macros.
2533 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
2534 Dominik Vogt <vogt@linux.vnet.ibm.com>
2537 * emit-rtl.c (init_emit): Remove ??? comment.
2538 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2539 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
2540 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
2541 alignment of 3 virtual registers to BITS_PER_WORD.
2543 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
2545 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
2547 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
2548 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
2550 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2553 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
2555 2017-12-13 Jakub Jelinek <jakub@redhat.com>
2557 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
2558 with the exception of debug begin stmt markers appear before
2562 * final.c (rest_of_handle_final): Call variable_tracking_main only
2563 if !flag_var_tracking.
2565 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2566 Jakub Jelinek <jakub@redhat.com>
2570 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
2571 labels debug stmts that can only appear after labels.
2573 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
2575 PR rtl-optimization/82398
2576 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
2577 EXPR_USEFULNESS in priority comparison.
2579 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
2581 PR rtl-optimization/83393
2582 * combine.c (move_deaths): If reg_stat points to a too new insn in
2583 last_death, do not use it: find the proper insn instead.
2585 2017-12-12 Jeff Law <law@redhat.com>
2587 PR tree-optimization/83298
2588 PR tree-optimization/83362
2589 PR tree-optimization/83383
2590 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
2591 push_value_range a public interface. Add new argument to
2592 record_ranges_from_stmt.
2593 * gimple-ssa-evrp-analyze.c
2594 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
2595 Update comments. Handle recording temporary equivalences.
2596 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
2597 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
2598 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
2599 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
2600 gimple-ssa-evrp-analyze.h.
2601 (record_temporary_equivalences_from_phis): Add new argument. When
2602 the PHI arg is an SSA_NAME, set the result's range to the range
2604 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
2605 from statements too.
2606 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
2607 Pass it down to children as needed.
2608 (thread_outgoing_edges): Likewise.
2609 (thread_across_edge): Likewise. Push/pop range state as needed.
2610 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
2612 2017-12-12 Julia Koval <julia.koval@intel.com>
2614 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
2615 (PTA_CANNONLAKE): Remove PTA_CLWB.
2617 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2620 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
2621 (vcondv2div2df): Likewise.
2622 (vconduv2dfv2di): Likewise.
2624 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2626 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
2627 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
2628 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
2629 ATTR_MATHFN_FPROUNDING_ERRNO.
2631 2017-12-12 Richard Biener <rguenther@suse.de>
2633 PR tree-optimization/83385
2634 * graphite-scop-detection.c (get_order, order): Remove.
2635 (bb_to_rpo): New global.
2637 (build_scops): Sort pbbs in RPO order.
2639 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
2641 * combine.c (simplify_set): Do not transform subregs to zero_extends
2642 if the destination mode is a vector mode.
2644 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2646 PR tree-optimization/83359
2647 * tree-cfg.h (fold_loop_internal_call): Declare.
2648 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
2649 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
2650 (find_loop_dist_alias): New function.
2651 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
2652 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
2655 PR tree-optimization/80631
2656 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
2657 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
2658 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
2659 hardcoding zero as the value if COND_EXPR is never true. For
2660 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
2661 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
2662 hardcoding MAX_EXPR as the reduction operation.
2663 (is_nonwrapping_integer_induction): Allow negative step.
2664 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
2665 vect_create_epilog_for_reduction, if no value is suitable, don't
2666 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
2668 2017-12-12 Richard Biener <rguenther@suse.de>
2670 PR tree-optimization/81889
2671 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
2672 range info from the non-wrapping IV instead of just the range
2675 2017-12-12 Julia Koval <julia.koval@intel.com>
2677 * config.gcc: Add vaesintrin.h.
2678 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2680 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2681 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2683 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2684 * config/i386/immintrin.h: Include vaesintrin.h.
2685 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2686 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2687 _mm_aesdec_epi128): New intrinsics.
2689 2017-12-12 Julia Koval <julia.koval@intel.com>
2691 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
2692 OPTION_MASK_ISA_VAES_UNSET): New.
2693 (ix86_handle_option): Handle -mvaes.
2694 * config/i386/cpuid.h: Define bit_VAES.
2695 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
2696 * config/i386/i386-c.c (__VAES__): New.
2697 * config/i386/i386.c (ix86_target_string): Add -mvaes.
2698 (ix86_valid_target_attribute_inner_p): Ditto.
2699 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
2700 * config/i386/i386.opt: Add -mvaes.
2701 * doc/invoke.texi: Ditto.
2703 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
2705 * debug.h (gcc_debug_hooks): Add inline_entry.
2706 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
2707 * debug.c (do_nothing_debug_hooks): Likewise.
2708 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2709 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
2710 (dwarf2_lineno_debug_hooks): Likewise.
2712 * common.opt (gstatement-frontiers): New, setting
2713 debug_nonbind_markers_p.
2714 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
2715 * toplev.c (process_options): Autodetect value for debug statement
2717 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
2718 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
2720 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
2721 markers. Integrate source bind into debug stmt expand loop.
2722 (pass_expand::execute): Check debug marker limit. Avoid deep
2723 TER and expand debug locations for debug bind insns only.
2724 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
2726 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
2727 * final.c (reemit_insn_block_notes): Take current block from
2728 nonbind markers. Declare note where it's first set.
2729 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
2730 begin stmt markers if enabled.
2731 (notice_source_line): Handle nonbind markers. Fail if their
2732 location is unknown or that of builtins.
2733 (rest_of_handle_final): Convert begin stmt markers to notes if
2734 var-tracking didn't run.
2735 (rest_of_clean_state): Skip begin stmt markers.
2736 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
2738 * function.c (allocate_struct_function): Set begin_stmt_markers.
2739 * function.h (struct function): Add debug_marker_count counter
2740 and debug_nonbind_markers flag.
2741 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
2742 * gimple-low.c (lower_function_body): Adjust
2743 debug_nonbind_markers.
2744 (lower_stmt): Drop or skip gimple debug stmts.
2745 (lower_try_catch): Skip debug stmts.
2746 * gimple.c (gimple_build_debug_begin_stmt): New.
2747 (gimple_copy): Increment debug_marker_count if copying one.
2748 * gimple.h (gimple_build_debug_begin_stmt): Declare.
2749 * gimplify.c (rexpr_location): New.
2750 (rexpr_has_location): New.
2751 (warn_switch_unreachable_r): Handle gimple debug stmts.
2752 (shortcut_cond_r): Call expr_location.
2754 (find_goto_label): New.
2755 (shortcut_cond_expr): Call expr_has_location, expr_location, and
2757 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
2759 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
2760 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
2761 (LANG_HOOKS_INITIALIZER): ... this.
2762 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
2763 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
2765 (update_ebb_live_info): Skip debug insn markers.
2766 * lra.c (debug_insn_static_data): Rename to...
2767 (debug_bind_static_data): ... this.
2768 (debug_marker_static_data): New.
2769 (lra_set_insn_recog_data): Select one of the above depending
2771 (lra_update_isn_regno_info): Don't assume debug insns have
2773 (push_insns): Skip debug insns.
2774 * lto-streamer-in.c (input_function): Drop debug stmts
2775 depending on active options. Adjust debug_nonbind_markers.
2776 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
2777 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
2778 begin stmt marker notes.
2779 (print_insn): Likewise.
2780 * recog.c (extract_insn): Recognize rtl for debug markers.
2781 * rtl.def (DEBUG_MARKER): New.
2782 * tree-inline.c: Include params.h.
2783 (remap_gimple_stmt): Handle nonbind markers.
2784 (maybe_move_debug_stmts_to_successors): Likewise.
2785 (copy_debug_stmt): Likewise.
2786 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
2787 markers regardless of no side effects.
2788 (tsi_link_before): Don't update container's side effects when adding
2789 a begin stmt marker.
2790 (tsi_link_after): Likewise.
2791 (expr_first): Skip begin stmt markers.
2792 (expr_last): Likewise.
2793 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
2794 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2795 Disregard nonbind markers.
2796 * tree.c (make_node_stat): Don't set side effects for begin stmt
2798 (build1_stat): Likewise.
2799 * tree.def (DEBUG_BEGIN_STMT): New.
2800 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
2801 * var-tracking.c (delete_debug_insns): Renamed to...
2802 (delete_vta_debug_insns): ... this.
2803 (reemit_marker_as_note): New.
2804 (vt_initialize): Reemit markers.
2805 (delete_vta_debug_insns): Likewise.
2806 (vt_debug_insns_local): Reemit or delete markers.
2807 (variable_tracking_main_1): Likewise.
2808 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
2809 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
2810 (gimple_debug_nonbind_marker_p): New.
2811 (gimple_build_debug_bind): Adjust.
2812 (gimple_build_debug_begin_stmt): New.
2813 * doc/invoke.texi (max-debug-marker-count): New param.
2814 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
2815 (debug_parameter_ref, debug_marker): New.
2816 (NOTE_INSN_BEGIN_STMT): New.
2817 (DEBUG_INSN): Describe begin stmt markers.
2819 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
2820 without debug insns, we wouldn't, but clean up debug insns
2821 after a control flow insn nevertheless.
2823 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
2824 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
2825 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
2826 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
2827 (rtl_tidy_fallthru_edge): Likewise.
2828 (rtl_verify_fallthru): Likewise.
2829 (rtl_verify_bb_layout): Likewise.
2830 (skip_insns_after_block): Likewise.
2831 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
2832 * dwarf2out.c: Include print-rtl.h.
2833 (dwarf2out_next_real_insn): New.
2834 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
2835 Dump debug binds in asm comments.
2836 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
2837 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
2838 callers to use gsi_start_nondebug_bb instead.
2839 (gsi_after_labels): Skip gimple debug stmts.
2840 (gsi_start_nondebug): New.
2841 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
2842 (proper_loop_form_for_interchange): Adjust.
2843 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
2844 * gimple.h (gimple_seq_last_nondebug_stmt): New.
2845 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
2846 (collect_fallthrough_labels): Likewise.
2847 (should_warn_for_implicit_fallthrough): Likewise.
2848 (warn_implicit_fallthrough_r): Likewise.
2849 (expand_FALLTHROUGH_r): Likewise.
2850 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
2851 (graphite_copy_stmts_from_block): Skip nonbind markers.
2852 * haifa-sched.c (sched_extend_bb): Skip debug insns.
2853 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
2854 * jump.c (clean_barriers): Skip debug insns.
2855 * omp-expand.c (expand_parallel_call): Skip debug insns.
2856 (expand_task_call): Likewise.
2857 (remove_exit_barrier): Likewise.
2858 (expand_omp_taskreg): Likewise.
2859 (expand_omp_for_init_counts): Likewise.
2860 (expand_omp_for_generic): Likewise.
2861 (expand_omp_for_static_nochunk): Likewise.
2862 (expand_omp_for_static_chunk): Likewise.
2863 (expand_omp_simd): Likewise.
2864 (expand_omp_taskloop_for_outer): Likewise.
2865 (expand_omp_taskloop_for_inner): Likewise.
2866 (expand_oacc_for): Likewise.
2867 (expand_omp_sections): Likewise.
2868 (expand_omp_single): Likewise.
2869 (expand_omp_synch): Likewise.
2870 (expand_omp_atomic_load): Likewise.
2871 (expand_omp_atomic_store): Likewise.
2872 (expand_omp_atomic_fetch_op): Likewise.
2873 (expand_omp_atomic_pipeline): Likewise.
2874 (expand_omp_atomic_mutex): Likewise.
2875 (expand_omp_target): Likewise.
2876 (grid_expand_omp_for_loop): Likewise.
2877 (grid_expand_target_grid_body): Likewise.
2878 (build_omp_regions_1): Likewise.
2879 * omp-low.c (check_combined_parallel): Skip debug stmts.
2880 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
2881 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
2882 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
2884 * tree-cfg.c (make_blobs_1): Skip debug stmts.
2885 (make_edges): Likewise.
2886 (cleanup_dead_labels): Likewise.
2887 (gimple_can_merge_blocks_p): Likewise.
2888 (stmt_starts_bb_p): Likewise.
2889 (gimple_block_label): Likewise.
2890 (gimple_redirect_edge_and_branch): Likewise.
2891 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
2893 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
2895 * tree-pretty-print (print_declaration): Omit initializer in slim
2897 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
2899 (eliminate_unnecessary_stmts): Stabilize block removal order.
2900 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
2901 * var-tracking.c (get_first_insn): New.
2902 (vt_emit_notes): Call it.
2903 (vt_initialize): Walk any insns before the first BB.
2904 (delete_debug_insns): Likewise.
2906 * gimple.h (enum gimple_debug_subcode): Add
2907 GIMPLE_DEBUG_BEGIN_STMT.
2908 (gimple_debug_begin_stmt_p): New.
2909 (gimple_debug_nonbind_marker_p): New.
2910 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
2911 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
2912 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
2913 * insn-notes.def (BEGIN_STMT): New.
2914 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
2915 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
2916 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
2917 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
2918 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
2919 (INSN_DEBUG_MARKER_KIND): New.
2920 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
2921 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
2922 (INSN_VAR_LOCATION_PTR): New.
2923 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
2925 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
2926 deps only if debug bind insns are enabled.
2927 (pass_expand::execute): Avoid deep TER and expand
2928 debug locations for debug bind insns only.
2929 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
2930 debug stmts special handling down to debug bind stmts.
2931 * combine.c (try_combine): Narrow debug insns special handling
2932 down to debug bind insns.
2933 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
2934 Narrow debug insns preexisting special handling down to debug
2936 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
2937 handling down to debug bind insns.
2938 * function.c (instantiate_virtual_regs): Skip debug markers,
2939 adjust handling of debug binds.
2940 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
2941 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
2942 * haifa-sched.c (schedule_insn): Narrow special handling of debug
2943 insns to debug bind insns.
2944 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
2945 special handling of debug stmts to debug bind stmts.
2946 * ipa-split.c (split_function): Likewise.
2947 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
2948 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
2950 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
2951 * regrename.c (build_def_use): Likewise.
2952 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
2953 (pass_cprop_hardreg): Narrow special casing of debug insns to
2955 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
2956 * reload1.c (reload): Likewise.
2957 * sese.c (sese_insert_phis_for_liveouts): Narrow special
2958 casing of debug stmts to debug bind stmts.
2959 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
2960 * ssa-iterators.h (num_imm_uses): Likewise.
2961 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
2962 debug stmts to debug bind stmts.
2963 * tree-inline.c (tree_function_versioning): Narrow special casing
2964 of debug stmts to debug bind stmts.
2965 * tree-loop-distribution.c (generate_loops_for_partition):
2966 Narrow special casing of debug stmts to debug bind stmts.
2967 * tree-sra.c (analyze_access_subtree): Narrow special casing
2968 of debug stmts to debug bind stmts.
2969 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
2970 stmts to debug bind stmts.
2971 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
2972 casing of debug stmts to debug bind stmts.
2973 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
2974 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
2975 casing of debug stmts to debug bind stmts.
2976 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2978 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
2979 debug stmts to debug bind stmts.
2980 (gimple_replace_ssa_lhs): Likewise.
2981 (insert_debug_temp_for_var_def): Likewise.
2982 (insert_debug_temps_for_defs): Likewise.
2983 (reset_debug_uses): Likewise.
2984 * tree-ssanames.c (release_ssa_name_fn): Likewise.
2985 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
2986 (adjust_debug_stmts): Likewise.
2987 (adjust_phi_and_debug_stmts): Likewise.
2988 (vect_do_peeling): Likewise.
2989 * tree-vect-loop.c (vect_transform_loop): Likewise.
2990 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
2991 * var-tracking.c (adjust_mems): Narrow special casing of debug
2992 insns to debug bind insns.
2993 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
2994 (compute_bb_dataflow, vt_find_locations): Likewise.
2995 (vt_expand_loc, emit_notes_for_changes): Likewise.
2996 (vt_init_cfa_base): Likewise.
2997 (vt_emit_notes): Likewise.
2998 (vt_initialize): Likewise.
2999 (vt_finalize): Likewise.
3001 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
3002 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
3003 (next_nonnote_nondebug_insn_bb): New.
3004 (prev_nonnote_nondebug_insn_bb): New.
3005 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
3006 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
3007 (prev_nonnote_nondebug_insn_bb): Declare.
3008 (next_nonnote_nondebug_insn_bb): Declare.
3009 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
3010 * cfgrtl.c (get_last_bb_insn): Likewise.
3011 * lra.c (push_insns): Likewise.
3013 2017-12-11 David Malcolm <dmalcolm@redhat.com>
3016 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
3018 * selftest.c (selftest::test_runner::test_runner): New ctor.
3019 (selftest::test_runner::~test_runner): New dtor.
3020 * selftest.h (class selftest::test_runner): New class.
3022 2017-12-11 Carl Love <cel@us.ibm.com>
3024 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
3025 vec_extract_fp32_from_shortl]): Add #defines.
3026 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
3027 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
3028 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
3029 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
3030 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
3031 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
3032 * doc/extend.texi: Add documentation for the added builtins.
3034 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
3036 PR rtl-optimization/80693
3037 PR rtl-optimization/81019
3038 PR rtl-optimization/81020
3039 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
3040 are not mentioned in i3. Place the REG_UNUSED note on i2,
3041 possibly modified to REG_DEAD, if it did not originate in i3.
3043 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3045 * recog.c (store_data_bypass_p_1): New function.
3046 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
3047 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
3049 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
3051 PR rtl-optimization/83361
3052 * ifcvt.c (if_convert): Call fixup_partitions.
3054 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
3056 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3057 early folding of splat_u{8,16,32}.
3059 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3061 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
3062 output_operand_lossage first argument with capital letter.
3063 (aarch64_override_options): Don't start error and sorry first argument
3064 with capital letter.
3066 2017-12-11 Andi Kleen <ak@linux.intel.com>
3068 PR gcov-profile/83355
3069 * auto-profile.c (string_table::get_index_by_decl): Don't
3070 recurse when abstract origin points to itself.
3072 2017-12-11 Bin Cheng <bin.cheng@arm.com>
3074 PR tree-optimization/83320
3075 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
3076 (prune_datarefs_not_in_loop): Ditto.
3078 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
3080 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
3082 2017-12-10 Jakub Jelinek <jakub@redhat.com>
3084 PR tree-optimization/83337
3085 * gimple-loop-interchange.cc (compute_access_stride): Handle
3086 bitfield DRs properly.
3088 2017-12-09 Jakub Jelinek <jakub@redhat.com>
3090 PR tree-optimization/83338
3091 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
3092 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
3093 vector of unsigned integers to vector of signed integers.
3095 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
3097 PR rtl-optimization/83317
3098 * lra-constraints.c (process_address_1): Add insn code check.
3100 2017-12-08 Michael Matz <matz@suse.de>
3102 Fix PR tree-optimization/83323
3103 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
3104 head-controlled loops and loop BBs.
3105 * common.opt (funroll-and-jam): Remove, instead ...
3106 (floop-unroll-and-jam): ... reuse this option.
3107 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
3108 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
3109 (-floop-unroll-and-jam): ... this option.
3111 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3113 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
3114 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
3115 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
3117 2017-12-08 Richard Biener <rguenther@suse.de>
3120 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
3121 handle accesses outside of zero-sized vars.
3123 2017-12-08 Martin Jambor <mjambor@suse.cz>
3125 PR tree-optimization/83141
3126 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
3127 test for MEM_REFs implicitely changing types with padding. Remove
3129 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
3131 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3133 * config/arc/arc.c (arc_attribute_table): Add exclusions to
3135 * config/avr/avr.c (avr_attribute_table): Likewise.
3136 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3137 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3138 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3139 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
3140 of struct attribute_spec.
3141 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3143 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
3146 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
3148 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
3151 * profile-count.c (profile_count::from_gcov_type): Move from
3152 profile-count.h; handle overflow.
3153 * profile-count.h (profile_count::from_gcov_type): Move offline.
3155 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
3157 PR rtl-optimization/83304
3158 * combine.c (move_deaths): If we do not know where a register died,
3161 2017-12-08 Richard Biener <rguenther@suse.de>
3163 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
3164 Provide -fopt-info-loop feedback when we interchange in a nest.
3166 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3168 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
3169 for armv6 ARM CPU IDs.
3171 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3173 * common/config/arm/arm-common.c: Include <algorithm>.
3174 (INCLUDE_VECTOR): Define.
3175 (compare_opt_names): New function.
3176 (arm_rewrite_selected_arch): Only strip out extensions that can be
3177 expressed through -mfpu. Sort the remaining extensions
3180 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3182 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
3183 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
3184 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
3185 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
3187 2017-06-08 Tristan Gingold <gindold@adacore.com>
3190 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
3191 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
3193 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3195 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
3196 of struct attribute_spec.
3198 2017-12-08 Julia Koval <julia.koval@intel.com>
3200 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
3201 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
3202 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
3203 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
3204 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
3205 _mm_maskz_dpwssds_epi32): Ditto.
3207 2017-12-08 Richard Biener <rguenther@suse.de>
3209 PR tree-optimization/81303
3210 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
3211 conditions try to create a comparison vector type matching
3212 the data vector type.
3213 (vectorizable_condition): Adjust.
3214 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
3215 Leave invariant conditions alone in case we can vectorize those.
3217 2017-12-08 Julia Koval <julia.koval@intel.com>
3219 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
3220 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
3221 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
3222 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
3223 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
3225 2017-12-08 Julia Koval <julia.koval@intel.com>
3227 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
3228 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
3229 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
3230 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
3231 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
3232 _mm_maskz_dpbusds_epi32): New intrinsics.
3234 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3236 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
3238 (ld<bh>io_signed): Likewise.
3239 (st<bhw>io): Likewise.
3240 * config/nios2/predicates.md (ldstio_memory_operand): Allow
3241 SMALL_INT12 constant integer operand.
3243 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3245 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
3247 (nios2_large_constant_allowed): ...this. Adjust uses.
3248 (nios2_plus_symbolic_constant_p): Rename to...
3249 (nios2_plus_large_constant_p): ...this. Adjust uses.
3250 (nios2_legitimate_address_p): Correct CONST_INT handling.
3251 (nios2_symbolic_memory_operand_p): Rename to...
3252 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
3253 (nios2_large_constant_p): Check for large constant integers too.
3254 (nios2_split_large_constant): Handle constant integers.
3255 (nios2_split_symbolic_memory_operand): Rename to...
3256 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
3257 (nios2_legitimize_constant_address): Handle constant integers.
3258 (r0rel_constant_p): Handle small constant integers.
3259 (nios2_print_operand_address): Handle r0-relative integer addresses.
3260 * config/nios2/nios2-protos.h: Adjust for renamed functions.
3261 * config/nios2/nios2.md: Adjust for renamed functions.
3263 2017-12-07 Andrew Waterman <andrew@sifive.com>
3265 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
3266 (TARGET_HAVE_SRODATA_SECTION): New define.
3267 (riscv_select_section): New function.
3269 2017-12-08 Joseph Myers <joseph@codesourcery.com>
3270 Alexander Monakov <amonakov@ispras.ru>
3271 Jakub Jelinek <jakub@redhat.com>
3274 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
3276 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3278 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
3279 a bool success value. Don't call output_operand_lossage here.
3280 (aarch64_print_ldpstp_address): Return a bool success value.
3281 (aarch64_print_operand_address): Call output_addr_const if
3282 aarch64_print_address_internal fails.
3283 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
3284 'y'; call output_operand_lossage instead. Call output_operand_lossage
3285 if aarch64_print_ldpstp_address fails.
3287 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3289 * tree-vector-builder.h
3290 (tree_vector_builder::binary_encoded_nelts): Declare.
3291 * tree-vector-builder.c
3292 (tree_vector_builder::binary_encoded_nelts): New function.
3293 * fold-const.c (negate_expr_p): Likewise.
3294 (operand_equal_p, fold_checksum_tree): Likewise.
3295 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
3296 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
3297 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
3298 (uniform_vector_p): Likewise.
3299 * varasm.c (const_hash_1, compare_constant): Likewise.
3300 * tree-ssa-ccp.c: Include tree-vector-builder.h.
3301 (valid_lattice_transition): Operate directly on the VECTOR_CST
3303 * ipa-icf.c: Include tree-vector-builder.h.
3304 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
3305 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
3307 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3309 * tree.c (build_vector): Delete.
3310 * tree.h (build_vector): Make static and move into the self-testing
3313 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3315 * vector-builder.h (vector_builder::derived): New const overload.
3316 (vector_builder::elt): New function.
3317 * tree-vector-builder.h (tree_vector_builder::type): New function.
3318 (tree_vector_builder::apply_step): Declare.
3319 * tree-vector-builder.c (tree_vector_builder::apply_step): New
3321 * gimple-fold.h (tree_vector_builder): Declare.
3322 (gimple_build_vector): Take a tree_vector_builder instead of a
3323 type and vector of elements.
3324 * gimple-fold.c (gimple_build_vector): Likewise.
3325 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
3327 (get_initial_defs_for_reduction): Likewise.
3328 (vectorizable_induction): Likewise.
3330 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3332 * tree-vector-builder.h
3333 (tree_vector_builder::new_binary_operation): Declare.
3334 * tree-vector-builder.c
3335 (tree_vector_builder::new_binary_operation): New function.
3336 * fold-const.c (fold_relational_const): Use it.
3337 (const_binop): Likewise. Check that both input vectors have
3338 the same number of elements, thus excluding things like WIDEN_SUM.
3339 Check whether it is possible to operate directly on the encodings
3342 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3344 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
3345 new_unary_operation, operating only on the encoded elements.
3346 (const_unop): Likewise.
3347 (exact_inverse): Likewise.
3348 (distributes_over_addition_p): New function.
3349 (const_binop): Use tree_vector_builder and new_unary_operation
3350 for combinations of VECTOR_CST and INTEGER_CST. Operate only
3351 on the encoded elements unless the encoding is strided and the
3352 operation does not distribute over addition.
3353 (fold_convert_const): Use tree_vector_builder and
3354 new_unary_operation. Operate only on the encoded elements
3355 for truncating integer conversions, or for non-stepped encodings.
3357 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3359 * config/sparc/sparc.c: Include tree-vector-builder.h.
3360 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
3361 * expmed.c: Include tree-vector-builder.h.
3362 (make_tree): Use tree_vector_builder instead of build_vector.
3363 * fold-const.c: Include tree-vector-builder.h.
3364 (const_binop): Use tree_vector_builder instead of build_vector.
3365 (const_unop): Likewise.
3366 (native_interpret_vector): Likewise.
3367 (fold_vec_perm): Likewise.
3368 (fold_ternary_loc): Likewise.
3369 * gimple-fold.c: Include tree-vector-builder.h.
3370 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
3372 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
3373 (simplify_vector_constructor): Use tree_vector_builder instead
3375 * tree-vect-generic.c: Include tree-vector-builder.h.
3376 (add_rshift): Use tree_vector_builder instead of build_vector.
3377 (expand_vector_divmod): Likewise.
3378 (optimize_vector_constructor): Likewise.
3379 * tree-vect-loop.c: Include tree-vector-builder.h.
3380 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
3381 of build_vector. Explicitly use a stepped encoding for
3383 * tree-vect-slp.c: Include tree-vector-builder.h.
3384 (vect_get_constant_vectors): Use tree_vector_builder instead
3386 (vect_transform_slp_perm_load): Likewise.
3387 (vect_schedule_slp_instance): Likewise.
3388 * tree-vect-stmts.c: Include tree-vector-builder.h.
3389 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
3390 (vect_gen_perm_mask_any): Likewise.
3391 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
3392 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
3394 (build_vector_from_val): Likewise. Explicitly use a duplicate
3397 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
3399 * doc/generic.texi (VECTOR_CST): Describe new representation of
3401 * vector-builder.h: New file.
3402 * tree-vector-builder.h: Likewise.
3403 * tree-vector-builder.c: Likewise.
3404 * Makefile.in (OBJS): Add tree-vector-builder.o.
3405 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
3406 * tree-core.h (tree_base): Add a vector_cst field to the u union.
3407 (tree_vector): Change the number of elements to
3408 vector_cst_encoded_nelts.
3409 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
3410 (VECTOR_CST_ELTS): Delete.
3411 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
3412 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
3413 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
3414 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
3415 (VECTOR_CST_ENCODED_ELT): Likewise.
3416 (vector_cst_encoded_nelts): New function.
3417 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3418 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3419 (vector_cst_int_elt, vector_cst_elt): Declare.
3420 * tree.c: Include tree-vector-builder.h.
3421 (tree_code_size): Abort if passed VECTOR_CST.
3422 (tree_size): Update for new VECTOR_CST layout.
3423 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3424 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3425 (build_vector): Use tree_vector_builder.
3426 (vector_cst_int_elt, vector_cst_elt): New functions.
3427 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
3428 encoded elements and then create the vector in the canonical form.
3429 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
3430 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
3431 (tree_c_tests): Call test_vector_cst_patterns.
3432 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
3434 (hash_tree): Likewise.
3435 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
3436 (streamer_write_tree_header): Likewise.
3437 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
3438 (streamer_alloc_tree): Likewise. Update call to make_vector.
3439 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
3441 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3443 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
3444 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
3445 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
3447 2017-12-07 Bin Cheng <bin.cheng@arm.com>
3448 Richard Biener <rguenther@suse.de>
3450 PR tree-optimization/81303
3451 * Makefile.in (gimple-loop-interchange.o): New object file.
3452 * common.opt (floop-interchange): Reuse the option from graphite.
3453 * doc/invoke.texi (-floop-interchange): Ditto. New document for
3454 -floop-interchange and mention it for -O3.
3455 * opts.c (default_options_table): Enable -floop-interchange at -O3.
3456 * gimple-loop-interchange.cc: New file.
3457 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
3458 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
3459 * passes.def (pass_linterchange): New pass.
3460 * timevar.def (TV_LINTERCHANGE): New time var.
3461 * tree-pass.h (make_pass_linterchange): New declaration.
3462 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
3463 interchange. Record IV before/after increment in new parameters.
3464 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
3465 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
3467 (check_reduction_path): ...New function here.
3468 * tree-vectorizer.h (check_reduction_path): New declaration.
3470 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
3473 PR rtl-optimization/80818
3474 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
3475 always early clobbered.
3476 * lra-lives.c (process_bb_lives): Check input hard regs for early
3477 clobbered non-operand hard reg.
3479 2017-12-07 Jakub Jelinek <jakub@redhat.com>
3482 * tree-cfg.c (verify_gimple_assign_binary): Don't require
3483 types_compatible_p, just that TYPE_MODE is the same.
3485 2017-12-07 Martin Sebor <msebor@redhat.com>
3488 * attribs.c (empty_attribute_table): Initialize new member of
3489 struct attribute_spec.
3490 (decl_attributes): Add argument. Handle mutually exclusive
3491 combinations of attributes.
3492 (selftests::test_attribute_exclusions): New function.
3493 (selftests::attribute_c_tests): Ditto.
3494 * attribs.h (decl_attributes): Add default argument.
3495 * selftest.h (attribute_c_tests): Declare.
3496 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
3497 * tree-core.h (attribute_spec::exclusions, exclude): New type and
3499 * doc/extend.texi (Common Function Attributes): Update const and pure.
3500 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
3501 of struct attribute_spec.
3502 * config/arc/arc.c (arc_attribute_table): Same.
3503 * config/arm/arm.c (arm_attribute_table): Same.
3504 * config/avr/avr.c ( avr_attribute_table): Same.
3505 * config/bfin/bfin.c (bfin_attribute_table): Same.
3506 * config/cr16/cr16.c (cr16_attribute_table): Same.
3507 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
3508 * config/h8300/h8300.c (h8300_attribute_table): Same.
3509 * config/i386/i386.c (ix86_attribute_table): Same.
3510 * config/ia64/ia64.c (ia64_attribute_table): Same.
3511 * config/m32c/m32c.c (m32c_attribute_table): Same.
3512 * config/m32r/m32r.c (m32r_attribute_table): Same.
3513 * config/m68k/m68k.c (m68k_attribute_table): Same.
3514 * config/mcore/mcore.c (mcore_attribute_table): Same.
3515 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
3516 * config/mips/mips.c (mips_attribute_table): Same.
3517 * config/msp430/msp430.c (msp430_attribute_table): Same.
3518 * config/nds32/nds32.c (nds32_attribute_table): Same.
3519 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
3520 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
3521 * config/rl78/rl78.c (rl78__attribute_table): Same.
3522 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
3523 * onfig/rx/rx.c (rx_attribute_table): Same.
3524 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
3525 * config/sh/sh.c (sh_attribute_table): Same.
3526 * config/sparc/sparc.c (sparc_attribute_table): Same.
3527 * config/spu/spu.c (spu_attribute_table): Same.
3528 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
3529 * config/v850/v850.c (v850_attribute_table): Same.
3530 * config/visium/visium.c (visium_attribute_table): Same.
3532 2017-12-07 Tamar Christina <tamar.christina@arm.com>
3535 * config/arm/arm.c (INCLUDE_STRING): Define.
3536 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
3537 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
3540 2017-12-07 Michael Matz <matz@suse.de>
3542 Add unroll and jam pass
3544 * gimple-loop-jam.c: New file.
3545 * Makefile.in (OBJS): Add gimple-loop-jam.o.
3546 * common.opt (funroll-and-jam): New option.
3547 * opts.c (default_options_table): Add unroll-and-jam at -O3.
3548 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
3549 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
3550 * passes.def: Add pass_loop_jam.
3551 * timevar.def (TV_LOOP_JAM): Add.
3552 * tree-pass.h (make_pass_loop_jam): Declare.
3553 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
3554 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
3555 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
3556 to flow_loop_tree_node_add.
3557 (duplicate_subloops, copy_loops_to): Append to sibling list.
3558 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
3559 * doc/invoke.texi (-funroll-and-jam): Document new option.
3560 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
3562 2017-12-07 Richard Biener <rguenther@suse.de>
3564 PR tree-optimization/83296
3565 PR tree-optimization/67769
3566 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
3567 flow sensitive info in an unrelated BB.
3568 (value_replacement): Use reset_flow_sensitive_info.
3569 (minmax_replacement): Reset flow sensitive info on the def
3570 we move. Do not reset flow sensitive info in the whole BB
3571 we move the stmt to.
3572 (abs_replacement): Likewise.
3574 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3577 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3578 rs6000_cpu to the given -mcpu=, or to the default processor.
3580 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3582 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
3583 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
3584 instead of rs6000_cpu_attr.
3585 (rs6000_adjust_cost): Ditto.
3586 (is_microcoded_insn): Ditto.
3587 (rs6000_adjust_priority): Ditto.
3588 (rs6000_issue_rate): Ditto.
3589 (rs6000_use_sched_lookahead): Ditto.
3590 (rs6000_use_sched_lookahead_guard): Ditto.
3591 (rs6000_sched_reorder): Ditto.
3592 (force_new_group): Ditto.
3593 * config/rs6000/rs6000.md (cpu attribute): Ditto.
3594 (group_ending_nop): Ditto.
3596 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3598 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
3599 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
3600 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
3601 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
3603 (rs6000_reassociation_width): Ditto.
3604 (rs6000_emit_epilogue): Ditto.
3605 (rs6000_adjust_cost): Ditto.
3606 (is_microcoded_insn): Ditto.
3607 (is_cracked_insn): Ditto.
3608 (rs6000_adjust_priority): Ditto.
3609 (rs6000_sched_reorder): Ditto.
3610 (rs6000_sched_reorder2): Ditto.
3611 (insn_must_be_first_in_group): Ditto.
3612 (insn_must_be_last_in_group): Ditto.
3613 (rs6000_register_move_cost): Ditto.
3614 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
3617 2017-12-07 Julia Koval <julia.koval@intel.com>
3619 * config.gcc: Add vaesintrin.h.
3620 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3622 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3623 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3625 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3626 * config/i386/immintrin.h: Include vaesintrin.h.
3627 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3628 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3629 _mm_aesdec_epi128): New intrinsics.
3631 2017-12-06 David Malcolm <dmalcolm@redhat.com>
3633 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
3634 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
3635 c-family/c-spellcheck.cc.
3636 (best_macro_match::best_macro_match): Likewise.
3638 (struct edit_distance_traits<cpp_hashnode *>): Move to
3639 c-family/c-spellcheck.h.
3640 (class best_macro_match): Likewise.
3642 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3644 PR tree-optimization/83293
3645 * gimple-ssa-strength-reduction.c (insert_initializers): Use
3646 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
3647 might insert into empty bb.
3650 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
3651 simplify for plus with :c added, and pointer_plus without that.
3652 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
3653 with undefined overflow and the conversion is not widening,
3654 perform negation in utype and only convert to type afterwards.
3655 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
3656 simplify for plus with :c added, and pointer_plus without that.
3657 If type is integral with undefined overflow and the conversion is
3658 not widening, perform minus in utype and only convert to type
3659 afterwards. Move the last pointer_diff_expr simplify into the
3662 2017-12-06 Martin Sebor <msebor@redhat.com>
3664 PR tree-optimization/82646
3665 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
3666 strncpy, not maxlen.
3668 2017-12-06 Martin Sebor <msebor@redhat.com>
3670 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
3673 PR tree-optimization/83075
3674 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
3675 strncat/strncpy don't change length of source string.
3677 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
3680 2017-11-29 Martin Aberg <maberg@gaisler.com>
3682 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3683 to prevent b2bst errata sequence.
3684 (sqrtdf2_fix): Likewise.
3686 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3688 PR tree-optimization/81945
3689 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
3690 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
3691 to dest_cfun has orig_loop_num set, either remap it to the new
3692 loop number if the loop got moved too, or clear it.
3694 2017-12-05 Steve Ellcey <sellcey@cavium.com>
3696 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
3698 (thunderx2t99_nothing): New insn reservation.
3699 (thunderx2t99_mrs): New insn reservation.
3700 (thunderx2t99_multiple): New insn reservation.
3701 (thunderx2t99_alu_basi): Add bfx to reservation.
3702 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
3704 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3707 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
3709 2017-12-05 Bin Cheng <bin.cheng@arm.com>
3711 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
3712 tree-ssa-pre.c::remove_dead_inserted_code.
3713 * tree-ssa-dce.h: New file.
3714 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
3715 (remove_dead_inserted_code): Move and rename to function
3716 tree-ssa-dce.c::simple_dce_from_worklist.
3717 (pass_pre::execute): Update use.
3719 2017-12-05 Richard Biener <rguenther@suse.de>
3721 PR tree-optimization/83277
3722 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
3723 to code-gen liveout vars.
3725 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
3727 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
3728 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
3729 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
3732 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
3734 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
3736 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
3737 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
3740 2017-12-05 Richard Biener <rguenther@suse.de>
3742 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
3743 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
3744 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
3745 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
3747 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3749 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
3751 2017-12-05 Martin Liska <mliska@suse.cz>
3752 Jakub Jelinek <jakub@redhat.com>
3754 * doc/invoke.texi: Document the options.
3755 * flag-types.h (enum sanitize_code): Add
3756 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3757 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
3758 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3759 * opts.c: Define new sanitizer options.
3760 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
3761 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
3763 2017-12-05 Julia Koval <julia.koval@intel.com>
3765 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
3766 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
3767 (ix86_handle_option): Handle -mavx512vnni.
3768 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
3769 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3770 * config/i386/i386-c (__AVX512VNNI__): New.
3771 * config/i386/i386.c (ix86_target_string): Handle new option.
3772 (ix86_valid_target_attribute_inner_p): Handle new option.
3773 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
3774 * config/i386/i386.opt (mavx512vnni): New option.
3776 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3779 * athlon.md: Disable for generic.
3780 * haswell.md: Enable for generic.
3781 * i386.c (ix86_sched_init_global): Add core hooks for generic.
3782 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
3784 (ix86_adjust_cost): Move generic to haswell path.
3786 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
3788 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
3789 instead of MEM_P in a couple more places. Fix formatting issues.
3791 2017-12-04 Jim Wilson <jimw@sifive.com>
3793 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
3794 instead of GP_REG_LAST-1.
3795 (riscv_adjust_libcall_cfi_prologue): Likewise.
3796 (riscv_adjust_libcall_cri_epilogue): Likewise.
3797 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
3800 2017-12-04 Luis Machado <luis.machado@linaro.org>
3802 * ipa-pure-const.c (check_decl): Add missing newline.
3803 (state_from_flags): Likewise.
3805 2017-12-04 Jeff Law <law@redhat.com>
3807 PR tree-optimizatin/78496
3808 * gimple-ssa-evrp-analyze.h
3809 (evrp_range_analyzer::get_vr_values): Simplify.
3810 * gimple-ssa-evrp-analyze.c: Corresponding changes.
3811 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
3812 and gimple-ssa-evrp-analyze.h.
3813 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
3814 (simplify_stmt_for_jump_threading): Copy a blob of code from
3815 tree-vrp.c to use ranges to simplify statements.
3816 (dom_opt_dom_walker::before_dom_children): Call
3817 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
3818 (dom_opt_dom_walker::after_dom_children): Similarly for
3819 evrp_range_analyzer::leave.
3820 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
3823 * gimple-ssa-evrp-analyze.c
3824 (evrp_range_analyzer::extract_range_from_stmt): Always use
3825 vr_values::update_value_range so preexisting range info is
3826 medged with new range info, even if the new range is VR_VARYING.
3828 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3830 * combine.c: Adjust comment.
3831 (use_crosses_set_p): Delete.
3832 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
3833 (try_combine): Ditto.
3835 2017-12-04 Richard Biener <rguenther@suse.de>
3837 PR tree-optimization/83255
3838 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
3839 Re-add zero-iteration check.
3841 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3843 PR rtl-optimization/83245
3844 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
3845 hard registers as earlyclobber, also if not in an asm.
3847 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3851 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3854 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3855 rs6000_cpu based on cpu_index, not tune_index.
3857 2017-12-04 Richard Biener <rguenther@suse.de>
3859 PR tree-optimization/83238
3860 * graphite-scop-detection.c (scop_detection::merge_sese): Make
3861 code match comment, rejecting invalid SESE regions.
3863 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
3865 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
3866 require base operand is a REG_POINTER prior to reload on targets
3867 with non-equivalent space registers.
3869 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3871 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
3872 (update_specialized_profile): Fix updating of counts.
3873 (perhaps_add_new_callers): Likewise.
3875 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3878 * x86-tune.def: Remove obsolette FIXMEs.
3879 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
3880 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
3881 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
3883 (X86_TUNE_PAD_RETURNS): Disable for generic.
3885 2017-12-02 Jakub Jelinek <jakub@redhat.com>
3887 PR tree-optimization/83170
3888 PR tree-optimization/83241
3889 * gimple-ssa-store-merging.c
3890 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
3891 gimple_vuse (ins_stmt) in case it has changed.
3892 (imm_store_chain_info::output_merged_store): Likewise.
3894 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
3898 * tree-cfg.c (pass_warn_function_return::execute): Handle
3899 __builtin_ubsan_handle_missing_return like __builtin_unreachable
3900 with BUILTINS_LOCATION.
3904 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
3905 is BLKmode for vector field with vector raw mode, use TYPE_MODE
3906 instead of DECL_MODE.
3908 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
3909 last argument to rtx pointer.
3910 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
3911 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
3912 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
3913 depending on the chosen ISAs.
3914 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
3915 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
3916 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
3918 * config/i386/sse.md (mov<mode>_internal): Likewise.
3919 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
3921 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3923 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
3925 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3927 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
3928 parameters from prototype.
3929 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
3930 parameters. Don't print a comment.
3931 (emit_fusion_gpr_load): Adjust.
3932 (emit_fusion_load_store): Adjust.
3933 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
3934 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
3935 comment on the second line.
3937 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3940 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3941 rs6000_cpu based on cpu_index, not tune_index.
3943 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3945 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
3946 which_alternative instead of which_alternative + 1.
3947 (output_asm_insn): Print an extra tab if the template is short.
3949 2017-12-01 Jim Wilson <jimw@sifive.com>
3951 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
3953 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
3954 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
3956 * doc/tm.texi: Regenerate.
3958 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
3961 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
3962 whether we can allocate pseudos before trying to fix an address.
3963 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
3964 memory address is indexed or indirect.
3965 (floatuns_<mode>si2_hw2): Likewise.
3967 2017-12-01 Jason Merrill <jason@redhat.com>
3969 * Makefile.in (TAGS): Add c-family/*.cc.
3971 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
3973 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
3974 (call_value_insn): Likewise.
3975 (sibcall_insn): Likewise.
3976 (sibcall_value_insn): Likewise.
3977 (movsi_aarch64): Likewise.
3978 (movdi_aarch64): Likewise.
3979 (add_losym_): Likewise.
3980 (ldr_got_small_): Likewise.
3981 (ldr_got_small_sidi): Likewise.
3982 (ldr_got_small_28k_): Likewise.
3983 (ldr_got_small_28k_sidi): Likewise.
3984 * config/aarch64/aarch64.c (aarch64_print_address_internal):
3985 Move output_addr_const to symbolic case. Add error check.
3987 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3989 * config/s390/predicates.md (plus16_Q_operand): New predicate.
3990 * config/s390/s390.md: Disable MVC merging peephole if it would
3991 disable operand forwarding.
3992 (new peephole2): Split MVCs if it would turn them into up to 2
3995 2017-12-01 Richard Biener <rguenther@suse.de>
3997 PR tree-optimization/83232
3998 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
3999 detection of same access. Instead of breaking the group here
4000 do not consider the duplicate. Add comment explaining real fix.
4002 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
4004 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
4006 2017-12-01 Sudakshina Das <sudi.das@arm.com>
4008 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
4010 2017-12-01 Jakub Jelinek <jakub@redhat.com>
4012 * function.h (struct function): Remove cilk_frame_decl,
4013 is_cilk_function and calls_cilk_spawn fields.
4014 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
4016 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
4017 cilk_elemental field.
4018 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
4019 * target.def: Adjust comment.
4020 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
4021 Don't test cilk_elemental.
4023 PR tree-optimization/83233
4024 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
4025 bswap_stat name for the struct.
4028 * tree.h (SWITCH_BREAK_LABEL_P): Define.
4029 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
4030 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
4031 SWITCH_BREAK_LABEL_P set on the label.
4032 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
4033 added for default case if it was missing and not all cases covered.
4034 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
4035 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
4037 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
4038 to avoid -Wimplicit-fallthrough warning.
4039 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
4040 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
4042 PR tree-optimization/83221
4043 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
4045 (init_reassoc): Formatting fix.
4048 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
4049 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
4053 * tree-cfg.c: Include asan.h.
4054 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
4056 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4058 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
4060 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4062 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
4063 vpcmpeqd instruction.
4065 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4067 * config/i386/i386.c (standard_sse_constant_opcode): Fix
4068 registers type for 128bit mode.
4070 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4072 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
4074 * gengtype-state.c (read_state_pair): Likewise.
4075 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
4076 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
4077 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
4078 * ifcvt.c (noce_try_store_flag_constants): Likewise.
4079 * tree-ssa-ccp.c (ccp_finalize): Likewise.
4080 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
4081 * builtins.c (fold_builtin_3): Likewise.
4082 * graphite-scop-detection.c
4083 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
4084 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
4087 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
4090 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
4091 parameter. Return true if ignore_decl_rtl_set_p is true and other
4092 conditions are satisfied.
4093 * asan.h (asan_protect_global): Add new parameter.
4094 * varasm.c (categorize_decl_for_section): Pass true as second parameter
4095 to asan_protect_global calls.
4097 2017-11-30 Jim Wilson <jimw@sifive.com>
4099 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
4100 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
4101 -mstrict-align, add info on default value. Delete redundant lines for
4102 -mabi. Add missing -mexplicit-relocs docs.
4104 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4106 * config/arc/arc.md (trap): New pattern.
4108 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4110 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
4111 instruction to end into a delay slot.
4112 * config/arc/arc.md (cond_delay_insn): Check if the instruction
4113 can be placed into a delay slot against reg_note.
4115 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4117 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
4118 labels number of usages.
4120 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4122 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
4124 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
4126 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4129 * internal-fn.c (expand_mul_overflow): Optimize unsigned
4130 multiplication by power of 2 constant into two shifts + comparison.
4132 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
4135 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
4137 2017-11-30 Richard Biener <rguenther@suse.de>
4139 PR tree-optimization/83202
4140 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
4141 allow_peel argument and guard peeling.
4142 (canonicalize_loop_induction_variables): Likewise.
4143 (canonicalize_induction_variables): Pass false.
4144 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
4145 peeling from cunrolli.
4147 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4149 * combine.c (try_combine): Print a message to dump file whenever
4150 I0, I1, or I2 cannot be combined into I3.
4152 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4154 PR rtl-optimization/83156
4155 PR rtl-optimization/82621
4156 * combine.c (try_combine): Don't split an I2 if one of the dests is
4157 set again before I3. Allow unused dests.
4159 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4161 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
4163 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
4165 PR rtl-optimization/80818
4166 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
4167 recursively. Use insn code for clobber.
4168 (lra_set_insn_recog_data): Pass the new arg to
4169 collect_non_operand_hard_regs.
4170 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
4172 (lra_update_insn_regno_info): Pass insn to
4173 add_regs_to_insn_regno_info.
4175 2017-11-29 Jim Wilson <jimw@sifive.com>
4176 Andrew Waterman <andrew@sifive.com>
4178 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
4179 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
4180 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
4181 use SINGLE_SHIFT_COST.
4182 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
4183 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
4185 2017-11-29 Julia Koval <julia.koval@intel.com>
4187 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
4188 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
4189 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
4190 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
4191 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
4192 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
4193 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
4194 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
4195 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
4196 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
4197 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
4198 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
4199 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
4200 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
4201 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
4202 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
4203 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
4204 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
4205 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
4206 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
4207 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
4208 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
4209 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
4210 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
4211 __builtin_ia32_vpshldv_v2di_mask,
4212 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
4213 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
4214 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
4216 2017-11-29 Julia Koval <julia.koval@intel.com>
4218 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
4219 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
4220 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
4221 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
4222 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
4223 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
4224 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
4225 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
4226 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
4227 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
4228 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
4229 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
4230 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
4231 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
4232 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
4233 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
4234 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
4235 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
4236 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4237 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
4238 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
4240 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4242 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
4243 movsi_pic_gotdata_op instruction as a load for the UT699 errata
4246 2017-11-29 Martin Aberg <maberg@gaisler.com>
4248 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4249 to prevent b2bst errata sequence.
4250 (sqrtdf2_fix): Likewise.
4252 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4254 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
4255 (div_sqrt_insn_p): New function.
4256 (sparc_do_work_around_errata): Insert NOP instructions to
4257 prevent sequences that could trigger the TN-0013 errata for
4258 certain LEON3 processors.
4259 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
4260 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
4261 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
4262 (in_branch_delay): Prevent div and sqrt in delay slot if
4264 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
4266 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4268 * config/sparc/sparc.c (atomic_insn_p): New function.
4269 (sparc_do_work_around_errata): Insert NOP instructions to
4270 prevent sequences that could trigger the TN-0010 errata for
4272 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
4273 instruction referable in atomic_insns_p.
4275 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4277 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
4278 (atomic_compare_and_swap_leon3_1): Likewise.
4281 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4283 * config/sparc/sparc.c (fpop_insn_p): New function.
4284 (sparc_do_work_around_errata): Insert NOP instructions to
4285 prevent sequences that could trigger the TN-0012 errata for
4287 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
4288 * config/sparc/sparc.md (fix_gr712rc): New attribute.
4289 (in_branch_annul_delay): Prevent floating-point instructions
4290 in delay slot of annulled integer branch.
4292 2017-11-29 Richard Biener <rguenther@suse.de>
4294 PR tree-optimization/83202
4295 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
4297 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
4298 nodes vectorized to the same stmts multiple times.
4299 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
4300 (vect_analyze_slp_instance): Adjust.
4301 (scalar_stmts_to_slp_tree_map_t): New typedef.
4302 (vect_schedule_slp_instance): Add a map recording the SLP node
4303 representing the vectorized stmts for a set of scalar stmts.
4304 Avoid code-generating redundancies.
4305 (vect_schedule_slp): Allocate map and pass it down.
4307 2017-11-29 Nathan Sidwell <nathan@acm.org>
4310 * tree.c (build_complex_type): Fix canonicalization. Only fill in
4313 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
4315 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
4316 cases for printing LDP/STP memory addresses.
4317 (aarch64_print_address_internal): Renamed from
4318 aarch64_print_operand_address, added parameter, add Pmode check.
4319 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
4320 (aarch64_print_operand_address): Indirect to
4321 aarch64_print_address_internal.
4322 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
4323 'y' operand output specifier.
4325 2017-11-29 Jakub Jelinek <jakub@redhat.com>
4328 * tree.c (build_simple_mem_ref_loc): Handle
4329 get_addr_base_and_unit_offset returning a MEM_REF.
4332 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
4335 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
4338 2017-11-28 David Malcolm <dmalcolm@redhat.com>
4340 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
4342 (layout::move_to_column): Likewise.
4344 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4347 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
4348 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
4350 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4351 Martin Jambor <mjambor@suse.cz>
4354 * tree.h (expr_type_first_operand_type_p): Declare
4355 * tree.c (expr_type_first_operand_type_p): New function.
4356 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
4357 (ipa_value_from_jfunc): Adjust declaration.
4358 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
4359 Use it as result type for arithmetics, unless it is NULL in which case
4360 be more conservative.
4361 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
4362 ipa_get_jf_pass_through_result.
4363 (propagate_vals_across_pass_through): Likewise.
4364 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
4365 is to propagate_vals_across_pass_through.
4366 (propagate_constants_across_call): Pass PARM_TYPE to
4367 propagate_scalar_across_jump_function.
4368 (find_more_scalar_values_for_callers_subset): Pass parameter type to
4369 ipa_value_from_jfunc.
4370 (cgraph_edge_brings_all_scalars_for_node): Likewise.
4371 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
4372 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
4373 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
4374 target_type, pass it to ipa_value_from_jfunc.
4375 (update_indirect_edges_after_inlining): Pass parameter type to
4376 try_make_edge_direct_simple_call.
4378 2017-11-28 Jeff Law <law@redhat.com>
4380 * gimple-ssa-evrp-analyze.c
4381 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
4382 refine ranges if scev_initialized_p returns true.
4383 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
4385 2017-11-28 Julia Koval <julia.koval@intel.com>
4387 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
4388 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
4389 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
4390 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
4391 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
4392 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
4393 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
4394 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
4395 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
4396 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
4397 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
4398 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
4399 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
4400 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
4401 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
4402 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
4403 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
4404 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
4405 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
4406 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
4407 __builtin_ia32_vpshrd_v2di_mask): New builtins.
4408 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
4410 2017-11-28 Julia Koval <julia.koval@intel.com>
4412 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
4413 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
4414 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
4415 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
4416 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
4417 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
4418 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
4419 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
4420 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
4421 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
4422 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
4423 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
4424 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
4425 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
4426 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
4427 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
4428 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
4429 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
4430 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
4431 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
4432 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
4433 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
4434 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
4435 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
4436 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
4437 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
4438 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
4439 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
4440 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
4441 __builtin_ia32_vpshld_v2di_mask): New builtins.
4442 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4443 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
4445 2017-11-28 Richard Biener <rguenther@suse.de>
4447 PR tree-optimization/80776
4448 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
4450 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
4452 (evrp_range_analyzer::record_ranges_from_incoming_edges):
4453 If the incoming edge is an effective fallthru because the other
4454 edge only reaches a __builtin_unreachable () then record ranges
4455 derived from the controlling condition in SSA info.
4456 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
4457 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4459 2017-11-28 Olivier Hainque <hainque@adacore.com>
4461 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
4462 on mingw build hosts.
4464 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4467 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
4469 2017-11-28 Richard Biener <rguenther@suse.de>
4472 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
4473 copies generated from memcpy use a character array as reference
4476 2017-11-28 Julia Koval <julia.koval@intel.com>
4477 Sebastian Peryt <sebastian.peryt@intel.com>
4479 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
4480 c-family/c-cilkplus.o, c-family/array-notation-common.o,
4481 cilk-common.o, cilk.h, cilk-common.c): Remove.
4483 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
4484 * builtins.c (is_builtin_name): Remove cilkplus condition.
4485 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
4486 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
4487 cilk-builtins.def, cilkplus.def): Remove.
4488 * cif-code.def (CILK_SPAWN): Remove.
4489 * cilk-builtins.def: Delete.
4490 * cilk-common.c: Ditto.
4492 * cilkplus.def: Ditto.
4493 * config/darwin.h (fcilkplus): Delete.
4494 * cppbuiltin.c: Ditto.
4495 * doc/extend.texi: Remove cilkplus doc.
4496 * doc/generic.texi: Ditto.
4497 * doc/invoke.texi: Ditto.
4498 * doc/passes.texi: Ditto.
4499 * gcc.c (fcilkplus): Remove.
4500 * gengtype.c (cilk.h): Remove.
4501 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
4503 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
4505 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
4507 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
4508 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
4509 cilkplus conditions.
4510 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
4511 inline_read_section): Ditto.
4512 * ipa-inline-analysis.c (cilk.h): Remove.
4513 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
4514 * lto-wrapper.c (merge_and_complain, append_compiler_options,
4515 append_linker_options): Remove condition for fcilkplus.
4516 * lto/lto-lang.c (cilk.h): Remove.
4517 (lto_init): Remove condition for fcilkplus.
4518 * omp-expand.c (expand_cilk_for_call): Delete.
4519 (expand_omp_taskreg, expand_omp_for_static_chunk,
4520 expand_omp_for): Remove cilkplus
4522 (expand_cilk_for): Delete.
4523 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
4524 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
4525 execute_lower_omp, diagnose_sb_0): Ditto.
4526 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
4527 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4528 * tree-nested.c: Ditto.
4529 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
4530 (dump_generic_node): Ditto.
4531 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4532 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
4534 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
4536 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4538 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
4539 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
4540 for vector mode and !TARGET_SIMD.
4542 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4544 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
4546 * tree.h (SWITCH_LABELS): Remove.
4547 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
4548 assert SWITCH_BODY is non-NULL.
4549 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
4551 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
4553 PR tree-optimization/80788
4554 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
4555 has TREE_OVERFLOW set, call drop_tree_overflow.
4557 2017-11-28 Richard Biener <rguenther@suse.de>
4559 PR tree-optimization/83158
4560 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
4562 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
4565 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
4566 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
4568 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
4570 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
4571 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
4573 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4576 * varasm.c (compare_constant): Take the mode of the constants into
4577 account when comparing floating point constants.
4579 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
4581 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
4582 from explicit instantiation of debug_helper.
4583 * vec.h (DEFINE_DEBUG_VEC): Ditto.
4585 2017-11-27 Richard Biener <rguenther@suse.de>
4587 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
4590 2017-11-27 Richard Biener <rguenther@suse.de>
4592 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
4593 (type_hash_canon): Also clear int_cst_hash_table entry for
4595 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
4597 2017-11-27 Tamar Christina <tamar.christina@arm.com>
4599 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4600 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4601 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4603 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4605 * hash-map.h (gt_cleare_cache): Avoid UB.
4607 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
4609 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
4610 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
4611 (decide_unroll_runtime_iterations): Likewise.
4612 (decide_unroll_stupid): Likewise.
4614 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4617 * config/i386/i386.md: Add a loop with incssp.
4619 2017-11-27 Martin Jambor <mjambor@suse.cz>
4621 PR tree-optimization/81248
4622 * tree-sra.c (splice_param_accesses): Remove size check.
4623 (decide_one_param_reduction): Fix size check.
4624 * gimple-pretty-print.c (dump_profile): Silence warning.
4625 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
4627 2017-11-27 Jakub Jelinek <jakub@redhat.com>
4630 * dbxout.c (lastlineno): New variable.
4631 (dbx_debug_hooks): Use dbxout_switch_text_section as
4632 switch_text_section debug hook.
4633 (dbxout_function_end): Switch to current_function_section
4634 rather than function_section. If crtl->has_bb_partition,
4635 output just one N_FUN, depending on in_cold_section_p.
4636 (dbxout_source_line): Remember last lineno in lastlineno.
4637 (dbxout_switch_text_section): New function.
4638 (dbxout_function_decl): Adjust dbxout_block caller.
4639 (dbx_block_with_cold_children): New function.
4640 (dbxout_block): Return true if any LBRAC/RBRAC have been
4641 emitted. Use dbx_block_with_cold_children at depth == 0
4642 in second partition. Add PARENT_BLOCKNUM argument, pass
4643 it optionally adjusted to children. Output LBRAC/RBRAC
4644 around recursive call only if the block is in the current
4645 partition, if not and anything was output, emit empty
4647 * final.c (final_scan_insn): Compute cold_function_name
4648 before calling switch_text_section debug hook. Call
4649 that hook even if dwarf2out_do_frame if not emitting
4653 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
4654 TREE_READONLY decls.
4656 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4658 PR rtl-optimization/82488
4659 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
4661 2017-11-26 Julia Koval <julia.koval@intel.com>
4663 * config/i386/i386.c (processor_target_table): Add skylake_cost for
4665 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
4668 2017-11-26 Julia Koval <julia.koval@intel.com>
4670 * config/i386/driver-i386.c (host_detect_local_cpu):
4671 Detect skylake-avx512.
4673 2017-11-26 Julia Koval <julia.koval@intel.com>
4675 * config.gcc: Add -march=cannonlake.
4676 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
4677 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
4678 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
4679 (PTA_CANNONLAKE): New.
4680 (processor_target_table): Add cannonlake.
4681 (ix86_option_override_internal): Ditto.
4682 (fold_builtin_cpu): Ditto.
4683 (get_builtin_code_for_version): Handle cannonlake.
4684 (M_INTEL_COREI7_CANNONLAKE): New.
4685 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
4686 * doc/invoke.texi: Add -march=cannonlake.
4688 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
4690 * plugin.c (add_new_plugin): Use platform-specific library extensions.
4691 (try_init_one_plugin): Alternative implementation for MinGW.
4692 * Makefile.in (plugin_implib): New.
4693 (gengtype-lex.c): Fix broken AIX workaround.
4694 * configure: Regenerate.
4695 * doc/plugins.texi: Document support for MinGW.
4697 2017-11-25 Jakub Jelinek <jakub@redhat.com>
4699 PR rtl-optimization/81553
4700 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
4701 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
4702 is a shift where C1 has different mode than the whole shift, use C1's
4703 mode for MULT rather than the shift's mode.
4706 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
4707 builtins not enabled in the currently selected ISA.
4709 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
4711 PR tree-optimization/71026
4712 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
4713 (insert_reciprocals): Change to insert reciprocals before a division
4714 by a square and to insert the square of a reciprocal.
4715 (execute_cse_reciprocals_1): Change to consider division by a square.
4716 (register_division_in): Add importance parameter.
4718 2017-11-24 Richard Biener <rguenther@suse.de>
4720 PR tree-optimization/82402
4721 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
4722 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
4724 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
4726 * match.pd (0-ptr): New transformation.
4728 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
4731 * ipa-inline.c (inline_small_functions): Set current badnes correctly
4732 when skipping checking.
4734 2017-11-24 Richard Biener <rguenther@suse.de>
4736 PR tree-optimization/83128
4737 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
4738 (vn_reference_lookup_3): Likewise.
4740 2017-11-24 Jakub Jelinek <jakub@redhat.com>
4743 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
4744 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
4747 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
4748 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
4749 check for the latter with gimple_call_builtin_p. Do not handle
4750 BUILT_IN_STPNCPY_CHK which is not a pass through call.
4752 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
4754 * config/arm/arm_neon.h: Fix pragma GCC push_options before
4757 2017-11-23 Julia Koval <julia.koval@intel.com>
4759 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
4760 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
4761 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
4762 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
4763 _mm512_maskz_expandloadu_epi16): New intrinsics.
4764 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
4765 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
4766 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
4767 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
4768 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
4769 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
4770 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
4771 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
4772 _mm256_maskz_expandloadu_epi8): New intrinsics.
4773 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
4774 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
4775 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
4776 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
4777 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
4778 * config/i386/sse.md (VI248_VLBW): New iterator.
4779 (expand<mode>_mask, expand<mode>_maskz): New patterns.
4781 2017-11-23 Julia Koval <julia.koval@intel.com>
4783 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
4784 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
4785 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
4786 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
4787 _mm512_mask_compressstoreu_epi16): New.
4788 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
4789 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
4790 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
4791 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
4792 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
4793 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
4794 _mm256_mask_compressstoreu_epi8): New.
4795 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
4796 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
4797 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
4798 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
4799 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
4800 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
4801 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
4802 __builtin_ia32_compresshi128_mask,
4803 __builtin_ia32_compressstoreuqi512_mask,
4804 __builtin_ia32_compressstoreuhi512_mask,
4805 __builtin_ia32_compressstoreuqi256_mask,
4806 __builtin_ia32_compressstoreuqi128_mask,
4807 __builtin_ia32_compressstoreuhi256_mask,
4808 __builtin_ia32_compressstoreuhi128_mask): New builtins.
4809 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
4811 (ix86_expand_special_args_builtin): Handle new types.
4812 (s4fma_expand): Handle new builtin array.
4813 * config/i386/immintrin.h: Include new headers.
4814 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
4815 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
4817 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4819 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
4820 fixes. Declare temp and g variables at the top in order to avoid
4821 {} in most of the cases.
4823 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
4825 * match.pd (ptr-0): New transformation.
4827 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4829 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
4830 (struct addr_mode_cost_table): New.
4831 (struct tune_params): Add field addr_mode_costs.
4832 * config/arm/arm.c (generic_addr_mode_costs): New.
4833 (arm_slowmul_tune): Initialise addr_mode_costs field.
4834 (arm_fastmul_tune): Likewise.
4835 (arm_strongarm_tune): Likewise.
4836 (arm_xscale_tune): Likewise.
4837 (arm_9e_tune): Likewise.
4838 (arm_marvell_pj4_tune): Likewise.
4839 (arm_v6t2_tune): Likewise.
4840 (arm_cortex_tune): Likewise.
4841 (arm_cortex_a8_tune): Likewise.
4842 (arm_cortex_a7_tune): Likewise.
4843 (arm_cortex_a15_tune): Likewise.
4844 (arm_cortex_a35_tune): Likewise.
4845 (arm_cortex_a53_tune): Likewise.
4846 (arm_cortex_a57_tune): Likewise.
4847 (arm_exynosm1_tune): Likewise.
4848 (arm_xgene1_tune): Likewise.
4849 (arm_cortex_a5_tune): Likewise.
4850 (arm_cortex_a9_tune): Likewise.
4851 (arm_cortex_a12_tune): Likewise.
4852 (arm_cortex_a73_tune): Likewise.
4853 (arm_v7m_tune): Likewise.
4854 (arm_cortex_m7_tune): Likewise.
4855 (arm_v6m_tune): Likewise.
4856 (arm_fa726te_tune): Likewise.
4857 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
4859 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4861 * config/arm/arm.c (arm_mem_costs): New function.
4862 (arm_rtx_costs_internal): Use arm_mem_costs.
4864 2017-11-23 Mark Wielaard <mark@klomp.org>
4866 * dwarf2out.c (init_sections_and_labels): Use generation to create
4867 unique ranges_section_label and ranges_base_label. Return generation.
4868 (output_rnglists): Add generation argument. Use generation to create
4869 unique ranges labels.
4870 (dwarf2out_finish): Get generation from init_sections_and_labels
4871 and pass generation to output_rnglists.
4873 2017-11-23 Mike Stump <mikestump@comcast.net>
4874 Eric Botcazou <ebotcazou@adacore.com>
4876 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
4877 * cfgloop.h (struct loop): Add unroll field.
4878 * function.h (struct function): Add has_unroll bitfield.
4879 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
4880 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
4881 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
4882 (pass_rtl_unroll_loops::gate): Likewise.
4883 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
4884 for which loop->unroll==1.
4885 (decide_unroll_constant_iterations): Use note for consistency and
4886 take loop->unroll into account. Return early if loop->unroll is set.
4887 Fix thinko in existing test.
4888 (decide_unroll_runtime_iterations): Use note for consistency and
4889 take loop->unroll into account.
4890 (decide_unroll_stupid): Likewise.
4891 * lto-streamer-in.c (input_cfg): Read loop->unroll.
4892 * lto-streamer-out.c (output_cfg): Write loop->unroll.
4893 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
4895 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
4896 (print_loop): Print loop->unroll if set.
4897 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
4898 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
4899 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
4901 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
4902 loop->unroll is set and smaller than the trip count. Otherwise bypass
4903 entirely the heuristics if loop->unroll is set. Remove dead note.
4904 Fix off-by-one bug in other note.
4905 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
4906 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
4908 (tree_unroll_loops_completely): Make static.
4909 (pass_complete_unroll::execute): Use correct type for variable.
4910 (pass_complete_unrolli::execute): Fix formatting.
4911 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
4913 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
4915 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
4916 enable when TARGET_PREFER_AVX128 is set.
4918 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4920 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
4922 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
4923 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
4925 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4927 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
4928 frequencies. Fix estimation of aggregate parameters.
4930 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4932 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
4935 2017-11-23 Tom de Vries <tom@codesourcery.com>
4937 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
4939 2017-11-23 Tom de Vries <tom@codesourcery.com>
4941 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
4944 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
4947 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
4948 sibcall_value_pcrel_fdpic): Use local variable instead of
4950 (calli_tbr_rel): Add missing operand 2.
4951 (call_valuei_tbr_rel): Add missing operand 3.
4953 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4956 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
4957 bitpos/bitsize covering the whole destination, use store_expr only if
4958 the complex mode is the same. Otherwise, use expand_normal and if
4959 it returns CONCAT, subreg each part separately instead of trying to
4960 subreg the whole result.
4962 2017-11-23 Richard Biener <rguenther@suse.de>
4964 PR tree-optimization/23094
4965 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
4966 come here from walking over backedges in the first iteration.
4967 (vn_reference_lookup_3): Skip clobbers that store the same value.
4969 2017-11-23 Richard Biener <rguenther@suse.de>
4971 PR tree-optimization/81403
4972 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
4973 a block we need a leader relative to.
4974 (phi_translate_1): For nary processing require a leader from
4975 get_representative_for given we run expression simplification
4976 using match-and-simplify. Remove previous fix.
4978 2017-11-22 Jeff Law <law@redhat.com>
4980 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
4981 Use new method allocate_value_range rather than accessing the
4982 vrp_value_range_pool data member directly.
4983 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
4984 to use extract_range_from_stmt method to avoid need for
4985 extract_range_from_assignment method.
4986 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
4987 method rather than setting values_propgated data member directly.
4988 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
4989 and values propagated data members and extract_range_from_assignment
4990 method. Reorder private data members to conform to standards.
4991 Add new methods set_lattice_propagation_complete and
4992 allocate_value_range.
4994 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
4996 PR rtl-optimization/83030
4997 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
4998 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
4999 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
5000 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
5001 CROSSING_JUMP_P flag is already set before setting it.
5002 * cfgrtl.c (fixup_partition_crossing): Likewise.
5003 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
5006 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5008 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
5009 Handle the case where both arguments are using gen_const_vec_series.
5011 2017-11-22 David Malcolm <dmalcolm@redhat.com>
5014 * pretty-print.c (pp_format): Move quoting implementation to
5015 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
5016 to pass address of "quote" local.
5017 (pp_begin_quote): New function.
5018 (pp_end_quote): New function.
5019 * pretty-print.h (printer_fn): Convert penultimate param from bool
5021 (pp_begin_quote): New decl.
5022 (pp_end_quote): New decl.
5023 * tree-diagnostic.c (default_tree_printer): Convert penultimate
5024 param from bool to bool *.
5025 * tree-diagnostic.h (default_tree_printer): Likewise.
5027 2017-11-22 Jeff Law <law@redhat.com>
5029 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
5030 Initialize vr_values.
5031 (evrp_range_analyzer::try_find_new_range): Call methods attached to
5032 vr_values via vr_values class instance rather than delegators.
5033 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
5034 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5035 (evrp_range_analyzer::push_value_range): Likewise.
5036 (evrp_range_analyzer::pop_value_range): Likewise.
5037 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
5038 most delegators. Those remaining are exposed as public interfaces.
5039 Make vr_values a pointer and private.
5040 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
5042 (evrp_range_analyzer::get_vr_value): New method.
5043 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
5044 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
5045 (class evrp_dom_walker): Attach evrp_folder class, initialize
5046 it in the ctor. Remove temporary delegators.
5047 (evrp_dom_walker::before_dom_children): Call methods in attached
5048 evrp_range_analyzer class via class instance pointer. Use
5049 free value_range_constant_singleton to remove need for
5050 op_with_constant_singleton_value delegator method. Do not
5051 create a vrp_prop class instance for every call! Narrow
5052 scope of a couple variables.
5053 (evrp_dom_walker::cleanup): Call methods in attached
5054 evrp_range_analyzer class via class instance pointer.
5055 * vr-values.h (class vr_values): Privatize many methods and
5058 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5060 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
5061 * cfgexpand.c (expand_debug_expr): Remove handling for them.
5062 * expr.c (expand_expr_real_2): Likewise.
5063 * fold-const.c (const_unop): Likewise.
5064 * optabs-tree.c (optab_for_tree_code): Likewise.
5065 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
5066 * tree-inline.c (estimate_operator_cost): Likewise.
5067 * tree-pretty-print.c (dump_generic_node): Likewise.
5068 (op_code_prio): Likewise.
5069 (op_symbol_code): Likewise.
5070 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
5071 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
5072 * internal-fn.c (direct_internal_fn_optab): New function.
5073 (direct_internal_fn_array, direct_internal_fn_supported_p
5074 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
5075 * fold-const-call.c (fold_const_reduction): New function.
5076 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
5078 * tree-vect-loop.c: Include internal-fn.h.
5079 (reduction_code_for_scalar_code): Rename to...
5080 (reduction_fn_for_scalar_code): ...this and return an internal
5082 (vect_model_reduction_cost): Take an internal_fn rather than
5084 (vect_create_epilog_for_reduction): Likewise. Build calls rather
5086 (vectorizable_reduction): Use internal functions rather than tree
5087 codes for the reduction operation. Update calls to the functions
5089 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
5090 Use calls to internal functions rather than REDUC tree codes.
5091 * config/aarch64/aarch64-simd.md: Update comment accordingly.
5093 2017-11-22 Olivier Hainque <hainque@adacore.com>
5095 * config/vxworks.c (vxworks_override_options): Pick default
5096 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
5097 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
5098 DWARF_GNAT_ENCODINGS_DEFAULT.
5099 * config/vxworksae.h: Likewise.
5101 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
5103 PR tree-optimization/83104
5104 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
5107 2017-11-22 Marek Polacek <polacek@redhat.com>
5108 H.J. Lu <hongjiu.lu@intel.com>
5109 Jason Merrill <jason@redhat.com>
5114 * calls.c (initialize_argument_information): Call
5115 warn_parameter_passing_abi target hook.
5116 (store_one_arg): Use 0 for empty record size. Don't push 0 size
5117 argument onto stack.
5118 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
5119 * common.opt: Update -fabi-version description.
5120 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
5121 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
5123 (ix86_is_empty_record): New function.
5124 (ix86_warn_parameter_passing_abi): New function.
5125 (TARGET_EMPTY_RECORD_P): Redefine.
5126 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
5127 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
5128 * doc/tm.texi: Regenerated.
5129 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
5130 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
5131 * dwarf2out.c (get_ultimate_context): Move to tree.c.
5132 * explow.c (hard_function_value): Call arg_int_size_in_bytes
5133 instead of int_size_in_bytes.
5134 * expr.c (copy_blkmode_to_reg): Likewise.
5135 * function.c (aggregate_value_p): Return 0 for empty types.
5136 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
5138 (locate_and_pad_parm): Call arg size_in_bytes instead
5140 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
5141 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
5142 * target.def (empty_record_p, warn_parameter_passing_abi): New target
5144 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
5145 (std_gimplify_va_arg_expr): Skip empty records. Call
5146 arg_size_in_bytes instead size_in_bytes.
5147 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
5148 * tree-core.h (tree_type_common): Add empty_flag.
5149 (tree_decl_common): Update comments.
5150 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
5152 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5153 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
5155 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5156 * tree.c (default_is_empty_type): New function.
5157 (default_is_empty_record): New function.
5158 (arg_int_size_in_bytes): New function.
5159 (arg_size_in_bytes): New function.
5160 (get_ultimate_context): New function.
5161 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
5162 TRANSLATION_UNIT_WARN_EMPTY_P.
5163 (default_is_empty_record, arg_int_size_in_bytes,
5164 arg_size_in_bytes, get_ultimate_context): Declare.
5166 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5168 * config/arm/arm.c (cmse_clear_registers): New function.
5169 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
5170 code by call to cmse_clear_registers.
5171 (cmse_nonsecure_entry_clear_before_return): Likewise.
5173 2017-11-22 Tamar Christina <tamar.christina@arm.com>
5175 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
5176 (vdot_s32, vdotq_s32): New.
5177 (vdot_lane_u32, vdotq_lane_u32): New.
5178 (vdot_lane_s32, vdotq_lane_s32): New.
5181 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5184 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
5185 for unsigned values with fewer HWIs than the precision.
5186 (test_overflow): New function.
5187 (wide_int_cc_tests): Call it.
5189 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5191 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
5192 has the same mode class as Pmode.
5194 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5196 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
5197 padding_bits_to_clear_ptr.
5198 (cmse_nonsecure_entry_clear_before_return): Likewise.
5200 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5202 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
5203 auto_sbitap instead of integer bitfield to control register needing
5206 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5208 PR tree-optimization/83044
5209 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
5210 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
5211 that need the upper bound. Subtract offset from
5212 get_addr_base_and_unit_offset only if positive and subtract it
5213 before division by eltsize rather than after it.
5216 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
5217 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
5218 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
5222 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
5223 expand_binop with *mul_widen_optab, make sure at least one of the
5224 operands doesn't have VOIDmode.
5227 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
5229 PR rtl-optimization/82044
5230 PR tree-optimization/82042
5231 * dse.c (record_store): Check for overflow.
5232 (check_mem_read_rtx): Properly check for overflow if width == -1, call
5233 add_wild_read instead of clear_rhs_from_active_local_stores on
5234 overflow and log it into dump_file.
5236 2017-11-22 Richard Biener <rguenther@suse.de>
5238 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
5239 fake edges to exit when looking for a place to insert.
5240 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
5242 (insert_into_preds_of_block): Commit edge insertion immediately,
5243 assert that doesn't require new BBs.
5244 (fini_pre): Release expressions.
5245 (pass_pre::execute): Shuffle things around a bit, if the fn
5246 is too large do not compute AVAIL either as this is really the
5249 2017-11-22 Richard Biener <rguenther@suse.de>
5251 PR tree-optimization/83089
5252 * tree-if-conv.c (pass_if_conversion::execute): If anything
5253 changed reset SCEV and free the number of iteration estimates.
5255 2017-11-21 Martin Sebor <msebor@redhat.com>
5257 PR tree-optimization/82945
5258 * calls.h (warn_nonstring_bound): Remove unused function.
5260 2017-11-21 Martin Sebor <msebor@redhat.com>
5262 PR tree-optimization/82945
5263 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
5264 * calls.h (maybe_warn_nonstring_arg): Declare new function.
5265 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
5267 (initialize_argument_information): Call maybe_warn_nonstring_arg.
5268 * calls.h (get_attr_nonstring_decl): Declare new function.
5269 * doc/extend.texi (attribute nonstring): Update.
5270 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
5271 get_attr_nonstring_decl and handle it.
5272 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
5273 detection of nul-termination.
5274 (strlen_to_stridx): Change to a pointer.
5275 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
5276 (pass_strlen::execute): Same.
5278 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
5280 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
5281 for the new option -mprefer-vector-width=[none|128|256|512].
5282 * config/i386/i386.c (ix86_target_string): remove old style options
5283 -mprefer-avx256 and make -mprefer-avx128 as alias.
5284 (ix86_option_override_internal): Apply defaults for the
5285 -mprefer-vector-width=[128|256] option.
5286 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
5287 Implement macros to work with -mprefer-vector-width=.
5288 * config/i386/i386.opt: Implemented option
5289 -mprefer-vector-width=[none|128|256|512].
5290 * doc/invoke.texi: Documentation for
5291 -mprefer-vector-width=[none|128|256|512].
5293 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
5295 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
5297 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
5299 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
5301 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
5303 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
5305 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
5306 * expr.c (expand_expr_real_2): Likewise.
5307 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
5308 fold_binary_loc): Likewise.
5309 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
5310 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
5311 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
5312 MINUS_EXPR transformations.
5313 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
5314 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
5315 * tree-inline.c (estimate_operator_cost): Likewise.
5316 * tree-pretty-print.c (dump_generic_node, op_code_prio,
5317 op_symbol_code): Likewise.
5318 * tree-vect-stmts.c (vectorizable_operation): Likewise.
5319 * vr-values.c (extract_range_from_binary_expr): Likewise.
5320 * varasm.c (initializer_constant_valid_p_1): Likewise.
5321 * tree.def: New tree code POINTER_DIFF_EXPR.
5323 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
5325 * config/i386/i386.md (*bswap<mode>2_movbe): Add
5326 integer suffix to movbe mnemonic.
5327 (*bswaphi2_movbe): Ditto.
5328 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
5330 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5333 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
5334 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
5335 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
5337 2017-11-21 Martin Liska <mliska@suse.cz>
5339 * tree-inline.c (expand_call_inline): Remove not needed
5342 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
5343 Jakub Jelinek <jakub@redhat.com>
5346 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
5347 Remove static keyword from f variable.
5349 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5351 PR tree-optimization/83086
5352 * gimple-ssa-store-merging.c
5353 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
5354 rather than n.base_addr.
5356 2017-11-21 Martin Liska <mliska@suse.cz>
5358 PR rtl-optimization/82044
5359 PR tree-optimization/82042
5360 * dse.c (check_mem_read_rtx): Check for overflow.
5362 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5364 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
5367 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5369 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
5370 and its corresponding call arg location note.
5372 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5373 Andrew Burgess <andrew.burgess@embecosm.com>
5375 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
5377 (arc_return_slot_offset): Likewise.
5378 (arc_eh_return_address_location): New declaration.
5379 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
5380 (MUST_SAVE_REGISTER): Add exception handler case.
5381 (MUST_SAVE_RETURN_ADDR): Likewise.
5382 (arc_frame_pointer_required): Likewise.
5383 (arc_frame_pointer_needed): New function.
5384 (arc_compute_frame_size): Changed.
5385 (arc_expand_prologue): Likewise.
5386 (arc_expand_epilogue): Likewise.
5387 (arc_initial_elimination_offset): Likewise.
5388 (arc_return_slot_offset): Delete.
5389 (arc_eh_return_address_location): New function.
5390 (arc_builtin_setjmp_frame_value): Likewise.
5391 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
5392 (EH_RETURN_STACKADJ_RTX): Define.
5393 (EH_RETURN_HANDLER_RTX): Likewise.
5394 * config/arc/arc.md (eh_return): Delete.
5396 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5398 * print-tree.h (debug_vec_tree): Remove prototype.
5399 * gdbinit.in (pvt): Remove macro.
5401 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5403 PR tree-optimization/83047
5404 * gimple-ssa-store-merging.c
5405 (imm_store_chain_info::output_merged_store): If the loads with the
5406 same vuse are in different basic blocks, for load_gsi pick a load
5407 location that is dominated by the other loads.
5410 * config/i386/i386.c (ix86_memmodel_check): Start
5411 -Winvalid-memory-model diagnostics with lowercase letter.
5414 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
5415 set in_cold_section_p to the partition containing loc_list->first.
5416 When seeing loc_list->last_before_switch node, update secname and
5417 perform range_across_switch second partition handling only after that.
5420 * run-rtl-passes.c: Include debug.h.
5421 (run_rtl_passes): Call debug_hooks->assembly_start.
5422 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
5426 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
5427 OPTAB_DIRECT in calls to expand_simple_binop.
5429 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5432 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
5434 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5437 * spellcheck.h (best_match::blithely_get_best_candidate): New
5440 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5442 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
5443 with lower case letter.
5445 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
5447 * config/i386/i386.md (bswaphi2): New expander.
5448 (*bswaphi2_movbe): New insn pattern.
5449 (bswaphi -> rorhi pepehole2): New peephole pattern.
5451 2017-11-20 Jeff Law <law@redhat.com>
5453 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
5454 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5455 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
5456 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
5457 gimple-ssa-evrp-analyze.h.
5459 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
5460 set BB_VISITED here.
5461 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
5463 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5465 * tree-predcom.c: Add general comment on Store-Store chains.
5466 (split_data_refs_to_components): Postpone clearing eliminate_store_p
5468 (get_chain_last_ref_at): Rename into...
5469 (get_chain_last_write_at): ...this.
5470 (get_chain_last_write_before_load): New function.
5471 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
5472 CT_STORE_STORE when write reference is added.
5473 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
5474 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
5475 (initialize_root_vars_store_elim_1): Ditto.
5476 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
5477 definition is created.
5478 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
5479 chain by replacing it with dominant stored value.
5481 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5483 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
5485 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
5487 * vr-values.c (extract_range_from_binary_expr): Use a full range
5490 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
5492 * config/arm/arm.md (R4_REGNUM): Define constant.
5493 (nonsecure_call_internal): Remove r4 clobber.
5494 (nonsecure_call_value_internal): Likewise.
5495 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
5496 clobber and resequence match_operands.
5497 (nonsecure_call_value_reg_thumb1_v5): Likewise.
5498 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
5499 (nonsecure_call_value_reg_thumb2): Likewise.
5501 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5503 PR tree-optimization/78821
5504 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
5505 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
5506 to the address of the base rather than the base itself.
5507 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
5508 (find_bswap_or_nop_finalize): New function.
5509 (find_bswap_or_nop): Use it.
5510 (bswap_replace): Return a tree rather than bool, change first
5511 argument from gimple * to gimple_stmt_iterator, allow inserting
5512 into an empty sequence, allow ins_stmt to be NULL - then emit
5513 all stmts into gsi. Fix up MEM_REF address gimplification.
5514 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
5515 (struct store_immediate_info): Add N and INS_STMT non-static
5517 (store_immediate_info::store_immediate_info): Initialize them
5518 from newly added ctor args.
5519 (merged_store_group::apply_stores): Formatting fixes. Sort by
5521 (stmts_may_clobber_ref_p): For stores call also
5522 refs_anti_dependent_p.
5523 (gather_bswap_load_refs): New function.
5524 (imm_store_chain_info::try_coalesce_bswap): New method.
5525 (imm_store_chain_info::coalesce_immediate_stores): Use it.
5526 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
5527 (imm_store_chain_info::output_merged_store): Fail if number of
5528 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
5529 and NOP_EXPR rhs_code.
5530 (pass_store_merging::process_store): Compute n and ins_stmt, if
5531 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
5532 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
5534 (pass_store_merging::execute): Calculate dominators.
5536 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
5537 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
5538 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
5539 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5540 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
5541 class pass_optimize_bswap, bswap_replace,
5542 pass_optimize_bswap::execute): Moved to ...
5543 * gimple-ssa-store-merging.c: ... this file.
5544 Include optabs-tree.h.
5545 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
5546 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5547 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
5548 anonymous namespace, remove static keywords.
5549 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
5550 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
5552 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
5555 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
5557 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
5559 * vec.h (debug_helper): New function.
5560 (DEFINE_DEBUG_VEC): New macro.
5561 * hash-set.h (debug_helper): New function.
5562 (DEFINE_DEBUG_HASH_SET): New macro.
5563 * cfg.c (debug_slim (edge)): New function.
5564 Call DEFINE_DEBUG_VEC for edges.
5565 Call DEFINE_DEBUG_HASH_SET for edges.
5566 * cfghooks.c (debug_slim (basic_block)): New function.
5567 Call DEFINE_DEBUG_VEC for basic blocks.
5568 Call DEFINE_DEBUG_HASH_SET for basic blocks.
5569 * print-tree.c (debug_slim): New function to handle trees.
5570 Call DEFINE_DEBUG_VEC for trees.
5571 Call DEFINE_DEBUG_HASH_SET for trees.
5572 (debug (vec<tree, va_gc>) &): Remove.
5573 (debug (<vec<tree, va_gc>) *): Remove.
5574 * print-rtl.c (debug_slim): New function to handle const_rtx.
5575 Call DEFINE_DEBUG_VEC for rtx_def.
5576 Call DEFINE_DEBUG_VEC for rtx_insn.
5577 Call DEFINE_DEBUG_HASH_SET for rtx_def.
5578 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
5579 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
5580 (debug (vec<rtx_insn *> *ptr): Remove.
5581 (debug_insn_vector): Remove.
5582 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
5584 2017-11-20 Tom de Vries <tom@codesourcery.com>
5586 PR rtl-optimization/82020
5587 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
5588 IF_THEN_ELSE condition.
5590 2017-11-19 Jeff Law <law@redhat.com>
5592 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
5593 of degenerates resulting from ignoring an edge.
5595 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5598 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
5600 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5603 * profile-count.c (profile_count::to_sreal_scale): Fix return value
5604 for uninitialied counts.
5606 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5609 * tree-inline.c (estimate_num_insns): Set to 1 at least.
5611 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5614 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
5616 2017-11-19 Tom de Vries <tom@codesourcery.com>
5618 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
5621 2017-11-19 Tom de Vries <tom@codesourcery.com>
5623 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
5626 2017-11-19 Tom de Vries <tom@codesourcery.com>
5628 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5631 2017-11-19 Tom de Vries <tom@codesourcery.com>
5633 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
5635 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
5637 2017-11-19 Tom de Vries <tom@codesourcery.com>
5639 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
5641 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
5642 semicolon after MCORE_EXPORT_NAME call.
5644 2017-11-19 Tom de Vries <tom@codesourcery.com>
5647 * vmsdbgout.c (vmsdbgout_early_finish): New function.
5648 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
5650 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
5652 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
5653 (cgraph_edge::clone): Cleanup updating of profile.
5654 * ipa-cp.c (update_profiling_info): Likewise.
5655 * ipa-inline-transform.c (inline_transform): Likewise.
5656 * ipa-inline.c (inline_small_functions): Add missing space to dump.
5657 * ipa-split.c (execute_split_functions): Do not split when function
5659 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
5660 * profile-count.c (profile_count::dump): Add global0.
5661 (profile_count::to_cgraph_frequency): Do not ICE when entry is
5663 (profile_count::to_sreal_scale): Likewise.
5664 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
5665 (profile_count::combine_with_ipa_count): New function.
5666 * profile-count.h (profile_guessed_global0adjusted): New.
5667 (profile_count::adjusted_zero): New.
5668 (profile_count::global0adjusted): New.
5669 (profile_count::combine_with_ipa_count): New.
5670 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
5671 correct profile of return block of split functions.
5672 (copy_cfg_body): Remove unused profile_count.
5673 (copy_body): Likewise.
5674 (expand_call_inline): Update.
5675 (tree_function_versioning): Update.
5677 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
5679 * hash-set.h (hash_set::empty): New.
5680 * tree-ssa-threadbackward.h: Delete.
5681 * tree-ssa-threadbackward.c (class thread_jumps): New.
5682 Move max_threaded_paths into class.
5683 (fsm_find_thread_path): Remove arguments that are now in class.
5684 (profitable_jump_thread_path): Rename to...
5685 (thread_jumps::profitable_jump_thread_path): ...this.
5686 (convert_and_register_jump_thread_path): Rename to...
5687 (thread_jumps::convert_and_register_current_path): ...this.
5688 (check_subpath_and_update_thread_path): Rename to...
5689 (thread_jumps::check_subpath_and_update_thread_path): ...this.
5690 (register_jump_thread_path_if_profitable): Rename to...
5691 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
5692 (handle_phi): Rename to...
5693 (thread_jumps::handle_phi): ...this.
5694 (handle_assignment): Rename to...
5695 (thread_jumps::handle_assignment): ...this.
5696 (fsm_find_control_statement_thread_paths): Rename to...
5697 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
5698 (find_jump_threads_backwards): Rename to...
5699 (thread_jumps::find_jump_threads_backwards): ...this.
5700 Initialize path local data.
5701 (pass_thread_jumps::execute): Call find_jump_threads_backwards
5702 from within thread_jumps class.
5703 (pass_early_thread_jumps::execute): Same.
5705 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5707 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
5709 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5711 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
5712 to canonical location.
5714 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5717 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
5719 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
5721 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
5723 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
5724 rcrt1.o%s/grcrt1.o%s for -static-pie.
5726 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5728 * i386.c (ix86_multiplication_cost, ix86_division_cost,
5729 ix86_shift_rotate_cost): Break out from ...
5730 (ix86_rtx_costs): ... here.
5731 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
5734 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5736 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
5737 when entry block was promoted unlikely.
5738 (estimate_bb_frequencies): Increase frequency scale.
5739 * profile-count.h (profile_count): Export precision info.
5741 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5743 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
5744 disturbin profile of entry block.
5746 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5748 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
5751 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5753 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
5754 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
5756 2017-11-17 Jeff Law <law@redhat.com>
5758 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
5759 from evrp_dom_walker class. Various methods moved into new class.
5760 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
5761 (evrp_range_analyzer::enter): New method.
5762 (evrp_range_analyzer::leave): New method.
5763 (evrp_dom_walker): Remove delegators no longer needed by this class.
5764 Replace vr_values data member with evrp_range_analyzer
5766 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
5767 method extracted from evrp_dom_walker::before_dom_children.
5768 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
5769 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
5771 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
5772 Add private copy constructor and move assignment operators.
5773 Privatize methods and class data where trivially possible.
5774 (evrp_dom_walker::cleanup): New function, extracted from
5775 execute_early_vrp. Simplify access to class data.
5777 * vr-values.h (get_output_for_vrp): Prototype.
5778 * vr-values.c (get_output_for_vrp): New function extracted from
5779 vrp_visit_assignment_or_call and extract_range_from_stmt.
5780 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
5782 2017-11-17 Luis Machado <luis.machado@linaro.org>
5784 * config/aarch64/aarch64.c
5785 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
5786 (qdf24xx_tunings) <autoprefetcher_model>: Set to
5787 tune_params::AUTOPREFETCHER_WEAK.
5789 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5792 * config/arm/arm.c (arm_valid_target_attribute_rec):
5793 Parse "arch=" and "+<ext>".
5794 (arm_valid_target_attribute_tree): Re-init global options.
5795 (arm_option_override): Make non-static.
5796 (arm_options_perform_arch_sanity_checks): Make errors fatal.
5797 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
5798 (__ARM_FEATURE_CRC32): Support undef.
5799 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
5800 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
5802 2017-11-17 David Malcolm <dmalcolm@redhat.com>
5804 * gdbinit.in (break-on-diagnostic): New command.
5806 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5808 * config/i386/i386.c (ix86_expand_epilogue): Change simple
5809 return to indirect jump for EH return if control-flow
5810 protection is enabled. Change explicit 'false' argument in
5811 pro_epilogue_adjust_stack with a value of
5813 * config/i386/i386.md (simple_return_indirect_internal):
5814 Remove SImode restriction to support 64-bit.
5816 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5818 * combine.c (added_notes_insn): New.
5819 (try_combine): Handle added_notes_insn like added_links_insn.
5820 Rewrite return value code.
5821 (distribute_notes): Set added_notes_insn to the earliest insn we added
5824 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5826 PR rtl-optimization/82621
5827 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
5828 dest of one of those SETs is unused.
5830 2017-11-17 Richard Biener <rguenther@suse.de>
5833 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
5834 * tree-pretty-print.c (dump_generic_node): Handle
5835 annot_expr_parallel_kind.
5836 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
5837 * gimplify.c (gimple_boolify): Likewise.
5839 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5841 * config.gcc (extra_headers): Add cet.h for x86 targets.
5842 * config/i386/cet.h: New file.
5843 * doc/install.texi: Add --enable-cet/--disable-cet.
5845 2017-11-17 Richard Biener <rguenther@suse.de>
5847 PR tree-optimization/83017
5848 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
5849 (gen_parallel_loop): Properly count iterations.
5850 (parallelize_loops): Handle loop->can_be_parallel independent
5851 of flag_loop_parallelize_all. Make static profitability test match
5853 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
5854 * invoke.texi (parloops-min-per-thread): Document.
5856 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
5858 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
5859 upstreaming review comments.
5861 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5863 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
5864 with fast unaligned access.
5865 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
5867 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
5869 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
5870 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
5872 2017-11-17 Richard Biener <rguenther@suse.de>
5874 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
5875 folding of references.
5877 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
5880 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
5881 of replacing call to strncmp with corresponding call to strcmp when
5884 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
5886 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
5887 option prefer-avx256 for skylake-avx512 configuration.
5888 * config/i386/i386.c (ix86_option_override_internal): Ditto.
5889 (get_builtin_code_for_version): Ditto.
5891 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5892 Monk Chiang <sh.chiang04@gmail.com>
5894 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
5895 (FIXED_REGISTERS): Reserve more register numbers.
5896 (CALL_USED_REGISTERS): Likewise.
5897 (REG_ALLOC_ORDER): Likewise.
5898 (REG_CLASS_CONTENTS): Likewise.
5899 (REGISTER_NAMES): Likewise.
5901 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5902 Kito Cheng <kito.cheng@gmail.com>
5904 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
5906 * config/nds32/iterators.md: Add vector mode iterators and attributes.
5908 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
5910 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
5911 parameter list for vec_splats.
5913 2017-11-16 Joseph Myers <joseph@codesourcery.com>
5915 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
5917 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
5919 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
5922 * doc/install.texi: Document incompatibility of Perl >=5.6.26
5923 with the required version of automake 1.11.6.
5925 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
5927 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
5929 (DU_C2_3_power9): Correct reservation combinations.
5930 (FP_DIV_power9, VEC_DIV_power9): New.
5931 (power9-alu): Split out rotate/shift...
5932 (power9-rot): ...to here, correct dispatch resource.
5933 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
5935 (power9-fp): Correct latency.
5936 (power9-sdiv): Add div/sqrt resource.
5937 (power9-ddiv): Correct latency, add div/sqrt resource.
5938 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
5939 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
5941 (power9-qpdiv, power9-qpmul): Adjust resource usage.
5943 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
5945 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
5946 switch statement mapping KF built-ins to TF built-ins if we don't
5947 have the proper ISA 3.0 assembler support.
5949 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5951 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
5952 (lower_emutls_function_body): Do not compute it.
5954 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5956 * ipa-split.c (split_bb_info): Turn time to sreal.
5957 (split_point): Likewise.
5958 (dump_split_point): Likewise.
5959 (fine_split_points): Likewise.
5960 (execute_split_functions): Only zero split_bbs; turn time to sreals.
5962 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5964 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
5967 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5969 * predict.c (combine_predictions_for_bb): Preserve zero predicted
5971 (expensive_function_p): Remove useless assert.
5972 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
5974 2017-11-16 Martin Sebor <msebor@redhat.com>
5976 PR tree-optimization/82588
5977 PR tree-optimization/82583
5978 * tree-vrp.c (check_array_ref): Handle flexible array members,
5979 string literals, and inner indices.
5980 (search_for_addr_array): Add detail to diagnostics.
5982 2017-11-16 Nathan Sidwell <nathan@acm.org>
5986 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
5987 (SET_DECL_ASSEMBLER_NAME): Forward to
5988 overwrite_decl_assembler_name.
5989 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
5990 (overwrite_decl_assembler_name): Declare.
5991 * tree.c (overwrite_decl_assembler_name): New.
5992 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
5993 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
5994 (LANG_HOOKS_INITIALIZER): Add it.
5995 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
5996 * langhooks.c (lhd_set_decl_assembler_name): Use
5997 SET_DECL_ASSEMBLER_NAME.
5998 (lhd_overwrite_decl_assembler_name): Default implementation.
6000 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6001 Jackson Woodruff <jackson.woodruff@arm.com>
6003 PR tree-optimization/71026
6004 * match.pd: Canonicalize constant multiplies in division.
6006 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6008 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
6009 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
6010 Remove OPT_fomit_frame_pointer entry.
6011 * common/config/arc/arc-common.c: Likewise.
6012 * common/config/arm/arm-common.c: Likewise.
6013 * common/config/avr/avr-common.c: Likewise.
6014 * common/config/c6x/c6x-common.c: Likewise.
6015 * common/config/cr16/cr16-common.c: Likewise.
6016 * common/config/cris/cris-common.c: Likewise.
6017 * common/config/epiphany/epiphany-common.c: Likewise.
6018 * common/config/fr30/fr30-common.c: Likewise.
6019 * common/config/frv/frv-common.c: Likewise.
6020 * common/config/ia64/ia64-common.c: Likewise.
6021 * common/config/iq2000/iq2000-common.c: Likewise.
6022 * common/config/lm32/lm32-common.c: Likewise.
6023 * common/config/m32r/m32r-common.c: Likewise.
6024 * common/config/mcore/mcore-common.c: Likewise.
6025 * common/config/microblaze/microblaze-common.c: Likewise.
6026 * common/config/mips/mips-common.c: Likewise.
6027 * common/config/mmix/mmix-common.c: Likewise.
6028 * common/config/mn10300/mn10300-common.c: Likewise.
6029 * common/config/nios2/nios2-common.c: Likewise.
6030 * common/config/pa/pa-common.c: Likewise.
6031 * common/config/pdp11/pdp11-common.c: Likewise.
6032 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
6033 * common/config/riscv/riscv-common.c: Likewise.
6034 * common/config/rs6000/rs6000-common.c: Likewise.
6035 * common/config/rx/rx-common.c: Likewise.
6036 * common/config/s390/s390-common.c: Likewise.
6037 * common/config/sh/sh-common.c: Likewise.
6038 * common/config/sparc/sparc-common.c: Likewise.
6039 * common/config/tilegx/tilegx-common.c: Likewise.
6040 * common/config/tilepro/tilepro-common.c: Likewise.
6041 * common/config/v850/v850-common.c: Likewise.
6042 * common/config/visium/visium-common.c: Likewise.
6043 * common/config/xstormy16/xstormy16-common.c: Likewise.
6044 * common/config/xtensa/xtensa-common.c: Likewise.
6045 * invoke.texi (-fomit-frame-pointer): Update documentation.
6047 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6049 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
6051 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6053 * cfg.c (scale_bbs_frequencies_int,
6054 cale_bbs_frequencies_gcov_type): Remove.
6055 * cfg.h (scale_bbs_frequencies_int,
6056 cale_bbs_frequencies_gcov_type): Remove.
6058 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6060 * tree-ssa-loop-manip.c
6061 (scale_dominated_blocks_in_loop): Update to profile counts.
6062 (tree_transform_and_unroll_loop): Likewise.
6064 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6066 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
6067 scale_bbs_frequencies_int.
6069 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6071 * final.c (compute_alignments): Use counts rather than frequencies.
6073 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6075 * cfgloopanal.c: Include sreal.h
6076 (average_num_loop_insns): Use counts and sreal for accounting.
6078 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6080 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
6083 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6084 Kito Cheng <kito.cheng@gmail.com>
6086 * config/nds32/constraints.md: Provide more constraints.
6087 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
6088 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
6089 support constraints usage.
6091 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6093 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
6095 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6096 Kito Cheng <kito.cheng@gmail.com>
6098 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
6099 * config/nds32/nds32.opt: Refine the layout.
6100 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
6101 TARGET_EXT_STRING): Support new options.
6102 * config/nds32/nds32.h: Likewise.
6103 * config/nds32/nds32.md: Likewise.
6104 * config/nds32/nds32-predicates.c: Likewise.
6105 * config/nds32/constraints.md: Likewise.
6106 * common/config/nds32/nds32-common.c: Likewise.
6108 2017-11-16 Julia Koval <julia.koval@intel.com>
6111 * config/i386/gfniintrin.h: Add sse check.
6112 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
6114 2017-11-16 Julia Koval <julia.koval@intel.com>
6116 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
6117 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
6118 (ix86_handle_option): Handle -mavx512vbmi2.
6119 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
6120 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
6121 * config/i386/i386-c.c (__AVX512VBMI2__): New.
6122 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
6123 (ix86_valid_target_attribute_inner_p): Ditto.
6124 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
6125 * config/i386/i386.opt (mavx512vbmi2): New option.
6126 * doc/invoke.texi: Add new option.
6128 2017-11-16 Julia Koval <julia.koval@intel.com>
6130 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
6131 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
6132 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
6133 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
6134 _mm512_gf2p8mul_epi8): New intrinsics.
6135 * config/i386/i386-builtin-types.def
6136 (V64QI_FTYPE_V64QI_V64QI): New type.
6137 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
6138 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
6139 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
6140 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
6141 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
6142 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
6144 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
6146 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
6147 explicitly as a stream of bytes.
6149 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6151 * config/rs6000/altivec.h (vec_xst_be): New #define.
6152 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
6153 and externalize from *altivec_vperm_<mode>_internal.
6154 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
6156 (XL_BE_V8HI): Likewise.
6157 (XL_BE_V4SI): Likewise.
6158 (XL_BE_V4SI): Likewise.
6159 (XL_BE_V2DI): Likewise.
6160 (XL_BE_V4SF): Likewise.
6161 (XL_BE_V2DF): Likewise.
6162 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
6163 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
6164 all array entries with these keys: VSX_BUILTIN_VEC_XL,
6165 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
6166 VSX_BUILTIN_VEC_XST_BE.
6167 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
6168 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
6170 (altivec_init_builtins): Replace conditional calls to def_builtin
6171 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
6172 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
6173 with unconditional calls. Remove calls to def_builtin for
6174 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
6175 __builtin_vec_xst_be.
6176 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
6177 to define_expand, and add alternate RTL generation for P8.
6178 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
6179 vsx_ld_elemrev_v8hi.
6180 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
6181 add alternate RTL generation for P8.
6182 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
6183 vsx_ld_elemrev_v16qi.
6184 (vsx_st_elemrev_v8hi): Convert define_insn
6185 to define_expand, and add alternate RTL generation for P8.
6186 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
6187 vsx_st_elemrev_v8hi.
6188 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
6189 add alternate RTL generation for P8.
6190 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
6191 vsx_st_elemrev_v16qi.
6193 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
6196 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
6197 TARGET_AVX512ER check.
6198 (ix86_option_override_internal): Set MASK_VZEROUPPER if
6199 neither -mzeroupper nor -mno-zeroupper is used and
6200 TARGET_EMIT_VZEROUPPER is set.
6201 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
6202 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
6204 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6206 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6207 folding of vector compares.
6208 (fold_build_vec_cmp): New helper function.
6209 (fold_compare_helper): New helper function.
6210 (builtin_function_type): Add compare builtins to the list of functions
6211 having unsigned arguments. Cosmetic updates to comment indentation.
6212 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
6213 the not+eq combination.
6215 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6217 PR tree-optimization/82726
6218 PR tree-optimization/70754
6219 * tree-predcom.c (order_drefs_by_pos): New function.
6220 (combine_chains): Move code setting has_max_use_after to...
6221 (try_combine_chains): ...here. New parameter. Sort combined chains
6222 according to position information.
6223 (tree_predictive_commoning_loop): Update call to above function.
6224 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
6226 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6228 PR tree-optimization/82726
6230 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6232 PR tree-optimization/70754
6233 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6234 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6235 combined stmt before it if not NULL.
6236 (combine_chains): Process refs reversely and compute dominance point
6240 2017-02-23 Bin Cheng <bin.cheng@arm.com>
6242 PR tree-optimization/79663
6243 * tree-predcom.c (combine_chains): Process refs in reverse order
6244 only for ZERO length chains, and add explaining comment.
6246 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6248 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
6249 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
6250 * doc/invoke.texi (ARM Options): Add armv8.3-a.
6252 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6254 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
6256 2017-11-15 Martin Liska <mliska@suse.cz>
6258 * tree-cfg.c (pass_warn_function_return::execute):
6259 Compare warn_return_type for greater than zero.
6261 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
6265 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
6266 to return true on Xeon and not on Xeon Phi.
6267 (ix86_check_avx256_register): Changed to ...
6268 (ix86_check_avx_upper_register): ... this. Add extra check for
6269 VALID_AVX512F_REG_OR_XI_MODE.
6270 (ix86_avx_u128_mode_needed): Changed
6271 ix86_check_avx256_register to ix86_check_avx_upper_register.
6272 (ix86_check_avx256_stores): Changed to ...
6273 (ix86_check_avx_upper_stores): ... this. Changed
6274 ix86_check_avx256_register to ix86_check_avx_upper_register.
6275 (ix86_avx_u128_mode_after): Changed
6276 avx_reg256_found to avx_upper_reg_found. Changed
6277 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
6278 (ix86_avx_u128_mode_entry): Changed
6279 ix86_check_avx256_register to ix86_check_avx_upper_register.
6280 (ix86_avx_u128_mode_exit): Ditto.
6281 * config/i386/i386.h: (host_detect_local_cpu): New define.
6283 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6285 * config/arm/xgene1.md (xgene1): Split into automatons
6286 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
6287 (xgene1_f_load): Adjust reservations and/or types.
6288 (xgene1_f_store): Likewise.
6289 (xgene1_load_pair): Likewise.
6290 (xgene1_store_pair): Likewise.
6291 (xgene1_fp_load1): Likewise.
6292 (xgene1_load1): Likewise.
6293 (xgene1_store1): Likewise.
6294 (xgene1_move): Likewise.
6295 (xgene1_alu): Likewise.
6296 (xgene1_simd): Likewise.
6297 (xgene1_bfm): Likewise.
6298 (xgene1_neon_load1): Likewise.
6299 (xgene1_neon_store1): Likewise.
6300 (xgene1_neon_logic): Likewise.
6301 (xgene1_neon_st1): Likewise.
6302 (xgene1_neon_ld1r): Likewise.
6303 (xgene1_alu_cond): Added.
6304 (xgene1_shift_reg): Likwise.
6305 (xgene1_bfx): Likewise.
6306 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
6308 2017-11-15 Jakub Jelinek <jakub@redhat.com>
6311 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
6313 (can_widen_mult_without_libcall): New function.
6314 (expand_mul_overflow): If only checking unsigned mul overflow,
6315 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
6316 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
6317 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
6318 (expand_DIVMOD): Formatting fix.
6319 * expmed.h (expand_mult): Add NO_LIBCALL argument.
6320 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
6321 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
6323 PR tree-optimization/82977
6324 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
6325 constructed temporary to strlen_to_stridx.put.
6327 2017-11-15 Martin Liska <mliska@suse.cz>
6329 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
6330 * configure: Regenerate.
6332 2017-11-15 Martin Liska <mliska@suse.cz>
6335 * config/sh/sh-mem.cc: Use proper probability for
6338 2017-11-14 Jeff Law <law@redhat.com>
6340 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
6341 the red zone for stack_clash_protection_final_dynamic_probe targets
6342 when the total dynamic stack size is zero bytes.
6344 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
6345 blocks is post order.
6347 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
6349 * dumpfile.h (TDF_COMPARE_DEBUG): New.
6350 * final.c (rest_of_clean_state): Set it for the
6351 -fcompare-debug dump.
6352 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
6353 class when TDF_COMPARE_DEBUG is set.
6355 * dwarf2out.c (gen_producer_string): Discard
6358 2017-11-15 Joseph Myers <joseph@codesourcery.com>
6361 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
6362 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
6363 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
6364 (__TGMATH_REAL_2_3): Remove macros.
6365 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
6366 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
6368 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
6370 (remquo): Define using __TGMATH_REAL_3.
6372 2017-11-14 Jeff Law <law@redhat.com>
6374 * vr-values.c: New file with contents extracted from tree-vrp.c.
6375 * Makefile.in (OBJS): Add vr-values.o
6376 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
6377 (set_value_range, set_and_canonicalize_value_range): Likewise.
6378 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
6379 (value_range_constant_singleton, symbolic_range_p): Likewise.
6380 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
6381 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
6382 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
6383 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
6384 (find_case_label_range, find_case_label_index): Likewise.
6385 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
6386 (range_int_cst_singleton_p, value_inside_range): Likewise.
6387 (get_single_symbol): Likewise.
6388 (switch_update): Move structure definition here.
6389 (to_remove_edges, to_update_switch_stmts): Provide externs.
6390 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
6391 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
6392 (vrp_val_is_min, set_value_range): Likewise.
6393 (set_and_canonicalize_value_range, copy_value_range): Likewise.
6394 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
6395 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
6396 (range_is_nonnull, range_int_cst_p): Likewwise.
6397 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
6398 (get_single_symbol, operand_less_p): Likewise
6399 (compare_values_warnv, compare_values): Likewise.
6400 (value_inside_range, value_range_constant_singleton): Likewise.
6401 (zero_nonzero_bitgs_from_vr): Likewise.
6402 (extract_range_from_binary_expr_1): Likewise.
6403 (overflow_comparison_p): Likewise.
6404 (to_remove_edges, to_update_switch_stmts): Likewise.
6405 (find_case_label-index, find_case_label_range): Likewise.
6406 (switch_update, set_value_range_to_nonnegative): Remove.
6407 (set_value_range_to_truthvalue): Likewise.
6408 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
6409 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
6410 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
6411 (find_case_label_ranges, test_for_singularity): Likewise.
6412 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
6413 (x_vr_values): Move to its remaining use site.
6415 2017-11-10 Jeff Law <law@redhat.com>
6417 * vr-values.h (VR_INITIALIZER): Move #define here.
6418 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
6419 * Makefile.in (OBJS): Add tree-evrp.o
6420 * tree-vrp.h (assert_info): Move structure definition here.
6421 (set_value_range_to_varying): Prototype.
6422 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6423 (infer_value_range, register_edge_assert_for): Likewise.
6424 (stmt_interesting_for_vrp): Likewise.
6425 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
6426 (set_value_range_to_varying): No longer static.
6427 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6428 (infer_value_range, register_edge_assert_for): Likewise.
6430 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6432 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
6433 generate the XXBRD instruction.
6435 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
6436 (rs6000_builtin_type_compatible): Treat _Float128 and long double
6437 as being compatible if -mabi=ieeelongdouble.
6438 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
6439 to setup float128 built-ins with hardware support.
6440 (BU_FLOAT128_HW_2): Likewise.
6441 (BU_FLOAT128_HW_3): Likewise.
6442 (BU_FLOAT128_HW_VSX_1): Likewise.
6443 (BU_FLOAT128_HW_VSX_2): Likewise.
6444 (scalar_extract_expq): Change float128 built-in functions to
6445 accommodate having both KFmode and TFmode functions. Use the
6446 KFmode variant as the default.
6447 (scalar_extract_sigq): Likewise.
6448 (scalar_test_neg_qp): Likewise.
6449 (scalar_insert_exp_q): Likewise.
6450 (scalar_insert_exp_qp): Likewise.
6451 (scalar_test_data_class_qp): Likewise.
6452 (sqrtf128_round_to_odd): Delete processing the round to odd
6453 built-in functions as special built-in functions, and define them
6454 as float128 built-ins. Use the KFmode variant as the default.
6455 (truncf128_round_to_odd): Likewise.
6456 (addf128_round_to_odd): Likewise.
6457 (subf128_round_to_odd): Likewise.
6458 (mulf128_round_to_odd): Likewise.
6459 (divf128_round_to_odd): Likewise.
6460 (fmaf128_round_to_odd): Likewise.
6461 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6462 support for KFmode and TFmode xststdcqp calls.
6463 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
6464 point, switch the built-in handlers for the get/set float128
6465 exponent, get float128 mantissa, float128 test built-ins, and the
6466 float128 round to odd built-in functions. Eliminate creating the
6467 float128 round to odd built-in functions as special built-ins.
6468 (rs6000_init_builtins): Eliminate special creation of the float128
6469 round to odd built-in functions.
6470 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
6471 function insns to support both TFmode and KFmode variants.
6472 (xsxsigqp_<mode>): Likewise.
6473 (xsiexpqpf_<mode>): Likewise.
6474 (xsiexpqp_<mode>): Likewise.
6475 (xststdcqp_<mode>): Likewise.
6476 (xststdcnegqp_<mode>): Likewise.
6477 (xststdcqp_<mode>): Likewise.
6479 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6481 * tree-ssa-threadupdate.c (compute_path_counts): Remove
6482 unused path_in_freq_ptr parameter.
6483 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
6485 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6487 * ipa-inline.c (edge_badness): Dump sreal frequency.
6488 (compute_inlined_call_time): Match natural implementaiton ...
6489 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
6490 forgotten division by CGRAPH_FREQ_BASE.
6492 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6494 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
6495 Solaris 11. Update comment.
6496 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
6498 * config/sol2.h (STARTFILE_SPEC): Likewise.
6499 * configure: Regenerate.
6501 2017-11-14 Carl Love <cel@us.ibm.com>
6503 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
6504 le_ and be_ prefixes to swap* variables. Remove
6505 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
6508 2017-11-14 Jason Merrill <jason@redhat.com>
6510 Support GTY((cache)) on hash_map.
6511 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
6512 (ggc_cache_remove): Override it instead of ggc_mx.
6513 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
6514 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
6515 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
6516 (simple_cache_map_traits): Override maybe_mx.
6517 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
6518 (hash_map): Friend gt_cleare_cache.
6519 (gt_cleare_cache): New.
6520 * tree.h (tree_cache_traits): New hash_map traits class.
6521 (tree_cache_map): New typedef.
6523 2017-11-14 Richard Biener <rguenther@suse.de>
6525 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
6526 paramter and handling.
6527 (cleanup_control_flow_bb): Likewise.
6528 (cleanup_control_flow_pre): New helper performing a DFS walk
6529 to call cleanup_control_flow_bb in PRE order.
6530 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
6531 via cleanup_control_flow_pre.
6533 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
6535 * config/aarch64/aarch64-simd.md
6536 (aarch64_simd_bsl<mode>_internal): Remove DImode.
6537 (*aarch64_simd_bsl<mode>_alt): Likewise.
6538 (aarch64_simd_bsldi_internal): New.
6539 (aarch64_simd_bsldi_alt): Likewise.
6541 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6543 * tracer.c (better_p): Do not compare frequencies.
6544 * reg-stack.c (better_edge): Likewise.
6545 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
6548 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6550 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
6551 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
6552 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
6553 * lto-streamer-in.c (input_function): Use update_max_bb_count.
6554 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
6555 * predict.c (maybe_hot_frequency_p): Inline to ...
6556 (maybe_hot_count_p): ... here; rewrite to counts.
6557 (counts_to_freqs): Rename to ...
6558 (update_max_bb_count): ... this one.
6559 (expensive_function_p): Use counts.
6560 (estimate_bb_frequencies): Update.
6561 (rebuild_frequencies): Update.
6562 * predict.h (counts_to_freqs): Rename to ...
6563 (update_max_bb_count): ... this one.
6564 * profile.c (compute_branch_probabilities): Add debug info
6565 * tree-inline.c (expand_call_inline): Update debug info.
6566 (optimize_inline_calls): Use update_max_bb_count..
6567 (tree_function_versioning): Use update_max_bb_count..
6568 * value-prof.c (gimple_value_profile_transformations):
6569 Do not use update_max_bb_count.
6571 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6573 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
6574 always use frequencies.
6576 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6578 * bb-reorder.c: Remove frequencies from comments.
6579 (better_edge_p): Use profile counts.
6580 (find_traces): Dump profile counts.
6581 (rotate_loop): Use profile counts.
6582 (find_traces_1_round): Likewise.
6583 (connect_better_edge_p): Use counts instead of probabilities for
6585 (copy_bb_p): Drop early check for non-0 frequency.
6586 (sanitize_hot_paths): Update comments.
6588 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6590 * ipa-split.c (struct split_point): Add count.
6591 (consider_split): Do not compute incoming frequency; compute incoming
6592 count and store it to split_point.
6593 (split_function): Set count of the call to split part correctly.
6595 2017-11-13 Carl Love <cel@us.ibm.com>
6597 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
6599 2017-11-13 Tom Tromey <tom@tromey.com>
6601 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
6603 2017-11-13 Carl Love <cel@us.ibm.com>
6605 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
6606 Add support for builtins:
6607 unsigned int vec_first_{,miss}_match_{,or_eos}index,
6608 vector {un,}signed {char,int,short},
6609 vector {un,}signed {char,int,short}) arguments.
6610 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
6611 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
6612 Add BU_P9V_AV_2 expansions for the builtins.
6613 * config/rs6000/altivec.h (vec_first_match_index,
6614 vec_first_mismatch_index, vec_first_match_or_eos_index,
6615 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
6616 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
6617 new extern declaration.
6618 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
6619 * config/rs6000/vsx.md (first_match_index_<mode>,
6620 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
6621 first_mismatch_or_eos_index_<mode>): Add define expand.
6622 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
6623 * doc/extend.texi: Update the built-in documenation file for the new
6626 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
6628 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
6629 into the min/max operations for _Float<N> and _Float<N>X types.
6631 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6634 * dwarf2out.c (do_eh_frame): New static variable.
6635 (dwarf2out_begin_prologue): Set it.
6636 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
6638 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6640 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
6642 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
6644 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6647 * debug.h (dwarf2out_do_eh_frame): Declare.
6648 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
6649 (dwarf2out_do_frame): Use it.
6650 (dwarf2out_do_cfi_asm): Likewise.
6651 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
6652 (dwarf2out_assembly_start): Likewise.
6653 (dwarf2out_begin_prologue): Fix comment.
6654 * toplev.c (compile_file): Always call dwarf2out_frame_finish
6655 if the target needs either debug or unwind DWARF2 info.
6656 * lto-opts.c (lto_write_options): Do not save -fexceptions,
6657 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
6658 -ftrapping-math, -ftrapv and -fwrapv.
6660 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6662 * cgraph.c (cgraph_edge::sreal_frequency): New function.
6663 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
6664 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
6665 (estimate_edge_size_and_time): Likewise.
6666 (ipa_merge_fn_summary_after_inlining): Likewise.
6667 * ipa-inline.c (cgraph_freq_base_rec): Remove.
6668 (compute_uninlined_call_time): Use sreal_frequency.
6669 (compute_inlined_call_time): Likewise.
6670 (ipa_inline): Do not initialize cgraph_freq_base_rec.
6671 * profile-count.c: Include sreal.h.
6672 (profile_count::to_sreal_scale): New.
6673 * profile-count.h: Forward declare sreal.
6674 (profile_count::to_sreal_scale): Declare.
6676 2017-11-13 Nathan Sidwell <nathan@acm.org>
6678 * diagnostic.c (maybe_line_and_column): New.
6679 (diagnostic_get_location_text): Use it.
6680 (diagnostic_report_current_module): Likewise.
6681 (test_diagnostic_get_location_text): Add tests.
6683 2017-11-13 Luis Machado <luis.machado@linaro.org>
6685 * doc/md.texi (Specifying processor pipeline description): Fix
6686 incorrect latency for the div instruction example.
6688 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6690 PR tree-optimization/78821
6691 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
6692 that bit_not_p is the same.
6693 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
6694 (split_group): Count precisely bit_not_p bits in each statement.
6695 (invert_op): New function.
6696 (imm_store_chain_info::output_merged_store): Use invert_op to
6697 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
6698 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
6700 2017-11-13 Martin Liska <mliska@suse.cz>
6702 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
6703 (struct source_info): Likewise.
6704 (add_branch_counts): Likewise.
6705 (add_line_counts): Likewise.
6706 (function_summary): Likewise.
6707 (output_intermediate_line): Likewise.
6708 (generate_results): Likewise.
6710 2017-11-13 Martin Liska <mliska@suse.cz>
6712 * gcov.c (struct block_info): Remove typedef for block_t.
6713 (struct line_info): Likewise.
6714 (line_info::has_block): Likewise.
6715 (EXIT_BLOCK): Likewise.
6716 (unblock): Likewise.
6717 (circuit): Likewise.
6718 (get_cycles_count): Likewise.
6719 (process_file): Likewise.
6720 (read_graph_file): Likewise.
6721 (solve_flow_graph): Likewise.
6722 (find_exception_blocks): Likewise.
6723 (add_line_counts): Likewise.
6724 (accumulate_line_info): Likewise.
6725 (output_line_details): Likewise.
6727 2017-11-13 Martin Liska <mliska@suse.cz>
6729 * gcov.c (struct arc_info): Remove typedef for arc_t.
6730 (struct line_info): Likewise.
6731 (add_branch_counts): Likewise.
6732 (output_branch_count): Likewise.
6733 (function_info::~function_info): Likewise.
6734 (circuit): Likewise.
6735 (output_intermediate_line): Likewise.
6736 (read_graph_file): Likewise.
6737 (solve_flow_graph): Likewise.
6738 (find_exception_blocks): Likewise.
6739 (add_line_counts): Likewise.
6740 (accumulate_line_info): Likewise.
6741 (output_line_details): Likewise.
6742 (output_function_details): Likewise.
6744 2017-11-13 Martin Liska <mliska@suse.cz>
6746 * gcov.c (struct function_info): Remove typedef for function_t.
6747 (struct source_info): Likewise.
6748 (source_info::get_functions_at_location): Likewise.
6749 (solve_flow_graph): Likewise.
6750 (find_exception_blocks): Likewise.
6751 (add_line_counts): Likewise.
6752 (output_intermediate_file): Likewise.
6753 (process_file): Likewise.
6754 (generate_results): Likewise.
6755 (release_structures): Likewise.
6756 (read_graph_file): Likewise.
6757 (read_count_file): Likewise.
6758 (accumulate_line_counts): Likewise.
6759 (output_lines): Likewise.
6761 2017-11-13 Martin Liska <mliska@suse.cz>
6763 * gcov.c (function_info::function_info): Remove num_counts
6764 and add vector<gcov_type>.
6765 (function_info::~function_info): Use the vector.
6766 (process_file): Likewise.
6767 (read_graph_file): Likewise.
6768 (read_count_file): Likewise.
6769 (solve_flow_graph): Likewise.
6771 2017-11-13 Martin Liska <mliska@suse.cz>
6773 * gcov.c (function_info::is_artificial): New function.
6774 (process_file): Erase all artificial early.
6775 (generate_results): Skip as all artificial are already
6778 2017-11-13 Martin Liska <mliska@suse.cz>
6780 * gcov.c (read_graph_file): Store to global vector of functions.
6781 (read_count_file): Iterate the vector.
6782 (process_file): Likewise.
6783 (generate_results): Likewise.
6784 (release_structures): Likewise.
6786 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6788 PR tree-optimization/82954
6789 * gimple-ssa-store-merging.c
6790 (imm_store_chain_info::coalesce_immediate_stores): If
6791 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
6793 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
6795 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6796 Upddate call to ENDIAN_LANE_N.
6797 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
6798 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
6799 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
6800 and use aarch64_endian_lane_rtx.
6801 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
6803 2017-11-12 Tom de Vries <tom@codesourcery.com>
6805 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
6807 2017-11-12 Tom de Vries <tom@codesourcery.com>
6809 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
6811 2017-11-12 Tom de Vries <tom@codesourcery.com>
6813 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
6815 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6816 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6817 * defaults.h (ASM_OUTPUT_LABELREF): Same.
6819 2017-11-11 Martin Sebor <msebor@redhat.com>
6822 * doc/extend.texi (attribute nonstring): Remove spurious argument.
6825 * prefic.c (translate_name): Replace strncpy with memcpy to
6826 avoid -Wstringop-truncation.
6828 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6830 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
6832 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6834 * predict.c (maybe_hot_frequency_p): Do not use cfun.
6836 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6838 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
6841 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
6843 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
6844 enable generating XXBRH if the value is in a vector register.
6845 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
6846 value is in a vector register.
6847 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
6848 register to register bswap64's instead of doing the GPR sequence
6849 used on previous machines.
6850 (bswapdi2_xxbrd): New insn.
6851 (bswapdi2_reg): Disallow on ISA 3.0.
6852 (register to register bswap64 splitter): Do not split the insn on
6853 ISA 3.0 systems that use XXBRD.
6855 2017-11-10 Martin Sebor <msebor@redhat.com>
6858 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
6860 (find_subframework_file): Same.
6862 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6864 * auto-profile.c (afdo_indirect_call): Drop frequency.
6865 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
6866 (cgraph_node::create_edge): Drop frequency argument.
6867 (cgraph_node::create_indirect_edge): Drop frequency argument.
6868 (cgraph_edge::make_speculative): Drop frequency arguments.
6869 (cgraph_edge::resolve_speculation): Do not update frequencies
6870 (cgraph_edge::dump_edge_flags): Do not dump frequency.
6871 (cgraph_node::dump): Check consistency in IPA mode.
6872 (cgraph_edge::maybe_hot_p): Use IPA counter.
6873 (cgraph_edge::verify_count_and_frequency): Rename to ...
6874 (cgraph_edge::verify_count): ... this one; drop frequency checking.
6875 (cgraph_node::verify_node): Update.
6876 * cgraph.h (struct cgraph_edge): Drop frequency.
6877 (cgraph_edge::frequency): New function.
6878 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
6880 (cgraph_edge::rebuild_edges): Likewise.
6881 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
6882 (duplicate_thunk_for_node): Do not pass frequency.
6883 (cgraph_node::create_clone): Scale only counts.
6884 (cgraph_node::create_virtual_clone): Do not pass frequency.
6885 (cgraph_node::create_edge_including_clones): Do not pass frequency.
6886 (cgraph_node::create_version_clone): Do not pass frequency.
6887 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
6888 (cgraph_node::expand_thunk): Do not pass frequency.
6889 (cgraph_node::create_wrapper): Do not pass frequency.
6890 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
6892 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
6893 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
6894 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
6895 (gather_caller_stats): Use frequency function.
6896 (ipcp_cloning_candidate_p): Use frequency function.
6897 (ipcp_propagate_stage): Use frequency function.
6898 (get_info_about_necessary_edges): Use frequency function.
6899 (update_profiling_info): Update only IPA profile.
6900 (update_specialized_profile): Use frequency functoin.
6901 (perhaps_add_new_callers): Update only IPA profile.
6902 * ipa-devirt.c (ipa_devirt): Use IPA profile.
6903 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
6904 (dump_ipa_call_summary): Use frequency function.
6905 (estimate_edge_size_and_time): Use frequency function.
6906 (ipa_merge_fn_summary_after_inlining): Use frequency function.
6907 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
6908 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
6909 (update_noncloned_counts): ... ths one; scale counts only.
6910 (clone_inlined_nodes): Do not scale frequency.
6911 (inline_call): Do not pass frequency.
6912 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
6913 (compute_inlined_call_time): Use IPA profile.
6914 (want_inline_small_function_p): Use IPA profile.
6915 (want_inline_self_recursive_call_p): Use IPA profile.
6916 (edge_badness): Use IPA profile.
6917 (lookup_recursive_calls): Use IPA profile.
6918 (recursive_inlining): Do not pass frequency.
6919 (resolve_noninline_speculation): Do not update frequency.
6920 (inline_small_functions): Collect max of IPA profile.
6921 (dump_overall_stats): Dump IPA porfile.
6922 (dump_inline_stats): Dump IPA porfile.
6923 (ipa_inline): Collect IPA stats.
6924 * ipa-inline.h (clone_inlined_nodes): Update prototype.
6925 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
6926 (ipa_propagate_frequency): Use frequency function.
6927 (ipa_profile): Cleanup.
6928 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
6929 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
6930 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
6931 (input_node): Do not stream frequency.
6932 (input_edge): Do not stream frequency.
6933 (merge_profile_summaries): Scale only IPA profiles.
6934 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
6935 * predict.c (drop_profile): Do not recompute frequency.
6936 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
6937 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
6938 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
6939 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
6940 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
6941 * tree-inline.c (copy_bb): Do not scale frequency.
6942 (expand_call_inline): Do not scale frequency.
6943 (tree_function_versioning): Do not scale frequency.
6944 * ubsan.c (ubsan_create_edge): Do not pass frequency.
6946 2017-11-10 Julia Koval <julia.koval@intel.com>
6948 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
6949 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
6950 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
6951 (_mm256_maskz_gf2p8affine_epi64_epi8)
6952 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
6953 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
6954 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
6955 (__builtin_ia32_vgf2p8affineqb_v32qi)
6956 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
6957 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
6959 2017-11-10 Tamar Christina <tamar.christina@arm.com>
6963 (arm_option_override): Refactor.
6964 (arm_option_reconfigure_globals): New.
6965 (arm_options_perform_arch_sanity_checks): New.
6966 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
6968 (arm_options_perform_arch_sanity_checks): Likewise
6970 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
6972 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
6973 (power9-qpmul): New.
6974 * rs6000/rs6000.md ("type" attr): Add qmul.
6975 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
6976 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
6977 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
6979 2017-11-10 Martin Sebor <msebor@redhat.com>
6982 * builtins.c (compute_objsize): Handle arrays that
6983 compute_builtin_object_size likes to fail for. Make extern.
6984 * builtins.h (compute_objsize): Declare.
6985 (check_strncpy_sizes): New function.
6986 (expand_builtin_strncpy): Call check_strncpy_sizes.
6987 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
6988 -Wstringop-truncation.
6989 (gimple_fold_builtin_strncat): Same.
6990 * gimple.c (gimple_build_call_from_tree): Set call location.
6991 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
6992 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
6993 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
6994 (handle_builtin_strlen): Use strlen_to_stridx.
6995 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
6997 Use strlen_to_stridx.
6998 (pass_strlen::execute): Release strlen_to_stridx.
6999 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
7000 (-Wstringop-truncation): Document new option.
7002 2017-11-10 Martin Liska <mliska@suse.cz>
7004 PR gcov-profile/82702
7005 * gcov.c (main): Handle intermediate files in a different
7007 (get_gcov_intermediate_filename): New function.
7008 (output_gcov_file): Remove support of intermediate files.
7009 (generate_results): Allocate intermediate file.
7010 (release_structures): Clean-up properly fn_end.
7011 (output_intermediate_file): Start iterating with line 1.
7013 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7015 PR tree-optimization/82929
7016 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7017 ops_swapped_p non-static data member.
7018 (store_immediate_info::store_immediate_info): Clear it.
7019 (imm_store_chain_info::coalesce_immediate_stores): If swapping
7020 ops set ops_swapped_p.
7021 (count_multiple_uses): Handle ops_swapped_p.
7023 2017-11-10 Martin Liska <mliska@suse.cz>
7025 * coverage.c (coverage_init): Stream information about
7026 support of has_unexecuted_blocks.
7027 * doc/gcov.texi: Document that.
7028 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
7029 * gcov.c (read_graph_file): Likewise.
7030 (output_line_beginning): Fix a small issue with
7033 2017-11-10 Bin Cheng <bin.cheng@arm.com>
7035 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
7036 reference of trivial component.
7038 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7041 * gimple-ssa-store-merging.c
7042 (pass_store_merging::terminate_all_aliasing_chains): For
7043 gimple_store_p stmts also call refs_output_dependent_p.
7045 PR rtl-optimization/82913
7046 * compare-elim.c (try_merge_compare): Punt if def_insn is not
7049 2017-11-09 Jeff Law <law@redhat.com>
7051 * vr-values.h: New file with vr_values class.
7052 * tree-vrp.c: Include vr-values.h
7053 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
7054 data objects into the vr_values class.
7055 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
7056 (get_value_range): Make it a member function within vr_values class.
7057 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
7058 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
7059 (op_with_constant_singleton_value_range): Likewise.
7060 (extract_range_for_var_from_comparison_expr): Likewise.
7061 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
7062 (extract_range_from_binary_expr): Likewise.
7063 (extract_range_from_unary_expr): Likewise.
7064 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
7065 (check_for_binary_op_overflow, extract_range_basic): Likewise.
7066 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
7067 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
7068 (compare_name_with_value, compare_names): Likewise.
7069 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
7070 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
7071 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
7072 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
7073 (extract_range_from_phi_node): Likewise.
7074 (simplify_truth_ops_using_ranges): Likewise.
7075 (simplify_div_or_mod_using_ranges): Likewise.
7076 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
7077 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
7078 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
7079 (simplify_float_conversion_using_ranges): Likewise.
7080 (simplify_internal_call_using_ranges): Likewise.
7081 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
7082 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
7083 poitner via x_vr_values for calls into gimple folder.
7084 (vrp_initialize_lattice): Make this the vr_values ctor.
7085 (vrp_free_lattice): Make this the vr_values dtor.
7086 (set_vr_value): New function.
7087 (class vrp_prop): Add vr_values data member. Add various member
7088 functions as well as member functions that delegate to vr_values.
7089 (check_array_ref): Make a member function within vrp_prop class.
7090 (search_for_addr_array, vrp_initialize): Likewise.
7091 (vrp_finalize): Likewise. Revamp to avoid direct access to
7092 vr_value, values_propagated, etc.
7093 (check_array_bounds): Extract vrp_prop class instance pointer from
7094 walk info structure. Use it to call member functions.
7095 (check_all_array_refs): Make a member function within vrp_prop class.
7096 Smuggle class instance pointer via walk info structure.
7097 (x_vr_values): New local static.
7098 (vrp_valueize): Use x_vr_values to get class instance.
7099 (vr_valueize_1): Likewise.
7100 (class vrp_folder): Add vr_values data member. Add various member
7101 functions as well as member functions that delegate to vr_values.
7102 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
7103 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
7104 class instance from vr_values. Use it to call member functions.
7105 (vrp_dom_walker): Add vr_values data member.
7106 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
7107 instance via x_vr_values.
7108 (identify_jump_threads): Accept vr_values as argument. Store
7109 it into the walker structure.
7110 (evrp_dom_walker): Add vr_values class data member. Add various
7112 (evrp_dom_walker::try_find_new_range): Use vr_values data
7113 member to access the memory allocator.
7114 (evrp_dom_walker::before_dom_children): Store vr_values class
7115 instance into the vrp_folder class.
7116 (evrp_dom_walker::push_value_range): Rework to avoid direct
7117 access to num_vr_values and vr_value.
7118 (evrp_dom_walker::pop_value_range): Likewise.
7119 (execute_early_vrp): Remove call to vrp_initialize_lattice.
7120 Use vr_values to get to dump_all_value_ranges member function.
7121 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
7122 and simplify_cond_using_ranges_2 via vrp_prop class instance.
7123 Pass vr_values class instance down to identify_jump_threads.
7124 Remove call to vrp_free_lattice.
7125 (debug_all_value_ranges): Remove.
7127 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
7128 (vrp_folder): Likewise.
7130 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
7131 Get it from the existing bitmap instead.
7132 (vrp_intersect_ranges_1): Likewise.
7134 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7136 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7138 (store_immediate_info::store_immediate_info): Add bitnotp argument,
7139 set bit_not_p to it.
7140 (imm_store_chain_info::coalesce_immediate_stores): Break group
7141 if bit_not_p is different.
7142 (count_multiple_uses, split_group,
7143 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
7144 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
7145 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
7146 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
7147 to store_immediate_info ctor.
7149 2017-11-09 Jim Wilson <jimw@sifive.com>
7151 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
7152 (scan_prog_file): Likewise.
7154 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7156 * bb-reorder.c (max_entry_frequency): Remove.
7157 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
7158 connect_traces, push_to_next_round_p): Remove prototypes.
7159 (find_traces_1_round): Use counts only.
7160 (push_to_next_round_p): Likewise.
7161 (find_traces): Likewise.
7162 (rotate_loop): Likewise.
7163 (find_traces_1_round): Likewise.
7164 (connect_traces): Likewise.
7165 (edge_order): Likewise.
7167 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
7169 * config/arm/arm.c (output_return_instruction): Add comments to
7170 indicate requirement for cmse_nonsecure_entry return to account
7171 for the size of clearing instruction output here.
7172 (thumb_exit): Likewise.
7173 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
7174 return in hardfloat mode.
7176 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
7178 * config/rs6000/rs6000.c (machine_function): Add a bool,
7179 "toc_is_wrapped_separately".
7180 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
7181 if it wasn't explicitly set or unset, we are optimizing for speed, and
7182 doing separate shrink-wrapping.
7183 (rs6000_get_separate_components): Enable the TOC component if
7184 saving the TOC register in the prologue.
7185 (rs6000_components_for_bb): Handle the TOC component.
7186 (rs6000_emit_prologue_components): Store the TOC register where needed.
7187 (rs6000_set_handled_components): Mark TOC as handled, if handled.
7188 (rs6000_emit_prologue): Don't save the TOC if that is already done.
7190 2017-11-09 Martin Jambor <mjambor@suse.cz>
7192 * ipa-param-manipulation.c: New file.
7193 * ipa-param-manipulation.h: Likewise.
7194 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
7195 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
7196 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
7197 (ipa_parm_adjustment): Likewise.
7198 (ipa_parm_adjustment_vec): Likewise.
7199 (ipa_get_vector_of_formal_parms): Moved declaration to
7200 ipa-param-manipulation.h.
7201 (ipa_get_vector_of_formal_parm_types): Likewise.
7202 (ipa_modify_formal_parameters): Likewise.
7203 (ipa_modify_call_arguments): Likewise.
7204 (ipa_combine_adjustments): Likewise.
7205 (ipa_dump_param_adjustments): Likewise.
7206 (ipa_modify_expr): Likewise.
7207 (ipa_get_adjustment_candidate): Likewise.
7208 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
7209 ipa-param-manipulation.c.
7210 (ipa_get_vector_of_formal_parm_types): Likewise.
7211 (ipa_modify_formal_parameters): Likewise.
7212 (ipa_modify_call_arguments): Likewise.
7213 (ipa_modify_expr): Likewise.
7214 (get_ssa_base_param): Likewise.
7215 (ipa_get_adjustment_candidate): Likewise.
7216 (index_in_adjustments_multiple_times_p): Likewise.
7217 (ipa_combine_adjustments): Likewise.
7218 (ipa_dump_param_adjustments): Likewise.
7219 * tree-sra.c: Also include ipa-param-manipulation.h
7220 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
7223 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7224 Alan Hayward <alan.hayward@arm.com>
7225 David Sherwood <david.sherwood@arm.com>
7227 * doc/sourcebuild.texi (vect_masked_store): Document.
7229 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7230 Alan Hayward <alan.hayward@arm.com>
7231 David Sherwood <david.sherwood@arm.com>
7233 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
7235 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7236 Alan Hayward <alan.hayward@arm.com>
7237 David Sherwood <david.sherwood@arm.com>
7239 * doc/sourcebuild.texi (vect_variable_length): Document.
7241 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7242 Alan Hayward <alan.hayward@arm.com>
7243 David Sherwood <david.sherwood@arm.com>
7245 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
7247 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7248 Alan Hayward <alan.hayward@arm.com>
7249 David Sherwood <david.sherwood@arm.com>
7251 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
7253 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7254 Alan Hayward <alan.hayward@arm.com>
7255 David Sherwood <david.sherwood@arm.com>
7257 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
7258 previously undocumented selectors.
7259 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
7261 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7263 * doc/rtl.texi (const_vector): Say that elements can be
7264 const_wide_ints too.
7265 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
7266 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
7267 (gen_vec_duplicate): Use it instead of CONSTANT_P.
7268 * optabs.c (expand_vector_broadcast): Likewise.
7270 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7271 Alan Hayward <alan.hayward@arm.com>
7272 David Sherwood <david.sherwood@arm.com>
7274 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
7275 scaled index even if the unscaled address was invalid.
7276 Don't increase the complexity of using a scale in that case.
7278 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7279 Alan Hayward <alan.hayward@arm.com>
7280 David Sherwood <david.sherwood@arm.com>
7282 * doc/rtl.texi: Rewrite the subreg rules so that they partition
7283 the inner register into REGMODE_NATURAL_SIZE bytes rather than
7284 UNITS_PER_WORD bytes.
7285 * emit-rtl.c (validate_subreg): Divide subregs into blocks
7286 based on REGMODE_NATURAL_SIZE of the inner mode.
7287 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
7288 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
7289 * expmed.c (lowpart_bit_field_p): Divide the value up into
7290 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
7291 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
7292 whether something is likely to occupy more than one register.
7294 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7297 * ipa-inline-transform.c (update_noncloned_frequencies): Use
7298 profile_count::adjust_for_ipa_scaling.
7299 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
7300 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
7302 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
7304 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7306 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
7307 (split_group): Add total_orig and total_new arguments, estimate the
7308 number of statements related to the store group without store merging
7309 and with store merging.
7310 (imm_store_chain_info::output_merged_store): Adjust split_group
7311 callers, punt if estimated number of statements with store merging
7312 is not smaller than estimated number of statements without it.
7314 (handled_load): Remove has_single_use checks.
7315 (pass_store_merging::process_store): Likewise.
7317 2017-11-09 Richard Biener <rguenther@suse.de>
7319 PR tree-optimization/82902
7320 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
7322 2017-11-09 Martin Liska <mliska@suse.cz>
7325 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
7328 2017-11-09 Martin Liska <mliska@suse.cz>
7330 PR tree-optimization/82669
7331 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
7333 2017-11-09 Martin Liska <mliska@suse.cz>
7335 PR gcov-profile/48463
7336 * coverage.c (coverage_begin_function): Output also end locus
7337 of a function and information whether the function is
7339 * gcov-dump.c (tag_function): Parse and print the information.
7340 * gcov.c (INCLUDE_MAP): Add include.
7341 (INCLUDE_SET): Likewise.
7342 (struct line_info): Move earlier in the source file because
7343 of vector<line_info> in function_info structure.
7344 (line_info::line_info): Likewise.
7345 (line_info::has_block): Likewise.
7346 (struct source_info): Add new member index.
7347 (source_info::get_functions_at_location): New function.
7348 (function_info::group_line_p): New function.
7349 (output_intermediate_line): New function.
7350 (output_intermediate_file): Use the mentioned function.
7351 (struct function_start): New.
7352 (struct function_start_pair_hash): Likewise.
7353 (process_file): Add code that identifies group functions.
7354 Assign lines either to global or function scope.
7355 (generate_results): Skip artificial functions.
7356 (find_source): Assign index for each source file.
7357 (read_graph_file): Read new flag artificial and end_line.
7358 (add_line_counts): Assign it either to global of function scope.
7359 (accumulate_line_counts): Isolate core of the function to
7360 accumulate_line_info and call it for both function and global
7362 (accumulate_line_info): New function.
7363 (output_line_beginning): Fix GNU coding style.
7364 (print_source_line): New function.
7365 (output_line_details): Likewise.
7366 (output_function_details): Likewise.
7367 (output_lines): Iterate both source (global) scope and function
7369 (struct function_line_start_cmp): New class.
7370 * doc/gcov.texi: Reflect changes in documentation.
7372 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7375 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
7376 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
7377 and similarly for not instead of neg.
7379 2017-11-08 Andi Kleen <ak@linux.intel.com>
7381 * config/i386/i386.opt: Add -mforce-indirect-call.
7382 * config/i386/predicates.md: Check for flag_force_indirect_call.
7383 * doc/invoke.texi: Document -mforce-indirect-call
7385 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
7387 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
7389 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
7390 config/riscv/riscv.c (predict.h): New include.
7391 (riscv_slow_unaligned_access_p): No longer static.
7392 (riscv_block_move_straight): Add require.
7393 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7395 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7398 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
7399 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
7400 nonimmediate_operand predicate for operand 1 instead of
7403 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7405 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
7407 * config/aarch64/constraints.md (Uml): New constraint.
7408 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
7411 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7413 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
7414 of two vec_duplicates into a vec_concat.
7416 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7418 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7419 Simplify vec_merge of vec_duplicate and vec_concat.
7420 * config/aarch64/constraints.md (Utq): New constraint.
7421 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
7424 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7426 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7427 Simplify vec_merge of vec_duplicate and const_vector.
7428 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
7430 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
7431 mode iterator. Update predicate on operand 1 to
7432 handle non-const_vec constants. Delete constraints.
7433 (*aarch64_combinez_be<mode>): Likewise for operand 2.
7435 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7437 PR tree-optimization/78821
7438 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
7440 (store_operand_info::store_operand_info): Initialize it to false.
7441 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
7442 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
7443 store in the group, and if chain_info is non-NULL, to ignore altogether
7445 (compatible_load_p): Fail if bit_not_p does not match.
7446 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
7447 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
7448 (pass_store_merging::process_store): Adjust
7449 terminate_all_aliasing_chains calls to pass NULL in all current spots,
7450 call terminate_all_aliasing_chains newly when adding a store into
7451 a chain with non-NULL chain_info.
7453 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7455 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
7457 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7459 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7461 (aarch64_layout_frame): Initialise emit_frame_chain.
7462 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
7463 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
7465 2017-11-08 Martin Liska <mliska@suse.cz>
7467 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
7468 of gimple_call_internal_p.
7470 2017-11-07 Tom de Vries <tom@codesourcery.com>
7472 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
7474 2017-11-07 Tom de Vries <tom@codesourcery.com>
7476 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
7479 2017-11-08 Martin Liska <mliska@suse.cz>
7482 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
7484 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7486 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
7488 2017-11-07 Jakub Jelinek <jakub@redhat.com>
7491 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
7492 (*cmp<mode>_ccz_1): New insn with $k alternative.
7495 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
7496 RTX_COMM_COMPARE as commutative as well.
7497 (ix86_binary_operator_ok): Formatting fix.
7498 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
7499 *<code><mode>3<mask_name><round_saeonly_name>,
7500 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
7501 *<s>mul<mode>3_highpart<mask_name>,
7502 *vec_widen_umult_even_v16si<mask_name>,
7503 *vec_widen_umult_even_v8si<mask_name>,
7504 *vec_widen_umult_even_v4si<mask_name>,
7505 *vec_widen_smult_even_v16si<mask_name>,
7506 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
7507 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
7508 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
7509 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
7510 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
7511 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
7512 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
7513 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
7514 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
7515 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
7516 ix86_binary_operator_ok. Formatting fixes.
7517 (*<plusminus_insn><mode>3<mask_name><round_name>,
7518 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
7521 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
7523 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
7524 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
7525 eq and ne if TARGET_ISEL.
7526 (cmp): New code_iterator.
7527 (UNS, UNSU_, UNSIK): New code_attrs.
7528 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
7529 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
7531 ("eq<mode>3"): ... this.
7532 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
7534 ("ne<mode>3"): ... this.
7536 2017-11-07 Julia Koval <julia.koval@intel.com>
7539 * common/config/i386/i386-common.c
7540 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
7541 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
7542 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
7543 * config/i386/i386.opt: Ditto.
7544 * config/i386/i386.c (ix86_target_string): Ditto.
7545 (ix86_option_override_internal): Ditto.
7546 (ix86_init_mpx_builtins): Move MPX to args2.
7547 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
7548 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7549 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7550 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7551 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7552 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7553 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
7555 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
7558 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
7559 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
7560 (zero-extendsidi peephole2): Remove peephole.
7562 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7565 * stor-layout.c: Include attribs.h.
7566 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
7567 explicit lookup of "aligned" attribute.
7569 2017-11-07 Andrew Waterman <andrew@sifive.com>
7571 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
7572 (riscv_expand_block_move): Likewise.
7573 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
7575 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
7576 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
7577 * config/riscv/riscv.c (riscv_block_move_straight): New function.
7578 (riscv_adjust_block_mem): Likewise.
7579 (riscv_block_move_loop): Likewise.
7580 (riscv_expand_block_move): Likewise.
7581 * config/riscv/riscv.md (movmemsi): New pattern.
7583 2017-11-07 Michael Clark <michaeljclark@mac.com>
7585 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
7586 (MUSL_DYNAMIC_LINKER): Likewise.
7588 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
7590 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
7591 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
7594 2017-11-07 Richard Biener <rguenther@suse.de>
7596 * match.pd: Fix build.
7598 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
7599 Jackson Woodruff <jackson.woodruff@arm.com>
7601 PR tree-optimization/71026
7602 * match.pd: Canonicalize negate in division.
7604 2017-11-07 Sudakshina Das <sudi.das@arm.com>
7607 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
7609 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7611 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
7612 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
7614 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7616 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
7617 non-scalar integral types.
7618 * match.pd (negate_expr_p): Handle MINUS_EXPR.
7619 (-(A-B), -(~A)): New transformations.
7621 2017-11-07 Tom de Vries <tom@codesourcery.com>
7623 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
7624 semicolon after "do {} while (0)".
7625 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7626 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7627 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7628 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7629 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7631 2017-11-07 Tom de Vries <tom@codesourcery.com>
7633 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
7634 after "do {} while (0)".
7635 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7636 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7637 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7638 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7639 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7641 2017-11-07 Tom de Vries <tom@codesourcery.com>
7644 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
7646 (arm_rtx_costs_internal): Add missing semicolon after
7647 HANDLE_NARROW_SHIFT_ARITH call.
7649 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7651 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7652 disable isel if it was not set explicitly.
7654 2017-11-06 James Bowman <james.bowman@ftdichip.com>
7656 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
7657 (add_type_attribute) likewise.
7659 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
7661 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
7662 of struct ix86_frame.
7663 (ix86_initial_elimination_offset): Likewise.
7664 (ix86_expand_split_stack_prologue): Likewise.
7666 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
7668 * tree-vrp.h (enum value_range_type): Update stale comment.
7670 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7671 Alan Hayward <alan.hayward@arm.com>
7672 David Sherwood <david.sherwood@arm.com>
7674 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
7675 (aarch64_expand_vec_perm_const): Take the number of units too.
7676 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
7677 (aarch64_expand_vec_perm_const): Likewise.
7678 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
7679 (vec_perm<mode>): Update accordingly.
7681 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7682 Alan Hayward <alan.hayward@arm.com>
7683 David Sherwood <david.sherwood@arm.com>
7685 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
7686 Take the number of units too.
7687 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
7688 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
7689 but check for a vector mode before rather than after the call.
7690 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7691 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
7692 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
7693 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
7694 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
7695 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
7696 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
7697 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
7698 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
7699 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
7700 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
7701 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
7702 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
7703 (aarch64_sqdmull2_n<mode>): Update accordingly.
7705 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7706 Alan Hayward <alan.hayward@arm.com>
7707 David Sherwood <david.sherwood@arm.com>
7709 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
7710 the number of units too.
7711 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
7712 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
7713 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
7714 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
7715 (vec_store_lanesxi<mode>): Update accordingly.
7717 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7718 Alan Hayward <alan.hayward@arm.com>
7719 David Sherwood <david.sherwood@arm.com>
7721 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
7722 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
7723 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
7724 of units rather than the mode.
7725 * config/aarch64/iterators.md (nunits): New mode attribute.
7726 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
7727 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
7728 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
7729 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
7730 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
7731 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
7732 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
7733 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
7734 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
7735 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
7736 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
7737 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
7738 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
7739 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
7740 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
7741 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
7742 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
7743 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
7744 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
7745 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
7746 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
7747 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
7748 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
7749 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
7750 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
7751 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
7752 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
7753 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
7754 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
7755 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
7756 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
7757 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
7758 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
7759 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
7760 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
7761 (aarch64_simd_vec_setv2di): Likewise.
7763 2017-11-06 Carl Love <cel@us.ibm.com>
7765 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
7767 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
7768 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
7770 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
7772 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
7773 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
7774 Add power 8 macro expansions.
7775 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
7776 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
7777 power 8 instructions. (VSX_XXBR): Add iterator.
7779 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
7781 * config/arm/arm.md (predicable_short_it): Change default to "no",
7782 improve documentation, remove uses that are identical to the default.
7783 (enabled_for_depr_it): Rename to enabled_for_short_it.
7784 * gcc/config/arm/arm-fixed.md (predicable_short_it):
7785 Remove default uses.
7786 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
7787 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
7788 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
7789 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
7791 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
7794 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
7795 float128 helper macros, which are no longer used after deleting
7796 the old 'q' built-in functions, and moving the round to odd
7797 built-in functions to being special built-in functions.
7798 (BU_FLOAT128_2): Likewise.
7799 (BU_FLOAT128_1_HW): Likewise.
7800 (BU_FLOAT128_2_HW): Likewise.
7801 (BU_FLOAT128_3_HW): Likewise.
7802 (FABSQ): Delete old 'q' built-in functions.
7803 (COPYSIGNQ): Likewise.
7804 (SQRTF128_ODD): Move round to odd built-in functions to be
7805 special built-in functions, so that we can handle
7806 -mabi=ieeelongdouble.
7807 (TRUNCF128_ODD): Likewise.
7808 (ADDF128_ODD): Likewise.
7809 (SUBF128_ODD): Likewise.
7810 (MULF128_ODD): Likewise.
7811 (DIVF128_ODD): Likewise.
7812 (FMAF128_ODD): Likewise.
7813 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
7814 built-in names to 'f128'.
7815 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
7816 old 'q' built-in functions, as the machine independent code for
7817 'f128' built-in functions handles this.
7818 (rs6000_expand_builtin): Add expansion for float128 round to odd
7819 functions, keying off on -mabi=ieeelongdouble of whether to use
7820 the KFmode or TFmode variant.
7821 (rs6000_init_builtins): Initialize the _Float128 round to odd
7823 * doc/extend.texi (PowerPC Built-in Functions): Document the old
7824 _Float128 'q' built-in functions are now mapped into the new
7825 'f128' built-in functions.
7827 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
7829 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
7830 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
7832 2017-11-06 David Malcolm <dmalcolm@redhat.com>
7835 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
7836 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
7837 * toplev.c: Include "ipa-fnsummary.h".
7838 (toplev::finalize): Call ipa_fnsummary_c_finalize.
7840 2017-11-06 Jakub Jelinek <jakub@redhat.com>
7842 PR tree-optimization/82838
7843 * gimple-ssa-store-merging.c
7844 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
7845 on a separate gimple_seq which is then appended to seq.
7847 2017-11-06 Jeff Law <law@redhat.com>
7850 * config/i386/i386.c (PROBE_INTERVAL): Remove.
7851 (get_probe_interval): New functions.
7852 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
7853 (ix86_adjust_stack_and_probe): Likewise.
7854 (output_adjust_stack_and_probe): Likewise.
7855 (ix86_emit_probe_stack_range): Likewise.
7856 (ix86_expand_prologue): Likewise.
7858 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7860 PR tree-optimization/82816
7861 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
7862 if the modes of the two types are the same.
7863 (convert_plusminus_to_widen): Likewise.
7865 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7867 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
7869 (usadv16qi): New define_expand.
7870 (usadv8hi): New define_expand.
7872 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
7875 * ipa-inline-transform.c (update_noncloned_frequencies): Always
7877 (inline_transform): Likewise.
7878 * predict.c (counts_to_freqs): Remove useless conditional.
7879 * profile-count.h (profile_count::apply_scale): Move sanity check.
7880 * tree-inline.c (copy_bb): Always scale.
7881 (copy_cfg_body): Likewise.
7883 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
7886 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
7888 (*cmp_ite0): Add enabled_for_depr_it attribute.
7889 (*cmp_ite1): Likewise.
7891 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7893 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
7896 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7898 * tree-vrp.c (vrp_int_const_binop): Return true on success and
7899 return the value by pointer.
7900 (extract_range_from_multiplicative_op_1): Update accordingly.
7901 Return as soon as an operation fails.
7903 2017-11-05 Tom de Vries <tom@codesourcery.com>
7906 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
7907 (DEF_SANITIZER_BUILTIN): ... here.
7908 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
7909 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
7911 2017-11-05 Tom de Vries <tom@codesourcery.com>
7914 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
7916 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
7917 ASM_OUTPUT_BEFORE_CASE_LABEL call.
7918 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
7920 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7921 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7922 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7924 2017-11-05 Tom de Vries <tom@codesourcery.com>
7927 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
7930 2017-11-04 Michael Clark <michaeljclark@mac.com>
7932 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
7933 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
7935 (*addsi3_extended): Likewise.
7936 (*addsi3_extended2): Likewise.
7937 (<optab>si3): Likewise.
7938 (<optab>di3): Likewise.
7939 (<optab><mode>3): Likewise.
7940 (<*optabe>si3_internal): Likewise.
7941 (zero_extendqi<SUPERQI:mode>2): Likewise.
7942 (*add<mode>hi3): Likewise.
7943 (*xor<mode>hi3): Likewise.
7944 (<optab>di3): Likewise.
7945 (*<optab>si3_extend): Likewise.
7946 (*sge<u>_<X:mode><GPR:mode>): Likewise.
7947 (*slt<u>_<X:mode><GPR:mode>): Likewise.
7948 (*sle<u>_<X:mode><GPR:mode>): Likewise.
7950 2017-11-04 Andrew Waterman <andrew@sifive.com>
7952 * config/riscv/riscv.c (riscv_option_override): Conditionally set
7953 TARGET_STRICT_ALIGN based upon -mtune argument.
7955 2017-11-04 Andrew Waterman <andrew@sifive.com>
7957 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
7959 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
7961 * config/i386/i386.c (choose_basereg): Use optional scratch
7962 register and add assertion.
7963 (x86_emit_outlined_ms2sysv_save): Use scratch register when
7964 needed, and don't allocate stack.
7965 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
7966 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
7967 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
7969 2017-11-03 Jeff Law <law@redhat.com>
7971 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
7972 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
7973 to probe at the start of a noreturn function.
7975 2017-11-03 Jakub Jelinek <jakub@redhat.com>
7977 PR tree-optimization/78821
7978 * gimple-ssa-store-merging.c: Update the file comment.
7979 (MAX_STORE_ALIAS_CHECKS): Define.
7980 (struct store_operand_info): New type.
7981 (store_operand_info::store_operand_info): New constructor.
7982 (struct store_immediate_info): Add rhs_code and ops data members.
7983 (store_immediate_info::store_immediate_info): Add rhscode, op0r
7984 and op1r arguments to the ctor, initialize corresponding data members.
7985 (struct merged_store_group): Add load_align_base and load_align
7987 (merged_store_group::merged_store_group): Initialize them.
7988 (merged_store_group::do_merge): Update them.
7989 (merged_store_group::apply_stores): Pick the constant for
7990 encode_tree_to_bitpos from one of the two operands, or skip
7991 encode_tree_to_bitpos if neither operand is a constant.
7992 (class pass_store_merging): Add process_store method decl. Remove
7993 bool argument from terminate_all_aliasing_chains method decl.
7994 (pass_store_merging::terminate_all_aliasing_chains): Remove
7995 var_offset_p argument and corresponding handling.
7996 (stmts_may_clobber_ref_p): New function.
7997 (compatible_load_p): New function.
7998 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
7999 if there is overlap and rhs_code is not INTEGER_CST. For
8000 non-overlapping stores terminate group if rhs is not mergeable.
8001 (get_alias_type_for_stmts): Change first argument from
8002 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
8003 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
8004 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
8006 (get_location_for_stmts): Change first argument from
8007 auto_vec<gimple *> & to vec<gimple *> &.
8008 (struct split_store): Remove orig_stmts data member, add orig_stores.
8009 (split_store::split_store): Create orig_stores rather than orig_stmts.
8010 (find_constituent_stmts): Renamed to ...
8011 (find_constituent_stores): ... this. Change second argument from
8012 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
8013 to info structures rather than the statements.
8014 (split_group): Rename ALLOW_UNALIGNED argument to
8015 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
8016 it. Adjust find_constituent_stores caller.
8017 (imm_store_chain_info::output_merged_store): Handle rhs_code other
8018 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
8019 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
8020 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
8021 (mem_valid_for_store_merging): New function.
8022 (handled_load): New function.
8023 (pass_store_merging::process_store): New method.
8024 (pass_store_merging::execute): Use process_store method. Adjust
8025 terminate_all_aliasing_chains caller.
8027 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8029 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
8030 Return true for more constants, symbols and label references.
8031 (aarch64_valid_floating_const): Remove unused function.
8033 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8036 * config/aarch64/aarch64.c (aarch64_layout_frame):
8037 Undo forcing of LR at bottom of frame.
8039 2017-11-03 Jeff Law <law@redhat.com>
8042 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
8043 for int_registers_saved.
8045 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
8046 extracted from tree-ssa-dom.c.
8047 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
8048 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
8049 (record_equivalences_from_incoming_edge): Add additional argument
8050 to single_pred_edge_ignoring_loop_edges call.
8051 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
8052 (uncprop_dom_walker::before_dom_children): Add additional argument
8053 to single_pred_edge_ignoring_loop_edges call.
8054 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
8055 single_pred_edge_ignoring_loop_edges rather than open coding.
8056 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
8058 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
8060 * match.pd (-(-A)): Rewrite.
8062 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8064 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
8065 (rs6000_emit_int_cmove): New declaration.
8066 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
8067 (rs6000_emit_sISEL): Delete.
8068 (rs6000_emit_int_cmove): Make non-static.
8069 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
8070 instead of rs6000_emit_sISEL.
8072 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
8074 * asan.c (create_cond_insert_point): Maintain profile.
8075 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
8077 * basic-block.h (struct basic_block_def): Remove frequency.
8078 (EDGE_FREQUENCY): Use to_frequency
8079 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
8081 (find_traces): Update to use to_frequency.
8082 (find_traces_1_round): Likewise; use only IPA counts.
8083 (bb_to_key): Likewise.
8084 (connect_traces): Use IPA counts only.
8085 (copy_bb_p): Update to use to_frequency.
8086 (fix_up_crossing_landing_pad): Likewise.
8087 (sanitize_hot_paths): Likewise.
8088 * bt-load.c (basic_block_freq): Likewise.
8089 * cfg.c (init_flow): Set count_max to uninitialized.
8090 (check_bb_profile): Remove frequencies; check counts.
8091 (dump_bb_info): Do not dump frequencies.
8092 (update_bb_profile_for_threading): Update counts only.
8093 (scale_bbs_frequencies_int): Likewise.
8094 (MAX_SAFE_MULTIPLIER): Remove.
8095 (scale_bbs_frequencies_gcov_type): Update counts only.
8096 (scale_bbs_frequencies_profile_count): Update counts only.
8097 (scale_bbs_frequencies): Update counts only.
8098 * cfg.h (struct control_flow_graph): Add count-max.
8099 (update_bb_profile_for_threading): Update prototype.
8100 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
8101 (find_many_sub_basic_blocks): Likewise.
8102 * cfgcleanup.c (try_forward_edges): Likewise.
8103 (try_crossjump_to_edge): Likewise.
8104 * cfgexpand.c (expand_gimple_cond): Likewise.
8105 (expand_gimple_tailcall): Likewise.
8106 (construct_init_block): Likewise.
8107 (construct_exit_block): Likewise.
8108 * cfghooks.c (verify_flow_info): Check consistency of counts.
8109 (dump_bb_for_graph): Do not dump frequencies.
8110 (split_block_1): Do not update frequencies.
8111 (split_edge): Do not update frequencies.
8112 (make_forwarder_block): Do not update frequencies.
8113 (duplicate_block): Do not update frequencies.
8114 (account_profile_record): Do not update frequencies.
8115 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
8116 for global heuristics.
8117 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
8118 (expected_loop_iterations_unbounded): Use counts only.
8119 * cfgloopmanip.c (scale_loop_profile): Simplify.
8120 (create_empty_loop_on_edge): Simplify
8122 (duplicate_loop_to_header_edge): Simplify
8123 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
8124 (update_br_prob_note): Take care of removing note when profile
8126 (relink_block_chain): Do not dump frequency.
8127 (rtl_account_profile_record): Use to_frequency.
8128 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
8129 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
8130 (cgraph_update_edges_for_call_stmt_node): Likewise.
8131 (cgraph_edge::verify_count_and_frequency): Update.
8132 (cgraph_node::verify_node): Temporarily disable frequency verification.
8133 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
8134 to_cgraph_frequency.
8135 (cgraph_edge::rebuild_edges): Convert to ipa counts.
8136 * cgraphunit.c (init_lowered_empty_function): Do not initialize
8138 (cgraph_node::expand_thunk): Update profile.
8139 * except.c (dw2_build_landing_pads): Do not update frequency.
8140 * final.c (compute_alignments): Use to_frequency.
8141 (dump_basic_block_info): Do not dump frequency.
8142 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
8143 (dump_gimple_bb_header): Do not dump frequency.
8144 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
8146 * gimple-streamer-in.c (input_bb): Do not stream frequency.
8147 * gimple-streamer-out.c (output_bb): Do not stream frequency.
8148 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
8149 (init_before_recovery): Do not update frequency.
8150 (sched_create_recovery_edges): Do not update frequency.
8151 * hsa-gen.c (convert_switch_statements): Do not update frequency.
8152 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
8153 (ipa_cp_c_finalize): Set max_count to uninitialized.
8154 * ipa-fnsummary.c (get_minimal_bb): Use counts.
8155 (param_change_prob): Use counts.
8156 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
8158 * ipa-split.c (consider_split): Use to_frequency.
8159 (split_function): Use to_frequency.
8160 * ira-build.c (loop_compare_func): Likewise.
8161 (mark_loops_for_removal): Likewise.
8162 (mark_all_loops_for_removal): Likewise.
8163 * loop-doloop.c (doloop_modify): Do not update frequency.
8164 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
8166 * lto-streamer-in.c (input_function): Update count_max.
8167 * omp-expand.c (expand_omp_taskreg): Update count_max.
8168 * omp-simd-clone.c (simd_clone_adjust): Update profile.
8169 * predict.c (maybe_hot_frequency_p): Use to_frequency.
8170 (maybe_hot_count_p): Use ipa counts only.
8171 (maybe_hot_bb_p): Simplify.
8172 (maybe_hot_edge_p): Simplify.
8173 (probably_never_executed): Do not take frequency argument.
8174 (probably_never_executed_bb_p): Do not pass frequency.
8175 (probably_never_executed_edge_p): Likewise.
8176 (combine_predictions_for_bb): Check that profile is nonzero.
8177 (propagate_freq): Do not set frequency.
8178 (drop_profile): Simplify.
8179 (counts_to_freqs): Simplify.
8180 (expensive_function_p): Use to_frequency.
8181 (propagate_unlikely_bbs_forward): Simplify.
8182 (determine_unlikely_bbs): Simplify.
8183 (estimate_bb_frequencies): Add hack to silence graphite issues.
8184 (compute_function_frequency): Use ipa counts.
8185 (pass_profile::execute): Update.
8186 (rebuild_frequencies): Use counts only.
8187 (force_edge_cold): Use counts only.
8188 * profile-count.c (profile_count::dump): Dump new count types.
8189 (profile_count::differs_from_p): Check compatiblity.
8190 (profile_count::to_frequency): New function.
8191 (profile_count::to_cgraph_frequency): New function.
8192 * profile-count.h (struct function): Declare.
8193 (enum profile_quality): Add profile_guessed_local and
8194 profile_guessed_global0.
8195 (class profile_proability): Decrease number of bits to 29;
8196 update from_reg_br_prob_note and to_reg_br_prob_note.
8197 (class profile_count: Update comment; decrease number of bits
8198 to 61. Check compatibility.
8199 (profile_count::compatible_p): New private member function.
8200 (profile_count::ipa_p): New member function.
8201 (profile_count::operator<): Handle global zero correctly.
8202 (profile_count::operator>): Handle global zero correctly.
8203 (profile_count::operator<=): Handle global zero correctly.
8204 (profile_count::operator>=): Handle global zero correctly.
8205 (profile_count::nonzero_p): New member function.
8206 (profile_count::force_nonzero): New member function.
8207 (profile_count::max): New member function.
8208 (profile_count::apply_scale): Handle IPA scalling.
8209 (profile_count::guessed_local): New member function.
8210 (profile_count::global0): New member function.
8211 (profile_count::ipa): New member function.
8212 (profile_count::to_frequency): Declare.
8213 (profile_count::to_cgraph_frequency): Declare.
8214 * profile.c (OVERLAP_BASE): Delete.
8215 (compute_frequency_overlap): Delete.
8216 (compute_branch_probabilities): Do not use compute_frequency_overlap.
8217 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
8218 * sched-ebb.c (rank): Use counts only.
8219 * shrink-wrap.c (handle_simple_exit): Use counts only.
8220 (try_shrink_wrapping): Use counts only.
8221 (place_prologue_for_one_component): Use counts only.
8222 * tracer.c (find_best_predecessor): Use to_frequency.
8223 (find_trace): Use to_frequency.
8224 (tail_duplicate): Use to_frequency.
8225 * trans-mem.c (expand_transaction): Do not update frequency.
8226 * tree-call-cdce.c: Do not update frequency.
8227 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
8228 (gimple_merge_blocks): Likewise.
8229 (gimple_split_edge): Likewise.
8230 (gimple_duplicate_sese_region): Likewise.
8231 (gimple_duplicate_sese_tail): Likewise.
8232 (move_sese_region_to_fn): Likewise.
8233 (gimple_account_profile_record): Likewise.
8234 (insert_cond_bb): Likewise.
8235 * tree-complex.c (expand_complex_div_wide): Likewise.
8236 * tree-eh.c (lower_resx): Update profile.
8237 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
8239 (initialize_cfun): Do not initialize frequencies
8240 (freqs_to_counts): Delete.
8241 (copy_cfg_body): Ignore count parameter.
8242 (copy_body): Update.
8243 (expand_call_inline): Update count_max.
8244 (optimize_inline_calls): Update count_max.
8245 (tree_function_versioning): Update count_max.
8246 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
8247 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
8249 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
8250 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
8251 (try_peel_loop): Likewise.
8252 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
8254 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
8255 (tree_transform_and_unroll_loop): Do not use frequencies
8256 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
8257 Use reliable prediction only.
8258 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
8259 * tree-ssa-sink.c (select_best_block): Use to_frequency.
8260 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
8261 probability scaling.
8262 * tree-ssa-threadupdate.c (create_block_for_threading): Do
8263 not update frequency
8264 (any_remaining_duplicated_blocks): Likewise.
8265 (update_profile): Likewise.
8266 (estimated_freqs_path): Delete.
8267 (freqs_to_counts_path): Delete.
8268 (clear_counts_path): Delete.
8269 (ssa_fix_duplicate_block_edges): Likewise.
8270 (duplicate_thread_path): Likewise.
8271 * tree-switch-conversion.c (gen_inbound_check): Use counts.
8272 * tree-tailcall.c (decrease_profile): Do not update frequency.
8273 (eliminate_tail_call): Likewise.
8274 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
8275 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
8276 (optimize_mask_stores): Likewise.
8277 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8278 * ubsan.c (ubsan_expand_null_ifn): Update profile.
8279 (ubsan_expand_ptr_ifn): Update profile.
8280 * value-prof.c (gimple_ic): Simplify.
8281 * value-prof.h (gimple_ic): Update prototype.
8282 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
8283 * ipa-inline.c (compute_uninlined_call_time): Be sure that
8285 (want_inline_self_recursive_call_p): Likewise.
8286 (resolve_noninline_speculation): Only cummulate defined counts.
8287 (inline_small_functions): Use nonzero_p.
8288 (ipa_inline): Do not access freed node.
8290 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8292 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8293 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
8295 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
8297 * config/riscv/riscv.c (riscv_legitimize_move): Handle
8298 non-legitimate address.
8300 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8302 * config/rs6000/rs6000.md (*lt0_disi): Delete.
8303 (*lt0_<mode>di, *lt0_<mode>si): New.
8305 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8307 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
8308 TARGET_PAIRED_FLOAT.
8310 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
8311 Jim Wilson <jim.wilson@linaro.org>
8313 * config/aarch64/aarch64-cores.def (saphira): New CPU.
8314 * config/aarch64/aarch64-tune.md: Regenerated.
8315 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
8316 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
8318 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
8320 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
8321 (arc_expand_prologue): Restore blink for millicode.
8322 * config/arc/linux.h (LINK_EH_SPEC): Defined.
8324 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
8327 * config/i386/i386.c (ix86_vector_duplicate_value): Use
8328 gen_vec_duplicate after forcing the scalar into a register.
8330 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
8332 * combine (try_combine): Print the insns input to try_combine to the
8335 2017-11-02 Steve Ellcey <sellcey@cavium.com>
8338 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
8339 Remove second argument from aarch64_process_target_attr call.
8340 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
8342 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
8344 (aarch64_handle_attr_arch): Remove second argument.
8345 (aarch64_handle_attr_cpu): Ditto.
8346 (aarch64_handle_attr_tune): Ditto.
8347 (aarch64_handle_attr_isa_flags): Ditto.
8348 (aarch64_process_one_target_attr): Ditto.
8349 (aarch64_process_target_attr): Ditto.
8350 (aarch64_option_valid_attribute_p): Remove second argument.
8351 on aarch64_process_target_attr call.
8353 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8355 * diagnostic.c: Include "selftest-diagnostic.h".
8356 (selftest::assert_location_text): New function.
8357 (selftest::test_diagnostic_get_location_text): New function.
8358 (selftest::diagnostic_c_tests): Call it.
8360 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8362 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
8363 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
8364 (class selftest::test_diagnostic_context): Move to...
8365 * selftest-diagnostic.c: New file.
8366 * selftest-diagnostic.h: New file.
8368 2017-11-02 James Bowman <james.bowman@ftdichip.com>
8370 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
8371 offset range for FT32B.
8372 * config/ft32/ft32.h: option "mcompress" enables relaxation.
8373 * config/ft32/ft32.md: Add TARGET_NOPM.
8374 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
8375 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
8377 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
8379 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
8381 2017-11-02 Jeff Law <law@redhat.com>
8383 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
8384 virtual keyword on FINAL OVERRIDE members.
8386 * tree-ssa-propagate.h (ssa_propagation_engine): Group
8387 virtuals together. Add virtual destructor.
8388 (substitute_and_fold_engine): Similarly.
8390 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
8392 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
8394 2017-11-02 Richard Biener <rguenther@suse.de>
8396 PR tree-optimization/82795
8397 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
8399 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8401 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
8403 (gcc_GAS_CHECK_FEATURE): Remove.
8404 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
8405 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
8406 * configure: Regenerate.
8408 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
8410 * config/arc/arc.c (hwloop_optimize): Account for empty
8413 2017-11-02 Richard Biener <rguenther@suse.de>
8416 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
8417 Truncate ARRAY_REF index and element size.
8419 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8421 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
8423 2017-11-01 Jeff Law <law@redhat.com>
8425 * tree-ssa-ccp.c (ccp_folder): New class derived from
8426 substitute_and_fold_engine.
8427 (ccp_folder::get_value): New member function.
8428 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
8429 (ccp_fold_stmt): Remove prototype.
8430 (ccp_finalize): Call substitute_and_fold from the ccp_class.
8431 * tree-ssa-copy.c (copy_folder): New class derived from
8432 substitute_and_fold_engine.
8433 (copy_folder::get_value): Renamed from get_value.
8434 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
8435 * tree-vrp.c (vrp_folder): New class derived from
8436 substitute_and_fold_engine.
8437 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
8438 (vrp_folder::get_value): New member function.
8439 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
8440 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
8441 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
8442 provide a class interface to folder/substitute routines.
8443 (ssa_prop_fold_stmt_fn): Remove typedef.
8444 (ssa_prop_get_value_fn): Likewise.
8445 (subsitute_and_fold): Remove prototype.
8446 (replace_uses_in): Likewise.
8447 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
8448 Renamed from replace_uses_in. Call the virtual member function
8449 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
8450 (substitute_and_fold_dom_walker): Remove initialization of
8451 data member entries for calbacks. Add substitute_and_fold_engine
8452 member and initialize it.
8453 (substitute_and_fold_dom_walker::before_dom_children0: Use the
8454 member functions for get_value, replace_phi_args_in c
8455 replace_uses_in, and fold_stmt calls.
8456 (substitute_and_fold_engine::substitute_and_fold): Renamed from
8457 substitute_and_fold. Remove assert. Update ctor call.
8459 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
8460 (ssa_prop_visit_phi_fn): Likewise.
8461 (class ssa_propagation_engine): New class to provide an interface
8463 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
8465 (ssa_prop_visit_phi): Likewise.
8466 (ssa_propagation_engine::simulate_stmt): Moved into class.
8467 Call visit_phi/visit_stmt from the class rather than via
8468 file scoped static variables.
8469 (ssa_propagation_engine::simulate_block): Moved into class.
8470 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
8471 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
8472 set file scoped statics for the visit_stmt/visit_phi callbacks.
8473 * tree-complex.c (complex_propagate): New class derived from
8474 ssa_propagation_engine.
8475 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
8476 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
8477 (tree_lower_complex): Call ssa_propagate via the complex_propagate
8479 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
8480 ssa_propagation_engine.
8481 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
8482 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
8483 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
8484 * tree-ssa-copy.c (copy_prop): New class derived from
8485 ssa_propagation_engine.
8486 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
8487 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
8488 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
8489 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
8490 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
8491 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
8492 (execute_vrp): Call ssa_propagate from the vrp_prop class.
8494 2017-11-01 Jakub Jelinek <jakub@redhat.com>
8496 PR rtl-optimization/82778
8497 PR rtl-optimization/82597
8498 * compare-elim.c (struct comparison): Add in_a_setter field.
8499 (find_comparison_dom_walker::before_dom_children): Remove killed
8500 bitmap and df_simulate_find_defs call, instead walk the defs.
8501 Compute last_setter and initialize in_a_setter. Merge definitions
8502 with first initialization for a few variables.
8503 (try_validate_parallel): Use insn_invalid_p instead of
8504 recog_memoized. Return insn rather than just the pattern.
8505 (try_merge_compare): Fix up comment. Don't uselessly test if
8506 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
8508 (execute_compare_elim_after_reload): Remove df_chain_add_problem
8511 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8512 Alan Hayward <alan.hayward@arm.com>
8513 David Sherwood <david.sherwood@arm.com>
8515 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
8516 aarch64_hard_regno_nregs to get the number of registers
8519 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8520 Alan Hayward <alan.hayward@arm.com>
8521 David Sherwood <david.sherwood@arm.com>
8523 * config/aarch64/constraints.md (Upl): Rename to...
8525 * config/aarch64/aarch64.md
8526 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
8529 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8530 Alan Hayward <alan.hayward@arm.com>
8531 David Sherwood <david.sherwood@arm.com>
8533 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
8534 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
8537 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8538 Alan Hayward <alan.hayward@arm.com>
8539 David Sherwood <david.sherwood@arm.com>
8541 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
8542 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
8543 (aarch64_evpc_dup): Generate rtl direcly, rather than using
8545 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
8546 of a single element.
8547 * config/aarch64/iterators.md: Add a comment above the permute
8548 unspecs to say that they are generated directly by
8549 aarch64_expand_vec_perm_const.
8550 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
8552 2017-11-01 Nathan Sidwell <nathan@acm.org>
8554 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
8556 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8558 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
8559 and medany code models, and describe what they do.
8561 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8563 Revert accidental duplicate:
8565 * combine.c (can_change_dest_mode): Reject changes in
8566 REGMODE_NATURAL_SIZE.
8568 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
8570 PR rtl-optimization/64682
8571 PR rtl-optimization/69567
8572 PR rtl-optimization/69737
8573 PR rtl-optimization/82683
8574 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
8575 register mentioned in the note, drop the note, unless it came from I3,
8576 in which case it should go to I3 again.
8578 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8580 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
8581 and return false if not.
8582 (clear_bytes_written_by, live_bytes_read): Update accordingly.
8584 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8586 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
8587 range is known to be empty.
8589 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8590 Alan Hayward <alan.hayward@arm.com>
8591 David Sherwood <david.sherwood@arm.com>
8593 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
8594 and CONST_VECTOR_NUNITS instead of computing the number of units from
8595 the byte sizes of the vector and element.
8596 (simplify_binary_operation_1): Likewise.
8597 (simplify_const_binary_operation): Likewise.
8598 (simplify_ternary_operation): Likewise.
8600 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8601 Alan Hayward <alan.hayward@arm.com>
8602 David Sherwood <david.sherwood@arm.com>
8604 * var-tracking.c (INT_MEM_OFFSET): Replace with...
8605 (int_mem_offset): ...this new function.
8606 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
8607 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
8608 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
8611 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8612 Alan Hayward <alan.hayward@arm.com>
8613 David Sherwood <david.sherwood@arm.com>
8615 * lower-subreg.c (interesting_mode_p): New function.
8616 (compute_costs, find_decomposable_subregs, decompose_register)
8617 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
8618 (resolve_clobber, dump_choices): Use it.
8620 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8621 Alan Hayward <alan.hayward@arm.com>
8622 David Sherwood <david.sherwood@arm.com>
8624 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
8626 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8627 Alan Hayward <alan.hayward@arm.com>
8628 David Sherwood <david.sherwood@arm.com>
8630 * alias.c (find_base_value, find_base_term): Only process integer
8631 truncations. Check the precision rather than the size.
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 * machmode.h (is_narrower_int_mode): New function
8638 * optabs.c (expand_float, expand_fix): Use it.
8639 * dwarf2out.c (rotate_loc_descriptor): Likewise.
8641 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8642 Alan Hayward <alan.hayward@arm.com>
8643 David Sherwood <david.sherwood@arm.com>
8645 * rtl.h (narrower_subreg_mode): New function.
8646 * ira-color.c (update_costs_from_allocno): Use it.
8648 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8649 Alan Hayward <alan.hayward@arm.com>
8650 David Sherwood <david.sherwood@arm.com>
8652 * optabs-query.h (convert_optab_p): New function, split out from...
8653 (convert_optab_handler): ...here.
8654 (widening_optab_handler): Delete.
8655 (find_widening_optab_handler): Remove permit_non_widening parameter.
8656 (find_widening_optab_handler_and_mode): Likewise. Provide an
8657 override that operates on mode class wrappers.
8658 * optabs-query.c (widening_optab_handler): Delete.
8659 (find_widening_optab_handler_and_mode): Remove permit_non_widening
8660 parameter. Assert that the two modes are the same class and that
8661 the "from" mode is narrower than the "to" mode. Use
8662 convert_optab_handler instead of widening_optab_handler.
8663 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
8664 instead of widening_optab_handler.
8665 * expr.c (expand_expr_real_2): Update calls to
8666 find_widening_optab_handler.
8667 * optabs.c (expand_widen_pattern_expr): Likewise.
8668 (expand_binop_directly): Take the insn_code as a parameter.
8669 (expand_binop): Only call find_widening_optab_handler for
8670 conversion optabs; use optab_handler otherwise. Update calls
8671 to find_widening_optab_handler and expand_binop_directly.
8672 Use convert_optab_handler instead of widening_optab_handler.
8673 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
8674 find_widening_optab_handler and use scalar_mode rather than
8676 (convert_plusminus_to_widen): Likewise.
8678 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8679 Alan Hayward <alan.hayward@arm.com>
8680 David Sherwood <david.sherwood@arm.com>
8682 * machmode.h (fixed_size_mode): New class.
8683 * rtl.h (get_pool_mode): Return fixed_size_mode.
8684 * gengtype.c (main): Add fixed_size_mode.
8685 * target.def (get_raw_result_mode): Return a fixed_size_mode.
8686 (get_raw_arg_mode): Likewise.
8687 * doc/tm.texi: Regenerate.
8688 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
8689 * targhooks.c (default_get_reg_raw_mode): Likewise.
8690 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
8691 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
8692 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
8693 (msp430_get_raw_result_mode): Likewise.
8694 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
8695 * dbxout.c (dbxout_parms): Require fixed-size modes.
8696 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
8697 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8698 * omp-low.c (lower_oacc_reductions): Likewise.
8699 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
8700 (simplify_subreg): Update accordingly.
8701 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
8702 (force_const_mem): Update accordingly. Return NULL_RTX for modes
8703 that aren't fixed-size.
8704 (get_pool_mode): Return a fixed_size_mode.
8705 (output_constant_pool_2): Take a fixed_size_mode.
8707 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8708 Alan Hayward <alan.hayward@arm.com>
8709 David Sherwood <david.sherwood@arm.com>
8711 * doc/rtl.texi (vec_series): Document.
8712 (const): Say that the operand can be a vec_series.
8713 * rtl.def (VEC_SERIES): New rtx code.
8714 * rtl.h (const_vec_series_p_1): Declare.
8715 (const_vec_series_p): New function.
8716 * emit-rtl.h (gen_const_vec_series): Declare.
8717 (gen_vec_series): Likewise.
8718 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
8719 (gen_vec_series): Likewise.
8720 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
8721 * simplify-rtx.c (simplify_unary_operation): Handle negations
8723 (simplify_binary_operation_series): New function.
8724 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
8725 (test_vector_ops_series): New function.
8726 (test_vector_ops): Call it.
8727 * config/powerpcspe/altivec.md (altivec_lvsl): Use
8728 gen_const_vec_series.
8729 (altivec_lvsr): Likewise.
8730 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
8732 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8733 Alan Hayward <alan.hayward@arm.com>
8734 David Sherwood <david.sherwood@arm.com>
8736 * doc/rtl.texi (const): Update description of address constants.
8737 Say that vector constants are allowed too.
8738 * common.md (E, F): Use CONSTANT_P instead of checking for
8740 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
8741 checking for CONST_VECTOR.
8742 * expmed.c (make_tree): Use build_vector_from_val for a CONST
8744 * expr.c (expand_expr_real_2): Check for vector modes instead
8745 of checking for CONST_VECTOR.
8746 * rtl.h (const_vec_p): New function.
8747 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
8748 (unwrap_const_vec_duplicate): Handle them here too.
8750 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8751 David Malcolm <dmalcolm@redhat.com>
8752 Alan Hayward <alan.hayward@arm.com>
8753 David Sherwood <david.sherwood@arm.com>
8755 * rtl.h (vec_duplicate_p): New function.
8756 * selftest-rtl.c (assert_rtx_eq_at): New function.
8757 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
8758 (assert_rtx_eq_at): Declare.
8759 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
8760 * selftest-run-tests.c (selftest::run_tests): Call it.
8761 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
8762 (simplify_unary_operation_1): Recursively handle vector duplicates.
8763 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
8765 (simplify_subreg): Handle subregs of vector duplicates.
8766 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
8767 (selftest::simplify_rtx_c_tests): New functions.
8769 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8770 Alan Hayward <alan.hayward@arm.com>
8771 David Sherwood <david.sherwood@arm.com>
8773 * emit-rtl.h (gen_const_vec_duplicate): Declare.
8774 (gen_vec_duplicate): Likewise.
8775 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
8777 (gen_const_vector): ...here.
8778 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
8779 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
8780 whose elements are all equal.
8781 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
8782 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
8783 (simplify_relational_operation): Likewise.
8784 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
8786 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
8787 (aarch64_expand_vector_init): Likewise.
8788 * config/arm/arm.c (neon_vdup_constant): Likewise.
8789 (neon_expand_vector_init): Likewise.
8790 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
8791 (arm_block_set_unaligned_vect): Likewise.
8792 (arm_block_set_aligned_vect): Likewise.
8793 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
8794 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
8795 (expand_vec_perm_even_odd_pack): Likewise.
8796 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
8797 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
8798 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
8799 gen_const_vec_duplicate.
8800 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
8801 * config/mips/mips.c (mips_gen_const_int_vector): Use
8802 gen_const_vec_duplicate.
8803 (mips_expand_vector_init): Use CONST0_RTX.
8804 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
8805 (define_split): Use gen_const_vec_duplicate.
8806 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
8807 (define_split): Use gen_const_vec_duplicate.
8808 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
8809 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
8810 * config/spu/spu.c (spu_const): Likewise.
8812 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8813 Alan Hayward <alan.hayward@arm.com>
8814 David Sherwood <david.sherwood@arm.com>
8816 * combine.c (can_change_dest_mode): Reject changes in
8817 REGMODE_NATURAL_SIZE.
8819 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
8821 * configure.ac (--enable-libssp): New.
8822 (gcc_cv_libc_provides_ssp): Check for explicit setting before
8823 trying to determine target-specific default. Adjust indentation.
8824 * configure: Regenerated.
8825 * doc/install.texi (Configuration): Expand --disable-libssp
8828 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
8830 config/i386/i386.c (ix86_expand_epilogue): Correct stack
8833 2017-10-31 Martin Jambor <mjambor@suse.cz>
8836 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
8838 2017-10-31 David Malcolm <dmalcolm@redhat.com>
8840 * auto-profile.c (autofdo_source_profile::read): Use
8841 UNKNOWN_LOCATION rather than 0.
8842 * diagnostic-core.h (warning_at_rich_loc): Rename to...
8843 (warning_at): ...this overload.
8844 (warning_at_rich_loc_n): Rename to...
8845 (warning_n): ...this overload.
8846 (error_at_rich_loc): Rename to...
8847 (error_at): ...this overload.
8848 (pedwarn_at_rich_loc): Rename to...
8849 (pedwarn): ...this overload.
8850 (permerror_at_rich_loc): Rename to...
8851 (permerror): ...this overload.
8852 (inform_at_rich_loc): Rename to...
8853 (inform): ...this overload.
8854 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
8855 (diagnostic_n_impl_richloc): Rename to...
8856 (diagnostic_n_impl): ...this rich_location *-based decl.
8857 (inform_at_rich_loc): Rename to...
8858 (inform): ...this, and add an assertion.
8859 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
8860 (warning_at_rich_loc): Rename to...
8861 (warning_at): ...this, and add an assertion.
8862 (warning_at_rich_loc_n): Rename to...
8863 (warning_n): ...this, and add an assertion.
8864 (warning_n): Update location_t-based implementation for removal of
8865 location_t-based diagnostic_n_impl.
8866 (pedwarn_at_rich_loc): Rename to...
8867 (pedwarn): ...this, and add an assertion.
8868 (permerror_at_rich_loc): Rename to...
8869 (permerror): ...this, and add an assertion.
8870 (error_n): Update for removal of location_t-based diagnostic_n_impl.
8871 (error_at_rich_loc): Rename to...
8872 (error_at): ...this, and add an assertion.
8873 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
8874 (driver::do_spec_on_infiles): Likewise.
8875 * substring-locations.c (format_warning_va): Update for renaming
8876 of inform_at_rich_loc.
8878 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
8880 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
8881 _Float<N>X built-in functions so that the variant without the
8882 "__builtin_" prefix is only enabled for the GNU C and Objective C
8883 languages when they are in non-strict ANSI/ISO mode.
8884 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
8885 * target.def (floatn_builtin_p): Add a target hook to control
8886 whether _Float<N> and _Float<N>X built-in functions without the
8887 "__builtin_" prefix are enabled, and return true for C and
8888 Objective C in the default hook. Include langhooks.h in
8890 * targhooks.h (default_floatn_builtin_p): Likewise.
8891 * targhooks.c (default_floatn_builtin_p): Likewise.
8892 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
8893 floatn_builtin_p target hook.
8894 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
8896 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
8897 Eric Botcazou <ebotcazou@adacore.com>
8899 PR rtl-optimization/81803
8900 * lra-constraints.c (curr_insn_transform): Also reload the whole
8901 register for a strict subreg no wider than a word if this is for
8902 a WORD_REGISTER_OPERATIONS target.
8904 2017-10-31 Jason Merrill <jason@redhat.com>
8906 * gdbinit.in: Skip over inlines from timevar.h.
8908 2017-10-31 Martin Liska <mliska@suse.cz>
8910 * doc/gcov.texi: Document new option.
8911 * gcov.c (print_usage): Likewise print it.
8912 (process_args): Support the argument.
8913 (format_count): New function.
8914 (format_gcov): Use the function.
8916 2017-10-31 Martin Liska <mliska@suse.cz>
8918 * gcov.c (struct name_map): do not use typedef.
8919 Define operator== and operator<.
8920 (name_search): Remove.
8921 (name_sort): Remove.
8922 (main): Do not allocate names.
8923 (process_file): Add vertical space.
8924 (generate_results): Use std::find.
8925 (release_structures): Do not release memory.
8926 (find_source): Use std::find.
8928 2017-10-31 Martin Liska <mliska@suse.cz>
8930 * gcov.c (struct line_info): Remove it's typedef.
8931 (line_info::line_info): Add proper ctor.
8932 (line_info::has_block): Do not use a typedef.
8933 (struct source_info): Do not use typedef.
8934 (circuit): Likewise.
8935 (get_cycles_count): Likewise.
8936 (output_intermediate_file): Iterate via vector iterator.
8937 (add_line_counts): Use std::vector methods.
8938 (accumulate_line_counts): Likewise.
8939 (output_lines): Likewise.
8941 2017-10-31 Martin Liska <mliska@suse.cz>
8943 * gcov.c (struct source_info): Remove typedef.
8944 (source_info::source_info): Add proper ctor.
8945 (accumulate_line_counts): Use struct, not it's typedef.
8946 (output_gcov_file): Likewise.
8947 (output_lines): Likewise.
8948 (main): Do not allocate an array.
8949 (output_intermediate_file): Use size of vector container.
8950 (process_file): Resize the vector.
8951 (generate_results): Do not preallocate, use newly added vector
8953 (release_structures): Do not release sources.
8954 (find_source): Use vector methods.
8955 (add_line_counts): Do not use typedef.
8957 2017-10-31 Martin Liska <mliska@suse.cz>
8959 * doc/gcov.texi: Document that.
8960 * gcov.c (add_line_counts): Mark lines with a non-executed
8962 (output_line_beginning): Handle such lines.
8963 (output_lines): Pass new argument.
8964 (output_intermediate_file): Print it in intermediate format.
8966 2017-10-31 Martin Liska <mliska@suse.cz>
8968 * color-macros.h: New file.
8969 * diagnostic-color.c: Factor out color related to macros to
8971 * doc/gcov.texi: Document -k option.
8972 * gcov.c (INCLUDE_STRING): Include string.h.
8973 (print_usage): Add -k option.
8974 (process_args): Parse it.
8975 (pad_count_string): New function.
8976 (output_line_beginning): Likewise.
8977 (DEFAULT_LINE_START): New macro.
8978 (output_lines): Support color output.
8980 2017-10-31 Martin Liska <mliska@suse.cz>
8982 PR gcov-profile/82633
8983 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
8984 their interaction with GCOV infrastructure.
8985 * configure.ac: Add -fkeep-{inline,static}-functions to
8987 * configure: Regenerate.
8989 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
8992 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
8994 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
8997 * config/rs6000/rs6000.md (allocate_stack): Force update interval
8998 into a register if it does not fit into an immediate offset field.
9000 2017-10-31 Olivier Hainque <hainque@adacore.com>
9002 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
9004 2017-10-31 Julia Koval <julia.koval@intel.com>
9006 * config.gcc: Add gfniintrin.h.
9007 * config/i386/gfniintrin.h: New.
9008 * config/i386/i386-builtin-types.def
9009 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
9010 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
9011 __builtin_ia32_vgf2p8affineinvqb_v32qi,
9012 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
9013 __builtin_ia32_vgf2p8affineinvqb_v16qi,
9014 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
9015 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
9016 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
9017 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
9018 V64QI_FTYPE_V64QI_V64QI_INT): New types.
9019 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
9020 * config/i386/immintrin.h: Include gfniintrin.h.
9021 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
9023 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
9025 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
9027 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9029 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
9030 (arm_ashldi3_1bit): Remove pattern.
9031 (ashrdi3): Remove shift by 1 expansion.
9032 (arm_ashrdi3_1bit): Remove pattern.
9033 (lshrdi3): Remove shift by 1 expansion.
9034 (arm_lshrdi3_1bit): Remove pattern.
9035 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
9036 cost of ashldi3 by 1.
9037 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
9038 (<shift>di3_neon): Likewise.
9040 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9042 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
9043 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
9044 and (*aarch64_simd_mov<VQ:mode>).
9045 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
9046 pattern alternative.
9047 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
9048 attributes to match pattern alternative.
9050 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
9052 * config.gcc (powerpc*-*-*): Add emmintrin.h.
9053 * config/rs6000/emmintrin.h: New file.
9054 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
9056 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9058 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
9059 * (movdi_vfp_cortexa8): Remove pattern.
9061 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9063 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
9065 (Specific, alpha*-dec-osf5.1): Remove.
9066 (Specific, mips-sgi-irix5): Remove.
9067 (Specific, mips-sgi-irix6): Remove.
9069 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9072 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
9073 arguments to clear_bit_region_be.
9075 2017-10-30 Jim Wilson <wilson@tuliptree.org>
9077 * gimplify.c: Include memmodel.h.
9079 2017-10-30 Martin Jambor <mjambor@suse.cz>
9081 * omp-grid.c (grid_attempt_target_gridification): Also insert a
9082 condition whether loop should be executed at all.
9084 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
9086 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
9087 gimple folding of vec_madd() intrinsics.
9088 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
9089 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
9090 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
9092 2017-10-30 Richard Biener <rguenther@suse.de>
9094 PR tree-optimization/82762
9096 2017-10-23 Richard Biener <rguenther@suse.de>
9098 PR tree-optimization/82129
9100 2017-08-01 Richard Biener <rguenther@suse.de>
9102 PR tree-optimization/81181
9103 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9104 (compute_antic): ... end of iteration here.
9106 2017-10-30 Joseph Myers <joseph@codesourcery.com>
9108 * doc/invoke.texi (C Dialect Options): Document -std=c17,
9109 -std=iso9899:2017 and -std=gnu17.
9110 * doc/standards.texi (C Language): Document C17 support.
9111 * doc/cpp.texi (Overview): Mention -std=c17.
9112 (Standard Predefined Macros): Document C11 and C17 values of
9113 __STDC_VERSION__. Do not refer to C99 support as incomplete.
9114 * doc/extend.texi (Inline): Do not list individual options for
9115 standards newer than C99.
9116 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
9118 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
9121 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
9123 * asan.c (asan_finish_file): Align asan globals array by shadow
9126 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9129 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
9130 (struct store_immediate_info): Add bitregion_start and bitregion_end
9132 (store_immediate_info::store_immediate_info): Add brs and bre
9133 arguments and initialize bitregion_{start,end} from those.
9134 (struct merged_store_group): Add bitregion_start, bitregion_end,
9135 align_base and mask fields. Drop unnecessary struct keyword from
9136 struct store_immediate_info. Add do_merge method.
9137 (clear_bit_region_be): Use memset instead of loop storing zeros.
9138 (merged_store_group::do_merge): New method.
9139 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
9140 stores as long as the surrounding bitregions have no gaps.
9141 (merged_store_group::merge_overlapping): Use do_merge.
9142 (merged_store_group::apply_stores): Test that bitregion_{start,end}
9143 is byte aligned, rather than requiring that start and width are
9144 byte aligned. Drop unnecessary struct keyword from
9145 struct store_immediate_info. Allocate and populate also mask array.
9146 Make start of the arrays relative to bitregion_start rather than
9147 start and size them according to bitregion_{end,start} difference.
9148 (struct imm_store_chain_info): Drop unnecessary struct keyword from
9149 struct store_immediate_info.
9150 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
9151 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
9152 struct keyword from struct store_immediate_info.
9153 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
9154 between stores as long as the surrounding bitregions have no gaps.
9156 (struct split_store): Add orig non-static data member.
9157 (split_store::split_store): Initialize orig to false.
9158 (find_constituent_stmts): Return store_immediate_info *, non-NULL
9159 if there is exactly a single original stmt. Change stmts argument
9160 to pointer from reference, if NULL, don't push anything to it. Add
9161 first argument, use it to optimize skipping over orig stmts that
9162 are known to be before bitpos already. Simplify.
9163 (split_group): Return unsigned int count how many stores are or
9164 would be needed rather than a bool. Add allow_unaligned argument.
9165 Change split_stores argument from reference to pointer, if NULL,
9166 only do a dry run computing how many stores would be produced.
9167 Rewritten algorithm to use both alignment and misalign if
9168 !allow_unaligned and handle bitfield stores with gaps.
9169 (imm_store_chain_info::output_merged_store): Set start_byte_pos
9170 from bitregion_start instead of start. Compute allow_unaligned
9171 here, if true, do 2 split_group dry runs to compute which one
9172 produces fewer stores and prefer aligned if equal. Punt if
9173 new count is bigger or equal than original before emitting any
9174 statements, rather than during that. Remove no longer needed
9175 new_ssa_names tracking. Replace num_stmts with
9176 split_stores.length (). Use 32-bit stack allocated entries
9177 in split_stores auto_vec. Try to reuse original store lhs/rhs1
9178 if possible. Handle bitfields with gaps.
9179 (pass_store_merging::execute): Ignore bitsize == 0 stores.
9180 Compute bitregion_{start,end} for the stores and construct
9181 store_immediate_info with that. Formatting fixes.
9183 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
9186 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
9187 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
9189 2017-10-29 Jim Wilson <wilson@tuliptree.org>
9191 * gimplify.c: Include tm_p.h.
9193 * common.opt (gcoff): Re-add as ignored option.
9194 (gcoff1, gcoff2, gcoff3): Likewise.
9196 * Makefile.in (OBJS): Delete sdbout.o.
9197 (GTFILES): Delete $(srcdir)/sdbout.c.
9198 * debug.h: Delete sdb_debug_hooks.
9199 * final.c: Delete sdbout.h include.
9200 (final_scan_insn): Delete SDB_DEBUG check.
9201 (rest_of_clean_state): Likewise.
9202 * output.h: Delete sdb_begin_function_line.
9205 * toplev.c: Delete sdbout.h include.
9206 (process_options): Delete SDB_DEBUG check.
9207 * tree-core.h (tree_type_common): Delete pointer field of
9209 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
9210 TYPE_SYMTAB_POINTER.
9211 * tree.h (TYPE_SYMTAB_POINTER): Delete.
9212 (TYPE_SYMTAB_IS_POINTER): Delete.
9213 (TYPE_SYMTAB_IS_DIE): Renumber.
9214 * xcoffout.c: Refer to former sdbout.c file.
9215 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
9217 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
9218 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
9219 (Debugging Options): Delete -gcoff.
9220 (-gstabs): Delete SDB reference.
9222 (-gcoff@var{level}): Delete.
9223 * doc/passes.texi (Debugging information output): Delete SDB and
9224 sdbout.c references.
9225 * doc/tm.texi: Regenerate.
9226 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
9227 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
9229 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
9230 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
9231 (SDB_DEBUGGING_INFO): Delete.
9232 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
9233 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
9234 * target.def (output_source_filename): Delete COFF reference.
9236 * common.opt (gcoff): Delete.
9237 (gxcoff+): Update Negative chain.
9238 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
9240 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
9241 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
9242 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
9244 (expand_function_start): Change sdb reference to past tense.
9245 (expand_function_end): Change sdb reference to past tense.
9246 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
9247 * opts.c (debug_type_names): Delete coff entry.
9248 (common_handle_option): Delete OPT_gcoff case.
9249 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
9251 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
9252 * config/cris/cris.h: Delete SDB reference in comment.
9253 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
9254 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
9255 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
9256 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
9258 (ix86_expand_prologue): Likewise.
9259 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
9260 * config/ia64/ia64.h: Likewise.
9261 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
9262 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
9264 * config/mmix/mmix.h: Likewise.
9265 * config/nds32/nds32.c: Likewise.
9266 * config/stormy/storym16.h: Likewise.
9267 * config/visium/visium.h: Likewise.
9268 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
9270 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
9272 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
9273 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
9274 FRAME_POINTER_REGNUM point at high end of local var area.
9276 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
9278 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
9279 Move comment around. Do not reset best_edge for a copiable
9280 destination if the copy would cause a partition change.
9281 (better_edge_p): Remove redundant check.
9283 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9285 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
9287 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
9289 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
9290 for math functions that have _Float<N> and _Float<N>X variants.
9291 (mathfn_built_in_2): Add support for math functions that have
9292 _Float<N> and _Float<N>X variants.
9293 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
9294 (expand_builtin_mathfn_ternary): Add support for fma with
9295 _Float<N> and _Float<N>X variants.
9296 (expand_builtin): Likewise.
9297 (fold_builtin_3): Likewise.
9298 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
9299 create math function _Float<N> and _Float<N>X variants as external
9301 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
9302 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
9303 the __builtin_ prefix and if not strict ansi, without the prefix.
9304 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
9305 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
9306 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
9307 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
9308 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
9309 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
9310 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
9311 function signatures for fma _Float<N> and _Float<N>X variants.
9312 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
9313 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
9314 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
9315 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
9316 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
9317 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
9318 * gencfn-macros.c (print_case_cfn): Add support for math functions
9319 that have _Float<N> and _Float<N>X variants.
9320 (print_define_operator_list): Likewise.
9321 (fltfn_suffixes): Likewise.
9323 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
9324 for math functions that have _Float<N> and _Float<N>X variants.
9325 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
9326 and _Float<N>X variants.
9327 (COPYSIGN): Likewise.
9330 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
9331 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
9333 (integer_valued_read_call_p): Likewise.
9334 * fold-const-call.c (fold_const_call_ss): Likewise.
9335 (fold_const_call_sss): Add support for copysign, fmin, and fmax
9336 _Float<N> and _Float<N>X variants.
9337 (fold_const_call_ssss): Add support for fma _Float<N> and
9338 _Float<N>X variants.
9339 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
9340 support for copysign and fma _Float<N> and _Float<N>X variants.
9341 (backprop::process_builtin_call_use): Likewise.
9342 * tree-call-cdce.c (can_test_argument_range); Add support for
9343 sqrt _Float<N> and _Float<N>X variants.
9344 (edom_only_function): Likewise.
9345 (get_no_error_domain): Likewise.
9346 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
9347 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
9348 copysign _Float<N> and _Float<N>X variants.
9349 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
9350 handled by machine independent code.
9351 (FMAF128): Likewise.
9352 * doc/cpp.texi (Common Predefined Macros): Document defining
9353 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
9354 fma _Float<N> and _Float<N>X variants.
9356 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9359 * config/i386/i386-modes.def (CCFPU): Remove definition.
9360 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
9361 (ix86_cc_modes_compatible): Ditto.
9362 (ix86_expand_carry_flag_compare): Ditto.
9363 (ix86_expand_int_movcc): Ditto.
9364 (ix86_expand_int_addcc): Ditto.
9365 (ix86_reverse_condition): Ditto.
9366 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
9367 Return true/false for unordered/ordered fp comparisons.
9368 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
9369 (ix86_prepare_fp_compare_args): Update for rename.
9370 (ix86_expand_fp_compare): Update for rename. Generate unordered
9371 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
9372 (ix86_expand_sse_compare_and_jump): Ditto.
9373 * config/i386/predicates.md (fcmov_comparison_operator):
9374 Remove CCFPU mode handling.
9375 (ix86_comparison_operator): Ditto.
9376 (ix86_carry_flag_operator): Ditto.
9377 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
9378 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
9379 (*cmpu<mode>_cc_i387): Ditto.
9380 (FPCMP): Remove mode iterator.
9381 (unord): Remove mode attribute.
9382 (unord_subst): New define_subst transformation
9383 (unord): New define_subst attribute.
9385 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
9386 (*cmpi<unord>xf_i387): Ditto.
9387 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
9388 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
9389 using unord_subst transformation.
9390 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
9391 (round_saeonly): Also handle CCFP mode.
9392 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
9393 Remove UNSPEC_SAHF unspec handling.
9395 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9397 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
9399 2017-10-27 Jeff Law <law@redhat.com>
9401 * gimple-ssa-sprintf.c: Include domwalk.h.
9402 (class sprintf_dom_walker): New class, derived from dom_walker.
9403 (sprintf_dom_walker::before_dom_children): New function.
9404 (struct call_info): Moved into sprintf_dom_walker class
9405 (compute_formath_length, handle_gimple_call): Likewise.
9406 (sprintf_length::execute): Call the dominator walker rather
9407 than walking the statements.
9409 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
9410 gimple statement locations.
9411 (check_array_bounds): Corresponding changes. Get the statement's
9412 location directly from wi->stmt.
9414 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
9417 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
9419 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9421 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
9422 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
9424 2017-10-27 Jakub Jelinek <jakub@redhat.com>
9427 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
9428 * config/i386/i386.c (maybe_get_pool_constant): Removed.
9429 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
9430 maybe_get_pool_constant.
9431 * config/i386/predicates.md (zero_extended_scalar_load_operand):
9434 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9436 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
9437 2.26 caveat. Update gas and gld versions.
9438 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
9441 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9443 * cgraph.h (set_malloc_flag): Declare.
9444 * cgraph.c (set_malloc_flag_1): New function.
9445 (set_malloc_flag): Likewise.
9446 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
9447 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
9449 (read_ipa_call_summary): Add support for reading is_return_callee.
9450 (write_ipa_call_summary): Stream is_return_callee.
9451 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
9452 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
9453 ipa-prop.h, ipa-fnsummary.h.
9454 (pure_const_names): Change to static.
9455 (malloc_state_e): Define.
9456 (malloc_state_names): Define.
9457 (funct_state_d): Add field malloc_state.
9458 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
9459 (check_retval_uses): New function.
9460 (malloc_candidate_p): Likewise.
9461 (analyze_function): Add support for malloc attribute.
9462 (pure_const_write_summary): Stream malloc_state.
9463 (pure_const_read_summary): Add support for reading malloc_state.
9464 (dump_malloc_lattice): New function.
9465 (propagate_malloc): New function.
9466 (warn_function_malloc): New function.
9467 (ipa_pure_const::execute): Call propagate_malloc and
9468 ipa_free_fn_summary.
9469 (pass_local_pure_const::execute): Add support for malloc attribute.
9470 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
9471 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
9473 2017-10-27 Martin Liska <mliska@suse.cz>
9475 PR gcov-profile/82457
9476 * doc/invoke.texi: Document that one needs a non-strict ISO mode
9477 for fork-like functions to be properly instrumented.
9479 2017-10-27 Richard Biener <rguenther@suse.de>
9482 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
9483 info when we redirected EH.
9485 2017-10-26 Michael Collison <michael.collison@arm.com>
9487 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
9489 (<optab>_trunchf<GPI:mode>2: New pattern.
9490 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
9491 * config/aarch64/iterators.md (wv): New mode attribute.
9492 (vf, VF): New mode attributes.
9493 (vgp, VGP): New mode attributes.
9494 (s): Update attribute with SImode and DImode prefixes.
9496 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9498 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
9499 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
9500 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
9501 (nios2_option_overide): Initialize it. Don't allow R0-relative
9502 addressing with PIC.
9503 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
9504 (nios2_symbolic_constant_p): Likewise.
9505 (nios2_legitimate_address_p): Likewise.
9506 (nios2_r0rel_section_name_p): New.
9507 (nios2_symbol_ref_in_r0rel_data_p): New.
9508 (nios2_emit_move_sequence): Handle r0rel_constant_p.
9509 (r0rel_constant_p): New.
9510 (nios2_print_operand_address): Handle r0rel_constant_p.
9511 (nios2_cdx_narrow_form_p): Likewise.
9512 * config/nios2/nios2.opt (mr0rel-sec=): New option.
9513 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
9514 (Nios II Options): Document -mr0rel-sec.
9516 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9518 * config/nios2/nios2.c: Include xregex.h.
9519 (nios2_gprel_sec_regex): New.
9520 (nios2_option_overide): Initialize it. Don't allow GP-relative
9521 addressing with PIC.
9522 (nios2_small_section_name_p): Check for regex match.
9523 * config/nios2/nios2.opt (mgprel-sec=): New option.
9524 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
9525 (Nios II Options): Document -mgprel-sec.
9527 2017-10-26 Jim Wilson <wilson@tuliptree.org>
9529 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
9531 2017-10-26 Tom de Vries <tom@codesourcery.com>
9533 PR tree-optimization/82707
9534 * gimple.c (gimple_copy): Fix unsharing of
9535 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
9537 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
9539 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
9540 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
9541 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
9542 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
9543 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
9544 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
9545 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
9546 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
9547 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
9548 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
9549 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
9550 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
9551 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
9552 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
9553 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
9554 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
9555 _mm512_mask_cmpunord_ps_mask): New intrinsics.
9557 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9559 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
9561 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
9562 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
9563 warning to rs6000.c. Remove the Undocumented flag, since it has
9565 (-mabi=ibmlongdouble): Likewise.
9566 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
9567 already set, set the default format for long double.
9568 (rs6000_debug_reg_global): Print whether long double is IBM or
9570 (rs6000_option_override_internal): Rework setting long double
9571 format. Only warn if the user is changing the long double default
9572 and they did not use -Wno-psabi.
9573 * doc/invoke.texi (PowerPC options): Update the documentation for
9574 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
9576 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9577 Alan Hayward <alan.hayward@arm.com>
9578 David Sherwood <david.sherwood@arm.com>
9580 * rtl.h (wider_subreg_mode): New function.
9581 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
9582 rather than an unsigned int *.
9583 * ira-color.c (regno_max_ref_width): Replace with...
9584 (regno_max_ref_mode): ...this new variable.
9585 (coalesced_pseudo_reg_slot_compare): Update accordingly.
9586 Use wider_subreg_mode.
9587 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
9588 rather than an unsigned int *.
9589 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
9590 (process_alt_operands): Likewise.
9591 (invariant_p): Likewise.
9592 * lra-spills.c (assign_mem_slot): Likewise.
9593 (add_pseudo_to_slot): Likewise.
9594 * lra.c (collect_non_operand_hard_regs): Likewise.
9595 (add_regs_to_insn_regno_info): Likewise.
9596 * reload1.c (regno_max_ref_width): Replace with...
9597 (regno_max_ref_mode): ...this new variable.
9598 (reload): Update accordingly. Update call to
9599 ira_sort_regnos_for_alter_reg.
9600 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
9601 (init_eliminable_invariants): Update to use regno_max_ref_mode.
9602 (scan_paradoxical_subregs): Likewise.
9604 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9606 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
9607 (aarch64_frame): Add emit_frame_chain boolean.
9608 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9609 Move eh_return case to aarch64_layout_frame.
9610 (aarch64_layout_frame): Initialize emit_frame_chain.
9611 (aarch64_expand_prologue): Use emit_frame_chain.
9613 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9615 * config/aarch64/aarch64.c (aarch64_layout_frame):
9616 Ensure LR is always stored at the bottom of the callee-saves.
9617 Remove rarely used frame layout which saves callee-saves at top of
9618 frame, so the store of LR can be used as a valid probe in all cases.
9620 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9622 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
9623 Improve unaligned TImode/TFmode base/offset split.
9625 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9626 Alan Hayward <alan.hayward@arm.com>
9627 David Sherwood <david.sherwood@arm.com>
9629 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
9630 * combine.c (find_single_use_1): Likewise.
9631 (expand_field_assignment): Likewise.
9632 (move_deaths): Likewise.
9633 * lra-constraints.c (simplify_operand_subreg): Likewise.
9634 (curr_insn_transform): Likewise.
9635 * lra.c (collect_non_operand_hard_regs): Likewise.
9636 (add_regs_to_insn_regno_info): Likewise.
9637 * rtlanal.c (reg_referenced_p): Likewise.
9638 (covers_regno_no_parallel_p): Likewise.
9640 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9642 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
9643 Don't print any bits outside the precision of the value.
9644 * wide-int.cc (test_printing): Add some new tests.
9646 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9648 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
9649 supports -xbrace_comment option.
9650 * configure: Regenerate.
9651 * config.in: Regenerate.
9652 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
9653 (ASM_CPU_SPEC): Use it.
9655 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9657 * target.def (static_rtx_alignment): New hook.
9658 * targhooks.h (default_static_rtx_alignment): Declare.
9659 * targhooks.c (default_static_rtx_alignment): New function.
9660 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
9661 * doc/tm.texi: Regenerate.
9662 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
9663 instead of targetm.constant_alignment. Remove call to
9665 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9666 (cris_preferred_mininum_alignment): New function, split out from...
9667 (cris_constant_alignment): ...here.
9668 (cris_static_rtx_alignment): New function.
9669 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
9671 (ix86_constant_alignment): ...here.
9672 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9673 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9674 (mmix_static_rtx_alignment): New function.
9675 * config/spu/spu.c (spu_static_rtx_alignment): New function.
9676 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9678 2017-10-26 Tamar Christina <tamar.christina@arm.com>
9681 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
9682 Add flag_trapping_math and flag_fp_int_builtin_inexact.
9684 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
9686 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
9687 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
9688 mark as a sign-extending load.
9689 (local_pic_load_u): Define.
9691 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
9694 * fold-const.c (operand_equal_for_comparison_p): Also return true
9695 if ARG0 is a simple variant of ARG1 with narrower precision.
9696 (fold_ternary_loc): Always pass unstripped operands to the predicate.
9698 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
9700 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
9702 * i386.h (processor_costs): Add gather_static, gather_per_elt,
9703 scatter_static, scatter_per_elt.
9704 * x86-tune-costs.h: Add new cost entries.
9706 2017-10-25 Richard Biener <rguenther@suse.de>
9708 * tree-ssa-sccvn.h (vn_eliminate): Declare.
9709 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
9710 class pass_fre): Move to ...
9711 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
9712 class pass_fre): ... here and adjust for statistics.
9714 2017-10-25 Jakub Jelinek <jakub@redhat.com>
9717 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
9718 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
9719 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
9720 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
9723 2017-10-25 Richard Biener <rguenther@suse.de>
9725 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
9726 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
9727 eliminate_push_avail, eliminate_insert): Move inside...
9728 (class eliminate_dom_walker): ... this class in preparation
9730 (fini_eliminate): Remove by merging with ...
9731 (eliminate): ... this function. Adjust for class changes.
9732 (pass_pre::execute): Remove fini_eliminate call.
9733 (pass_fre::execute): Likewise.
9735 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9738 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
9739 (VPERMI2, VPERMI2I): New mode iterators.
9740 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
9741 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
9743 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
9744 mode iterator. Remove 3 old define_insn patterns.
9745 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
9746 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
9747 VPERMI2 mode iterator, remove the other two expanders.
9748 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
9749 to use VPERMI2 mode iterator, add another alternative for vpermi2*
9750 instructions, remove the other two patterns.
9751 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
9752 mode iterator, remove the other two patterns.
9753 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
9754 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
9755 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
9756 and adjust argument order accordingly.
9757 (ix86_expand_vec_perm): Adjust caller.
9758 (expand_vec_perm_1): Likewise.
9759 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
9760 (expand_vec_perm_vpermt2_vpshub2): ... this.
9761 (ix86_expand_vec_perm_const_1): Adjust caller.
9762 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
9765 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
9766 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
9767 (vec_shl_<mode>): Remove unused expander.
9768 (avx512bw_<shift_insn><mode>3): New define_insn.
9769 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
9770 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
9772 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
9775 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
9778 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9780 PR rtl-optimization/82396
9781 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
9782 (autopref_multipass_init): Simplify initialization.
9783 (autopref_rank_data): Simplify sort order.
9784 * gcc/sched-int.h (autopref_multipass_data_): Remove
9785 multi_mem_insn_p, min_offset and max_offset.
9787 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9790 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9791 Check special value of flag_omit_frame_pointer.
9792 (aarch64_can_eliminate): Likewise.
9793 (aarch64_override_options_after_change_1): Simplify handling of
9794 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
9796 2017-10-24 Richard Biener <rguenther@suse.de>
9798 PR tree-optimization/82697
9799 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
9800 zero for conditional load and unconditional store.
9802 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9804 * doc/install.texi: Document bootstrap-cet.
9806 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9809 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
9810 ENDBR instruction at function entrance if function is only
9813 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9816 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
9817 patterns to better describe from which operation the CF is computed.
9818 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
9819 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
9820 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
9821 is 0, use _0 suffixed expanders instead of emitting a comparison
9824 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
9826 * config/i386/i386.md(*movsf_internal, *movdf_internal):
9827 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
9829 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
9832 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
9833 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
9834 * loop-iv.c (iv_get_reaching_def): Likewise.
9835 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
9836 variable is promoted and the partition contains undefined values.
9838 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9840 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
9842 (nios2_address_cost): Define.
9843 (nios2_legitimize_address): Recognize (exp + constant) directly.
9844 (TARGET_ADDRESS_COST): Define.
9846 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9848 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
9849 (nios2_symbolic_memory_operand_p): Declare.
9850 (nios2_split_large_constant): Declare.
9851 (nios2_split_symbolic_memory_operand): Declare.
9852 * config/nios2/nios2.c: Adjust includes.
9853 (nios2_symbolic_constant_allowed): New.
9854 (nios2_symbolic_constant_p): New.
9855 (nios2_plus_symbolic_constant_p): New.
9856 (nios2_valid_addr_expr_p): Recognize addresses involving
9858 (nios2_legitimate_address_p): Likewise, also LO_SUM.
9859 (nios2_symbolic_memory_operand_p): New.
9860 (nios2_large_constant_p): New.
9861 (nios2_split_large_constant): New.
9862 (nios2_split_plus_large_constant): New.
9863 (nios2_split_symbolic_memory_operand): New.
9864 (nios2_legitimize_address): Code refactoring. Handle addresses
9865 involving symbolic constants.
9866 (nios2_emit_move_sequence): Likewise.
9867 (nios2_print_operand): Improve error output.
9868 (nios2_print_operand_address): Handle LO_SUM.
9869 (nios2_cdx_narrow_form_p): Likewise.
9870 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
9871 operands involving symbolic constants.
9872 (movhi_internal, movsi_internal): Likewise.
9873 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
9874 (extendhisi2, extendqi<mode>2): Likewise.
9876 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9878 * tree-pass.h (PROP_rtl_split_insns): Define.
9879 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
9881 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9883 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
9885 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9888 * target.def (const_not_ok_for_debug_p): Default to
9889 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
9890 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
9891 * targhooks.c (default_const_not_ok_for_debug_p): New function.
9892 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
9893 which targetm.const_not_ok_for_debug_p returned true.
9894 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
9896 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
9898 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
9899 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
9900 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
9902 (ix86_const_not_ok_for_debug_p): New function.
9903 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
9904 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
9906 2017-10-23 David Malcolm <dmalcolm@redhat.com>
9909 * system.h: Conditionally include "unique-ptr.h" if
9910 INCLUDE_UNIQUE_PTR is defined.
9911 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
9912 of defining INCLUDE_UNIQUE_PTR before including "system.h".
9914 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
9916 * config/rl78/rl78.md: New define_expand "subdi3".
9918 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
9921 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
9922 DF_REF_INSN if DF_REF_INSN_INFO is false.
9924 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
9926 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
9927 xmm_move instead of sse_move.
9928 (sse_store_index): New function.
9929 (ix86_register_move_cost): Be more sensible about mismatch stall;
9930 model AVX moves correctly; make difference between sse->integer and
9932 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
9933 moves; make difference between SSE and AVX.
9934 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
9935 and zmm_move. Increase size of sse load and store tables;
9936 add unaligned load and store tables; add ssemmx_to_integer.
9937 * x86-tune-costs.h: Update all entries according to real
9938 move latencies from Agner Fog's manual and chip documentation.
9940 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9943 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
9944 * config/i386/constraints.md (Wf): New constraint.
9945 * config/i386/i386.md (UNSPEC_SBB): New unspec.
9946 (cmp<dwi>_doubleword): Removed.
9947 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
9948 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
9949 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
9950 expand with cmp<dwi>_doubleword. For LTU and GEU use
9951 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
9953 * common.opt (gcolumn-info): Enable by default.
9954 * doc/invoke.texi (gcolumn-info): Document new default.
9956 2017-10-23 Richard Biener <rguenther@suse.de>
9958 PR tree-optimization/82672
9959 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
9960 Fold the stmt if we propagated into it.
9962 2017-10-23 Richard Biener <rguenther@suse.de>
9964 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
9965 (bitmap_remove_expr_from_set): ... this. All callers call this
9966 for non-constant values.
9967 (bitmap_set_subtract): Rename to...
9968 (bitmap_set_subtract_expressions): ... this. Adjust and
9970 (bitmap_set_contains_value): Remove superfluous check.
9971 (bitmap_set_replace_value): Inline into single caller ...
9972 (bitmap_value_replace_in_set): ... here and simplify.
9973 (dependent_clean): Merge into ...
9974 (clean): ... this using an overload. Adjust.
9975 (prune_clobbered_mems): Adjust.
9976 (compute_antic_aux): Likewise.
9977 (compute_partial_antic_aux): Likewise.
9979 2017-10-23 Richard Biener <rguenther@suse.de>
9981 PR tree-optimization/82129
9983 2017-08-01 Richard Biener <rguenther@suse.de>
9985 PR tree-optimization/81181
9986 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9987 (compute_antic): ... end of iteration here.
9989 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9991 * target.def (starting_frame_offset): New hook.
9992 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
9993 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
9994 * doc/tm.texi.in: Regenerate.
9995 * hooks.h (hook_hwi_void_0): Declare.
9996 * hooks.c (hook_hwi_void_0): New function.
9997 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
9998 STARTING_FRAME_OFFSET.
9999 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
10000 * reload1.c (reload): Likewise.
10001 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
10002 instead of STARTING_FRAME_OFFSET.
10003 * function.c (try_fit_stack_local): Likewise.
10004 (assign_stack_local_1): Likewise
10005 (instantiate_virtual_regs): Likewise.
10006 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
10007 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
10008 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
10009 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
10010 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
10011 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
10012 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
10013 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
10014 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
10015 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
10016 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
10017 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
10018 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
10019 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
10020 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
10021 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
10022 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
10023 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
10024 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
10025 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
10026 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
10027 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
10028 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
10029 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
10030 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
10031 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
10032 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
10033 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
10034 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
10035 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
10036 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
10037 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
10038 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
10039 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
10040 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
10041 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
10042 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
10043 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
10044 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
10045 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
10046 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
10047 * config/avr/avr.c (avr_starting_frame_offset): Make static and
10048 return a HOST_WIDE_INT.
10049 (avr_builtin_setjmp_frame_value): Use it instead of
10050 STARTING_FRAME_OFFSET.
10051 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10052 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
10053 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
10055 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10056 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
10057 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
10058 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10059 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
10060 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
10061 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10062 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
10063 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
10064 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10065 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
10066 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
10068 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10069 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
10070 * config/mips/mips.c (mips_compute_frame_info): Refer to
10071 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10072 (mips_starting_frame_offset): New function.
10073 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10074 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
10075 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
10076 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
10077 and return a HOST_WIDE_INT.
10078 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10079 (mmix_initial_elimination_offset): Refer to
10080 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10081 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
10082 * config/pa/pa.c (pa_starting_frame_offset): New function.
10083 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
10084 (pa_expand_prologue): Likewise.
10085 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10086 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
10087 !FRAME_GROWS_DOWNWARD handling to...
10088 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10089 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
10090 !FRAME_GROWS_DOWNWARD handling to...
10091 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10092 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
10093 !FRAME_GROWS_DOWNWARD handling to...
10094 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10095 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
10097 (rs6000_starting_frame_offset): New function.
10098 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
10099 !FRAME_GROWS_DOWNWARD handling to...
10100 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10101 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
10102 !FRAME_GROWS_DOWNWARD handling to...
10103 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10104 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
10105 !FRAME_GROWS_DOWNWARD handling to...
10106 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10107 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
10108 (rs6000_starting_frame_offset): New function.
10109 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
10110 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
10111 * config/vax/vax.c (vax_starting_frame_offset): New function.
10112 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
10113 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10114 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
10115 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
10116 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10117 * system.h (STARTING_FRAME_OFFSET): Poison.
10119 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10121 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10122 SCALAR_TYPE_MODE instead of TYPE_MODE.
10124 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10125 Alan Hayward <alan.hayward@arm.com>
10126 David Sherwood <david.sherwood@arm.com>
10128 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
10130 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10131 Alan Hayward <alan.hayward@arm.com>
10132 David Sherwood <david.sherwood@arm.com>
10134 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
10136 2017-10-23 Richard Biener <rguenther@suse.de>
10138 PR tree-optimization/82129
10139 * tree-ssa-pre.c (bitmap_set_and): Remove.
10140 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
10141 canonicalizing expressions in the set to those with lowest
10142 ID rather than taking that from the first edge.
10144 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10146 * combine.c (rtx_equal_for_field_assignment_p): Use
10147 byte_lowpart_offset.
10149 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10150 Alan Hayward <alan.hayward@arm.com>
10151 David Sherwood <david.sherwood@arm.com>
10153 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
10154 to a SUBREG_PROMOTED_VAR.
10156 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10157 Alan Hayward <alan.hayward@arm.com>
10158 David Sherwood <david.sherwood@arm.com>
10160 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
10161 (expand_debug_source_expr): Likewise.
10162 * combine.c (combine_simplify_rtx): Likewise.
10163 * cse.c (fold_rtx): Likewise.
10164 * optabs.c (expand_float): Likewise.
10165 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10166 (simplify_binary_operation_1): Likewise.
10168 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10169 Alan Hayward <alan.hayward@arm.com>
10170 David Sherwood <david.sherwood@arm.com>
10172 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
10173 (record_promoted_value): Likewise.
10174 * expr.c (expand_expr_real_2): Likewise.
10175 * ree.c (update_reg_equal_equiv_notes): Likewise.
10176 (combine_set_extension): Likewise.
10177 * rtlanal.c (low_bitmask_len): Likewise.
10178 * simplify-rtx.c (neg_const_int): Likewise.
10179 (simplify_binary_operation_1): Likewise.
10181 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10182 Alan Hayward <alan.hayward@arm.com>
10183 David Sherwood <david.sherwood@arm.com>
10185 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
10186 * regcprop.c (maybe_mode_change): Likewise.
10187 * reload1.c (alter_reg): Likewise.
10189 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10191 * inchash.h (inchash::hash::add_wide_int): New function.
10192 * lto-streamer-out.c (hash_tree): Use it.
10194 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10196 * inchash.h (inchash::hash::add_wide_int): Rename to...
10197 (inchash::hash::add_hwi): ...this.
10198 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
10199 (polymorphic_call_target_hasher::hash): Likewise.
10200 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
10201 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
10202 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
10203 * lto-streamer-out.c (hash_tree): Likewise.
10204 * optc-save-gen.awk: Likewise.
10205 * tree.c (add_expr): Likewise.
10207 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10210 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
10211 for ordered inequality comparisons even with TARGET_IEEE_FP.
10213 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10216 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
10217 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10218 Expand with cmp<dwi>_doubleword.
10220 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10222 * extend.texi: Add x86 specific to 'nocf_check' attribute.
10223 List CET intrinsics.
10224 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
10225 specific to -fcf-protection option.
10227 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10229 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
10230 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
10231 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
10232 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
10233 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
10234 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
10235 (extra_objs): Add cet.o for Linux/x86 targets.
10236 (tmake_file): Add i386/t-cet for Linux/x86 targets.
10237 * config/i386/cet.c: New file.
10238 * config/i386/cetintrin.h: Likewise.
10239 * config/i386/t-cet: Likewise.
10240 * config/i386/cpuid.h (bit_SHSTK): New.
10241 (bit_IBT): Likewise.
10242 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
10243 pass IBT and SHSTK bits.
10244 * config/i386/i386-builtin-types.def
10245 (VOID_FTYPE_UNSIGNED_PVOID): New.
10246 (VOID_FTYPE_UINT64_PVOID): Likewise.
10247 * config/i386/i386-builtin.def: Add CET intrinsics.
10248 * config/i386/i386-c.c (ix86_target_macros_internal): Add
10249 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
10250 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
10251 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
10253 * config/i386/i386.c (rest_of_insert_endbranch): New.
10254 (pass_data_insert_endbranch): Likewise.
10255 (pass_insert_endbranch): Likewise.
10256 (make_pass_insert_endbranch): Likewise.
10257 (ix86_notrack_prefixed_insn_p): Likewise.
10258 (ix86_target_string): Add -mibt, -mshstk flags.
10259 (ix86_option_override_internal): Add flag_cf_protection
10261 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
10262 (ix86_print_operand): Add 'notrack' prefix output.
10263 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
10264 (ix86_expand_builtin): Expand CET intrinsics.
10265 (x86_output_mi_thunk): Add 'endbranch' instruction.
10266 * config/i386/i386.h (TARGET_IBT): New.
10267 (TARGET_IBT_P): Likewise.
10268 (TARGET_SHSTK): Likewise.
10269 (TARGET_SHSTK_P): Likewise.
10270 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
10271 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
10272 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
10273 (builtin_setjmp_setup): New pattern.
10274 (builtin_longjmp): Likewise.
10275 (rdssp<mode>): Likewise.
10276 (incssp<mode>): Likewise.
10277 (saveprevssp): Likewise.
10278 (rstorssp): Likewise.
10279 (wrss<mode>): Likewise.
10280 (wruss<mode>): Likewise.
10281 (setssbsy): Likewise.
10282 (clrssbsy): Likewise.
10283 (nop_endbr): Likewise.
10284 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
10286 * config/i386/immintrin.h: Include <cetintrin.h>.
10287 * config/i386/linux-common.h
10288 (file_end_indicate_exec_stack_and_cet): New prototype.
10289 (TARGET_ASM_FILE_END): New.
10291 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10293 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
10294 latencies instead of having separate table; make difference between
10295 integer and float costs.
10296 * i386.h (processor_costs): Remove scalar_stmt_cost,
10297 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
10298 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
10300 * x86-tune-costs.h: Remove entries which has been removed in
10301 procesor_costs from all tables; make cond_taken_branch_cost
10302 and cond_not_taken_branch_cost COST_N_INSNS based.
10304 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10306 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
10308 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10310 * config/i386/i386.md (isa): Remove fma_avx512f.
10311 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
10312 <avx512>_fmadd_<mode>_mask3<round_name>,
10313 <avx512>_fmsub_<mode>_mask<round_name>,
10314 <avx512>_fmsub_<mode>_mask3<round_name>,
10315 <avx512>_fnmadd_<mode>_mask<round_name>,
10316 <avx512>_fnmadd_<mode>_mask3<round_name>,
10317 <avx512>_fnmsub_<mode>_mask<round_name>,
10318 <avx512>_fnmsub_<mode>_mask3<round_name>,
10319 <avx512>_fmaddsub_<mode>_mask<round_name>,
10320 <avx512>_fmaddsub_<mode>_mask3<round_name>,
10321 <avx512>_fmsubadd_<mode>_mask<round_name>,
10322 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
10323 (*vec_widen_umult_even_v16si<mask_name>,
10324 *vec_widen_smult_even_v16si<mask_name>): Likewise.
10325 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
10327 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10329 * extend.texi: Add 'nocf_check' documentation.
10330 * gimple.texi: Add second parameter to
10331 gimple_build_call_from_tree.
10332 * invoke.texi: Add -fcf-protection documentation.
10333 * rtl.texi: Add REG_CALL_NOTRACK documenation.
10335 2017-10-20 Richard Biener <rguenther@suse.de>
10337 PR tree-optimization/82473
10338 * tree-vect-loop.c (vectorizable_reduction): Properly get at
10339 the largest input type.
10341 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10343 * c-attribs.c (handle_nocf_check_attribute): New function.
10344 (c_common_attribute_table): Add 'nocf_check' handling.
10345 * gimple-parser.c: Add second argument NULL to
10346 gimple_build_call_from_tree.
10347 * attrib.c (comp_type_attributes): Check nocf_check attribute.
10348 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
10350 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
10351 * common.opt: Add fcf-protection flag.
10352 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
10353 * flag-types.h: Add enum cf_protection_level.
10354 * gimple.c (gimple_build_call_from_tree): Add second parameter.
10355 Add 'nocf_check' attribute propagation to gimple call.
10356 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
10357 (gimple_build_call_from_tree): Update prototype.
10358 (gimple_call_nocf_check_p): New function.
10359 (gimple_call_set_nocf_check): Likewise.
10360 * gimplify.c: Add second argument to gimple_build_call_from_tree.
10361 * ipa-icf.c: Add nocf_check attribute in statement hash.
10362 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
10363 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
10364 * toplev.c (process_options): Add flag_cf_protection handling.
10366 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10368 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
10370 2017-10-20 Richard Biener <rguenther@suse.de>
10372 PR tree-optimization/82603
10373 * tree-if-conv.c (predicate_mem_writes): Make sure to only
10374 remove false predicated stores.
10376 2017-10-20 Richard Biener <rguenther@suse.de>
10378 * graphite-isl-ast-to-gimple.c
10379 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
10380 Remove return value and simplify, dump copied stmt after lhs
10382 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
10383 Reduce dump verbosity.
10384 (gsi_insert_earliest): Likewise.
10385 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
10386 * graphite.c (print_global_statistics): Adjust dumping.
10387 (print_graphite_scop_statistics): Likewise.
10388 (print_graphite_statistics): Do not dump loops here.
10389 (graphite_transform_loops): But here.
10391 2017-10-20 Nicolas Roche <roche@adacore.com>
10393 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
10394 * configure: Regenerate.
10396 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10399 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
10400 functions when optimizing replace GIMPLE_RETURN stmts with
10401 calls to __builtin_unreachable ().
10404 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
10405 for -fsanitize=thread link of executables.
10406 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
10407 link of executables.
10410 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
10411 New mode iterators.
10412 (<shift_insn><mode>3<mask_name>): Change the last of the 3
10413 define_insns for logical vector shifts to use VI248_AVX512BW
10414 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
10415 condition, useless isa and prefix attributes. Change the first
10416 2 of these define_insns to ...
10417 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
10418 define_insn for avx512vl.
10419 (<shift_insn><mode>3): ... and this, new define_insn without
10420 masking for non-avx512vl.
10423 * config/i386/sse.md (*andnot<mode>3,
10424 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
10425 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
10426 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
10427 not applied use empty suffix even for TARGET_AVX512VL.
10428 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
10429 is applied, supply evex,evex or evex,evex,evex instead of just
10432 2017-10-20 Julia Koval <julia.koval@intel.com>
10434 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
10435 (OPTION_MASK_ISA_GFNI_UNSET): New.
10436 (ix86_handle_option): Handle OPT_mgfni.
10437 * config/i386/cpuid.h (bit_GFNI): New.
10438 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
10439 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
10440 * config/i386/i386.c (ix86_target_string): Add -mgfni.
10441 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
10442 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
10443 * config/i386/i386.opt: Add mgfni.
10445 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
10447 * config/msp430/msp430.c (msp430_option_override): Disable
10448 -fdelete-null-pointer-checks.
10449 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
10451 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10453 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
10454 of x87 and SSE instructions.
10456 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10458 * asan.c (create_cond_insert_point): Do not update edge count.
10459 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
10460 (afdo_propagate_circuit): Likewise.
10461 (afdo_calculate_branch_prob): Likewise.
10462 (afdo_annotate_cfg): Likewise.
10463 * basic-block.h (struct edge_def): Remove count.
10464 (edge_def::count): New accessor.
10465 * bb-reorder.c (rotate_loop): Update.
10466 (find_traces_1_round): Update.
10467 (connect_traces): Update.
10468 (sanitize_hot_paths): Update.
10469 * cfg.c (unchecked_make_edge): Update.
10470 (make_single_succ_edge): Update.
10471 (check_bb_profile): Update.
10472 (dump_edge_info): Update.
10473 (update_bb_profile_for_threading): Update.
10474 (scale_bbs_frequencies_int): Update.
10475 (scale_bbs_frequencies_gcov_type): Update.
10476 (scale_bbs_frequencies_profile_count): Update.
10477 (scale_bbs_frequencies): Update.
10478 * cfganal.c (connect_infinite_loops_to_exit): Update.
10479 * cfgbuild.c (compute_outgoing_frequencies): Update.
10480 (find_many_sub_basic_blocks): Update.
10481 * cfgcleanup.c (try_forward_edges): Update.
10482 (try_crossjump_to_edge): Update
10483 * cfgexpand.c (expand_gimple_cond): Update
10484 (expand_gimple_tailcall): Update
10485 (construct_exit_block): Update
10486 * cfghooks.c (verify_flow_info): Update
10487 (redirect_edge_succ_nodup): Update
10488 (split_edge): Update
10489 (make_forwarder_block): Update
10490 (duplicate_block): Update
10491 (account_profile_record): Update
10492 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10493 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10494 * cfgloopmanip.c (scale_loop_profile): Update.
10496 (lv_adjust_loop_entry_edge): Update.
10497 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10498 (force_nonfallthru_and_redirect): Update.
10499 (purge_dead_edges): Update.
10500 (rtl_flow_call_edges_add): Update.
10501 * cgraphunit.c (init_lowered_empty_function): Update.
10502 (cgraph_node::expand_thunk): Update.
10503 * gimple-pretty-print.c (dump_probability): Update.
10504 (dump_edge_probability): Update.
10505 * gimple-ssa-isolate-paths.c (isolate_path): Update.
10506 * haifa-sched.c (sched_create_recovery_edges): Update.
10507 * hsa-gen.c (convert_switch_statements): Update.
10508 * ifcvt.c (dead_or_predicable): Update.
10509 * ipa-inline-transform.c (inline_transform): Update.
10510 * ipa-split.c (split_function): Update.
10511 * ipa-utils.c (ipa_merge_profiles): Update.
10512 * loop-doloop.c (add_test): Update.
10513 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
10514 * lto-streamer-in.c (input_cfg): Update.
10515 (input_function): Update.
10516 * lto-streamer-out.c (output_cfg): Update.
10517 * modulo-sched.c (sms_schedule): Update.
10518 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10519 * predict.c (maybe_hot_edge_p): Update.
10520 (unlikely_executed_edge_p): Update.
10521 (probably_never_executed_edge_p): Update.
10522 (dump_prediction): Update.
10523 (drop_profile): Update.
10524 (propagate_unlikely_bbs_forward): Update.
10525 (determine_unlikely_bbs): Update.
10526 (force_edge_cold): Update.
10527 * profile.c (compute_branch_probabilities): Update.
10528 * reg-stack.c (better_edge): Update.
10529 * shrink-wrap.c (handle_simple_exit): Update.
10530 * tracer.c (better_p): Update.
10531 * trans-mem.c (expand_transaction): Update.
10532 (split_bb_make_tm_edge): Update.
10533 * tree-call-cdce.c: Update.
10534 * tree-cfg.c (gimple_find_sub_bbs): Update.
10535 (gimple_split_edge): Update.
10536 (gimple_duplicate_sese_region): Update.
10537 (gimple_duplicate_sese_tail): Update.
10538 (gimple_flow_call_edges_add): Update.
10539 (insert_cond_bb): Update.
10540 (execute_fixup_cfg): Update.
10541 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
10542 * tree-complex.c (expand_complex_div_wide): Update.
10543 * tree-eh.c (lower_resx): Update.
10544 (unsplit_eh): Update.
10545 (cleanup_empty_eh_move_lp): Update.
10546 * tree-inline.c (copy_edges_for_bb): Update.
10547 (freqs_to_counts): Update.
10548 (copy_cfg_body): Update.
10549 * tree-ssa-dce.c (remove_dead_stmt): Update.
10550 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10551 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10552 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10553 (unloop_loops): Update.
10554 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10555 * tree-ssa-loop-split.c (connect_loops): Update.
10556 (split_loop): Update.
10557 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10558 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10559 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10560 * tree-ssa-reassoc.c (branch_fixup): Update.
10561 * tree-ssa-tail-merge.c (replace_block_by): Update.
10562 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10563 (compute_path_counts): Update.
10564 (update_profile): Update.
10565 (recompute_probabilities): Update.
10566 (update_joiner_offpath_counts): Update.
10567 (estimated_freqs_path): Update.
10568 (freqs_to_counts_path): Update.
10569 (clear_counts_path): Update.
10570 (ssa_fix_duplicate_block_edges): Update.
10571 (duplicate_thread_path): Update.
10572 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10573 (case_bit_test_cmp): Update.
10574 (collect_switch_conv_info): Update.
10575 (gen_inbound_check): Update.
10576 (do_jump_if_equal): Update.
10577 (emit_cmp_and_jump_insns): Update.
10578 * tree-tailcall.c (decrease_profile): Update.
10579 (eliminate_tail_call): Update.
10580 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10581 (vect_do_peeling): Update.
10582 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10583 * ubsan.c (ubsan_expand_null_ifn): Update.
10584 (ubsan_expand_ptr_ifn): Update.
10585 * value-prof.c (gimple_divmod_fixed_value): Update.
10586 (gimple_mod_pow2): Update.
10587 (gimple_mod_subtract): Update.
10588 (gimple_ic): Update.
10589 (gimple_stringop_fixed_value): Update.
10591 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10594 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
10596 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
10598 PR rtl-optimization/82395
10599 * ira-color.c (allocno_priority_compare_func): Fix comparison step
10600 based on non_spilled_static_chain_regno_p.
10602 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10604 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
10605 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
10606 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
10608 2017-10-19 Martin Sebor <msebor@redhat.com>
10610 PR tree-optimization/82596
10611 * tree.c (array_at_struct_end_p): Handle STRING_CST.
10613 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10615 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
10616 (get_mem_refs_of_builtin_call): Likewise.
10617 * builtins.c (expand_builtin_apply): Adjust call to
10618 allocate_dynamic_stack_space.
10619 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
10620 the third argument to allocate_dynamic_stack_space, otherwise -1.
10621 (expand_builtin): Deal with all alloca variants.
10622 (is_inexpensive_builtin): Likewise.
10623 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
10624 * calls.c (special_function_p): Deal with all alloca variants.
10625 (initialize_argument_information): Adjust call to
10626 allocate_dynamic_stack_space.
10627 (expand_call): Likewise.
10628 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
10629 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
10630 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
10631 use it for the stack usage computation.
10632 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
10633 * function.c (gimplify_parameters): Call build_alloca_call_expr.
10634 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
10635 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
10636 (in_loop_p): Remove first argument and useless check.
10637 (pass_walloca::execute): Remove useless test and adjust call to above.
10638 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
10639 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
10640 (gimplify_call_expr): Deal with all alloca variants.
10641 * hsa-gen.c (gen_hsa_alloca): Likewise.
10642 (gen_hsa_insns_for_call): Likewise.
10643 * ipa-pure-const.c (special_builtin_state): Likewise.
10644 * tree-chkp.c (chkp_build_returned_bound): Likewise.
10645 * tree-object-size.c (alloc_object_size): Likewise.
10646 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
10647 (call_may_clobber_ref_p_1): Likewise.
10648 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
10649 (ccp_fold_stmt): Likewise.
10650 (optimize_stack_restore): Likewise.
10651 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
10652 (mark_all_reaching_defs_necessary_1): Likewise.
10653 (propagate_necessity): Likewise.
10654 (eliminate_unnecessary_stmts): Likewise.
10655 * tree.c (build_common_builtin_nodes): Build
10656 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
10657 (build_alloca_call_expr): New function.
10658 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
10659 (CASE_BUILT_IN_ALLOCA): Likewise.
10660 (build_alloca_call_expr): Declare.
10661 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
10663 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10666 * dwarf2out.c (new_die_raw): New static inline function.
10667 (new_die): Use it to create the DIE.
10668 (add_AT_external_die_ref): Likewise.
10669 (clone_die): Likewise.
10670 (clone_as_declaration): Likewise.
10671 (dwarf2out_vms_debug_main_pointer): Likewise.
10672 (base_type_die): Likewise. Remove early return for corner cases.
10673 Do not call add_pubtype on the DIE here.
10674 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
10675 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
10676 typedefs for base types with DW_AT_endianity. Make sure a DIE with
10677 native order exists for base types, attach the DIE manually and call
10678 add_pubtype on it. Do not equate a reverse order DIE to the type.
10680 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
10682 * config/arm/arm.c (align_ok_ldrd_strd): New function.
10683 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
10685 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
10687 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10689 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
10690 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
10691 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
10692 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
10693 * opts.c (sanitizer_opts): Add builtin.
10694 * ubsan.c (instrument_builtin): New function.
10695 (pass_ubsan::execute): Call it.
10696 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
10697 * doc/invoke.texi: Document -fsanitize=builtin.
10699 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
10700 builtins, store max (log2 (align), 0) into uchar field instead of
10701 align into uptr field.
10702 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
10703 store uchar 0 field instead of uptr 0 field.
10704 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
10705 instead of passing one address of struct with 2 locations pass
10706 two addresses of structs with 1 location each.
10707 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
10708 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
10709 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
10710 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
10711 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
10712 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
10713 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
10714 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
10716 2017-10-19 Martin Liska <mliska@suse.cz>
10719 * file-find.c (remove_prefix): Remove.
10720 * file-find.h (remove_prefix): Likewise.
10721 * gcc-ar.c: Remove smartness of lookup.
10723 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
10725 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
10726 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
10727 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
10729 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10732 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
10733 (setcc + and to xor + setcc): New peephole2.
10735 2017-10-19 Tom de Vries <tom@codesourcery.com>
10737 * doc/sourcebuild.texi (Test Directives, Variants of
10738 dg-require-support): Add dg-require-stack-size.
10740 2017-10-19 Martin Liska <mliska@suse.cz>
10743 * gimplify.c (gimplify_decl_expr): Do not instrument variables
10744 that have a large alignment.
10745 (gimplify_target_expr): Likewise.
10747 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
10749 PR rtl-optimization/82602
10750 * ira.c (rtx_moveable_p): Return false for volatile asm.
10752 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
10755 * config/i386/i386-modes.def (CCGZ): New CC mode.
10756 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
10757 * config/i386/predicates.md (ix86_comparison_operator):
10759 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10760 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
10761 with double-word subtraction.
10762 (put_condition_code): Handle CCGZmode.
10764 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
10766 * wide-int.cc (debug (const wide_int &)): New.
10767 (debug (const wide_int *)): New.
10768 (debug (const widest_int &)): New.
10769 (debug (const widest_int *)): New.
10771 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
10773 PR middle-end/82556
10774 * lra-constraints.c (curr_insn_transform): Use non-input operand
10775 instead of output one for matched reload.
10777 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10779 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
10780 (tree-ssa-loop-ivopts.h): New header file.
10781 (struct builtin_info): New fields.
10782 (classify_builtin_1): Compute and record base and offset parts for
10783 memset builtin partition by calling strip_offset.
10784 (offset_cmp, fuse_memset_builtins): New functions.
10785 (finalize_partitions): Fuse adjacent memset partitions by calling
10787 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
10788 Expose the interface.
10789 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
10791 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10793 PR tree-optimization/82574
10794 * tree-loop-distribution.c (find_single_drs): New parameter. Check
10795 that data reference must be executed exactly once per iteration
10796 against the outermost loop in nest.
10797 (classify_partition): Update call to above function.
10799 2017-10-18 Richard Biener <rguenther@suse.de>
10801 PR tree-optimization/82591
10802 * graphite.c (graphite_transform_loops): Move code gen message
10804 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10805 Here. Handle scop_to_isl_ast failing.
10806 (scop_to_isl_ast): Limit the number of ISL operations.
10808 2017-10-18 Richard Biener <rguenther@suse.de>
10810 * graphite-isl-ast-to-gimple.c
10811 (translate_isl_ast_to_gimple::set_rename): Simplify.
10812 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
10813 (graphite_copy_stmts_from_block): ... here.
10814 (copy_bb_and_scalar_dependences): Simplify.
10815 (add_parameters_to_ivs_params): Canonicalize.
10816 (generate_entry_out_of_ssa_copies): Simplify.
10817 * graphite-sese-to-poly.c (extract_affine_name): Simplify
10818 by passing in ISL dimension.
10819 (parameter_index_in_region_1): Rename to ...
10820 (parameter_index_in_region): ... this.
10821 (extract_affine): Adjust assert, pass down parameter index.
10822 (add_param_constraints): Use range-info when available.
10823 (build_scop_context): Adjust.
10824 * sese.c (new_sese_info): Adjust.
10825 (free_sese_info): Likewise.
10826 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
10827 Remove unused typedefs.
10828 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
10830 2017-10-18 Martin Liska <mliska@suse.cz>
10832 * combine.c (simplify_compare_const): Add gcc_fallthrough.
10834 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10836 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
10837 (s390_sched_init): Do not reset s390_sched_state if we entered the
10838 current basic block via a fallthru edge and all others are unlikely.
10840 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10842 * config/s390/s390.c (NUM_SIDES): New variable.
10843 (LONGRUNNING_THRESHOLD): New variable.
10844 (LATENCY_FACTOR): New variable.
10845 (s390_sched_score): Decrease score for long-running instructions on
10847 (s390_sched_variable_issue): Perform bookkeeping for long-running
10850 2017-10-18 Richard Biener <rguenther@suse.de>
10852 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
10853 Simplify with removal of the parameter rename map.
10854 (set_rename): Likewise.
10855 (should_copy_to_new_region): Likewise.
10856 (graphite_copy_stmts_from_block): Likewise.
10857 (copy_bb_and_scalar_dependences): Remove initialization of
10858 unused copied_bb_map.
10859 (copy_def): Remove.
10860 (copy_internal_parameters): Likewise.
10861 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
10862 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
10863 Use INTEGRAL_TYPE_P.
10864 (parameter_index_in_region_1): Rename to ...
10865 (assign_parameter_index_in_region): ... this. Assert we have
10866 a parameter we handle.
10867 (scan_tree_for_params): Adjust.
10868 * sese.h (parameter_rename_map_t): Remove.
10869 (struct sese_info_t): Remove unused parameter_rename_map and
10870 copied_bb_map members.
10871 * sese.c (new_sese_info): Adjust.
10872 (free_sese_info): Likewise.
10874 2017-10-18 Martin Liska <mliska@suse.cz>
10877 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
10878 on an abnormal edge.
10880 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
10882 * doc/invoke.texi (ffunction-sections and fdata-sections):
10885 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10887 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
10888 the use statement can throw internally.
10890 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10892 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
10893 any RTX present on the RHS of a SET.
10894 * compare-elim.c (try_eliminate_compare): Restore comment.
10896 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10898 * langhooks.h (struct lang_hooks): Document that tree_size langhook
10899 may be also called on tcc_type nodes.
10900 * langhooks.c (lhd_tree_size): Likewise.
10902 2017-10-17 David Malcolm <dmalcolm@redhat.com>
10904 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
10905 format_warning_at_substring.
10906 (maybe_warn): Convert source_range * param to a location_t. Pass
10907 UNKNOWN_LOCATION rather than NULL to fmtwarn.
10908 (format_directive): Remove code to extract source_ranges and
10909 source_range * in favor of just a location_t.
10910 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
10912 * substring-locations.c (format_warning_va): Convert
10913 source_range * param to a location_t.
10914 (format_warning_at_substring): Likewise.
10915 * substring-locations.h (format_warning_va): Likewise.
10916 (format_warning_at_substring): Likewise.
10918 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
10920 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
10922 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
10923 and scatter/gather ops.
10925 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
10926 vec_gather_load and vec_scatter_store.
10927 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
10928 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
10929 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
10930 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
10931 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
10932 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
10934 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
10936 * reg-stack.c (compare_for_stack_reg): Add bool argument.
10937 Detect FTST instruction and handle its register pops. Only pop
10938 second operand if can_pop_second_op is true.
10939 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
10940 set can_pop_second_op to false in the compare_for_stack_reg call.
10942 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
10943 output_fp_compare for stack register operands.
10944 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
10945 instructions here. Do not emit stack register pops here. Assert
10946 that FCOMPP pops next to top stack register. Rewrite function.
10948 2017-10-17 Nathan Sidwell <nathan@acm.org>
10950 PR middle-end/82577
10951 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
10952 use DECL_ASSEMBLER_NAME_RAW.
10954 PR middle-end/82546
10955 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
10958 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
10959 Wilco Dijkstra <wilco.dijkstra@arm.com>
10961 * builtins.c (expand_builtin_update_setjmp_buf): Add a
10962 converstion to Pmode from the buf_addr.
10964 2017-10-17 Richard Biener <rguenther@suse.de>
10966 * graphite-dependences.c (scop_get_reads_and_writes): Change
10967 output parameters to references.
10969 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
10971 PR 71026/tree-optimization
10972 * fold-const.c (distribute_real_division): Removed.
10973 (fold_binary_loc): Remove calls to distribute_real_divison.
10975 2017-10-17 Richard Biener <rguenther@suse.de>
10977 * graphite-scop-detection.c
10978 (scop_detection::stmt_has_simple_data_refs_p): Always use
10979 the full nest as region.
10980 (try_generate_gimple_bb): Likewise.
10981 * sese.c (scalar_evolution_in_region): Simplify now that
10982 SCEV can handle instantiation in regions.
10983 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
10984 in the non-loop part of a function if requested.
10986 2017-10-17 Richard Biener <rguenther@suse.de>
10988 PR tree-optimization/82563
10989 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
10991 (graphite_regenerate_ast_isl): Call it.
10992 * graphite-scop-detection.c (build_scops): Remove entry edge split.
10994 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10996 PR tree-optimization/82549
10997 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
10998 Formatting fixes. Instead of calling make_bit_field_ref with negative
11001 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
11003 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
11004 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
11005 _mm_maskz_reduce_ss): New.
11006 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
11007 __builtin_ia32_reducess_mask): Ditto..
11008 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
11009 * config/i386/sse.md (reduces<mode>): Renamed to ...
11010 (reduces<mode><mask_scalar_name>): ... this.
11011 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
11013 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
11014 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
11016 2017-10-16 David Malcolm <dmalcolm@redhat.com>
11018 * Makefile.in (OBJS): Add unique-ptr-tests.o.
11019 * selftest-run-tests.c (selftest::run_tests): Call
11020 selftest::unique_ptr_tests_cc_tests.
11021 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
11022 * unique-ptr-tests.cc: New file.
11024 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
11027 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11029 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11030 (make_hard_regno_born, make_hard_regno_dead): Update
11031 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
11033 2017-10-16 Jeff Law <law@redhat.com>
11035 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11037 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
11039 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
11041 2017-10-16 Olivier Hainque <hainque@adacore.com>
11043 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
11044 with_cpu if we were configured for an e500v2 target cpu name.
11046 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
11048 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
11049 * doc/invoke.texi: Document +nodsp as a valid extension for
11052 2017-10-16 Martin Liska <mliska@suse.cz>
11054 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
11055 (test_set_range): Likewise.
11056 (test_range_functions): Rename to ...
11057 (test_bit_in_range): ... this.
11058 (sbitmap_c_tests): Add new test.
11060 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11062 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
11064 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
11065 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
11067 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11069 * config/aarch64/aarch64-builtins.c
11070 (aarch64_types_quadopu_lane_qualifiers): New.
11071 (TYPES_QUADOPU_LANE): New.
11072 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
11073 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
11074 (aarch64_<sur>dot_laneq<vsi2qi>): New.
11075 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
11076 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
11077 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
11078 (Vdottype, DOTPROD): New.
11079 (sur): Add SDOT and UDOT.
11081 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11083 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
11084 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
11085 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
11086 Add TARGET_DOTPROD.
11087 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
11088 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
11089 Enable TARGET_DOTPROD.
11090 (cortex-a75.cortex-a55): Likewise.
11091 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
11093 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11095 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
11096 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
11098 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
11100 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
11101 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
11102 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
11103 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
11104 * config/arm/types.md (neon_dot, neon_dot_q): New.
11105 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
11107 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11109 * config/arm/arm.h (TARGET_DOTPROD): New.
11110 * config/arm/arm.c (arm_arch_dotprod): New.
11111 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
11112 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
11113 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
11114 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
11115 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
11116 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
11117 * doc/invoke.texi (armv8.2-a): Document dotprod
11119 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11121 * i386.c (ix86_vec_cost): New function.
11122 (ix86_rtx_costs): Handle vector operations better.
11123 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
11124 * x86-tune-costs.h: Add new costs to all tables.
11126 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11128 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
11130 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
11131 divsd, sqrtss and sqrtsd
11132 * x86-tune-costs.h: Add new entries to all costs.
11133 (znver1_cost): Fix to match real instruction latencies.
11135 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11136 Michael Collison <michael.collison@arm.com>
11138 * compare-elim.c: Include emit-rtl.h.
11139 (can_merge_compare_into_arith): New function.
11140 (try_validate_parallel): Likewise.
11141 (try_merge_compare): Likewise.
11142 (try_eliminate_compare): Call the above when no previous clobber
11144 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
11147 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11149 PR middle-end/62263
11150 PR middle-end/82498
11151 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
11152 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
11154 PR middle-end/62263
11155 PR middle-end/82498
11156 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
11157 to be any operand_equal_p operands. For & (B - 1) require
11158 B to be power of 2. Recognize
11159 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
11161 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11164 * optabs.c (expand_memory_blockage): Fix call of
11165 targetm.have_memory_blockage.
11167 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11170 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
11171 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
11172 objects to extra_objs instead of overwriting it.
11174 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11176 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
11177 Use any_fp_register_operand as operand[3] predicate. Simplify
11178 equality test for operands[2] and operands[4] memory location.
11179 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
11180 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
11181 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
11182 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
11183 any_fp_register_operand as operand[1] predicate. Simplify
11184 equality test for operands[0] and operands[3] memory location.
11185 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
11186 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
11187 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
11189 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11191 * target-insns.def: Add memory_blockage.
11192 * optabs.c (expand_memory_blockage): New function.
11193 (expand_asm_memory_barrier): Rename ...
11194 (expand_asm_memory_blockage): ... to this.
11195 (expand_mem_thread_fence): Call expand_memory_blockage
11196 instead of expand_asm_memory_barrier.
11197 (expand_mem_singnal_fence): Ditto.
11198 (expand_atomic_load): Ditto.
11199 (expand_atomic_store): Ditto.
11200 * doc/md.texi (Standard Pattern Names For Generation):
11201 Document memory_blockage instruction pattern.
11203 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
11205 * config/rl78/rl78.c (rl78_emit_libcall): New function.
11206 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
11207 * config/rl78/rl78.md: New define_expand "adddi3".
11209 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11211 * cfghooks.c (verify_flow_info): Disable check that all probabilities
11214 2017-10-13 Jeff Law <law@redhat.com>
11216 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
11218 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11221 * internal-fn.c (expand_mul_overflow): If both operands have
11222 the same highpart of -1 or 0 and the topmost bit of lowpart
11223 is different, overflow is if res <= 0 rather than res < 0.
11225 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
11227 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
11228 TARGET_P9_VECTOR code for unaligned_load case.
11230 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11232 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
11234 2017-10-13 Nathan Sidwell <nathan@acm.org>
11236 * tree-core.h (tree_contains_struct): Make bool.
11237 * tree.c (tree_contains_struct): Likewise.
11238 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
11239 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
11240 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
11241 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
11242 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
11244 2017-10-13 Richard Biener <rguenther@suse.de>
11246 * graphite-isl-ast-to-gimple.c
11247 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
11248 parameters and dominance check.
11249 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
11250 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
11251 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
11252 Do not update SSA form here or do intermediate IL verification.
11253 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
11254 (graphite_initialize): Remove check on the number of loops in
11255 the function and inline into graphite_transform_loops.
11256 (graphite_finalize): Inline into graphite_transform_loops.
11257 (graphite_transform_loops): Perform SSA update and IL verification
11259 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
11261 2017-10-13 Richard Biener <rguenther@suse.de>
11263 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
11264 graphite_expression_type_precision): Avoid global constructor
11266 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
11267 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
11268 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
11269 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
11270 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
11271 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
11273 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
11276 * config/i386/i386.h (ix86_red_zone_size): New.
11277 * config/i386/i386.md (push peephole2s): Replace
11278 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
11280 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11281 Alan Hayward <alan.hayward@arm.com>
11282 David Sherwood <david.sherwood@arm.com>
11284 * combine.c (can_change_dest_mode): Reject changes in
11285 REGMODE_NATURAL_SIZE.
11287 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11288 Alan Hayward <alan.hayward@arm.com>
11289 David Sherwood <david.sherwood@arm.com>
11291 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
11292 (expand_debug_source_expr): Likewise.
11293 * combine.c (combine_simplify_rtx): Likewise.
11294 * cse.c (fold_rtx): Likewise.
11295 * fwprop.c (canonicalize_address): Likewise.
11296 * targhooks.c (default_shift_truncation_mask): Likewise.
11298 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11299 Alan Hayward <alan.hayward@arm.com>
11300 David Sherwood <david.sherwood@arm.com>
11302 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
11303 (widened_mode): Likewise.
11304 (expand_unop): Likewise.
11305 * ree.c (transform_ifelse): Likewise.
11306 (merge_def_and_ext): Likewise.
11307 (combine_reaching_defs): Likewise.
11308 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11310 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11311 Alan Hayward <alan.hayward@arm.com>
11312 David Sherwood <david.sherwood@arm.com>
11314 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
11315 * combine.c (gen_lowpart_for_combine): Likewise.
11316 * dwarf2out.c (rtl_for_decl_location): Likewise.
11317 * final.c (alter_subreg): Likewise.
11318 * rtlhooks.c (gen_lowpart_general): Likewise.
11319 (gen_lowpart_if_possible): Likewise.
11321 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11322 Alan Hayward <alan.hayward@arm.com>
11323 David Sherwood <david.sherwood@arm.com>
11325 * calls.c (expand_call): Use subreg_lowpart_offset.
11326 * cse.c (cse_insn): Likewise.
11327 * regcprop.c (copy_value): Likewise.
11328 (copyprop_hardreg_forward_1): Likewise.
11330 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11333 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
11334 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
11335 =Q constraints to +Q and into insn condition add check
11336 that operands[0] and operands[1] are equal.
11337 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
11338 =Q constraints to +Q and into insn condition add check
11339 that operands[0] is equal to either operands[1] or operands[2].
11342 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
11343 instead of handling MINUS_EXPR twice (once for each argument),
11344 canonicalize operand order and handle just once, use rtype where
11345 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
11348 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
11349 any values of __C while still being pattern recognizable as a simple
11350 rotate instruction.
11352 2017-10-13 Richard Biener <rguenther@suse.de>
11354 PR tree-optimization/82451
11356 2017-10-02 Richard Biener <rguenther@suse.de>
11358 PR tree-optimization/82355
11359 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11360 a mapping for the enclosing loop but avoid generating one for
11361 the loop tree root.
11362 (copy_bb_and_scalar_dependences): Remove premature codegen
11363 error on PHIs in blocks duplicated into multiple places.
11364 * graphite-scop-detection.c
11365 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11366 in the region use it as loop and nest to analyze the DR in.
11367 (try_generate_gimple_bb): Likewise.
11368 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11369 (add_loop_constraints): For blocks in a loop not in the region
11370 create a dimension with a single iteration.
11371 * sese.h (gbb_loop_at_index): Remove assert.
11373 * cfgloop.c (loop_preheader_edge): For the loop tree root
11374 return the single successor of the entry block.
11375 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11376 Reset the SCEV hashtable and niters.
11377 * graphite-scop-detection.c
11378 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
11379 assert that we only have POLYNOMIAL_CHREC that vary in loops
11380 contained in the region.
11381 (scop_detection::graphite_can_represent_expr): Adjust.
11382 (scop_detection::stmt_has_simple_data_refs_p): For loops
11383 not in the region set loop to NULL. The nest is now the
11384 entry edge to the region.
11385 (try_generate_gimple_bb): Likewise.
11386 * sese.c (scalar_evolution_in_region): Adjust for
11387 instantiate_scev change.
11388 * tree-data-ref.h (graphite_find_data_references_in_stmt):
11389 Make nest parameter the edge into the region.
11390 (create_data_ref): Likewise.
11391 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
11392 entry edge into a region and adjust instantiate_scev calls.
11393 (create_data_ref): Likewise.
11394 (graphite_find_data_references_in_stmt): Likewise.
11395 (find_data_references_in_stmt): Pass the loop preheader edge
11396 from the nest argument.
11397 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
11398 parameter the edge into the region.
11399 (instantiate_parameters): Use the loop preheader edge as entry.
11400 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
11402 (get_instantiated_value_entry): Make instantiate_below parameter
11403 the edge into the region.
11404 (instantiate_scev_name): Likewise. Adjust dominance checks,
11405 when we cannot use loop-based instantiation instantiate by
11406 walking use-def chains.
11407 (instantiate_scev_poly): Adjust.
11408 (instantiate_scev_binary): Likewise.
11409 (instantiate_scev_convert): Likewise.
11410 (instantiate_scev_not): Likewise.
11411 (instantiate_array_ref): Remove.
11412 (instantiate_scev_3): Likewise.
11413 (instantiate_scev_2): Likewise.
11414 (instantiate_scev_1): Likewise.
11415 (instantiate_scev_r): Do not blindly handle N-operand trees.
11416 Do not instantiate array-refs. Handle all constants and invariants.
11417 (instantiate_scev): Make instantiate_below parameter
11418 the edge into the region.
11419 (resolve_mixers): Use the loop preheader edge for the region
11420 parameter to instantiate_scev_r.
11421 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
11423 2017-10-13 Richard Biener <rguenther@suse.de>
11425 PR tree-optimization/82525
11426 * graphite-isl-ast-to-gimple.c
11427 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
11429 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
11430 Fail code generation when we cannot represent the isl integer.
11431 (binary_op_to_tree): Elide modulo operations that are no-ops
11432 in the type we code generate. Remove now superfluous code
11435 2017-10-13 Richard Biener <rguenther@suse.de>
11437 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
11438 (scop_detection::harmful_loop_in_region): Remove premature
11439 IV type restriction.
11440 (scop_detection::graphite_can_represent_scev): We can handle
11441 pointer IVs just fine.
11443 2017-10-13 Alan Modra <amodra@gmail.com>
11445 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
11446 "Clobbers and Scratch Registers". Add paragraph on
11447 alternative to clobbers for scratch registers and OpenBLAS
11450 2017-10-13 Alan Modra <amodra@gmail.com>
11452 * doc/extend.texi (Clobbers): Correct vax example. Delete old
11453 example of a memory input for a string of known length. Move
11454 commentary out of table. Add a number of new examples
11455 covering array memory inputs.
11457 2017-10-12 Martin Liska <mliska@suse.cz>
11459 PR tree-optimization/82493
11460 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
11461 (test_range_functions): New function.
11462 (sbitmap_c_tests): Likewise.
11463 * selftest-run-tests.c (selftest::run_tests): Run new tests.
11464 * selftest.h (sbitmap_c_tests): New function.
11466 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11468 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11470 * config/rs6000/amo.h: Fix spacing issue.
11472 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11475 * config/i386/i386.md (*ashl<mode>3_mask_1,
11476 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
11477 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
11480 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11482 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
11483 (profile_probability): Set max_probability
11484 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
11486 * profile-count.c (profile_probability::differs_from_p): Do not
11487 rely on max_probaiblity == 10000
11489 2017-10-12 Jeff Law <law@redhat.com>
11491 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
11494 2017-10-12 Martin Sebor <msebor@redhat.com>
11498 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
11499 (handle_alias_pairs): Call it.
11500 * common.opt (-Wattribute-alias): New option.
11501 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
11502 * doc/invoke.texi (-Wattribute-alias): Document.
11504 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
11507 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11509 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11511 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11512 (make_hard_regno_born, make_hard_regno_dead): Update
11513 bb_killed_pseudos and bb_gen_pseudos.
11515 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11517 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
11519 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
11521 * config/alpha/alpha.c (alpha_split_conditional_move):
11522 Use std::swap instead of manually swapping.
11523 (alpha_stdarg_optimize_hook): Ditto.
11524 (alpha_canonicalize_comparison): Ditto.
11526 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11528 * tree-loop-distribution.c (struct builtin_info): New struct.
11529 (struct partition): Refactor fields into struct builtin_info.
11530 (partition_free): Free struct builtin_info.
11531 (build_size_arg_loc, build_addr_arg_loc): Delete.
11532 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
11533 information from struct builtin_info.
11534 (find_single_drs): New function refactored from classify_partition.
11535 Also moved builtin validity checks to this function.
11536 (compute_access_range, alloc_builtin): New functions.
11537 (classify_builtin_st, classify_builtin_ldst): New functions.
11538 (classify_partition): Refactor code into functions find_single_drs,
11539 classify_builtin_st and classify_builtin_ldst.
11540 (distribute_loop): Don't do runtime alias check when distributing
11542 (find_seed_stmts_for_distribution): New function.
11543 (pass_loop_distribution::execute): Refactor code finding seed
11544 stmts into above function. Support distribution for the innermost
11545 two-level loop nest. Adjust dump information.
11547 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11549 * tree-loop-distribution.c: Adjust the general comment.
11550 (NUM_PARTITION_THRESHOLD): New macro.
11551 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
11552 (classify_partition): Skip builtin pattern of loop nest's inner loop.
11553 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
11554 in call to build_partition_graph.
11555 (finalize_partitions): New parameter. Make loop distribution more
11556 conservative by fusing more partitions.
11557 (distribute_loop): Don't do runtime alias check in case of loop nest
11559 (find_seed_stmts_for_distribution): New function.
11560 (prepare_perfect_loop_nest): New function.
11561 (pass_loop_distribution::execute): Refactor code finding seed stmts
11562 and loop nest into above functions. Support loop nest distribution.
11563 Adjust dump information accordingly.
11565 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11567 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
11568 and set PTYPE_SEQUENTIAL for merged partition.
11570 2017-10-12 Richard Biener <rguenther@suse.de>
11572 PR tree-optimization/69728
11574 2017-09-19 Richard Biener <rguenther@suse.de>
11576 PR tree-optimization/69728
11577 * graphite-sese-to-poly.c (schedule_error): New global.
11578 (add_loop_schedule): Handle empty domain by failing the
11580 (build_original_schedule): Handle schedule_error.
11582 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
11583 domain by returning an unchanged schedule.
11585 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11587 * genrecog.c (validate_pattern): For VEC_SELECT verify that
11588 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
11590 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
11592 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
11595 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11598 * expr.c (store_field): Don't optimize away bitsize == 0 store
11599 from CALL_EXPR with addressable return type.
11601 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11603 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
11604 * config/rs6000/rs6000.md (sel): Delete mode attribute.
11605 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
11606 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
11607 TARGET_ISEL instead of TARGET_ISEL<sel>.
11609 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
11611 * config/rs6000/rs6000.c
11612 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
11614 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11616 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
11618 (reg_or_cint_operand): Fix comment.
11619 (reg_or_zero_operand): New predicate.
11620 * config/rs6000/rs6000-protos.h (output_isel): Delete.
11621 * config/rs6000/rs6000.c (output_isel): Delete.
11622 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
11623 instead of reg_or_cint_operand. Output instruction directly (not via
11625 (isel_unsigned_<mode>): Ditto.
11626 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
11627 gpc_reg_operand. Add an instruction alternative for this. Output
11628 instruction directly.
11629 (*isel_reversed_unsigned_<mode>): Ditto.
11631 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11633 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
11634 (TARGET_CANONICALIZE_COMPARISON): Define.
11636 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
11639 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
11640 Check whether the dest is REG before adding REG_EQUIV note.
11642 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11645 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11647 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11648 (make_hard_regno_born, make_hard_regno_dead): Update
11649 bb_killed_pseudos and bb_gen_pseudos.
11651 2017-10-11 Nathan Sidwell <nathan@acm.org>
11653 * incpath.h (enum incpath_kind): Name enum, prefix values.
11654 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
11655 * incpath.c (heads, tails): Use INC_MAX.
11656 (add_env_var_paths, add_standard_paths): Use incpath_kind.
11657 (merge_include_chains, split_quote_chain,
11658 register_include_chains): Update incpath_kind names.
11659 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
11660 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
11662 (add_framework_path, darwin_register_objc_includes): Likewise.
11663 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
11665 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11667 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
11668 Do not use float_operator operator predicate.
11669 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
11670 * config/i386/predicates.md (float_operator): Remove predicate.
11672 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11674 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
11675 (*jccxf_i387): Ditto.
11676 (*jcc<mode>_i387): Ditto.
11677 (*jccu<mode>_i387): Ditto.
11678 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
11679 (*jcc_*_i387 splitters): Remove.
11680 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
11681 * config/i386/i386.c (ix86_split_fp_branch): Remove.
11682 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
11685 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11687 * profile-count.h (slow_safe_scale_64bit): New function.
11688 (safe_scale_64bit): New inline.
11689 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
11690 * profile-count.c: Include wide-int.h
11691 (slow_safe_scale_64bit): New.
11693 2017-10-11 Nathan Sidwell <nathan@acm.org>
11695 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
11696 HAS_DECL_ASSEMBLER_NAME_P.
11697 * gimple-expr.c (gimple_decl_printable_name: Check
11698 HAS_DECL_ASSEMBLER_NAME_P too.
11699 * ipa-utils.h (type_in_anonymous_namespace_p): Check
11700 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
11701 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
11702 * passes.c (rest_of_decl_compilation): Check
11703 HAS_DECL_ASSEMBLER_NAME_P too.
11704 * recog.c (verify_changes): Likewise.
11705 * tree-pretty-print.c (dump_decl_name): Likewise.
11706 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
11708 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
11709 (SET_DECL_ASSEMBLER_NAME): Use it.
11710 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
11711 (COPY_DECL_ASSEMBLER_NAME): Likewise.
11712 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
11714 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11716 * config.gcc (i386, x86_64): Add extra objects.
11717 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
11718 (ix86_min_insn_size): Declare.
11719 (ix86_issue_rate): Declare.
11720 (ix86_adjust_cost): Declare.
11721 (ia32_multipass_dfa_lookahead): Declare.
11722 (ix86_macro_fusion_p): Declare.
11723 (ix86_macro_fusion_pair_p): Declare.
11724 (ix86_bd_has_dispatch): Declare.
11725 (ix86_bd_do_dispatch): Declare.
11726 (ix86_core2i7_init_hooks): Declare.
11727 (ix86_atom_sched_reorder): Declare.
11728 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
11729 (COSTS_N_BYTES): Move to x86-tune-costs.h.
11730 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
11731 (rip_relative_addr_p): Rename to ...
11732 (ix86_rip_relative_addr_p): ... this one; export.
11733 (memory_address_length): Update.
11734 (ix86_issue_rate): Move to x86-tune-sched.c.
11735 (ix86_flags_dependent): Move to x86-tune-sched.c.
11736 (ix86_agi_dependent): Move to x86-tune-sched.c.
11737 (exact_dependency_1): Move to x86-tune-sched.c.
11738 (exact_store_load_dependency): Move to x86-tune-sched.c.
11739 (ix86_adjust_cost): Move to x86-tune-sched.c.
11740 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
11741 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
11742 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
11743 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
11744 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
11745 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
11746 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
11747 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
11748 (min_insn_size): Rename to ...
11749 (ix86_min_insn_size): ... this one; export.
11750 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
11751 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
11752 (core2i7_first_cycle_multipass_backtrack): Move to
11753 x86-tune-sched-core.c.
11754 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
11755 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
11756 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
11757 (ix86_avoid_jump_mispredicts): Update.
11758 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
11759 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
11760 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
11761 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
11762 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
11763 (MAX_INSN): Move to ix86-tune-sched-bd.c.
11764 (MAX_IMM): Move to ix86-tune-sched-bd.c.
11765 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
11766 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
11767 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
11768 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
11769 (MAX_STORE): Move to ix86-tune-sched-bd.c.
11770 (BIG): Move to ix86-tune-sched-bd.c.
11771 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
11772 (enum insn_path): Move to ix86-tune-sched-bd.c.
11773 (get_mem_group): Move to ix86-tune-sched-bd.c.
11774 (is_cmp): Move to ix86-tune-sched-bd.c.
11775 (dispatch_violation): Move to ix86-tune-sched-bd.c.
11776 (is_branch): Move to ix86-tune-sched-bd.c.
11777 (is_prefetch): Move to ix86-tune-sched-bd.c.
11778 (init_window): Move to ix86-tune-sched-bd.c.
11779 (allocate_window): Move to ix86-tune-sched-bd.c.
11780 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
11781 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
11782 (process_end_window): Move to ix86-tune-sched-bd.c.
11783 (allocate_next_window): Move to ix86-tune-sched-bd.c.
11784 (find_constant): Move to ix86-tune-sched-bd.c.
11785 (get_num_immediates): Move to ix86-tune-sched-bd.c.
11786 (has_immediate): Move to ix86-tune-sched-bd.c.
11787 (get_insn_path): Move to ix86-tune-sched-bd.c.
11788 (get_insn_group): Move to ix86-tune-sched-bd.c.
11789 (count_num_restricted): Move to ix86-tune-sched-bd.c.
11790 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
11791 (add_insn_window): Move to ix86-tune-sched-bd.c.
11792 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
11793 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
11794 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
11795 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
11796 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
11797 (do_dispatch): Move to ix86-tune-sched-bd.c.
11798 (has_dispatch): Move to ix86-tune-sched-bd.c.
11799 * config/i386/t-i386: Add new object files.
11800 * config/i386/x86-tune-costs.h: New file.
11801 * config/i386/x86-tune-sched-atom.c: New file.
11802 * config/i386/x86-tune-sched-bd.c: New file.
11803 * config/i386/x86-tune-sched-core.c: New file.
11804 * config/i386/x86-tune-sched.c: New file.
11806 2017-10-11 Liu Hao <lh_mouse@126.com>
11808 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
11809 the generic version below instead.
11810 (should_colorize): Recognize Windows consoles as terminals
11812 * pretty-print.c [__MINGW32__] (write_all): New function.
11813 [__MINGW32__] (find_esc_head): Likewise.
11814 [__MINGW32__] (find_esc_terminator): Likewise.
11815 [__MINGW32__] (eat_esc_sequence): Likewise.
11816 [__MINGW32__] (mingw_ansi_fputs): New function that handles
11818 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
11821 2017-10-11 Richard Biener <rguenther@suse.de>
11823 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
11824 Properly call analyze_scalar_evolution with the loop of the stmt.
11826 2017-10-11 Richard Biener <rguenther@suse.de>
11828 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
11829 * tree-core.h (tree_base): Add chrec_var union member.
11830 * tree.h (CHREC_VAR): Remove.
11831 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
11832 * tree-chrec.h (build_polynomial_chrec): Adjust.
11833 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
11834 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
11836 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
11838 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
11839 * match.pd: ... here.
11840 ((T) X == (T) Y): Relax condition.
11842 2017-10-11 Bin Cheng <bin.cheng@arm.com>
11844 PR tree-optimization/82472
11845 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
11847 (break_alias_scc_partitions): Update postorder number.
11849 2017-10-11 Martin Liska <mliska@suse.cz>
11852 * opts.c (parse_no_sanitize_attribute): Do not use error_value
11854 * opts.h (parse_no_sanitize_attribute): Remove last argument.
11856 2017-10-11 Martin Liska <mliska@suse.cz>
11858 * print-rtl.c (print_insn): Move declaration of idbuf
11859 to same scope as name.
11861 2017-10-11 Martin Liska <mliska@suse.cz>
11866 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11867 volatile arguments.
11869 2017-10-11 Martin Liska <mliska@suse.cz>
11872 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11873 volatile arguments.
11875 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
11877 * config.gcc (default_gnu_indirect_function): Default to yes for
11878 arm*-*-linux* with glibc.
11880 2017-10-11 Richard Biener <rguenther@suse.de>
11882 * tree-scalar-evolution.c (get_scalar_evolution): Handle
11883 default-defs and types we do not want to analyze.
11884 (interpret_loop_phi): Replace unreachable code with an assert.
11885 (compute_scalar_evolution_in_loop): Remove and inline ...
11886 (analyze_scalar_evolution_1): ... here, replacing condition with
11887 what makes the intent clearer. Remove handling of cases
11888 get_scalar_evolution now handles.
11890 2017-10-10 Jim Wilson <wilson@tuliptree.org>
11892 PR rtl-optimization/81434
11893 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
11894 comment for main loop. In sched_group_found if, also add checks for
11895 pass and min_cost_group.
11897 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
11899 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
11900 (rs6000_insn_cost): New function.
11901 * config/rs6000/rs6000.md (cost): New attribute.
11903 2017-10-10 Jakub Jelinek <jakub@redhat.com>
11904 H.J. Lu <hongjiu.lu@intel.com>
11908 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
11909 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
11910 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
11911 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
11912 to OPTION_MASK_ISA_AVX512VL - builtins that have both
11913 OPTION_MASK_ISA_MMX and some other bit set require both
11914 mmx and the ISAs without the mmx bit.
11915 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
11916 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
11917 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
11918 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
11919 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
11920 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
11921 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
11922 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
11923 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
11924 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
11925 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
11926 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
11927 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
11928 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
11929 Add OPTION_MASK_ISA_MMX.
11931 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
11933 * config.gcc (armv7*-*-freebsd*): New target.
11934 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
11936 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
11938 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
11939 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
11942 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
11944 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
11945 a template parameter.
11946 (WIDE_INT_REF_FOR): Update accordingly.
11947 * tree.h (wi::int_traits <const_tree>): Delete.
11948 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
11949 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
11950 (wi::tree_to_wide_ref): New typedef.
11951 (wi::to_wide): New function.
11952 * calls.c (get_size_range): Use wi::to_wide when operating on
11953 trees as wide_ints.
11954 * cgraph.c (cgraph_node::create_thunk): Likewise.
11955 * config/i386/i386.c (ix86_data_alignment): Likewise.
11956 (ix86_local_alignment): Likewise.
11957 * dbxout.c (stabstr_O): Likewise.
11958 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
11959 * expr.c (const_vector_from_tree): Likewise.
11960 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
11961 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
11962 (fold_negate_expr_1, int_const_binop_1, const_binop)
11963 (fold_convert_const_int_from_real, optimize_bit_field_compare)
11964 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
11965 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
11966 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
11967 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
11968 (fold_not_const, round_up_loc): Likewise.
11969 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
11970 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
11971 (alloca_call_type): Likewise.
11972 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
11973 * godump.c (go_output_typedef): Likewise.
11974 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
11975 * internal-fn.c (get_min_precision): Likewise.
11976 * ipa-cp.c (ipcp_store_vr_results): Likewise.
11977 * ipa-polymorphic-call.c
11978 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
11979 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
11980 (ipa_modify_call_arguments): Likewise.
11981 * match.pd: Likewise.
11982 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
11983 * print-tree.c (print_node_brief, print_node): Likewise.
11984 * stmt.c (expand_case): Likewise.
11985 * stor-layout.c (layout_type): Likewise.
11986 * tree-affine.c (tree_to_aff_combination): Likewise.
11987 * tree-cfg.c (group_case_labels_stmt): Likewise.
11988 * tree-data-ref.c (dr_analyze_indices): Likewise.
11989 (prune_runtime_alias_test_list): Likewise.
11990 * tree-dump.c (dequeue_and_dump): Likewise.
11991 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
11992 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
11993 * tree-pretty-print.c (dump_generic_node): Likewise.
11994 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
11995 (simple_iv_with_niters): Likewise.
11996 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
11997 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
11998 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
11999 * tree-ssa-loop-niter.c (split_to_var_and_offset)
12000 (refine_value_range_using_guard, number_of_iterations_ne_max)
12001 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
12002 (get_cst_init_from_scev, record_nonwrapping_iv)
12003 (scev_var_range_cant_overflow): Likewise.
12004 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
12005 * tree-ssa-pre.c (compute_avail): Likewise.
12006 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
12007 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
12008 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
12009 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
12010 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
12011 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
12012 (dump_case_nodes, try_switch_expansion): Likewise.
12013 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
12014 (vect_do_peeling): Likewise.
12015 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
12016 * tree-vect-stmts.c (vectorizable_load): Likewise.
12017 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
12018 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
12019 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
12020 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
12021 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
12022 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
12023 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
12024 (evrp_dom_walker::before_dom_children): Likewise.
12025 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
12026 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
12027 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
12028 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
12029 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
12030 (get_range_pos_neg): Likewise.
12031 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
12032 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
12033 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
12034 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
12035 * config/avr/avr.c (avr_fold_builtin): Likewise.
12036 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
12037 * config/msp430/msp430.c (msp430_attr): Likewise.
12038 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
12039 * config/powerpcspe/powerpcspe-c.c
12040 (altivec_resolve_overloaded_builtin): Likewise.
12041 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
12042 (rs6000_expand_ternop_builtin): Likewise.
12043 * config/rs6000/rs6000-c.c
12044 (altivec_resolve_overloaded_builtin): Likewise.
12045 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
12046 (rs6000_expand_ternop_builtin): Likewise.
12047 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
12049 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12051 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
12052 when copying loop nest with only one inner loop.
12054 2017-10-10 Richard Biener <rguenther@suse.de>
12056 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
12057 blocks if SCEV is active.
12058 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
12060 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
12061 (scev_initialize): Assert we are not yet initialized.
12063 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12065 * tree-loop-distribution.c (generate_loops_for_partition): Remove
12066 inner loop's exit stmt by making it always exit the loop, otherwise
12067 we would generate an infinite empty loop.
12069 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12071 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
12072 renaming variables in new preheader if it's deleted.
12074 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12076 * tree-loop-distribution.c (struct partition): Remove unused field
12077 loops of the structure.
12078 (partition_alloc, partition_free): Ditto.
12079 (build_rdg_partition_for_vertex): Ditto.
12081 2017-10-09 Jeff Law <law@redhat.com>
12083 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
12084 return type to match prototype and documentation.
12086 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12088 * config/rs6000/rs6000.c (processor_costs): Move to ...
12089 * config/rs6000/rs6000.h: ... here.
12090 (rs6000_cost): Declare.
12092 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
12094 * except.c (setjmp_fn): New global variable.
12095 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
12096 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
12097 if DONT_USE_BUILTIN_SETJMP is defined.
12099 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12101 * target.def (insn_cost): New hook.
12102 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
12103 * doc/tm.texi: Regenerate.
12104 * rtlanal.c (insn_cost): Use the new hook.
12106 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12108 * combine.c (combine_validate_cost): Compute the new insn_cost,
12109 not just pattern_cost.
12110 (try_combine): Adjust comment.
12112 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12114 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
12116 * combine.c (uid_insn_cost): Adjust comment.
12117 (combine_validate_cost): Adjust comment. Use pattern_cost instead
12119 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
12120 * dse.c (find_shift_sequence): Ditto.
12121 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
12122 (bb_valid_for_noce_process_p): Use pattern_cost.
12123 * rtl.h (insn_rtx_cost): Delete.
12124 (pattern_cost): New prototype.
12125 (insn_cost): New prototype.
12126 * rtlanal.c (insn_rtx_cost): Rename to...
12127 (pattern_cost): ... this.
12130 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
12132 * config/i386/i386.md (*jcc_2): Remove insn pattern.
12133 (*jcc<mode>_0_r_i387): Ditto.
12134 (*jccxf_r_i387): Ditto.
12135 (*jcc<mode>_r_i387): Ditto.
12136 (*jccu<mode>_r_i387): Ditto.
12137 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
12138 (*jcc): Rename from *jcc_1.
12140 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12142 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
12143 deferred rescans after the lvx/stvx recombination pre-pass.
12145 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
12147 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
12148 memory operation instruction support.
12149 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
12150 (rs6000-ibm-aix[789]*): Likewise.
12151 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
12152 Document new functions.
12154 2017-10-09 Richard Biener <rguenther@suse.de>
12156 PR tree-optimization/82397
12157 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
12158 equality only for semantically equal trees.
12160 2017-10-09 Richard Biener <rguenther@suse.de>
12162 PR tree-optimization/82449
12163 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
12164 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
12165 allow constant addresses.
12166 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
12169 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12171 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
12174 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12177 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
12180 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12183 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
12185 2017-10-09 Jakub Jelinek <jakub@redhat.com>
12188 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
12189 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
12191 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
12193 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
12194 (WI_BINARY_PREDICATE_RESULT): Likewise.
12195 (wi::binary_traits::operator_result): New type.
12196 (wi::binary_traits::predicate_result): Likewise.
12197 (generic_wide_int::operator~, unary generic_wide_int::operator-)
12198 (generic_wide_int::operator==, generic_wide_int::operator!=)
12199 (generic_wide_int::operator&, generic_wide_int::and_not)
12200 (generic_wide_int::operator|, generic_wide_int::or_not)
12201 (generic_wide_int::operator^, generic_wide_int::operator+
12202 (binary generic_wide_int::operator-, generic_wide_int::operator*):
12204 (operator~, unary operator-, operator==, operator!=, operator&)
12205 (operator|, operator^, operator+, binary operator-, operator*): New
12207 * expr.c (get_inner_reference): Use wi::bit_and_not.
12208 * fold-const.c (fold_binary_loc): Likewise.
12209 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
12210 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
12211 (bit_value_binop): Likewise.
12212 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
12213 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
12214 (extract_range_from_binary_expr_1): Likewise.
12215 (masked_increment): Likewise.
12216 (simplify_bit_ops_using_ranges): Likewise.
12218 2017-10-09 Martin Jambor <mjambor@suse.cz>
12221 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
12222 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
12223 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
12225 (hsa_fixup_mov_insn_type): New function.
12226 (hsa_op_with_type::get_in_type): Use it.
12227 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
12228 immediates in an assert.
12229 (hsa_op_with_type::extend_int_to_32bit): New method.
12230 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
12231 types. Convert to dest type if necessary.
12232 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
12233 (reg_for_gimple_ssa): Pass false as min32int to
12234 hsa_type_for_scalar_tree_type.
12235 (gen_hsa_addr): Fixup type when creating addresable temporary.
12236 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
12237 (gen_hsa_unary_operation): Extend operands and convert to dest type if
12238 necessary. Call hsa_fixup_mov_insn_type.
12239 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
12240 extend operands and convert to dest type if necessary.
12241 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
12242 to dest type if necessary.
12243 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
12244 if conversion nt necessary and size matches.
12245 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
12246 to dest type if necessary.
12247 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
12248 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
12250 (gen_hsa_clrsb): Likewise.
12251 (gen_hsa_ffs): Likewise.
12252 (gen_hsa_divmod): Extend operands and convert to dest type if
12254 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
12256 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12258 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
12259 Remove empty default arguments. Use a brace block as output
12261 (conditional return): Ditto.
12263 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
12264 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
12266 (group_ending_nop): Ditto.
12267 (doloop_end): Ditto.
12268 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
12269 (splitters for those): Ditto.
12271 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12273 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
12274 a conditional jump (and the compare for it) so that pc_rtx is the
12276 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
12277 for the deleted and renamed ctr<mode>_internal[234] patterns.
12278 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
12279 Delete second conditional return pattern.
12280 (ctr<mode>_internal2): Delete this second bdnz pattern.
12281 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
12282 (ctr<mode>_internal4): Delete this second bdz pattern.
12284 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12286 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
12287 (always_initialized_rtx_for_ssa_name_p): New predicate.
12288 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
12289 (finish_out_of_ssa): Free new field of SA.
12290 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
12291 * tree-ssa-coalesce.c: Include tree-ssa.h.
12292 (get_parm_default_def_partitions): Remove extern keyword.
12293 (get_undefined_value_partitions): New function.
12294 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
12295 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
12296 uninitialized bits.
12297 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
12299 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12301 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
12303 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12305 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
12306 for targets that preffer 128bit.
12308 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12310 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
12312 2017-10-08 Olivier Hainque <hainque@adacore.com>
12314 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
12315 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
12317 (thumb_set_return_address): Likewise.
12319 2017-10-08 Olivier Hainque <hainque@adacore.com>
12321 * common/config/arm/arm-common.c (arm_except_unwind_info):
12322 Handle DWARF2_UNWIND_INFO.
12324 2017-10-07 Michael Collison <michael.collison@arm.com>
12326 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
12329 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
12331 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
12332 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
12333 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
12334 defined, force the creation of a new block for a dispatch label.
12336 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
12338 * invoke.texi (Wsuggest-attribute=cold): Document.
12339 * common.opt (Wsuggest-attribute=cold): New
12340 * ipa-pure-const.c (warn_function_cold): New function.
12341 * predict.c (compute_function_frequency): Use it.
12342 * predict.h (warn_function_cold): Declare.
12344 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
12346 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
12349 2017-10-06 Martin Liska <mliska@suse.cz>
12351 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
12352 keyword for member functions.
12353 (struct sanopt_tree_couple): New struct.
12354 (struct sanopt_tree_couple_hash): New function.
12355 (struct sanopt_ctx): Add new hash_map.
12356 (has_dominating_ubsan_ptr_check): New function.
12357 (record_ubsan_ptr_check_stmt): Likewise.
12358 (maybe_optimize_ubsan_ptr_ifn): Likewise.
12359 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
12360 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
12362 2017-10-06 Sudakshina Das <sudi.das@arm.com>
12365 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
12366 aarch64_simd_valid_immediate on CONST_VECTORs.
12367 (aarch64_reg_or_bic_imm): Likewise.
12369 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
12371 PR rtl-optimization/82396
12372 * haifa-sched.c (ready_sort_real): Disable qsort checking.
12374 2017-10-06 Sebastian Pop <sebpop@gmail.com>
12376 * graphite-dependences.c (scop_get_reads): Move code to...
12377 (scop_get_must_writes): Move code to...
12378 (scop_get_may_writes): Move code to...
12379 (scop_get_reads_and_writes): ... here.
12380 (scop_get_dependences): Call scop_get_reads_and_writes.
12382 2017-10-06 Jakub Jelinek <jakub@redhat.com>
12384 PR tree-optimization/82434
12385 * fold-const.h (can_native_encode_type_p,
12386 can_native_encode_string_p): Remove.
12387 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
12388 don't encode anything, just return what would be otherwise returned.
12389 (native_encode_fixed, native_encode_complex, native_encode_vector):
12391 (native_encode_string): Likewise. Inline by hand
12392 can_native_encode_string_p.
12393 (can_native_encode_type_p): Remove.
12394 (can_native_encode_string_p): Remove.
12395 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
12396 STRING_CSTs using can_native_encode_string_p, test all
12397 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
12398 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
12399 argument from native_encode_expr.
12400 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
12401 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
12404 2017-10-06 Richard Biener <rguenther@suse.de>
12406 PR tree-optimization/82397
12407 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
12408 operand_equal_p but rely on data_ref_compare_tree for detecting
12410 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
12411 to match up with dr_group_sort_cmp.
12413 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12416 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
12418 * config/s390/s390-builtin-types.def: Regenerate.
12420 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12423 * config/s390/s390-builtin-types.def: Regenerate.
12424 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
12425 Change flag from B_VXE to B_VX.
12426 (s390_vec_min_dbl): Remove B_VXE flag.
12428 2017-10-06 Richard Biener <rguenther@suse.de>
12430 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
12431 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
12432 translate_isl_ast_to_gimple::is_valid_rename,
12433 translate_isl_ast_to_gimple::get_rename,
12434 translate_isl_ast_to_gimple::get_def_bb_for_const,
12435 translate_isl_ast_to_gimple::get_new_name,
12436 translate_isl_ast_to_gimple::collect_all_ssa_names,
12437 translate_isl_ast_to_gimple::copy_loop_phi_args,
12438 translate_isl_ast_to_gimple::collect_all_ssa_names,
12439 translate_isl_ast_to_gimple::copy_loop_phi_args,
12440 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
12441 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
12442 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
12443 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
12444 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
12445 translate_isl_ast_to_gimple::copy_cond_phi_args,
12446 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
12447 translate_isl_ast_to_gimple::edge_for_new_close_phis,
12448 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
12449 translate_isl_ast_to_gimple::rename_uses,
12450 translate_isl_ast_to_gimple::rename_all_uses): Remove.
12451 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
12452 (set_rename_for_each_def): Likewise.
12453 (graphite_copy_stmts_from_block): Handle debug stmt resetting
12454 here. Handle rewriting SCEV analyzable uses here.
12455 (copy_bb_and_scalar_dependences): Generate code for PHI
12457 (graphite_regenerate_ast_isl): Adjust.
12458 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
12459 (add_write, add_read): New functions.
12460 (build_cross_bb_scalars_def): Use it and simplify.
12461 (build_cross_bb_scalars_use): Likewise.
12462 (graphite_find_cross_bb_scalar_vars): Inline into...
12463 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
12464 simulating out-of-SSA. Compute liveout and add dependencies.
12465 (build_scops): Force an empty entry block.
12466 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
12468 (sese_build_liveouts): Declare.
12469 (sese_trivially_empty_bb_p): Likewise.
12470 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
12471 compute liveout and debug_liveout.
12472 (sese_bad_liveouts_use): Remove.
12473 (sese_reset_debug_liveouts_bb): Likewise.
12474 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
12475 (sese_build_liveouts): Build liveout and debug_liveout and store
12477 (new_sese_info): Adjust.
12478 (free_sese_info): Likewise.
12479 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
12480 do not build liveout here.
12481 (move_sese_in_condition): Adjust region entry.
12482 (scev_analyzable_p): Match up with chrec_apply requirements.
12483 (sese_trivially_empty_bb_p): New.
12484 * tree-into-ssa.c (get_reaching_def): Properly support generating
12485 default-defs for incremental rewrite of anonymous names.
12487 2017-10-06 Richard Biener <rguenther@suse.de>
12489 * graphite-sese-to-poly.c (extract_affine): For casts increasing
12490 precision do not perform modulo reduction.
12492 2017-10-06 Richard Biener <rguenther@suse.de>
12494 PR tree-optimization/82436
12495 * tree-vect-slp.c (vect_supported_load_permutation_p): More
12496 conservatively choose the vectorization factor when checking
12497 whether we can perform the required load permutation.
12498 (vect_transform_slp_perm_load): Assert when we may not fail.
12500 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
12502 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
12503 message for incompatible -msdata=* and -mcall-* options.
12505 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12507 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
12508 rate for post-reload scheduling.
12510 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12512 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
12514 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12516 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
12517 to improve monte carlo in scimark.
12519 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12521 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
12522 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
12523 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
12524 pentium4_cost, nocona_cost): Set reassociation width to 1.
12525 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
12526 width to 2 for fp operations and 1 otherwise.
12527 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
12529 (atom_cost): Set reassociation width to 2.
12530 (slm_cost, generic_cost): Set fp reassociation width
12531 to 2 and 1 otherwise.
12532 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
12533 (core_cost): Set fp reassociation width to 4 and vector to 2.
12534 (ix86_reassociation_width): Rewrite using cost table; special case
12535 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
12536 and TARGET_AVX128_OPTIMAL.
12537 * config/i386/i386.h (processor_costs): Add
12538 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
12539 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
12540 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
12541 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
12542 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
12543 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
12545 2017-10-05 Nathan Sidwell <nathan@acm.org>
12547 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
12549 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12551 * config/arm/arm.c (arm_test_fpu_data): New.
12552 (arm_run_selftests): Call arm_test_fpu_data.
12554 2017-10-04 Nathan Sidwell <nathan@acm.org>
12556 * toplev.c (toplev::main): Remove excess parens on pretty_printer
12558 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
12560 2017-10-04 Sudakshina Das <sudi.das@arm.com>
12562 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
12563 check type for aarch64_simd_valid_immediate.
12564 (aarch64_output_simd_mov_immediate): Update prototype.
12565 (aarch64_simd_valid_immediate): Update prototype.
12566 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
12567 support for ORR-immediate.
12568 (and<mode>3): modified pattern to add support for BIC-immediate.
12569 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
12570 now checks for valid immediate for BIC and ORR based on new enum
12572 (aarch64_output_simd_mov_immediate): Function now used to output
12573 BIC/ORR imm as well based on new enum argument.
12574 * config/aarch64/constraints.md (Do): New vector immediate constraint.
12576 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
12577 (aarch64_reg_or_bic_imm): Likewise.
12579 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12581 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
12582 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
12584 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12588 PR rtl-optimization/82396
12589 * haifa-sched.c (autopref_multipass_init): Simplify
12591 (autopref_rank_data): Simplify sort order.
12592 * sched-int.h (autopref_multipass_data_): Remove
12593 multi_mem_insn_p, min_offset and max_offset.
12595 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12597 * doc/sourcebuild.texi: Document vect_peeling_profitable.
12599 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12601 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
12602 vect_doubleint_cvt.
12604 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12606 * doc/sourcebuild.texi: Document vect_long_mult.
12608 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
12610 PR tree-optimization/82413
12611 * fold-const.c (build_range_check): Use widest_int when comparing
12612 the maximum ETYPE value with HIGH.
12614 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12616 PR rtl-optimization/82396
12617 * haifa-sched.c (autopref_multipass_init): Simplify
12619 (autopref_rank_data): Simplify sort order.
12620 * sched-int.h (autopref_multipass_data_): Remove
12621 multi_mem_insn_p, min_offset and max_offset.
12623 2017-10-04 Jakub Jelinek <jakub@redhat.com>
12625 PR tree-optimization/82381
12626 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
12627 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
12630 PR tree-optimization/82374
12631 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
12632 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
12633 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
12634 current_function_decl to the new decl.
12636 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
12638 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
12639 helper macro for IEEE float128 hardware built-in functions.
12640 (SQRTF128_ODD): Add built-in functions with the round-to-odd
12642 (TRUNCF128_ODD): Likewise.
12643 (ADDF128_ODD): Likewise.
12644 (SUBF128_ODD): Likewise.
12645 (MULF128_ODD): Likewise.
12646 (DIVF128_ODD): Likewise.
12647 (FMAF128_ODD): Likewise.
12648 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
12649 UNSPEC_TRUNC_ROUND_TO_ODD.
12650 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
12651 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
12652 floating point round to odd instructions.
12653 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
12654 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
12655 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
12656 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
12657 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
12658 (trunc<mode>sf2_hw): Change the truncate with round to odd
12659 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
12660 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
12661 to odd hardware instructions.
12662 (sub<mode>3_odd): Likewise.
12663 (mul<mode>3_odd): Likewise.
12664 (div<mode>3_odd): Likewise.
12665 (sqrt<mode>2_odd): Likewise.
12666 (fma<mode>4_odd): Likewise.
12667 (fms<mode>4_odd): Likewise.
12668 (nfma<mode>4_odd): Likewise.
12669 (nfms<mode>4_odd): Likewise.
12670 (trunc<mode>df2_odd): Change the truncate with round to odd
12671 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
12673 * doc/extend.texi (PowerPC built-in functions): Update documentation
12674 for existing IEEE float128-bit built-in functions. Add built-in
12675 functions that generate the IEEE 128-bit floating point round to
12678 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
12680 PR rtl-optimization/77729
12681 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
12682 to (X&(C1&~C2))|C2 transformations.
12684 2017-10-03 Martin Jambor <mjambor@suse.cz>
12686 PR tree-optimization/82363
12687 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
12688 mismatch, mark lacc written regardless of racc.
12690 2017-10-03 Jakub Jelinek <jakub@redhat.com>
12692 PR tree-optimization/82381
12693 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
12694 stmt_to_insert nor wheather SSA_NAMEs are default defs.
12695 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
12696 fallthrough into reassoc_stmt_dominates_stmt_p.
12699 * combine.c (combine_instructions): Don't combine in unreachable
12702 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
12705 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
12706 function to not use the have_cpu variable. Do not set cpu_index,
12707 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
12708 or the default cpu.
12709 (rs6000_valid_attribute_p): Remove duplicate initializations of
12710 old_optimize and func_optimize.
12711 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
12712 (rs6000_activate_target_options): Make global.
12713 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
12716 2017-10-02 Jakub Jelinek <jakub@redhat.com>
12718 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
12719 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
12720 Set *poffset to 0 and *psize and *pmax_size to -1 if
12721 *poffset + *psize overflows in HOST_WIDE_INT.
12723 PR tree-optimization/82387
12724 PR tree-optimization/82388
12725 PR tree-optimization/82389
12726 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
12727 instead of live_bytes non-NULL.
12729 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
12732 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
12735 2017-10-02 Richard Biener <rguenther@suse.de>
12737 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
12738 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
12739 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
12741 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12743 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
12744 requested precision matches the type's.
12745 * calls.c (alloc_max_size): Calculate the new candidate size as
12746 a widest_int and use wi::to_widest when comparing it with the
12747 current candidate size.
12748 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
12749 zero rather than integer_zero_node.
12750 * match.pd: Check for a no-op conversion before using wi::add
12751 rather than after. Use tree_to_uhwi when summing small shift
12752 counts into an unsigned int.
12754 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12755 Alan Hayward <alan.hayward@arm.com>
12756 David Sherwood <david.sherwood@arm.com>
12759 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
12760 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
12761 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
12762 POINTER_AND_FP_REGS.
12764 2017-10-02 Richard Biener <rguenther@suse.de>
12766 PR tree-optimization/82355
12767 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12768 a mapping for the enclosing loop but avoid generating one for
12769 the loop tree root.
12770 (copy_bb_and_scalar_dependences): Remove premature codegen
12771 error on PHIs in blocks duplicated into multiple places.
12772 * graphite-scop-detection.c
12773 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12774 in the region use it as loop and nest to analyze the DR in.
12775 (try_generate_gimple_bb): Likewise.
12776 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12777 (add_loop_constraints): For blocks in a loop not in the region
12778 create a dimension with a single iteration.
12779 * sese.h (gbb_loop_at_index): Remove assert.
12781 2017-10-01 Kevin Buettner <kevinb@redhat.com>
12783 * omp-expand.c (adjust_context_scope): New function.
12784 (expand_parallel_call): Call adjust_context_scope.
12786 2017-10-01 Jeff Law <law@redhat.com>
12788 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
12789 dom_opt_dom_walker class with direct access to private members.
12790 Add comments. Call test_for_singularity.
12791 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
12792 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
12793 m_dummy_cond anymore.
12794 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
12796 (pass_dominator:execute): Build the dummy_cond here and pass it
12797 to the dom_opt_dom_walker ctor.
12798 (test_for_singularity): New function.
12800 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
12801 Maya Rashish <coypu@sdf.org>
12803 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
12804 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
12805 (alpha*-*-netbsd*) Use nbsd_tm_file.
12806 (arm*-*-netbsdelf*) Likewise.
12807 (i[34567]86-*-netbsdelf*) Likewise.
12808 (x86_64-*-netbsd*) Likewise.
12809 (mips*-*-netbsd*) Likewise.
12810 (powerpc-*-netbsd*) Likewise.
12811 (sh*-*-netbsd*) Likewise.
12812 (sparc-*-netbsdelf*) Likewise.
12813 (sparc64-*-netbsd*) Likewise.
12814 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
12816 (vax-*-netbsdelf*) Likewise.
12817 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
12818 (UINT_FAST8_TYPE) Likewise.
12819 (INT_FAST16_TYPE) Check CHAR_FAST16.
12820 (UINT_FAST16_TYPE) Likewise.
12822 2017-09-30 Jakub Jelinek <jakub@redhat.com>
12825 * config/i386/i386.md
12826 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
12827 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
12828 *divmodsi4_zext_2): New define_insn_and_split.
12829 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
12830 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
12831 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
12832 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
12833 New define_insn_and_split.
12834 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
12835 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
12836 operands[1] having DImode when mode is SImode.
12838 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
12839 always SImode for DIV and MOD in REG_EQUAL notes.
12841 2017-09-29 Yury Gribov <tetra2005@gmail.com>
12843 PR middle-end/82319
12844 * match.pd: Fix handling of NaNs in pattern.
12846 2017-09-29 Jeff Law <law@redhat.com>
12848 * sbitmap.c (bitmap_bit_in_range_p): New function.
12849 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
12850 * tree-ssa-dse.c (live_bytes_read): New function.
12851 (dse_classify_store): Ignore reads of dead bytes.
12853 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
12854 typos and whitespace errors.
12855 * config/i386/predicates.md (address_no_seg_operand): Likewise.
12856 * config/s390/s390.c (s390_emit_prologue): Likewise.
12858 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12861 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
12862 with a symbol for LRA.
12864 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12866 PR rtl-optimization/82338
12867 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
12869 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
12871 * genmodes.c (calc_wider_mode): Suppress qsort macro.
12872 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
12873 (qsort_chk): Declare.
12874 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
12875 (qsort_chk): New function.
12877 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12879 PR tree-optimization/82337
12880 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12881 phi definition if the PHI result appears in an abnormal PHI.
12882 (find_basis_for_base_expr): Don't record a basis if the LHS of the
12883 basis appears in an abnormal PHI.
12885 2017-09-29 Richard Biener <rguenther@suse.de>
12887 * graphite-isl-ast-to-gimple.c
12888 (translate_isl_ast_to_gimple::set_codegen_error): New function.
12889 (binary_op_to_tree): Use it.
12890 (get_rename_from_scev): Likewise.
12891 (copy_loop_phi_nodes): Likewise.
12892 (copy_bb_and_scalar_dependences): Likewise.
12893 (translate_pending_phi_nodes): Likewise.
12895 2017-09-29 Jakub Jelinek <jakub@redhat.com>
12898 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
12899 for movabsq $(i32 << shift), r64.
12901 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
12903 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
12904 index when encoding %esp as %rsp to avoid 0x67 prefix.
12906 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
12908 * config/i386/i386.md (*movsf_internal, *movdf_internal):
12909 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
12911 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
12913 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
12914 Extensions with more than 16 double VFP registers.
12915 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
12916 to_clear_mask and all code related to it. Replace the remaining
12917 entry by a sbitmap and adapt code accordingly.
12919 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
12921 * brig-builtins.def: Change pure attributes to const.
12923 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12925 * config.gcc (default_gnu_indirect_function): Default to yes for
12926 sparc*-*-linux* with glibc.
12928 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12930 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
12931 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
12932 when creating .init_array and .fini_array sections with priority
12935 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
12938 * config/aarch64/aarch64.c
12939 (aarch64_builtin_support_vector_misalignment): Always return false
12940 when misalignment is unknown.
12942 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12944 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
12945 this function to return false if the definition used by the swap
12946 instruction is artificial, or if the memory address from which the
12947 constant value is loaded is not represented by a base address held
12948 in a register or if the base address register is a frame or stack
12949 pointer. Additionally, return false if the base address of the
12950 loaded constant is a SYMBOL_REF but is not considered to be a
12952 (replace_swapped_load_constant): New function.
12953 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
12954 loaded constant vector with a load of a swapped constant vector.
12956 2017-09-27 Carl Love <cel@us.ibm.com>
12958 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
12959 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
12960 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
12963 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
12965 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
12966 first, always call autopref_rank_data otherwise.
12968 2017-09-27 Richard Biener <rguenther@suse.de>
12970 * graphite-scop-detection.c (find_scop_parameters): Move
12971 loop bound handling ...
12972 (gather_bbs::before_dom_children): ... here, avoiding the need
12973 to build scop_info->loop_nest.
12974 (record_loop_in_sese): Remove.
12975 * sese.h (sese_info_t::loop_nest): Remove.
12976 * sese.c (new_sese_info): Do not allocate loop_nest.
12977 (free_sese_info): Do not free loop_nest.
12979 2017-09-27 Jakub Jelinek <jakub@redhat.com>
12982 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
12983 lhs from calls if the lhs has addressable type.
12985 2017-09-27 Richard Biener <rguenther@suse.de>
12987 * graphite.h (scop::max_alias_set): New member.
12988 * graphite-scop-detection.c: Remove references to non-existing
12989 --param in comments.
12990 (build_alias_sets): Record the maximum alias set used for drs.
12991 (build_scops): Support zero as unlimited for
12992 --param graphite-max-arrays-per-scop.
12993 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
12994 and inline into ...
12995 (build_poly_sr_1): ... here. Compute alias set based on the
12996 maximum alias set used for drs rather than
12997 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
12999 2017-09-27 Richard Biener <rguenther@suse.de>
13001 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
13002 --param loop-block-tile-size=0 to disable tiling.
13004 2017-09-27 Richard Biener <rguenther@suse.de>
13006 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
13007 (graphite-max-nb-scop-params): Document special value zero.
13008 * domwalk.h (dom_walker::STOP): New symbolical constant.
13009 (dom_walker::dom_walker): Add optional parameter for bb to
13011 (dom_walker::~dom_walker): Declare.
13012 (dom_walker::before_dom_children): Document STOP return value.
13013 (dom_walker::m_user_bb_to_rpo): New member.
13014 (dom_walker::m_bb_to_rpo): Likewise.
13015 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
13016 mapping here if not provided by the user.
13017 (dom_walker::~dom_walker): Free bb to RPO mapping if not
13018 provided by the user.
13019 (dom_walker::STOP): Define.
13020 (dom_walker::walk): Do not compute bb to RPO mapping here.
13021 Support STOP return value from before_dom_children to stop
13023 * graphite-optimize-isl.c (optimize_isl): If the schedule
13024 is the same still generate code if -fgraphite-identity
13025 or -floop-parallelize-all are given.
13026 * graphite-scop-detection.c: Include cfganal.h.
13027 (gather_bbs::gather_bbs): Get and pass through bb to RPO
13029 (gather_bbs::before_dom_children): Return STOP for BBs
13031 (build_scops): Compute bb to RPO mapping and pass it to
13032 the domwalk. Treat --param graphite-max-nb-scop-params=0
13033 as not limiting the number of params.
13034 * graphite.c (graphite_initialize): Remove limit on the
13035 number of basic-blocks in a function.
13036 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
13037 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
13038 default value of 10.
13040 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13042 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
13043 Adjust code to eliminate needing to do the shift right 32-bits
13044 operation after XSCVDPSPN.
13046 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13048 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
13049 ((X / Y) != 0 -> X >= Y): Likewise.
13051 2017-09-26 Carl Love <cel@us.ibm.com>
13053 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
13054 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
13055 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
13056 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
13057 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
13058 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
13059 definitions and overloading.
13060 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
13061 statement for P9V_BUILTIN_XST_LEN_R.
13062 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
13063 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
13064 define_expand and define_insn for the instructions and builtins.
13065 * doc/extend.texi: Update the built-in documentation file for the new
13066 built-in functions.
13067 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
13068 define_insn for the instructions
13070 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
13073 * gcc/config/netbsd-protos.h: New file.
13074 * gcc/config/netbsd.c: New file.
13075 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
13076 * gcc/config/t-netbsd: New file.
13077 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
13078 (tmake_file) Add t-netbsd.
13079 (extra_objs) Add netbsd.o.
13081 2017-09-26 Janus Weil <janus@gcc.gnu.org>
13085 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
13087 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13089 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
13090 sign extension from a vector register to a GPR by doing a 32-bit
13091 direct move and then an EXTSW.
13092 (extendsi<mode>2 splitter): Likewise.
13093 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
13094 right or vector extract after doing XSCVDPSPN. Use
13095 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
13097 (movdi_from_sf_zero_ext): Likewise.
13098 (reload_gpr_from_vsxsf): Likewise.
13099 (p8_mfvsrd_4_disf): Delete, no longer used.
13100 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
13101 then needing to move the SFmode to a GPR to use the XSCVDPSP
13102 instruction instead of FRSP and XSCVDPSPN.
13103 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
13104 it is adjacent to the other XSCVSPDP insns.
13105 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
13106 SFmode to be in traditional Altivec registers.
13107 (vsx_xscvdpspn): Eliminate useless alternative constraint.
13108 (vsx_xscvspdpn): Likewise.
13109 (vsx_xscvspdpn_scalar): Likewise.
13111 2017-09-26 Martin Jambor <mjambor@suse.cz>
13113 * tree-sra.c (compare_access_positions): Put integral types first,
13114 stabilize sorting of integral types, remove conditions putting
13115 non-full-precision integers last.
13116 (sort_and_splice_var_accesses): Disable scalarization if a
13117 non-integert would be represented by a non-full-precision integer.
13119 2017-09-26 Joseph Myers <joseph@codesourcery.com>
13121 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
13122 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
13123 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
13124 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
13125 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
13126 conditionals inside the function instead of around it. Call
13127 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
13128 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
13130 2017-09-26 Richard Biener <rguenther@suse.de>
13132 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
13134 (scop_detection::build_scop_breadth): ... this. Removed.
13135 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
13136 (scop_detection::harmful_stmt_in_bb): Likewise.
13137 (scop_detection::graphite_can_represent_stmt): Likewise.
13138 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
13139 (scop_detection::can_represent_loop): Remove recursion, fold in ...
13140 (scop_detection::can_represent_loop_1): ... this. Removed.
13141 (scop_detection::harmful_loop_in_region): Simplify after inlining
13142 the above and remove more quadraticness.
13143 (build_scops): Adjust.
13144 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
13147 2017-09-26 Jakub Jelinek <jakub@redhat.com>
13150 * config/i386/i386.c (ix86_print_operand_address_as): Only test
13151 REGNO (base) == SP_REG if base is a REG.
13153 PR middle-end/35691
13154 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
13155 if it is different SSA_NAME.
13156 (optimize_range_tests_cmp_bitwise): New function.
13157 (optimize_range_tests): Call it.
13159 2017-09-26 Richard Biener <rguenther@suse.de>
13161 PR tree-optimization/82321
13162 * graphite.c (canonicalize_loop_closed_ssa): Properly check
13163 for the def being inside the loop.
13165 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13167 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
13169 * config/s390/s390-builtins.def: Fix constraint on op4.
13171 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13173 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
13174 independent expanders.
13175 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
13176 ("vec_ordered", "vec_unordered"): New expanders.
13178 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13180 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
13183 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13185 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
13186 vec_unpacks_lo_v16qi.
13187 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
13189 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13191 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
13192 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
13193 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
13195 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13197 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
13199 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
13201 ("*vec_slb<mode>"): New insn pattern.
13202 ("vec_shr_<mode>"): New expander.
13203 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
13204 and force the shift count operand to V16QImode.
13205 ("vec_srb<mode>"): Set shift count mode to V16QI.
13207 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13209 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
13210 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
13211 ("vec_widen_smult_hi_<mode>"): New expander definitions.
13213 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
13216 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
13218 2017-09-26 Richard Biener <rguenther@suse.de>
13220 PR tree-optimization/82320
13221 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
13224 2017-09-25 Jeff Law <law@redhat.com>
13226 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
13227 prototype for new argument.
13228 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
13229 mostly extracted from rs6000_emit_allocate_stack.
13230 (rs6000_emit_probe_stack_range_stack_clash): New function.
13231 (rs6000_emit_allocate_stack): Call
13232 rs6000_emit_probe_stack_range_stack_clash as needed.
13233 (rs6000_emit_probe_stack_range): Add additional argument
13234 to call to gen_probe_stack_range{si,di}.
13235 (output_probe_stack_range): New.
13236 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
13237 (output_probe_stack_range_stack_clash): New.
13238 (rs6000_emit_prologue): Emit notes into dump file as requested.
13239 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
13240 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
13241 Add additional operand and pass it to output_probe_stack_range.
13243 2017-09-25 Bin Cheng <bin.cheng@arm.com>
13245 PR tree-optimization/82163
13246 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
13247 (checking_verify_loop_closed_ssa): New parameter.
13248 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
13249 (check_loop_closed_ssa_stmt): Delete.
13250 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
13251 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
13252 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
13255 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
13257 * brig-builtins.def: Treat HSAIL barrier builtins as
13258 setjmp/longjump style functions.
13260 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13262 * target.def (constant_alignment): New hook.
13263 * defaults.h (CONSTANT_ALIGNMENT): Delete.
13264 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
13265 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
13266 * doc/tm.texi: Regenerate.
13267 * targhooks.h (default_constant_alignment): Declare.
13268 (constant_alignment_word_strings): Likewise.
13269 * targhooks.c (default_constant_alignment): New function.
13270 (constant_alignment_word_strings): Likewise.
13271 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
13272 instead of CONSTANT_ALIGNMENT.
13273 * varasm.c (align_variable, get_variable_align, build_constant_desc)
13274 (force_const_mem): Likewise.
13275 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
13276 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
13277 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
13278 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13279 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
13281 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
13282 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13283 constant_alignment_word_strings.
13284 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
13285 (CONSTANT_ALIGNMENT): Likewise.
13286 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13287 (arm_constant_alignment): New function.
13288 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
13289 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13290 constant_alignment_word_strings.
13291 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
13292 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13293 constant_alignment_word_strings.
13294 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
13295 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13296 (cris_constant_alignment): New function.
13297 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
13298 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13299 (epiphany_constant_alignment): New function.
13300 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
13301 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13302 constant_alignment_word_strings.
13303 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
13304 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13305 constant_alignment_word_strings.
13306 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
13307 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13308 constant_alignment_word_strings.
13309 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
13310 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
13311 * config/i386/i386.c (ix86_constant_alignment): Make static.
13312 Use the same interface as the target hook.
13313 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13314 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
13315 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13316 constant_alignment_word_strings.
13317 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
13318 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
13319 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13320 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
13321 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13322 constant_alignment_word_strings.
13323 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
13324 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13325 constant_alignment_word_strings.
13326 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
13327 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13328 constant_alignment_word_strings.
13329 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
13330 * config/microblaze/microblaze.c (microblaze_constant_alignment):
13332 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13333 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
13334 * config/mips/mips.c (mips_constant_alignment): New function.
13335 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13336 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
13337 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
13338 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13339 (mmix_constant_alignment): Make static. Use the same interface
13340 as the target hook.
13341 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
13342 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13343 constant_alignment_word_strings.
13344 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
13345 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13346 constant_alignment_word_strings.
13347 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
13348 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13349 constant_alignment_word_strings.
13350 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
13351 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13352 (rs6000_constant_alignment): New function.
13353 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
13354 * config/riscv/riscv.c (riscv_constant_alignment): New function.
13355 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13356 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
13357 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13358 (rs6000_constant_alignment): New function.
13359 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
13360 * config/s390/s390.c (s390_constant_alignment): New function.
13361 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13362 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
13363 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13364 constant_alignment_word_strings.
13365 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
13366 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13367 (sparc_constant_alignment): New function.
13368 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
13369 * config/spu/spu.c (spu_constant_alignment): New function.
13370 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13371 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
13372 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13373 constant_alignment_word_strings.
13374 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
13375 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13376 constant_alignment_word_strings.
13377 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
13378 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13379 constant_alignment_word_strings.
13380 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
13381 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13382 (visium_constant_alignment): New function.
13383 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
13384 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13385 (xtensa_constant_alignment): New function.
13386 * system.h (CONSTANT_ALIGNMENT): Poison.
13388 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
13390 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13391 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
13392 (rs6000_builtin_valid_without_lhs): New helper function.
13393 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13394 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
13396 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13398 * target.h (vec_perm_indices): Use unsigned short rather than
13400 (auto_vec_perm_indices): Likewise.
13401 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
13402 Use unsigned int rather than unsigned char.
13403 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
13405 2017-09-25 Richard Biener <rguenther@suse.de>
13407 * cfgloop.h (sort_sibling_loops): Declare.
13408 * cfgloop.c (sort_sibling_loops_cmp): New helper.
13409 (sort_sibling_loops): New function sorting the sibling loop list
13411 * graphite.c (graphite_transform_loops): Sort sibling loops.
13413 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
13415 * target.def (vec_perm_const_ok): Change sel parameter to
13417 * optabs-query.c (can_vec_perm_p): Update accordingly.
13418 * doc/tm.texi: Regenerate.
13419 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
13420 auto_vec_perm_indices and remove separate nelt field.
13421 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
13422 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
13423 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
13424 (aarch64_expand_vec_perm_const): Update accordingly.
13425 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
13426 to vec_perm_indices.
13427 * config/arm/arm.c (expand_vec_perm_d): Change perm to
13428 auto_vec_perm_indices and remove separate nelt field.
13429 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
13430 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
13431 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
13433 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
13434 to vec_perm_indices.
13435 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
13436 sel to vec_perm_indices.
13437 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
13438 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
13439 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
13441 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
13444 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
13447 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
13448 on the FUNCTION_DECL function context if it has a DIE that is a
13451 2017-09-25 Richard Biener <rguenther@suse.de>
13453 PR tree-optimization/82285
13454 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
13457 2017-09-25 Tom de Vries <tom@codesourcery.com>
13461 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
13464 2017-09-25 Richard Biener <rguenther@suse.de>
13466 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
13467 ISL errors other than isl_error_quota happen. Dump if the
13468 schedule is the same.
13469 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
13470 errors instead of aborting inside ISL.
13472 2017-09-25 Iain Sandoe <iain@codesourcery.com>
13475 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
13476 of libgcc_eh for m64.
13477 * config/i386/darwin64.h: Likewise.
13479 2017-09-25 Richard Biener <rguenther@suse.de>
13481 PR middle-end/82144
13482 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
13483 attribute for incomplete types nor twice for complete ones.
13485 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
13488 * config/i386/i386.c (ix86_print_operand_address_as): Encode
13489 %esp as %rsp to avoid 0x67 prefix if there is no index or base
13492 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
13495 * config/i386/i386.opt (mprefer-avx256): Use
13496 ix86_target_flags variable.
13497 * config/i386/i386.c (ix86_target_string): Move
13498 -mprefer-avx256 to flag2_opts.
13500 2017-09-22 Jakub Jelinek <jakub@redhat.com>
13502 PR middle-end/35691
13503 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
13504 and x != -1 | y != -1 into (x & y) != -1.
13506 2017-09-22 Steve Ellcey <sellcey@cavium.com>
13508 * config.gcc: Add new case statement to set
13509 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
13510 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
13511 s390x-*-linux* case statements. Added aarch64 to the list of
13512 supported architectures.
13514 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13516 PR tree-optimization/82289
13517 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
13518 STMT_VINFO_RELEVANT_P.
13520 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13521 Alan Hayward <alan.hayward@arm.com>
13522 David Sherwood <david.sherwood@arm.com>
13524 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
13525 for VR_RANGE only; don't allow VR_ANTI_RANGE.
13526 (extract_range_from_binary_expr_1): Don't call
13527 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
13529 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13530 Alan Hayward <alan.hayward@arm.com>
13531 David Sherwood <david.sherwood@arm.com>
13533 * target.def (preferred_vector_alignment): New hook.
13534 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
13536 * doc/tm.texi: Regenerate.
13537 * targhooks.h (default_preferred_vector_alignment): Declare.
13538 * targhooks.c (default_preferred_vector_alignment): New function.
13539 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
13541 (DR_TARGET_ALIGNMENT): New macro.
13542 (aligned_access_p): Update commentary.
13543 (vect_known_alignment_in_bytes): New function.
13544 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
13546 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
13547 Calculate the misalignment based on the target alignment rather than
13549 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
13550 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
13551 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
13552 the target alignment, rather than masking the element misalignment
13553 with the number of elements in a vector. Also use the target
13554 alignment when calculating the maximum number of peels.
13555 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
13556 instead of TYPE_ALIGN_UNIT.
13557 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
13558 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
13559 (vect_create_addr_base_for_vector_ref): Update call accordingly.
13560 (vect_create_data_ref_ptr): Likewise.
13561 (vect_setup_realignment): Realign by ANDing with
13562 -DR_TARGET_MISALIGNMENT.
13563 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
13564 the number of peels based on DR_TARGET_ALIGNMENT.
13565 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
13566 with the guaranteed alignment boundary when deciding whether
13568 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
13569 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
13570 (ensure_base_align): Remove stmt_info parameter. Get the
13571 target base alignment from DR_TARGET_ALIGNMENT.
13572 (vectorizable_store): Update call accordingly. Interpret
13573 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
13575 (vectorizable_load): Likewise.
13577 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13578 Alan Hayward <alan.hayward@arm.com>
13579 David Sherwood <david.sherwood@arm.com>
13581 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
13582 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
13583 (vect_enhance_data_refs_alignment): Likewise.
13585 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13587 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13588 error. Only quit immediately if parsing is complete.
13589 (BEGIN): Initialize fatal_err and parse_done.
13590 (begin fpu, end fpu): Check number of arguments.
13591 (begin arch, end arch): Likewise.
13592 (begin cpu, end cpu): Likewise.
13593 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13594 (optalias): Likewise.
13596 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13598 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13599 * config/arm/arm-isa.h: Delete. Move definitions to ...
13600 * arm-cpus.in: ... here. Use new feature and fgroup values.
13601 * config/arm/arm.c (arm_option_override): Use lower case for feature
13603 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13604 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13605 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13607 (print_isa_bits_for): New function.
13608 (gen_isa): New function.
13609 (gen_comm_data): Use print_isa_bits_for.
13610 (define feature): New keyword.
13611 (define fgroup): New keyword.
13612 * config/arm/t-arm (TM_H): Remove.
13613 (GTM_H): Add arm-isa.h.
13614 (arm-isa.h): Add rule to generate file.
13615 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13616 case for feature bit names.
13618 2017-09-22 Richard Biener <rguenther@suse.de>
13620 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
13622 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
13623 print of no dependency loops ...
13624 * graphite.c (graphite_transform_loops): ... here.
13625 (canonicalize_loop_closed_ssa_form): Work from inner to outer
13627 (same_close_phi_node, remove_duplicate_close_phi,
13628 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
13629 (canonicalize_loop_closed_ssa): ... here and simplify.
13630 * graphite-optimize-isl.c: Include tree-vectorizer.h.
13631 (optimize_isl): Use dump_printf_loc to tell when we stopped
13632 optimizing because of an ISL timeout.
13634 2017-09-22 Richard Biener <rguenther@suse.de>
13636 PR tree-optimization/82291
13637 * tree-if-conv.c (predicate_mem_writes): Make sure to
13638 remove writes in blocks predicated with false.
13640 2017-09-22 Richard Biener <rguenther@suse.de>
13642 * sese.c: Include cfganal.h.
13643 (if_region_set_false_region): Remove.
13644 (create_if_region_on_edge): Likewise.
13645 (move_sese_in_condition): Re-implement without destroying
13648 2017-09-22 Richard Biener <rguenther@suse.de>
13650 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
13651 Verify both BBs contain loop PHI nodes before dispatching to
13652 copy_loop_phi_args.
13653 (graphite_regenerate_ast_isl): Do not recompute dominators,
13654 do not verify three times. Restructure for clarity.
13655 * graphite-scop-detection.c (same_close_phi_node,
13656 remove_duplicate_close_phi, make_close_phi_nodes_unique,
13657 defined_in_loop_p, canonicalize_loop_closed_ssa,
13658 canonicalize_loop_closed_ssa_form): Simplify, remove excess
13659 checking and SSA rewrite, move to ...
13660 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
13661 (graphite_initialize): Do not pass in ctx, do not reset the
13662 SCEV cache, compute only dominators.
13663 (graphite_transform_loops): Allocate ISL ctx after
13664 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
13665 Maintain post-dominators only around build_scops.
13666 * sese.c (if_region_set_false_region): Make static. Free
13667 and recompute dominators.
13668 (move_sese_in_condition): Assert we don't get called with
13669 post-dominators computed.
13670 * sese.h (if_region_set_false_region): Remove.
13672 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
13674 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
13675 mode attribute for TARGET_AVX512VL.
13677 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
13679 * config/i386/i386.opt (mprefer-avx256): New option.
13680 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
13682 (ix86_preferred_simd_mode): Return 256-bit AVX modes
13683 for TARGET_PREFER_AVX256.
13684 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
13686 2017-09-21 Jeff Law <law@redhat.com>
13688 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
13689 Fix dump output if the only stack space is for pushed registers.
13691 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13693 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
13696 2017-09-21 Martin Sebor <msebor@redhat.com>
13699 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
13700 code (in C++) or code that triggers warnings.
13702 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
13704 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
13706 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
13708 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
13709 * sched-rgn.c: Ditto.
13710 * sel-sched-ir.c: Ditto.
13712 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
13714 * toplev.h (set_random_seed): Adjust return type.
13715 * toplev.c (init_local_tick): Move eager initialization of random_seed
13716 to get_random_seed. Adjust comment.
13717 (init_random_seed): Inline to get_random_seed, delete.
13718 (get_random_seed): Initialize random_seed lazily.
13719 (set_random_seed): Do not return previous value.
13720 (print_switch_value): Do not call get_random_seed.
13722 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
13724 * cgraph.c (delete_function_version): New, broken out from...
13725 (cgraph_node::delete_function_version): ...here. Rename to
13726 cgraph_node::delete_function_version_by_decl. Update all uses.
13727 (cgraph_node::remove): Call delete_function_version.
13729 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13732 * tree-inline.c (expand_call_inline): Emit clobber stmts for
13733 VAR_DECLs to which addressable non-volatile parameters are mapped
13734 and for id->retvar after the return value assignment. Clear
13735 id->retval and id->retbnd after inlining.
13737 2017-09-21 Richard Biener <rguenther@suse.de>
13739 PR tree-optimization/82276
13740 PR tree-optimization/82244
13741 * tree-vrp.c (build_assert_expr_for): Set
13742 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
13744 (remove_range_assertions): Revert earlier change.
13746 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
13749 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
13751 2017-09-21 Richard Biener <rguenther@suse.de>
13753 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
13754 Restore valid IL after code generation errors.
13755 * graphite.c (graphite_transform_loops): Diagnose code
13756 generation issues as MSG_MISSED_OPTIMIZATION and continue
13757 with processing SCOPs.
13759 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13760 Alan Hayward <alan.hayward@arm.com>
13761 David Sherwood <david.sherwood@arm.com>
13763 * calls.c (compute_argument_addresses): Use simplify_gen_binary
13764 rather than choosing between plus_constant and gen_rtx_<CODE>.
13765 * expr.c (emit_push_insn): Likewise.
13766 (expand_expr_real_2): Likewise.
13768 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13769 Alan Hayward <alan.hayward@arm.com>
13770 David Sherwood <david.sherwood@arm.com>
13772 * loop-unroll.c (split_iv): Call copy_rtx on the step.
13774 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13775 Alan Hayward <alan.hayward@arm.com>
13776 David Sherwood <david.sherwood@arm.com>
13778 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
13779 calling tree_to_uhwi.
13781 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13782 Alan Hayward <alan.hayward@arm.com>
13783 David Sherwood <david.sherwood@arm.com>
13785 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
13786 INTEGER_CST rather than a negative test for ADDR_EXPR.
13788 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13789 Alan Hayward <alan.hayward@arm.com>
13790 David Sherwood <david.sherwood@arm.com>
13792 * tree-vrp.c (extract_range_from_binary_expr_1): Check
13793 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
13795 2017-09-21 Richard Biener <rguenther@suse.de>
13797 PR tree-optimization/71351
13798 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
13799 graphite_create_new_loop_guard): Remove, fold remaining parts
13801 (translate_isl_ast_node_for): ... here and simplify.
13803 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13806 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
13807 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
13808 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
13809 alternative always use QI mode, for -Os imov (=R,R) alternative
13810 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
13813 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13814 Jeff Law <law@redhat.com>
13816 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
13817 (allocate_stack_space): New function, partially extracted from
13818 s390_emit_prologue.
13819 (s390_emit_prologue): Track offset to most recent stack probe.
13820 Code to allocate space moved into allocate_stack_space.
13821 Dump actions when no stack is allocated.
13822 (s390_prologue_plus_offset): New function.
13823 (s390_emit_stack_probe): Likewise.
13825 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
13827 * common.opt (Wa, Wl, Wp, g, gz=): Add
13829 (gno-column-info): Remove.
13830 (gcolumn-info): Drop RejectNegative.
13831 (gno-): New prefix.
13832 (gno-record-gcc-switches): Remove.
13833 (grecord-gcc-switches): Drop RejectNegative.
13834 (gno-split-dwarf): Remove.
13835 (gsplit-dwarf): Drop RejectNegative.
13836 (gno-strict-dwarf): Remove.
13837 (gstrict-dwarf): Drop RejectNegative.
13838 * config/darwin.opt (gfull, gused): Add RejectNegative.
13839 * dwarf2out.c (gen_producer_string): Drop
13840 gno-record-gcc-switches handler.
13841 * optc-gen.awk: Add g to prefixes with negative forms.
13842 * opts-common.c (remapping_prefix_p): New.
13843 (find_opt): Check it.
13844 (generate_canonical_option): Test g prefix.
13845 (option_map): Add -gno- mapping.
13846 (add_misspelling_candidates): Check remapping_prefix_p.
13848 2017-09-20 Jeff Law <law@redhat.com>
13850 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
13851 thinko in stack clash protection support.
13853 * explow.c (compute_stack_clash_protection_loop_data): Use
13854 CONST_INT_P instead of explicit test. Verify object is a
13855 CONST_INT_P before looking at INTVAL.
13856 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
13857 instead of explicit test.
13859 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
13862 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
13863 address instead of to r1 and r11.
13865 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
13867 * config.gcc: Support "knm".
13868 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
13869 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13871 * config/i386/i386.c (m_KNM): Define.
13872 (processor_target_table): Add "knm".
13874 (ix86_option_override_internal): Add "knm".
13875 (ix86_issue_rate): Add PROCESSOR_KNM.
13876 (ix86_adjust_cost): Ditto.
13877 (ia32_multipass_dfa_lookahead): Ditto.
13878 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
13879 (fold_builtin_cpu): Add M_INTEL_KNM.
13880 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
13881 (processor_type): Add PROCESSOR_KNM.
13882 * config/i386/x86-tune.def: Add m_KNM.
13883 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
13885 2017-09-20 Richard Biener <rguenther@suse.de>
13887 PR tree-optimization/80213
13888 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
13889 are allowed in empty BBs as well.
13890 (canonicalize_loop_closed_ssa): Also look for other complex
13892 (scop_detection::get_sese): Include the loop-closed PHI block
13894 (scop_detection::merge_sese): Remove code adding extra blocks.
13895 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
13896 (build_scops): Assert the final returned scop is invalid.
13898 2017-09-20 Richard Biener <rguenther@suse.de>
13900 PR tree-optimization/82264
13901 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
13903 (vn_phi_lookup): Likewise.
13904 (vn_phi_insert): Likewise.
13906 2017-09-20 Jakub Jelinek <jakub@redhat.com>
13908 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
13909 that fits into uhwi or shwi, add DW_AT_const_value regardless
13910 of early_dwarf without going through RTL, using add_AT_unsigned
13913 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
13914 (DEBUG_LTO_ABBREV_SECTION): Likewise.
13915 (DEBUG_LTO_MACINFO_SECTION): Likewise.
13916 (DEBUG_MACRO_SECTION): Likewise.
13917 (DEBUG_LTO_MACRO_SECTION): Likewise.
13918 (DEBUG_STR_DWO_SECTION): Likewise.
13919 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
13920 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
13921 (DEBUG_LTO_DWO_LINE_SECTION): Define.
13922 (DEBUG_LTO_LINE_STR_SECTION): Define.
13923 (init_sections_and_labels): Initialize debug_line_str_section
13924 variable. Initialize debug_loc_section for -gdwarf-5 to
13925 DEBUG_LOCLISTS_SECTION. Formatting fixes.
13927 2017-09-20 Richard Biener <rguenther@suse.de>
13929 * graphite-sese-to-poly.c (extract_affine): Properly handle
13930 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
13932 2017-09-20 Richard Biener <rguenther@suse.de>
13934 PR tree-optimization/81373
13935 * graphite-scop-detection.c (build_cross_bb_scalars_def):
13936 Force SESE live-out defs to be handled even if they are
13939 2017-09-19 Jeff Law <law@redhat.com>
13941 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
13942 nothing for stack adjustments with REG_STACK_CHECK.
13943 * sched-deps.c (parse_add_or_inc): Reject insns with
13944 REG_STACK_CHECK from dependency breaking.
13945 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
13946 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
13947 * reg-notes.def (STACK_CHECK): New note.
13949 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
13950 (ix86_expand_prologue): Dump stack clash info as needed.
13951 Call ix86_adjust_stack_and_probe_stack_clash as needed.
13953 * function.c (dump_stack_clash_frame_info): New function.
13954 * function.h (dump_stack_clash_frame_info): Prototype.
13955 (enum stack_clash_probes): New enum.
13957 * config/alpha/alpha.c (alpha_expand_prologue): Also check
13958 flag_stack_clash_protection.
13959 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
13960 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
13961 (arm_frame_pointer_required): Likewise.
13962 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
13963 (ia64_expand_prologue): Likewise.
13964 * config/mips/mips.c (mips_expand_prologue): Likewise.
13965 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
13966 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13967 (sparc_flat_expand_prologue): Likewise.
13968 * config/spu/spu.c (spu_expand_prologue): Likewise.
13970 * explow.c: Include "params.h".
13971 (anti_adjust_stack_and_probe_stack_clash): New function.
13972 (get_stack_check_protect): Likewise.
13973 (compute_stack_clash_protection_loop_data): Likewise.
13974 (emit_stack_clash_protection_loop_start): Likewise.
13975 (emit_stack_clash_protection_loop_end): Likewise.
13976 (allocate_dynamic_stack_space): Use get_stack_check_protect.
13977 Use anti_adjust_stack_and_probe_stack_clash.
13978 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
13979 (emit_stack_clash_protection_loop_start): Likewise.
13980 (emit_stack_clash_protection_loop_end): Likewise.
13981 * rtl.h (get_stack_check_protect): Prototype.
13982 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
13983 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
13984 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
13986 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
13988 * doc/tm.texi: Rebuilt.
13989 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
13990 get_stack_check_protect.
13991 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
13992 * config/arm/arm.c (arm_expand_prologue): Likewise.
13993 (arm_frame_pointer_required): Likewise.
13994 * config/i386/i386.c (ix86_expand_prologue): Likewise.
13995 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
13996 * config/mips/mips.c (mips_expand_prologue): Likewise.
13997 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
13998 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
13999 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14000 (sparc_flat_expand_prologue): Likewise.
14002 * common.opt (-fstack-clash-protection): New option.
14003 * flag-types.h (enum stack_check_type): Note difference between
14004 -fstack-check= and -fstack-clash-protection.
14005 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
14006 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
14007 * toplev.c (process_options): Issue warnings/errors for cases
14008 not handled with -fstack-clash-protection.
14009 * doc/invoke.texi (-fstack-clash-protection): Document new option.
14010 (-fstack-check): Note additional problem with -fstack-check=generic.
14011 Note that -fstack-check is primarily for Ada and refer users
14012 to -fstack-clash-protection for stack-clash-protection.
14013 Document new params for stack clash protection.
14015 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
14017 * config/i386/i386.c (ix86_split_long_move): Do not handle
14018 address used for LEA in a special way.
14020 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
14022 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
14024 2017-09-19 Martin Sebor <msebor@redhat.com>
14027 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
14028 of incompatible types.
14030 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
14032 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14033 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
14034 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14035 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
14037 2017-09-19 Richard Biener <rguenther@suse.de>
14039 PR tree-optimization/82244
14040 * tree-vrp.c (remove_range_assertions): Do not propagate
14041 a constant to abnormals but replace the assert with a copy.
14043 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
14045 PR rtl-optimization/57878
14046 PR rtl-optimization/68988
14047 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
14048 avoidance test involving non_reload_pseudos. Move frequency test
14049 below the general fragmentation avoidance test.
14051 2017-09-19 Richard Biener <rguenther@suse.de>
14053 PR tree-optimization/69728
14054 * graphite-sese-to-poly.c (schedule_error): New global.
14055 (add_loop_schedule): Handle empty domain by failing the
14057 (build_original_schedule): Handle schedule_error.
14059 2017-09-19 Richard Biener <rguenther@suse.de>
14061 * graphite-scop-detection.c (scop_detection::can_represent_loop):
14062 Do not iterate to sibling loops but only to siblings of inner
14065 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
14068 * config/m68k/m68k.md (moveq feeding equality comparison): Check
14069 that the registers are different.
14071 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
14073 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
14074 to processor_model and "amdfam17h" to arch_names_table.
14075 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
14077 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14080 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
14082 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14083 Alan Hayward <alan.hayward@arm.com>
14084 David Sherwood <david.sherwood@arm.com>
14086 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
14088 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
14090 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
14091 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
14092 vect_schedule_slp_instance.
14093 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
14094 Update call to vect_slp_analyze_node_operations. Simplify return
14096 (vect_slp_analyze_bb_1): Update call accordingly.
14097 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
14098 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
14099 (vect_schedule_slp): Update call accordingly.
14101 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14102 Alan Hayward <alan.hayward@arm.com>
14103 David Sherwood <david.sherwood@arm.com>
14105 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
14106 with types that aren't in fact scalar.
14108 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14110 * tree-vect-slp.c (vect_record_max_nunits): New function,
14112 (vect_build_slp_tree_1): ...here.
14113 (vect_build_slp_tree_2): Call it for phis too.
14115 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14117 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
14118 to vect_get_vec_def_for_operand when getting the mask operand.
14120 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14121 Alan Hayward <alan.hayward@arm.com>
14122 David Sherwood <david.sherwood@arm.com>
14124 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
14127 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14128 Alan Hayward <alan.hayward@arm.com>
14129 David Sherwood <david.sherwood@arm.com>
14131 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
14132 calculation for vector booleans.
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-stmts.c (can_vectorize_live_stmts): New function,
14140 (vect_transform_stmt): ...here.
14141 (vect_analyze_stmt): Use it instead of calling
14142 vectorizable_live_operation directly.
14144 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
14146 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
14147 non-SIMT targets in acc vector loops.
14149 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
14151 * configure.ac: Add arc and check if assembler supports gdwarf2.
14152 * configure: Regenerate.
14154 2017-09-18 Richard Biener <rguenther@suse.de>
14156 PR tree-optimization/82220
14157 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
14158 epilogue niters from the min_profitable_iters compute.
14160 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14163 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
14164 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
14165 (ix86_init_pic_reg): Revert 2017-09-01 changes.
14167 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
14170 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
14171 switching to a new text section.
14173 2017-09-18 Richard Biener <rguenther@suse.de>
14175 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
14177 (build_alias_set): Reject aliases with no access function.
14179 2017-09-18 Richard Biener <rguenther@suse.de>
14181 PR tree-optimization/79622
14182 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
14184 (build_cross_bb_scalars_use): Likewise.
14186 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
14188 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
14190 2017-09-18 Alan Modra <amodra@gmail.com>
14193 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
14194 stack_pointer_rtx for count 0. Update comments. Break up
14195 large rtl expression.
14197 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
14199 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
14200 Increase to 20 bytes.
14201 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
14202 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
14203 or avx version of the stub.
14205 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
14208 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
14209 compute the minimum stack alignment. Also update preferred stack
14210 boundary for leaf functions.
14212 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
14214 PR tree-optimization/82228
14215 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
14218 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14220 * common/config/nds32/nds32-common.c
14221 (nds32_option_optimization_table): Refine formatting.
14222 (nds32_option_optimization_table): Use -fsched-pressure and
14223 -fomit-frame-pointer for specific optimization level.
14225 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14227 * config/nds32/nds32.c: Refine formatting and comments.
14228 * config/nds32/nds32.h: Likewise.
14229 * config/nds32/nds32.md: Likewise.
14230 * config/nds32/nds32-cost.c: Likewise.
14231 * config/nds32/nds32-isr.c: Likewise.
14232 * config/nds32/nds32-md-auxiliary.c: Likewise.
14233 * config/nds32/nds32-multiple.md: Likewise.
14234 * config/nds32/nds32-predicates.c: Likewise.
14236 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
14237 Jakub Jelinek <jakub@redhat.com>
14239 Add support for -std=c++2a.
14240 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
14242 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
14244 2017-09-15 Steve Ellcey <sellcey@cavium.com>
14247 * doc/extend.texi (Common Function Attributes): Add
14248 references to ARM, AArch64, and S/390 specific attributes.
14249 (Function Specific Option Pragmas): Add AArch64 and S/390
14250 to list of back ends that support the target pragma.
14252 2017-09-15 Nathan Sidwell <nathan@acm.org>
14254 * doc/standards.texi: Fix C++17 description. Update URLs for
14257 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
14259 * common.opt (Wcast-align=strict): New warning option.
14260 * doc/invoke.texi: Document -Wcast-align=strict.
14262 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
14264 * cgraph.h (cgraph_thunk_info): Add comments.
14265 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
14266 assert for VIRTUAL_* arguments stricter.
14268 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
14270 PR tree-optimization/71026
14271 * match.pd: Move RDIV patterns from fold-const.c
14272 * fold-const.c (distribute_real_division): Removed.
14273 (fold_binary_loc): Remove calls to distribute_real_divison.
14275 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14277 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
14278 c++1z and gnu++1z as deprecated. Change other references to
14279 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
14280 Change -Wc++1z-compat to -Wc++17-compat.
14281 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
14282 * dwarf2out.c (highest_c_language): Handle C++17.
14283 (gen_compile_unit_die): Likewise.
14285 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14287 PR rtl-optimization/82192
14288 * combine.c (make_extraction): Don't look through non-paradoxical
14289 SUBREGs or TRUNCATE if pos + len is or might be bigger than
14292 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14293 Alan Hayward <alan.hayward@arm.com>
14294 David Sherwood <david.sherwood@arm.com>
14296 * target.def (function_arg_offset): New hook.
14297 * targhooks.h (default_function_arg_offset): Declare.
14298 * targhooks.c (default_function_arg_offset): New function.
14299 * function.c (locate_and_pad_parm): Use
14300 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
14301 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
14302 (TARGET_FUNCTION_ARG_OFFSET): ...this.
14303 * doc/tm.texi: Regenerate.
14304 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
14305 * config/spu/spu.c (spu_function_arg_offset): New function.
14306 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
14307 * system.h (FUNCTION_ARG_OFFSET): Poison.
14309 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14310 Alan Hayard <alan.hayward@arm.com>
14311 David Sherwood <david.sherwood@arm.com>
14313 * target.def (truly_noop_truncation): New hook.
14314 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
14315 than TRULY_NOOP_TRUNCATION.
14316 * hooks.h (hook_bool_uint_uint_true): Declare.
14317 * hooks.c (hook_bool_uint_uint_true): New function.
14318 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
14319 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
14320 * doc/tm.texi: Regenerate.
14321 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
14322 rather than TRULY_NOOP_TRUNCATION in comments.
14323 (simplify_comparison): Likewise.
14324 (record_truncated_value): Likewise.
14325 * expmed.c (extract_bit_field_1): Likewise.
14326 (extract_split_bit_field): Likewise.
14327 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
14328 instead of TRULY_NOOP_TRUNCATION.
14329 * function.c (assign_parm_setup_block): Likewise.
14330 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
14331 * rtlhooks.c: Include target.h.
14332 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
14333 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
14334 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
14335 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
14336 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
14337 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
14338 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
14339 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
14340 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
14341 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
14342 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
14343 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
14344 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
14345 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
14346 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
14347 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
14348 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
14349 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
14350 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
14351 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
14352 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
14353 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
14354 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
14355 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
14356 * config/mips/mips.c (mips_truly_noop_truncation): New function.
14357 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14358 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14359 rather than TRULY_NOOP_TRUNCATION in comments.
14360 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
14361 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
14362 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
14363 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
14364 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
14365 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
14366 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
14367 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
14368 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
14369 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
14370 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
14371 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14372 rather than TRULY_NOOP_TRUNCATION in comments.
14373 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
14374 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
14375 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
14376 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
14377 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
14378 TRULY_NOOP_TRUNCATION condition.
14379 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
14380 (TRULY_NOOP_TRUNCATION): Delete.
14381 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
14382 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
14383 * config/spu/spu.c (spu_truly_noop_truncation): New function.
14384 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14385 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
14386 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
14387 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
14388 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14389 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14390 rather than TRULY_NOOP_TRUNCATION in comments.
14391 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
14392 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
14393 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
14394 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
14395 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
14396 * system.h (TRULY_NOOP_TRUNCATION): Poison.
14398 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
14401 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
14402 (*cmp_ior): Likewise.
14403 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
14404 (*ior_scc_scc_cmp): Likewise.
14405 (*and_scc_scc): Likewise.
14406 (*and_scc_scc_cmp): Likewise.
14408 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14409 Alan Hayard <alan.hayward@arm.com>
14410 David Sherwood <david.sherwood@arm.com>
14412 * target.def (can_change_mode_class): New hook.
14413 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
14414 (hard_regno_nregs): Likewise.
14415 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
14416 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
14417 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
14418 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
14419 (LOAD_EXTEND_OP): Update accordingly.
14420 * doc/tm.texi: Regenerate.
14421 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
14422 CANNOT_CHANGE_MODE_CLASS.
14423 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
14424 (REG_CAN_CHANGE_MODE_P): ...this new macro.
14425 * combine.c (simplify_set): Update accordingly.
14426 * emit-rtl.c (validate_subreg): Likewise.
14427 * recog.c (general_operand): Likewise.
14428 * regcprop.c (mode_change_ok): Likewise.
14429 * reload1.c (choose_reload_regs): Likewise.
14430 (inherit_piecemeal_p): Likewise.
14431 * rtlanal.c (simplify_subreg_regno): Likewise.
14432 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
14433 instead of CANNOT_CHANGE_MODE_CLASS.
14434 (reload_cse_simplify_operands): Likewise.
14435 * reload.c (push_reload): Use targetm.can_change_mode_class
14436 instead of CANNOT_CHANGE_MODE_CLASS.
14437 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
14438 REG_CANNOT_CHANGE_MODE_P.
14439 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14440 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
14441 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14442 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14443 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14444 (arm_can_change_mode_class): New function.
14445 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14446 than CANNOT_CHANGE_MODE_CLASS in comments.
14447 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14448 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
14449 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
14450 (ix86_can_change_mode_class): ...this new function, inverting the
14451 sense of the return value.
14452 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14453 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14454 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14455 (ia64_can_change_mode_class): New function.
14456 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14457 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
14458 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
14459 (m32c_can_change_mode_class): ...this new function, inverting the
14460 sense of the return value.
14461 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14462 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14463 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
14464 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
14465 (mips_can_change_mode_class): ...this new function, inverting the
14466 sense of the return value.
14467 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14468 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14469 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14470 (msp430_can_change_mode_class): New function.
14471 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14472 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
14473 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14474 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14475 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14476 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
14477 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14478 (pa_cannot_change_mode_class): Replace with...
14479 (pa_can_change_mode_class): ...this new function, inverting the
14480 sense of the return value.
14481 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14482 than CANNOT_CHANGE_MODE_CLASS in comments.
14483 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14484 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
14485 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14486 (pdp11_cannot_change_mode_class): Replace with...
14487 (pdp11_can_change_mode_class): ...this new function, inverting the
14488 sense of the return value.
14489 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14490 * config/powerpcspe/powerpcspe-protos.h
14491 (rs6000_cannot_change_mode_class_ptr): Delete.
14492 * config/powerpcspe/powerpcspe.c
14493 (rs6000_cannot_change_mode_class_ptr): Delete.
14494 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14495 (rs6000_option_override_internal): Assign to
14496 targetm.can_change_mode_class instead of
14497 rs6000_cannot_change_mode_class_ptr.
14498 (rs6000_cannot_change_mode_class): Replace with...
14499 (rs6000_can_change_mode_class): ...this new function, inverting the
14500 sense of the return value.
14501 (rs6000_debug_cannot_change_mode_class): Replace with...
14502 (rs6000_debug_can_change_mode_class): ...this new function.
14503 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14504 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
14505 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14506 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14507 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
14509 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
14510 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14511 (rs6000_option_override_internal): Assign to
14512 targetm.can_change_mode_class instead of
14513 rs6000_cannot_change_mode_class_ptr.
14514 (rs6000_cannot_change_mode_class): Replace with...
14515 (rs6000_can_change_mode_class): ...this new function, inverting the
14516 sense of the return value.
14517 (rs6000_debug_cannot_change_mode_class): Replace with...
14518 (rs6000_debug_can_change_mode_class): ...this new function.
14519 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14520 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
14521 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
14522 (s390_can_change_mode_class): ...this new function, inverting the
14523 sense of the return value.
14524 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14525 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14526 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
14527 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14528 (sh_cannot_change_mode_class): Replace with...
14529 (sh_can_change_mode_class): ...this new function, inverting the
14530 sense of the return value.
14531 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14532 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14533 (sparc_can_change_mode_class): New function.
14534 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14535 * config/spu/spu.c (spu_can_change_mode_class): New function.
14536 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14537 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14538 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14539 (visium_can_change_mode_class): New function.
14540 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
14542 2017-09-15 Richard Biener <rguenther@suse.de>
14544 PR tree-optimization/82217
14545 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
14546 but not undefined case.
14548 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14551 * postreload.c (reload_cse_simplify_operands): Skip
14552 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
14554 2017-09-15 Richard Biener <rguenther@suse.de>
14556 PR tree-optimization/68823
14557 * graphite-scop-detection.c (build_alias_set): If we have a
14558 possible dependence check whether we can handle them by just
14559 looking at the DRs DR_ACCESS_FNs.
14560 (build_scops): If build_alias_set fails, fail the SCOP.
14562 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
14564 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
14565 to support float128 built-in functions that require the ISA 3.0
14567 (BU_FLOAT128_3_HW): Likewise.
14568 (SQRTF128): Add support for the IEEE 128-bit square root and fma
14569 built-in functions.
14570 (FMAF128): Likewise.
14572 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14573 support for built-in functions that need the ISA 3.0 IEEE 128-bit
14574 floating point instructions.
14575 (rs6000_invalid_builtin): Likewise.
14576 (rs6000_builtin_mask_names): Likewise.
14577 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
14578 (RS6000_BTM_FLOAT128_HW): Likewise.
14579 (RS6000_BTM_COMMON): Likewise.
14580 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
14582 * doc/extend.texi (RS/6000 built-in functions): Document the
14583 IEEE 128-bit floating point square root and fused multiply-add
14584 built-in functions.
14586 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
14588 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
14589 reg (r2) isn't in the set of registers defined in the prologue.
14591 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14592 Alan Hayward <alan.hayward@arm.com>
14593 David Sherwood <david.sherwood@arm.com>
14595 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
14596 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
14597 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
14598 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
14599 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
14601 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
14602 max_vectorization_factor.
14603 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
14605 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14606 Alan Hayward <alan.hayward@arm.com>
14607 David Sherwood <david.sherwood@arm.com>
14609 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
14610 (vect_worthwhile_without_simd_p): Declare.
14611 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
14612 (vectorizable_reduction): Use it.
14613 * tree-vect-stmts.c (vectorizable_shift): Likewise.
14614 (vectorizable_operation): Likewise.
14616 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14617 Alan Hayward <alan.hayward@arm.com>
14618 David Sherwood <david.sherwood@arm.com>
14620 * tree-vectorizer.h (vect_get_num_copies): New function.
14621 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
14622 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14623 (vectorizable_induction): Likewise.
14624 (vectorizable_live_operation): Likewise.
14625 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14626 (vectorizable_bswap): Likewise.
14627 (vectorizable_call): Likewise.
14628 (vectorizable_conversion): Likewise.
14629 (vectorizable_assignment): Likewise.
14630 (vectorizable_shift): Likewise.
14631 (vectorizable_operation): Likewise.
14632 (vectorizable_store): Likewise.
14633 (vectorizable_load): Likewise.
14634 (vectorizable_condition): Likewise.
14635 (vectorizable_comparison): Likewise.
14636 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
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-vect-loop.c (vectorizable_induction): Use gimple_build instead
14643 of vect_init_vector.
14645 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14646 Alan Hayward <alan.hayward@arm.com>
14647 David Sherwood <david.sherwood@arm.com>
14649 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
14650 an inline wrapper that provides a location.
14651 (gimple_build_vector): Likewise.
14652 * gimple-fold.c (gimple_build_vector_from_val): New function.
14653 (gimple_build_vector): Likewise.
14654 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
14655 functions to build the initial value. Always return a gimple value.
14656 (get_initial_defs_for_reduction): Likewise. Only compute
14658 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
14659 vect_init_vector on the results from get_initial_def(s)_for_reduction.
14660 (vectorizable_induction): Use gimple_build_vector rather than
14663 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14664 Alan Hayward <alan.hayward@arm.com>
14665 David Sherwood <david.sherwood@arm.com>
14667 * target.h (vec_perm_indices): New typedef.
14668 (auto_vec_perm_indices): Likewise.
14669 * optabs-query.h: Include target.h
14670 (can_vec_perm_p): Take a vec_perm_indices *.
14671 * optabs-query.c (can_vec_perm_p): Likewise.
14672 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
14673 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14674 * tree-vect-generic.c (lower_vec_perm): Likewise.
14675 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
14676 (vect_grouped_load_supported): Likewise.
14677 (vect_shift_permute_load_chain): Likewise.
14678 (vect_permute_store_chain): Use auto_vec_perm_indices.
14679 (vect_permute_load_chain): Likewise.
14680 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
14681 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
14682 Update uses of can_vec_perm_p.
14683 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
14684 mode with a number of elements. Take a vec_perm_indices *.
14685 (vect_create_epilog_for_reduction): Update accordingly.
14686 Use auto_vec_perm_indices.
14687 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
14688 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
14689 (vect_transform_slp_perm_load): Likewise.
14690 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
14691 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
14692 (vect_gen_perm_mask_checked): Likewise.
14693 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
14694 (vect_gen_perm_mask_checked): Likewise.
14695 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
14696 (vectorizable_store): Likewise.
14697 (vectorizable_load): Likewise.
14698 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
14699 (vectorizable_bswap): Likewise.
14701 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14702 Alan Hayward <alan.hayward@arm.com>
14703 David Sherwood <david.sherwood@arm.com>
14705 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
14706 * tree.c (build_vector): Likewise.
14707 (build_vector_from_ctor): Update accordingly.
14708 (build_vector_from_val): Likewise.
14709 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14710 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14711 * tree-vect-generic.c (add_rshift): Likewise.
14712 (expand_vector_divmod): Likewise.
14713 (optimize_vector_constructor): Likewise.
14714 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
14715 (vect_transform_slp_perm_load): Likewise.
14716 (vect_schedule_slp_instance): Likewise.
14717 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
14718 (vectorizable_call): Likewise.
14719 (vect_gen_perm_mask_any): Likewise. Add elements in order.
14720 * expmed.c (make_tree): Likewise.
14721 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
14722 a vector passed to build_vector.
14723 (fold_convert_const): Likewise.
14724 (exact_inverse): Likewise.
14725 (fold_ternary_loc): Likewise.
14726 (fold_relational_const): Likewise.
14727 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
14728 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
14729 (const_unop): Likewise. Store the reduction accumulator in a
14730 variable rather than an array.
14731 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
14732 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
14733 the new vector, rather than constructing it after the input arrays.
14734 (native_interpret_vector): Use auto_vec<tree> when building
14735 a vector passed to build_vector. Add elements in order.
14736 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
14737 auto_vec<tree> when building a vector passed to build_vector.
14738 (vect_create_epilog_for_reduction): Likewise.
14739 (vectorizable_induction): Likewise.
14740 (get_initial_def_for_reduction): Likewise. Fix indentation of
14742 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
14744 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
14745 passed to build_vector.
14747 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14748 Alan Hayward <alan.hayward@arm.com>
14749 David Sherwood <david.sherwood@arm.com>
14751 * tree-core.h (tree_base::u): Add an "nelts" field.
14752 (tree_vector): Use VECTOR_CST_NELTS as the length.
14753 * tree.c (tree_size): Likewise.
14754 (make_vector): Initialize VECTOR_CST_NELTS.
14755 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
14756 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
14757 TYPE_VECTOR_SUBPARTS.
14758 * expr.c (const_vector_mask_from_tree): Consistently use "units"
14759 as the number of units, setting it from VECTOR_CST_NELTS.
14760 (const_vector_from_tree): Likewise.
14761 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
14762 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
14763 (fold_negate_expr_1): Likewise.
14764 (fold_convert_const): Likewise.
14765 (const_binop): Likewise. Differentiate the number of output and
14767 (const_unop): Likewise.
14768 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
14769 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
14770 in cases that did the opposite.
14772 2017-09-14 Richard Biener <rguenther@suse.de>
14774 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
14777 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
14779 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
14781 2017-09-14 Jakub Jelinek <jakub@redhat.com>
14784 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
14785 if and where to split a bb, except for splitting before debug insn
14786 sequences followed by non-label real insn. Delete debug insns
14787 in between basic blocks.
14789 * combine.c (make_compound_operation_int): Formatting fixes.
14791 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
14792 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14793 * config/netbsd.h (LINK_EH_SPEC): Likewise.
14794 * config/sol2.h (LINK_EH_SPEC): Likewise.
14795 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14796 * config/s390/linux.h (LINK_SPEC): Likewise.
14797 * config/freebsd.h (LINK_EH_SPEC): Likewise.
14798 * config/openbsd.h (LINK_EH_SPEC): Likewise.
14799 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14800 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
14801 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
14802 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14803 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
14804 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
14806 2017-09-13 Jakub Jelinek <jakub@redhat.com>
14808 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
14810 (ENDFILE_LINUX_SPEC): Likewise.
14811 (LINK_EH_SPEC): Likewise.
14812 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
14813 (LINK_OS_LINUX_SPEC32): Likewise.
14814 (LINK_OS_LINUX_SPEC64): Likewise.
14815 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
14816 (LINK_OS_LINUX_SPEC): Likewise.
14818 2017-09-13 Martin Liska <mliska@suse.cz>
14820 PR middle-end/82154
14821 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
14822 CASE_HIGH is NULL_TREE.
14824 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14825 Alan Hayward <alan.hayward@arm.com>
14826 David Sherwood <david.sherwood@arm.com>
14828 * target.def (secondary_memory_needed): New hook.
14829 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
14830 instead of SECONDARY_MEMORY_NEEDED.
14831 (secondary_memory_needed_mode): Likewise.
14832 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
14833 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
14834 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
14835 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
14836 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
14837 * doc/tm.texi: Regenerate.
14838 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
14839 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
14840 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14841 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
14842 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
14843 * config/i386/i386.c (inline_secondary_memory_needed): Put the
14844 mode argument first and change the reg_class arguments to reg_class_t.
14845 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
14846 Make static. Update the call to inline_secondary_memory_needed.
14847 (ix86_register_move_cost): Update the call to
14848 inline_secondary_memory_needed.
14849 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14850 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
14852 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
14853 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14855 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
14856 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
14857 * config/mips/mips.c (mips_secondary_memory_needed): Make static
14858 and match hook interface. Add comment from mips.h.
14859 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14860 * config/mmix/mmix.md (truncdfsf2): Refer to
14861 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14863 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
14864 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
14865 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14866 (pa_secondary_memory_needed): New function.
14867 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
14868 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
14869 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14870 (pdp11_secondary_memory_needed): Make static and match hook interface.
14871 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
14872 * config/powerpcspe/powerpcspe-protos.h
14873 (rs6000_secondary_memory_needed_ptr): Delete.
14874 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
14876 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14877 (rs6000_option_override_internal): Assign to
14878 targetm.secondary_memory_needed rather than
14879 rs6000_secondary_memory_needed_ptr.
14880 (rs6000_secondary_memory_needed): Match hook interface.
14881 (rs6000_debug_secondary_memory_needed): Likewise.
14882 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
14883 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
14884 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
14885 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14886 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
14887 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
14889 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
14890 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14891 (rs6000_option_override_internal): Assign to
14892 targetm.secondary_memory_needed rather than
14893 rs6000_secondary_memory_needed_ptr.
14894 (rs6000_secondary_memory_needed): Match hook interface.
14895 (rs6000_debug_secondary_memory_needed): Likewise.
14896 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
14897 * config/s390/s390.c (s390_secondary_memory_needed): New function.
14898 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14899 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
14900 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14901 (sparc_secondary_memory_needed): New function.
14902 * lra-constraints.c (check_and_process_move): Refer to
14903 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14905 (curr_insn_transform): Likewise.
14906 (process_alt_operands): Use targetm.secondary_memory_needed
14907 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14908 (check_secondary_memory_needed_p): Likewise.
14909 (choose_split_class): Likewise.
14910 * reload.c: Unconditionally include code that was previously
14911 conditional on SECONDARY_MEMORY_NEEDED.
14912 (push_secondary_reload): Use targetm.secondary_memory_needed
14913 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14914 (push_reload): Likewise.
14915 * reload1.c: Unconditionally include code that was previously
14916 conditional on SECONDARY_MEMORY_NEEDED.
14917 (choose_reload_regs): Use targetm.secondary_memory_needed
14918 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14919 (gen_reload): Likewise.
14920 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
14922 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14923 Alan Hayward <alan.hayward@arm.com>
14924 David Sherwood <david.sherwood@arm.com>
14926 * target.def (secondary_memory_needed_mode): New hook:
14927 * targhooks.c (default_secondary_memory_needed_mode): Declare.
14928 * targhooks.h (default_secondary_memory_needed_mode): New function.
14929 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
14930 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
14931 * doc/tm.texi: Regenerate.
14932 * lra-constraints.c (check_and_process_move): Use
14933 targetm.secondary_memory_needed_mode instead of
14934 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
14935 (curr_insn_transform): Likewise.
14936 * reload.c (get_secondary_mem): Likewise.
14937 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14938 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
14940 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14941 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14942 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
14943 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14944 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
14946 * config/powerpcspe/powerpcspe-protos.h
14947 (rs6000_secondary_memory_needed_mode): Delete.
14948 * config/powerpcspe/powerpcspe.c
14949 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14950 (rs6000_secondary_memory_needed_mode): Make static.
14951 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14952 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
14954 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14956 (rs6000_secondary_memory_needed_mode): Make static.
14957 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14958 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
14959 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14960 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14961 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14963 (sparc_secondary_memory_needed_mode): New function.
14964 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
14966 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
14968 * config/aarch64/constraints.md (Umq): New constraint.
14969 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
14971 (mov<mode>): Update condition.
14973 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14975 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
14976 when bitposition is the same.
14978 2017-09-13 Richard Biener <rguenther@suse.de>
14980 * dwarf2out.c (output_die_symbol): Remove.
14981 (output_die): Do not output a DIEs symbol.
14983 2017-09-13 Richard Biener <rguenther@suse.de>
14985 PR middle-end/82128
14986 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
14987 default-def to avoid breaking iterator update with the weird
14988 interaction with cgraph_update_edges_for_call_stmt_node.
14990 2017-09-13 Richard Biener <rguenther@suse.de>
14992 * tree-cfg.c (verify_gimple_assign_binary): Add verification
14993 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
14994 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
14995 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
14997 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
14999 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
15000 Disable pc relative literal load irrespective of
15001 TARGET_FIX_ERR_A53_84341 for default.
15003 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
15005 * config/sparc/sparc.c (output_return): Output the source location of
15006 the insn in the delay slot, if any.
15007 (output_sibcall): Likewise.
15009 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
15012 * common.opt (-static-pie): New alias.
15013 (shared): Negate static-pie.
15014 (-no-pie): Update help text.
15016 (static-pie): New option.
15017 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
15018 -static-pie support.
15019 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
15020 (LINK_EH_SPEC): Likewise.
15021 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15022 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15023 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15024 * gcc.c (LINK_COMMAND_SPEC): Likewise.
15025 (init_gcc_specs): Likewise.
15026 (init_spec): Likewise.
15027 (display_help): Update help message for -pie.
15028 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
15031 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
15033 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
15034 (movdi_aarch64): Likewise.
15035 (movti_aarch64): Likewise.
15037 2017-09-12 Simon Wright <simon@pushface.org>
15040 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
15041 calculation of the minor version, always output as 0.
15043 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15046 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15047 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
15048 on it early, rather than manual conversion late. For
15049 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
15050 instead of performing manual conversion.
15052 2017-09-12 Carl Love <cel@us.ibm.com>
15054 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
15055 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
15057 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15058 VMULOSW): Add definitions.
15059 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15060 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15061 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15062 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15063 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15065 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15067 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
15069 (movti_aarch64): Likewise.
15070 (movdf_aarch64): Likewise.
15071 (movtf_aarch64): Likewise.
15072 (load_pairdi): Likewise.
15073 (store_pairdi): Likewise.
15074 (load_pairdf): Likewise.
15075 (store_pairdf): Likewise.
15076 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
15077 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
15078 (ldr_got_small_<mode>): Likewise.
15079 (ldr_got_small_28k_<mode>): Likewise.
15080 (ldr_got_tiny): Likewise.
15081 * config/aarch64/iterators.md (ldst_sz): New.
15082 (ldpstp_sz): Likewise.
15083 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
15085 (thunderx_load): Split load_8 to load_16.
15086 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
15088 (thunderx2t99_storepair_basic): Split store_8 to store_16.
15089 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
15090 (xgene1_store_pair): Split store_8 to store_16.
15091 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
15092 (falkor_st_0_st_sd): Split store_8 to store_16.
15094 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15096 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
15097 and store1/2/3/4 to store_4/8/12/16.
15098 * config/aarch64/aarch64.md: Update for rename.
15099 * config/arm/arm.md: Likewise.: Likewise.
15100 * config/arm/arm.c: Likewise.
15101 * config/arm/thumb1.md: Likewise.
15102 * config/arm/thumb2.md: Likewise.
15103 * config/arm/vfp.md: Likewise.
15104 * config/arm/arm-generic.md: Likewise.
15105 * config/arm/arm1020e.md: Likewise.
15106 * config/arm/arm1026ejs.md: Likewise.
15107 * config/arm/arm1136jfs.md: Likewise.
15108 * config/arm/arm926ejs.md: Likewise.
15109 * config/arm/cortex-a15.md: Likewise.
15110 * config/arm/cortex-a17.md: Likewise.
15111 * config/arm/cortex-a5.md: Likewise.
15112 * config/arm/cortex-a53.md: Likewise.
15113 * config/arm/cortex-a57.md: Likewise.
15114 * config/arm/cortex-a7.md: Likewise.
15115 * config/arm/cortex-a8.md: Likewise.
15116 * config/arm/cortex-a9.md: Likewise.
15117 * config/arm/cortex-m4.md: Likewise.
15118 * config/arm/cortex-m7.md: Likewise.
15119 * config/arm/cortex-r4.md: Likewise.
15120 * config/arm/exynos-m1.md: Likewise.
15121 * config/arm/fa526.md: Likewise.
15122 * config/arm/fa606te.md: Likewise.
15123 * config/arm/fa626te.md: Likewise.
15124 * config/arm/fa726te.md: Likewise.
15125 * config/arm/fmp626.md: Likewise.
15126 * config/arm/iwmmxt.md: Likewise.
15127 * config/arm/ldmstm.md: Likewise.
15128 * config/arm/marvell-pj4.md: Likewise.
15129 * config/arm/xgene1.md: Likewise.
15130 * config/aarch64/thunderx.md: Likewise.
15131 * config/aarch64/thunderx2t99.md: Likewise.
15132 * config/aarch64/falkor.md: Likewise.
15134 2017-09-12 Martin Liska <mliska@suse.cz>
15136 * attribs.c (private_lookup_attribute): New function.
15137 * attribs.h (private_lookup_attribute): Declared here.
15138 (lookup_attribute): Called from this place.
15140 2017-09-12 Richard Biener <rguenther@suse.de>
15142 PR tree-optimization/82157
15143 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
15144 stmts with side-effects.
15146 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15147 Alan Hayward <alan.hayward@arm.com>
15148 David Sherwood <david.sherwood@arm.com>
15150 * target.def (hard_regno_nregs): New hook.
15151 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
15152 * targhooks.h (default_hard_regno_nregs): Declare.
15153 * targhooks.c (default_hard_regno_nregs): New function.
15154 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
15155 (TARGET_HARD_REGNO_NREGS): ...this hook.
15156 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
15157 (CLASS_MAX_NREGS): Likewise.
15158 * doc/tm.texi: Regenerate.
15159 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
15160 instead of HARD_REGNO_NREGS.
15161 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
15162 HARD_REGNO_NREGS in the comment.
15163 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
15164 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
15165 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
15166 Return an unsigned int.
15167 (TARGET_HARD_REGNO_NREGS): Redefine.
15168 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
15169 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
15170 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15171 (arc_hard_regno_nregs): New function.
15172 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
15173 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
15174 (arm_hard_regno_nregs): New function.
15175 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
15176 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
15177 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
15178 (TARGET_HARD_REGNO_NREGS): Redefine.
15179 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
15180 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
15181 (HARD_REGNO_NREGS): Delete.
15182 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
15183 (cr16_hard_regno_nregs): New function.
15184 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
15185 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
15186 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
15187 (cris_hard_regno_nregs): New function.
15188 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
15189 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
15190 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15191 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
15192 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
15193 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
15194 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
15195 (frv_hard_regno_nregs): Make static. Take and return an
15197 (frv_class_max_nregs): Remove outdated copy of documentation.
15198 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
15199 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
15200 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
15201 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
15202 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
15203 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
15204 (TARGET_HARD_REGNO_NREGS): Redefine.
15205 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
15206 (CLASS_MAX_NREGS): Update comment.
15207 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
15208 (ia64_hard_regno_nregs): New function.
15209 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
15210 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
15211 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
15212 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
15213 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
15215 (m32c_hard_regno_nregs): Likewise. Make static.
15216 (TARGET_HARD_REGNO_NREGS): Redefine.
15217 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
15218 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
15219 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
15220 (m68k_hard_regno_nregs): New function.
15221 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
15222 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
15223 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
15224 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
15225 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
15226 Take and return an unsigned int.
15227 (TARGET_HARD_REGNO_NREGS): Redefine.
15228 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
15229 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15230 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
15231 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
15232 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
15233 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
15234 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
15235 (msp430_hard_regno_nregs): Make static. Take and return an
15237 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
15238 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
15239 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
15240 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
15241 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
15242 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
15243 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
15244 (TARGET_HARD_REGNO_NREGS): Redefine.
15245 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
15246 (PA_HARD_REGNO_NREGS): ...this.
15247 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
15248 (PA_HARD_REGNO_NREGS): ...this.
15249 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15250 (pa_hard_regno_nregs): New function.
15251 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
15252 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
15253 (pdp11_hard_regno_nregs): New function.
15254 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
15255 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
15256 (rs6000_hard_regno_nregs_hook): New function.
15257 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
15258 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
15259 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
15260 Take and return an unsigned int. Move earlier in file.
15261 (TARGET_HARD_REGNO_NREGS): Redefine.
15262 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
15263 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
15264 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
15265 (rl78_hard_regno_nregs): Make static. Take and return an
15267 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
15268 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
15269 (rs6000_hard_regno_nregs_hook): New function.
15270 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
15271 * config/rx/rx.c (rx_hard_regno_nregs): New function.
15272 (TARGET_HARD_REGNO_NREGS): Redefine.
15273 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
15274 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
15275 instead of HARD_REGNO_NREGS.
15276 (s390_hard_regno_nregs): New function.
15277 (s390_hard_regno_mode_ok): Add comment from s390.h.
15278 (TARGET_HARD_REGNO_NREGS): Redefine.
15279 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
15280 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
15281 (sh_hard_regno_nregs): New function.
15282 (sh_pass_in_reg_p): Use it.
15283 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
15284 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15285 (sparc_hard_regno_nregs): New function.
15286 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
15287 * config/spu/spu.c (spu_hard_regno_nregs): New function.
15288 (spu_function_arg_advance): Use it, supplying a valid register number.
15289 (TARGET_HARD_REGNO_NREGS): Redefine.
15290 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
15291 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
15292 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
15293 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
15294 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
15295 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
15296 (CLASS_MAX_NREGS): Remove copy of old documentation.
15297 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
15298 (visium_hard_regno_nregs): New function.
15299 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
15300 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
15301 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15302 xtensa_hard_regno_nregs): New function.
15303 * system.h (HARD_REGNO_NREGS): Poison.
15305 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15307 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
15308 hard_regno_nregs instead of HARD_REGNO_NREGS.
15309 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
15310 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
15311 (c6x_expand_epilogue): Likewise.
15312 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
15313 (frv_read_iacc_argument): Likewise.
15314 * config/sh/sh.c: Include regs.h.
15315 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15316 (regs_used): Likewise.
15317 (output_stack_adjust): Likewise.
15318 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
15319 * expmed.c: Include regs.h.
15320 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15321 * ree.c: Include regs.h.
15322 (combine_reaching_defs): Use hard_regno_nregs instead of
15324 (add_removable_extension): Likewise.
15326 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15328 * regs.h (hard_regno_nregs): Turn into a function.
15329 (end_hard_regno): Update accordingly.
15330 * caller-save.c (setup_save_areas): Likewise.
15331 (save_call_clobbered_regs): Likewise.
15332 (replace_reg_with_saved_mem): Likewise.
15333 (insert_restore): Likewise.
15334 (insert_save): Likewise.
15335 * combine.c (can_change_dest_mode): Likewise.
15336 (move_deaths): Likewise.
15337 (distribute_notes): Likewise.
15338 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
15339 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
15340 (rs6000_split_multireg_move): Likewise.
15341 (rs6000_register_move_cost): Likewise.
15342 (rs6000_memory_move_cost): Likewise.
15343 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
15344 (rs6000_split_multireg_move): Likewise.
15345 (rs6000_register_move_cost): Likewise.
15346 (rs6000_memory_move_cost): Likewise.
15347 * cselib.c (cselib_reset_table): Likewise.
15348 (cselib_lookup_1): Likewise.
15349 * emit-rtl.c (set_mode_and_regno): Likewise.
15350 * function.c (aggregate_value_p): Likewise.
15351 * ira-color.c (setup_profitable_hard_regs): Likewise.
15352 (check_hard_reg_p): Likewise.
15353 (calculate_saved_nregs): Likewise.
15354 (assign_hard_reg): Likewise.
15355 (improve_allocation): Likewise.
15356 (calculate_spill_cost): Likewise.
15357 * ira-emit.c (modify_move_list): Likewise.
15358 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
15359 (ira_hard_reg_in_set_p): Likewise.
15360 * ira.c (setup_reg_mode_hard_regset): Likewise.
15361 (clarify_prohibited_class_mode_regs): Likewise.
15362 (check_allocation): Likewise.
15363 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15364 (lra_setup_reg_renumber): Likewise.
15365 (setup_try_hard_regno_pseudos): Likewise.
15366 (spill_for): Likewise.
15367 (assign_hard_regno): Likewise.
15368 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
15369 * lra-constraints.c (in_class_p): Likewise.
15370 (lra_constraint_offset): Likewise.
15371 (simplify_operand_subreg): Likewise.
15372 (lra_constraints): Likewise.
15373 (split_reg): Likewise.
15374 (split_if_necessary): Likewise.
15375 (invariant_p): Likewise.
15376 (inherit_in_ebb): Likewise.
15377 * lra-lives.c (process_bb_lives): Likewise.
15378 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
15379 (get_hard_regs): Likewise.
15380 (do_remat): Likewise.
15381 * lra-spills.c (assign_spill_hard_regs): Likewise.
15382 * mode-switching.c (create_pre_exit): Likewise.
15383 * postreload.c (reload_combine_recognize_pattern): Likewise.
15384 * recog.c (peep2_find_free_register): Likewise.
15385 * regcprop.c (kill_value_regno): Likewise.
15386 (set_value_regno): Likewise.
15387 (copy_value): Likewise.
15388 (maybe_mode_change): Likewise.
15389 (find_oldest_value_reg): Likewise.
15390 (copyprop_hardreg_forward_1): Likewise.
15391 * regrename.c (check_new_reg_p): Likewise.
15392 (regrename_do_replace): Likewise.
15393 * reload.c (push_reload): Likewise.
15394 (combine_reloads): Likewise.
15395 (find_dummy_reload): Likewise.
15396 (operands_match_p): Likewise.
15397 (find_reloads): Likewise.
15398 (find_equiv_reg): Likewise.
15399 (reload_adjust_reg_for_mode): Likewise.
15400 * reload1.c (count_pseudo): Likewise.
15401 (count_spilled_pseudo): Likewise.
15402 (find_reg): Likewise.
15403 (clear_reload_reg_in_use): Likewise.
15404 (free_for_value_p): Likewise.
15405 (allocate_reload_reg): Likewise.
15406 (choose_reload_regs): Likewise.
15407 (reload_adjust_reg_for_temp): Likewise.
15408 (emit_reload_insns): Likewise.
15409 (delete_output_reload): Likewise.
15410 * rtlanal.c (subreg_get_info): Likewise.
15411 * sched-deps.c (sched_analyze_reg): Likewise.
15412 * sel-sched.c (init_regs_for_mode): Likewise.
15413 (mark_unavailable_hard_regs): Likewise.
15414 (choose_best_reg_1): Likewise.
15415 (verify_target_availability): Likewise.
15416 * valtrack.c (dead_debug_insert_temp): Likewise.
15417 * var-tracking.c (track_loc_p): Likewise.
15418 (emit_note_insn_var_location): Likewise.
15419 * varasm.c (make_decl_rtl): Likewise.
15420 * reginfo.c (choose_hard_reg_mode): Likewise.
15421 (init_reg_modes_target): Refer directly to
15422 this_target_regs->x_hard_regno_nregs.
15424 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15426 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
15427 instead of hard_regno_nregs.
15429 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15431 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
15432 end_hard_regno instead of hard_regno_nregs.
15433 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
15434 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
15435 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
15436 * ira-color.c (improve_allocation): Likewise.
15437 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15438 * lra-lives.c (mark_regno_live): Likewise.
15439 (mark_regno_dead): Likewise.
15440 * lra-remat.c (operand_to_remat): Likewise.
15441 * lra.c (collect_non_operand_hard_regs): Likewise.
15442 * postreload.c (reload_combine_note_store): Likewise.
15443 (move2add_valid_value_p): Likewise.
15444 * reload.c (regno_clobbered_p): Likewise.
15446 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15448 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
15450 * config/v850/v850.c (v850_reorg): Likewise.
15451 * reload.c (refers_to_regno_for_reload_p): Likewise.
15452 (find_equiv_reg): Likewise.
15453 * reload1.c (reload_reg_reaches_end_p): Likewise.
15455 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15457 * caller-save.c (add_used_regs): Use REG_NREGS instead of
15459 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
15460 * config/arm/arm.c (output_move_neon): Likewise.
15461 (arm_attr_length_move_neon): Likewise.
15462 (neon_split_vcombine): Likewise.
15463 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
15464 (c6x_mark_reg_written): Likewise.
15465 (c6x_dwarf_register_span): Likewise.
15466 * config/i386/i386.c (ix86_save_reg): Likewise.
15467 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
15468 (rws_access_reg): Likewise.
15469 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
15470 * mode-switching.c (create_pre_exit): Likewise.
15471 * ree.c (combine_reaching_defs): Likewise.
15472 (add_removable_extension): Likewise.
15473 * regcprop.c (find_oldest_value_reg): Likewise.
15474 (copyprop_hardreg_forward_1): Likewise.
15475 * reload.c (reload_inner_reg_of_subreg): Likewise.
15476 (push_reload): Likewise.
15477 (combine_reloads): Likewise.
15478 (find_dummy_reload): Likewise.
15479 (reload_adjust_reg_for_mode): Likewise.
15480 * reload1.c (find_reload_regs): Likewise.
15481 (forget_old_reloads_1): Likewise.
15482 (reload_reg_free_for_value_p): Likewise.
15483 (reload_adjust_reg_for_temp): Likewise.
15484 (emit_reload_insns): Likewise.
15485 (delete_output_reload): Likewise.
15486 * sel-sched.c (choose_best_reg_1): Likewise.
15487 (choose_best_pseudo_reg): Likewise.
15489 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15490 Alan Hayward <alan.hayward@arm.com>
15491 David Sherwood <david.sherwood@arm.com>
15493 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
15494 * target.def (slow_unaligned_access): New hook.
15495 * targhooks.h (default_slow_unaligned_access): Declare.
15496 * targhooks.c (default_slow_unaligned_access): New function.
15497 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
15498 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
15499 * doc/tm.texi: Regenerate.
15500 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
15501 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
15502 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
15504 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
15505 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
15507 (rs6000_slow_unaligned_access): New function.
15508 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15509 (expand_block_compare): Likewise.
15510 (expand_strn_compare): Likewise.
15511 (rs6000_rtx_costs): Likewise.
15512 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
15513 (riscv_slow_unaligned_access): Likewise.
15514 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
15515 (riscv_slow_unaligned_access_p): ...this and make static.
15516 (riscv_option_override): Update accordingly.
15517 (riscv_slow_unaligned_access): New function.
15518 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15519 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
15520 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15521 (rs6000_slow_unaligned_access): New function.
15522 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15523 (rs6000_rtx_costs): Likewise.
15524 * config/rs6000/rs6000-string.c (expand_block_compare)
15525 (expand_strn_compare): Use targetm.slow_unaligned_access instead
15526 of SLOW_UNALIGNED_ACCESS.
15527 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
15528 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
15529 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
15530 of SLOW_UNALIGNED_ACCESS.
15531 * expmed.c (simple_mem_bitfield_p): Likewise.
15532 * expr.c (alignment_for_piecewise_move): Likewise.
15533 (emit_group_load_1): Likewise.
15534 (emit_group_store): Likewise.
15535 (copy_blkmode_from_reg): Likewise.
15536 (emit_push_insn): Likewise.
15537 (expand_assignment): Likewise.
15538 (store_field): Likewise.
15539 (expand_expr_real_1): Likewise.
15540 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
15541 * lra-constraints.c (simplify_operand_subreg): Likewise.
15542 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15543 * gimple-ssa-store-merging.c: Likewise in block comment at start
15545 * tree-ssa-strlen.c: Include target.h.
15546 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
15547 of SLOW_UNALIGNED_ACCESS.
15548 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
15550 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15552 PR rtl-optimization/82185
15553 * expmed.c (emit_store_flag_int): Only test tem if it has been
15556 2017-09-12 Richard Biener <rguenther@suse.de>
15558 PR middle-end/82149
15559 * match.pd ((FTYPE) N CMP CST): Fix typo.
15561 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
15563 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
15565 (mips_near_type_p): Add 'short_call' attribute as a synonym
15567 * doc/extend.texi (short_call): Document new function attribute.
15569 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15572 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
15573 assertion check that in the condition.
15574 (get_atomic_generic_size): Likewise. Before testing if parameter
15575 has pointer type, if it has array type, call for C++
15576 default_conversion to perform array-to-pointer conversion.
15578 2017-09-12 Richard Biener <rguenther@suse.de>
15580 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
15581 for operations we cannot scalarize.
15583 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
15585 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
15586 vectors heap vectors. Clean up comments.
15587 Make visited_bbs a reference.
15588 (profitable_jump_thread_path): Make GC
15589 vectors heap vectors. Clean up comments.
15591 (convert_and_register_jump_thread_path): Make GC vectors heap
15593 (check_subpath_and_update_thread_path): Same. Clean up comments.
15594 Make visited_bbs a reference.
15595 (handle_phi): Abstract common code to to
15596 register_jump_thread_path_if_profitable.
15597 Rename VAR_BB to DEF_BB.
15599 Make GC vectors heap vectors.
15600 Make visited_bbs a reference.
15601 (handle_assignment): Same.
15602 (register_jump_thread_path_if_profitable): New.
15603 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
15605 Make GC vectors heap vectors. Clean up comments.
15606 Make visited_bbs a reference.
15607 (find_jump_threads_backwards): Make visited_bbs live in the stack.
15608 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
15611 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
15614 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
15615 words of E_DImode object are reachable by xtensa_uimm8x4 access.
15617 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
15619 Revert r251800 and r251799.
15621 2017-09-11 Martin Jambor <mjambor@suse.cz>
15624 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
15625 arguments in advance.
15626 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
15627 use it to find predecessor edges.
15628 (naive_outof_ssa): Collect vector of predecessors.
15630 2017-09-08 Jason Merrill <jason@redhat.com>
15632 PR c++/70029 - ICE with ref-qualifier and -flto
15633 * langhooks.h (struct lang_hooks_for_types): Add
15634 copy_lang_qualifiers.
15635 * attribs.c (build_type_attribute_qual_variant): Use it.
15636 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
15638 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
15639 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
15641 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
15644 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
15645 (*mulsi3_sp64): New instruction.
15646 (mulsi3): New expander.
15648 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
15650 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
15652 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15654 * sancov.c: Include memmodel.h.
15656 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
15659 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
15662 2017-09-07 Carl Love <cel@us.ibm.com>
15664 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
15665 the sldi instruction.
15667 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
15669 * sancov.c: Include tm_p.h.
15671 2017-09-07 Jakub Jelinek <jakub@redhat.com>
15674 * output.h (switch_to_other_text_partition): New declaration.
15675 * varasm.c (switch_to_other_text_partition): New function.
15676 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
15677 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
15678 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
15679 to the other text partition before emitting LCL label and switch back
15680 after emitting the word after it.
15682 2017-09-07 Richard Biener <rguenther@suse.de>
15684 * passes.def (pass_split_crit_edges): Remove instance before PRE.
15685 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
15686 critical edges here, after loop init.
15687 (pass_data_pre): Remove PROP_no_crit_edges flags.
15688 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
15689 for valueization of call args to avoid leaking VN_TOP.
15690 (visit_use): Assert we do not visit default defs.
15691 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
15692 Use error_mark_node to more easily detect leaking VN_TOP.
15693 All default-defs are varying, not VN_TOP. Mark them visited.
15694 (run_scc_vn): Make code match comment.
15696 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15698 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
15699 OPTION_MASK_FLOAT128_KEYWORD.
15700 (POWERPC_MASKS): Likewise.
15701 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
15702 support for the -mfloat128-type option, and make -mfloat128
15703 default on PowerPC Linux systems. Define or undefine
15704 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
15705 Define __float128 to be __ieee128 if IEEE 128-bit support is
15706 enabled, or undefine it.
15707 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
15708 Delete defining __FLOAT128_TYPE__.
15709 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
15710 -mfloat128-type option and make -mfloat128 default on PowerPC
15712 (TARGET_FLOAT128_TYPE): Likewise.
15713 (-mfloat128-type): Likewise.
15714 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15715 Delete the -mfloat128-type option and make -mfloat128 default on
15716 PowerPC Linux systems. Always use __ieee128 to be the keyword for
15717 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
15718 128-bit floating point is enabled. Change tests from using
15719 -mfloat128-type to -mfloat128.
15720 (rs6000_mangle_type): Use the correct mangling for the __float128
15721 type even if normal long double is restricted to 64-bits.
15722 (floatn_mode): Enable the _Float128 type by default on VSX Linux
15724 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
15725 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
15726 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
15728 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
15729 documentation for -mfloat128.
15731 2017-09-06 Olivier Hainque <hainque@adacore.com>
15733 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
15735 2017-09-06 Wish Wu <wishwu007@gmail.com>
15736 Jakub Jelinek <jakub@redhat.com>
15738 * asan.c (initialize_sanitizer_builtins): Add
15739 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
15740 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
15741 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
15742 BT_FN_VOID_UINT64_PTR variables.
15743 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
15744 (BT_FN_VOID_UINT16_UINT16): Likewise.
15745 (BT_FN_VOID_UINT32_UINT32): Likewise.
15746 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
15747 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
15748 (BT_FN_VOID_UINT64_PTR): Likewise.
15749 * common.opt (flag_sanitize_coverage): New variable.
15750 (fsanitize-coverage=trace-pc): Remove.
15751 (fsanitize-coverage=): Add.
15752 * flag-types.h (enum sanitize_coverage_code): New enum.
15753 * fold-const.c (fold_range_test): Disable non-short-circuit
15754 optimization if flag_sanitize_coverage.
15755 (fold_truth_andor): Likewise.
15756 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15757 * opts.c (COVERAGE_SANITIZER_OPT): Define.
15758 (coverage_sanitizer_opts): New array.
15759 (get_closest_sanitizer_option): Add OPTS argument, handle also
15760 OPT_fsanitize_coverage_.
15761 (parse_sanitizer_options): Adjusted to also handle
15762 OPT_fsanitize_coverage_.
15763 (common_handle_option): Add OPT_fsanitize_coverage_.
15764 * sancov.c (instrument_comparison, instrument_switch): New function.
15765 (sancov_pass): Add trace-cmp support.
15766 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
15767 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
15768 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
15769 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
15770 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
15771 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
15772 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
15773 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
15774 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
15775 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
15777 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15779 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
15780 error. Only quit immediately if parsing is complete.
15781 (BEGIN): Initialize fatal_err and parse_done.
15782 (begin fpu, end fpu): Check number of arguments.
15783 (begin arch, end arch): Likewise.
15784 (begin cpu, end cpu): Likewise.
15785 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
15786 (optalias): Likewise.
15788 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15790 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
15791 * config/arm/arm-isa.h: Delete. Move definitions to ...
15792 * arm-cpus.in: ... here. Use new feature and fgroup values.
15793 * config/arm/arm.c (arm_option_override): Use lower case for feature
15795 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
15796 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
15797 * config/arm/parsecpu.awk (END): Add new command 'isa'.
15799 (print_isa_bits_for): New function.
15800 (gen_isa): New function.
15801 (gen_comm_data): Use print_isa_bits_for.
15802 (define feature): New keyword.
15803 (define fgroup): New keyword.
15804 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
15805 (arm-isa.h): Add rule to generate file.
15806 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
15807 case for feature bit names.
15809 2017-09-06 Richard Biener <rguenther@suse.de>
15811 * tree-ssa-pre.c (NECESSARY): Remove.
15812 (create_expression_by_pieces): Do not touch pass-local flags.
15813 (insert_into_preds_of_block): Likewise.
15814 (do_pre_regular_insertion): Likewise.
15815 (eliminate_insert): Likewise.
15816 (eliminate_dom_walker::before_dom_children): Likewise.
15817 (fini_eliminate): Do not look at inserted_exprs.
15818 (mark_operand_necessary): Remove.
15819 (remove_dead_inserted_code): Replace with simple work-list
15820 algorithm based on inserted_exprs and SSA uses.
15821 (pass_pre::execute): Re-order fini_eliminate and
15822 remove_dead_inserted_code.
15824 2017-09-06 Olivier Hainque <hainque@adacore.com>
15826 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
15827 for VxWorks 7. Adjust surrounding comments.
15829 2017-09-06 Richard Biener <rguenther@suse.de>
15831 * gimple-ssa-strength-reduction.c
15832 (find_candidates_dom_walker::before_dom_children): Also allow
15835 2017-09-06 Richard Biener <rguenther@suse.de>
15837 PR tree-optimization/82108
15838 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
15839 for gap in the non-permutation SLP case.
15841 2017-09-06 Martin Jambor <mjambor@suse.cz>
15843 PR tree-optimization/82078
15844 * tree-sra.c (sort_and_splice_var_accesses): Move call to
15845 add_access_to_work_queue...
15846 (build_accesses_from_assign): ...here.
15847 (propagate_all_subaccesses): Make sure racc is the group
15848 representative, if there is one.
15850 2017-09-06 Jakub Jelinek <jakub@redhat.com>
15852 PR middle-end/82095
15853 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
15856 2017-09-06 Richard Biener <rguenther@suse.de>
15858 * gimple-ssa-strength-reduction.c
15859 (find_candidates_dom_walker::before_doom_children): Use a
15860 type and not a mode check.
15862 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
15865 * config/arm/predicates.md (arm_general_adddi_operand): Create new
15867 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
15869 2017-09-05 Jeff Law <law@redhat.com>
15871 PR tree-optimization/64910
15872 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
15873 cases where we have 3 or more operands.
15875 2017-09-05 Jakub Jelinek <jakub@redhat.com>
15877 PR middle-end/81768
15878 * omp-low.c (lower_omp_for): Recompute tree invariant if
15879 gimple_omp_for_initial/final is ADDR_EXPR.
15881 PR middle-end/81768
15882 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
15883 into gimple val before gimplification fo the COND_EXPR.
15885 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
15887 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
15888 REGION_COPY argument.
15889 (thread_through_all_blocks): Remove unused argument to
15890 duplicate_thread_path.
15892 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15893 Alan Hayward <alan.hayward@arm.com>
15894 David Sherwood <david.sherwood@arm.com>
15896 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
15897 Take a scalar_mode rather than a machine_mode.
15898 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15899 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
15900 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15901 (aarch64_gen_adjusted_ldpstp): Likewise.
15902 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
15904 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15905 Alan Hayward <alan.hayward@arm.com>
15906 David Sherwood <david.sherwood@arm.com>
15908 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
15909 Take a scalar_int_mode instead of a machine_mode.
15910 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15911 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15912 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15913 (aarch64_simd_attr_length_rglist): Delete.
15914 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
15915 a scalar_int_mode instead of a machine_mode.
15916 (aarch64_add_offset): Likewise.
15917 (aarch64_internal_mov_immediate): Likewise
15918 (aarch64_add_constant_internal): Likewise.
15919 (aarch64_add_constant): Likewise.
15920 (aarch64_movw_imm): Likewise.
15921 (aarch64_rtx_arith_op_extract_p): Likewise.
15922 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15923 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15924 Remove assert that the mode isn't a vector.
15925 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15926 (aarch64_expand_mov_immediate): Update calls after above changes.
15927 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
15928 (aarch64_and_bitmask_imm): Check for scalar integer modes.
15929 (aarch64_move_imm): Likewise.
15930 (aarch64_can_const_movi_rtx_p): Likewise.
15931 (aarch64_strip_extend): Likewise.
15932 (aarch64_extr_rtx_p): Likewise.
15933 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
15934 a CONST_INT when the mode parameter is VOIDmode.
15935 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
15937 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15939 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
15940 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15941 (bitwise_type_for_mode): Update accordingly.
15943 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15945 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
15946 * stor-layout.c (mode_for_size_tree): Likewise.
15947 (mode_for_array): Update accordingly.
15948 (layout_decl): Likewise.
15949 (compute_record_mode): Likewise. Only set the mode once.
15951 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15953 * target.def (get_mask_mode): Change return type to opt_mode.
15955 * doc/tm.texi: Regenerate.
15956 * targhooks.h (default_get_mask_mode): Return an opt_mode.
15957 * targhooks.c (default_get_mask_mode): Likewise.
15958 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15959 * optabs-query.c (can_vec_mask_load_store_p): Update use of
15960 targetm.get_mask_mode.
15961 * tree.c (build_truth_vector_type): Likewise.
15963 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15965 * machmode.h (mode_for_vector): Return an opt_mode.
15966 * stor-layout.c (mode_for_vector): Likewise.
15967 (mode_for_int_vector): Update accordingly.
15968 (layout_type): Likewise.
15969 * config/i386/i386.c (emit_memmov): Likewise.
15970 (ix86_expand_set_or_movmem): Likewise.
15971 (ix86_expand_vector_init): Likewise.
15972 (ix86_get_mask_mode): Likewise.
15973 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
15975 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
15976 * expmed.c (extract_bit_field_1): Likewise.
15977 * expr.c (expand_expr_real_2): Likewise.
15978 * optabs-query.c (can_vec_perm_p): Likewise.
15979 (can_vec_mask_load_store_p): Likewise.
15980 * optabs.c (expand_vec_perm): Likewise.
15981 * targhooks.c (default_get_mask_mode): Likewise.
15982 * tree-vect-stmts.c (vectorizable_store): Likewise.
15983 (vectorizable_load): Likewise.
15984 (get_vectype_for_scalar_type_and_size): Likewise.
15986 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15988 * machmode.h (mode_for_int_vector): New function.
15989 * stor-layout.c (mode_for_int_vector): Likewise.
15990 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
15991 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
15992 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15993 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15994 (s390_expand_vcond): Likewise.
15996 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15998 * machmode.h (opt_machine_mode): New type.
15999 (opt_mode<T>): Allow construction from anything that can be
16001 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
16002 (mode_for_size): Return an opt_machine_mode.
16003 * stor-layout.c (mode_for_size): Likewise.
16004 (mode_for_size_tree): Update call accordingly.
16005 (bitwise_mode_for_mode): Likewise.
16006 (make_fract_type): Likewise.
16007 (make_accum_type): Likewise.
16008 * caller-save.c (replace_reg_with_saved_mem): Update call
16010 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16011 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16012 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16013 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16014 * expmed.c (extract_bit_field_1): Likewise.
16015 * reload.c (get_secondary_mem): Likewise.
16016 * varasm.c (assemble_integer): Likewise.
16017 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
16020 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16022 * machmode.h (decimal_float_mode_for_size): New function.
16023 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
16024 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
16025 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
16026 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
16027 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
16028 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
16030 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16032 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
16033 (get_builtin_sync_mode): Likewise.
16034 (expand_ifn_atomic_compare_exchange): Likewise.
16035 (expand_builtin_atomic_clear): Likewise.
16036 (expand_builtin_atomic_test_and_set): Likewise.
16037 (fold_builtin_atomic_always_lock_free): Likewise.
16038 * calls.c (compute_argument_addresses): Likewise.
16039 (emit_library_call_value_1): Likewise.
16040 (store_one_arg): Likewise.
16041 * combine.c (combine_instructions): Likewise.
16042 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
16043 * config/arm/arm.c (arm_function_value): Likewise.
16044 (aapcs_allocate_return_reg): Likewise.
16045 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
16046 * config/i386/i386.c (construct_container): Likewise.
16047 (ix86_gimplify_va_arg): Likewise.
16048 (ix86_expand_sse_cmp): Likewise.
16049 (emit_memmov): Likewise.
16050 (emit_memset): Likewise.
16051 (expand_small_movmem_or_setmem): Likewise.
16052 (ix86_expand_pextr): Likewise.
16053 (ix86_expand_pinsr): Likewise.
16054 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
16055 * config/microblaze/microblaze.c (microblaze_block_move_straight):
16057 * config/mips/mips.c (mips_function_value_1) Likewise.
16058 (mips_block_move_straight): Likewise.
16059 (mips_expand_ins_as_unaligned_store): Likewise.
16060 * config/powerpcspe/powerpcspe.c
16061 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16062 (rs6000_darwin64_record_arg_flush): Likewise.
16063 * config/rs6000/rs6000.c
16064 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16065 (rs6000_darwin64_record_arg_flush): Likewise.
16066 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
16067 (sparc_function_value_1): Likewise.
16068 * config/spu/spu.c (adjust_operand): Likewise.
16069 (spu_emit_branch_or_set): Likewise.
16070 (arith_immediate_p): Likewise.
16071 * emit-rtl.c (gen_lowpart_common): Likewise.
16072 * expr.c (expand_expr_real_1): Likewise.
16073 * function.c (assign_parm_setup_block): Likewise.
16074 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
16075 * reload1.c (alter_reg): Likewise.
16076 * stor-layout.c (mode_for_vector): Likewise.
16077 (layout_type): Likewise.
16079 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16081 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
16082 (spu_convert_move): Likewise.
16083 * lower-subreg.c (resolve_simple_move): Likewise.
16085 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16088 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
16089 define_insn to a define_expand.
16090 (altivec_vsum2sws_direct): New define_insn.
16091 (altivec_vsumsws): Convert from a define_insn to a define_expand.
16093 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16095 * config/arm/arm.c (arm_option_params_internal): Improve setting of
16098 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
16103 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
16104 realign stack if stack alignment needed is less than incoming
16107 2017-09-05 Marek Polacek <polacek@redhat.com>
16110 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
16113 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16115 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
16117 2017-09-05 Richard Biener <rguenther@suse.de>
16119 PR tree-optimization/82084
16120 * fold-const.c (can_native_encode_string_p): Handle wide characters.
16122 2017-09-05 Richard Biener <rguenther@suse.de>
16124 PR tree-optimization/82102
16125 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
16127 2017-09-05 Martin Liska <mliska@suse.cz>
16129 PR tree-optimization/82032
16130 * tree-cfg.c (generate_range_test): New function.
16131 * tree-cfg.h (generate_range_test): Declared here.
16132 * tree-cfgcleanup.c (convert_single_case_switch): New function.
16133 (cleanup_control_expr_graph): Use it.
16134 * tree-switch-conversion.c (try_switch_expansion): Remove
16136 (emit_case_nodes): Use generate_range_test.
16138 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16141 * config/i386/i386.md (*<btsc><mode>_mask): Add
16142 TARGET_USE_BT to insn constraint.
16143 (*btr<mode>_mask): Ditto.
16145 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
16147 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
16148 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
16150 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
16153 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
16154 TARGET_NEON and TARGET_IWMMXT.
16155 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
16156 TARGET_NEON and TARGET_IWMMXT.
16157 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
16159 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16161 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
16162 (ix86_rewrite_tls_address): Ditto.
16163 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
16164 (ix86_rewrite_tls_address_1): Ditto.
16165 (ix86_rewrite_tls_address): Ditto.
16166 * config/i386/predicates.md (tls_address_pattern): New predicate.
16167 * config/i386/i386.md (TLS address splitter): New splitter.
16169 2017-09-04 Richard Biener <rguenther@suse.de>
16171 PR tree-optimization/82084
16172 * fold-const.h (can_native_encode_string_p): Declare.
16173 * fold-const.c (can_native_encode_string_p): Factor out from ...
16174 (native_encode_string): ... here.
16175 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
16176 vectorizing stores from constants we later cannot handle.
16178 2017-09-04 Marek Polacek <polacek@redhat.com>
16181 * doc/invoke.texi: Update -Wtautological-compare documentation.
16183 2017-09-04 Jeff Law <law@redhat.com>
16185 PR tree-optimization/64910
16186 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
16187 swap the first and last operand if the last is a constant.
16189 2017-09-04 Marek Polacek <polacek@redhat.com>
16192 * convert.c (do_narrow): When sanitizing signed integer overflows,
16193 bail out for signed types.
16194 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
16196 2017-09-04 Richard Biener <rguenther@suse.de>
16198 PR tree-optimization/82060
16199 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16200 Move devirtualization after stmt folding and before EH/AB/noreturn
16201 cleanup to get the stmt refs canonicalized. Use a bool instead
16202 of gimple_modified_p since that doesn't work for NOPs. Schedule
16203 NOPs generated by folding for removal.
16205 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16206 Alan Hayward <alan.hayward@arm.com>
16207 David Sherwood <david.sherwood@arm.com>
16209 * coretypes.h (pad_direction): New enum.
16210 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
16211 (FUNCTION_ARG_PADDING): Likewise.
16212 * target.def (function_arg_padding): New hook.
16213 * targhooks.h (default_function_arg_padding): Declare.
16214 * targhooks.c (default_function_arg_padding): New function.
16215 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
16216 (TARGET_FUNCTION_ARG_PADDING): ...this.
16217 * doc/tm.texi: Regenerate.
16218 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
16219 instead of direction.
16220 (compute_argument_addresses): Likewise.
16221 (load_register_parameters): Likewise.
16222 (emit_library_call_value_1): Likewise.
16223 (store_one_arg): Use targetm.calls.function_arg_padding instead
16224 of FUNCTION_ARG_PADDING.
16225 (must_pass_in_stack_var_size_or_pad): Likewise.
16226 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
16227 (emit_group_store): Likewise.
16228 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
16229 instead of FUNCTION_ARG_PADDING.
16230 (emit_push_insn): Likewise, and propagate enum change throughout
16232 * function.h (direction): Delete.
16233 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
16235 * function.c (assign_parm_find_stack_rtl): Likewise.
16236 (assign_parm_setup_block_p): Likewise.
16237 (assign_parm_setup_block): Likewise.
16238 (gimplify_parameters): Likewise.
16239 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
16240 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
16242 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
16243 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16244 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
16245 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
16246 (aarch64_function_arg_padding): ...this new function.
16247 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
16248 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16249 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
16250 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16251 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
16252 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16253 (arm_pad_arg_upward): Replace with...
16254 (arm_function_arg_padding): ...this new function.
16255 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
16257 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
16258 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
16259 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16260 (ia64_hpux_function_arg_padding): Replace with...
16261 (ia64_function_arg_padding): ...this new function. Use pad_direction
16262 instead of direction. Check for TARGET_HPUX.
16263 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
16264 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16265 (iq2000_function_arg_padding): New function.
16266 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
16267 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
16268 (mips_function_arg_padding): ...this new function.
16269 (mips_pad_reg_upward): Update accordingly.
16270 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16271 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
16272 targetm.calls.function_arg_padding.
16273 (FUNCTION_ARG_PADDING): Delete.
16274 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16275 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
16276 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16277 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
16278 (nios2_block_reg_padding): Return pad_direction instead of direction.
16279 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
16280 instead of direction.
16281 (nios2_function_arg_padding): Likewise. Make static.
16282 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16283 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
16284 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
16285 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
16286 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16287 (pa_function_arg_padding): Make static. Return pad_direction instead
16289 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
16290 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16291 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
16292 instead of direction. Use targetm.calls.function_arg_padding.
16293 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
16294 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16295 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
16296 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
16297 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
16299 (function_arg_padding): Rename to...
16300 (rs6000_function_arg_padding): ...this. Make static. Return
16301 pad_direction instead of direction.
16302 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16303 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
16304 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16305 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
16306 instead of direction. Use targetm.calls.function_arg_padding.
16307 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
16308 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16309 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
16310 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
16311 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16312 (function_arg_padding): Rename to...
16313 (rs6000_function_arg_padding): ...this. Make static. Return
16314 pad_direction instead of direction.
16315 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16316 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
16317 * config/s390/s390.c (s390_function_arg_padding): New function.
16318 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16319 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
16320 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
16321 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16322 (function_arg_padding): Rename to...
16323 (sparc_function_arg_padding): ...this. Make static. Return
16324 pad_direction instead of direction.
16325 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
16326 * config/spu/spu.c (spu_function_arg_padding): New function.
16327 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16328 * system.h (FUNCTION_ARG_PADDING): Poison.
16330 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16331 Alan Hayward <alan.hayward@arm.com>
16332 David Sherwood <david.sherwood@arm.com>
16334 * target.def (modes_tieable_p): New hook.
16335 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
16336 (TARGET_MODES_TIEABLE_P): ...this.
16337 * doc/tm.texi.in: Regenerate.
16338 * hooks.h (hook_bool_mode_mode_true): Declare.
16339 * hooks.c (hook_bool_mode_mode_true): New function.
16340 * combine.c (subst): Use targetm.modes_tieable_p instead of
16342 * dse.c (find_shift_sequence): Likewise.
16343 * expmed.c (extract_low_bits): Likewise.
16344 * lower-subreg.c: Include target.h.
16345 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
16347 * rtlanal.c (rtx_cost): Likewise.
16348 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
16349 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
16350 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
16351 (TARGET_MODES_TIEABLE_P): Redefine.
16352 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
16353 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
16354 (TARGET_MODES_TIEABLE_P): Redefine.
16355 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
16356 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
16357 (arc_modes_tieable_p): New function.
16358 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
16359 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
16360 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
16361 (arm_modes_tieable_p): Make static.
16362 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
16363 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
16364 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
16365 (TARGET_MODES_TIEABLE_P): Redefine.
16366 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
16367 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
16368 (TARGET_MODES_TIEABLE_P): Redefine.
16369 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
16370 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
16371 (cr16_modes_tieable_p): New function.
16372 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
16373 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
16374 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
16375 (TRULY_NOOP_TRUNCATION): Update comment.
16376 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
16377 (TRULY_NOOP_TRUNCATION): Update comment.
16378 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
16379 (frv_modes_tieable_p): New function.
16380 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
16381 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
16382 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
16383 (TARGET_MODES_TIEABLE_P): Redefine.
16384 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
16385 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
16386 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
16387 (TARGET_MODES_TIEABLE_P): Redefine.
16388 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
16389 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
16390 (ia64_modes_tieable_p): New function.
16391 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
16392 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
16393 (iq2000_modes_tieable_p): New function.
16394 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
16395 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
16396 (lm32_modes_tieable_p): New function.
16397 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
16398 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
16399 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
16400 (TARGET_MODES_TIEABLE_P): Redefine.
16401 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
16402 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
16403 (m32r_modes_tieable_p): New function.
16404 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
16405 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
16406 (m68k_modes_tieable_p): New function.
16407 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
16408 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
16409 (mcore_modes_tieable_p): New function.
16410 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
16411 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
16413 (TARGET_MODES_TIEABLE_P): Redefine.
16414 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
16415 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
16416 * config/mips/mips.c (mips_modes_tieable_p): Make static.
16417 (TARGET_MODES_TIEABLE_P): Redefine.
16418 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
16419 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
16420 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
16421 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
16422 (mn10300_modes_tieable_p): ...this and make static.
16423 (TARGET_MODES_TIEABLE_P): Redefine.
16424 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
16425 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
16426 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
16427 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
16428 (msp430_modes_tieable_p): Make static.
16429 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
16430 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
16431 (TARGET_MODES_TIEABLE_P): Redefine.
16432 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
16433 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
16434 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
16435 (TARGET_MODES_TIEABLE_P): Redefine.
16436 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
16437 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
16438 * config/pa/pa.c (pa_modes_tieable_p): Make static.
16439 (TARGET_MODES_TIEABLE_P): Redefine.
16440 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
16441 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
16442 (pdp11_modes_tieable_p): New function.
16443 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
16444 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
16445 (rs6000_modes_tieable_p): New function.
16446 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16447 * config/powerpcspe/powerpcspe.md: Update comment.
16448 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
16449 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
16450 (TARGET_MODES_TIEABLE_P): Redefine.
16451 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
16452 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
16453 (rl78_modes_tieable_p): New function.
16454 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
16455 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
16456 (rs6000_modes_tieable_p): New function.
16457 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16458 * config/rs6000/rs6000.md: Update comment.
16459 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
16460 * config/rx/rx.c (rx_modes_tieable_p): New function.
16461 (TARGET_MODES_TIEABLE_P): Redefine.
16462 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
16463 * config/s390/s390.c (s390_modes_tieable_p): New function.
16464 (TARGET_MODES_TIEABLE_P): Redefine.
16465 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
16466 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
16467 (sh_modes_tieable_p): New function.
16468 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
16469 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
16470 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
16471 (sparc_modes_tieable_p): Make static.
16472 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
16473 * config/spu/spu.c (spu_modes_tieable_p): New function.
16474 (TARGET_MODES_TIEABLE_P): Redefine.
16475 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
16476 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
16477 (TARGET_MODES_TIEABLE_P): Redefine.
16478 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
16479 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
16480 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
16481 * config/v850/v850.c (v850_modes_tieable_p): New function.
16482 (TARGET_MODES_TIEABLE_P): Redefine.
16483 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
16484 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
16485 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
16486 (visium_modes_tieable_p): New function.
16487 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
16488 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
16489 (xtensa_modes_tieable_p): New function.
16490 * system.h (MODES_TIEABLE_P): Poison.
16492 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16493 Alan Hayward <alan.hayward@arm.com>
16494 David Sherwood <david.sherwood@arm.com>
16496 * target.def (hard_regno_mode_ok): New hook.
16497 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
16498 (TARGET_HARD_REGNO_MODE_OK): ...this.
16499 * doc/tm.texi.in: Regenerate.
16500 * hooks.h (hook_bool_uint_mode_true): Declare.
16501 * hooks.c (hook_bool_uint_mode_true): New function.
16502 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
16503 HARD_REGNO_MODE_OK.
16504 * genpreds.c (write_insn_preds_c): Add an include of target.h.
16505 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
16506 instead of HARD_REGNO_MODE_OK.
16507 * caller-save.c: Include target.h.
16508 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
16509 HARD_REGNO_MODE_OK.
16510 * combine.c (can_combine_p): Likewise.
16511 (combinable_i3pat): Likewise.
16512 (can_change_dest_mode): Likewise.
16513 * expr.c (init_expr_target): Likewise.
16514 (convert_move): Likewise.
16515 (convert_modes): Likewise.
16516 * ira.c (setup_prohibited_class_mode_regs): Likewise.
16517 (setup_prohibited_mode_move_regs): Likewise.
16518 * ira.h (target_ira): Likewise.
16519 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16520 * lra-constraints.c (process_alt_operands): Likewise.
16521 (split_reg): Likewise.
16522 * recog.c (peep2_find_free_register): Likewise.
16523 * ree.c (combine_reaching_defs): Likewise.
16524 * regcprop.c (maybe_mode_change): Likewise.
16525 * reginfo.c (init_reg_sets_1): Likewise.
16526 (choose_hard_reg_mode): Likewise.
16527 (simplifiable_subregs): Likewise.
16528 * regrename.c (check_new_reg_p): Likewise.
16529 * reload.c (find_valid_class): Likewise.
16530 (find_valid_class_1): Likewise.
16531 (reload_inner_reg_of_subreg): Likewise.
16532 (push_reload): Likewise.
16533 (combine_reloads): Likewise.
16534 (find_dummy_reload): Likewise.
16535 (find_reloads): Likewise.
16536 * reload1.c (find_reg): Likewise.
16537 (set_reload_reg): Likewise.
16538 (allocate_reload_reg): Likewise.
16539 (choose_reload_regs): Likewise.
16540 (reload_adjust_reg_for_temp): Likewise.
16541 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
16542 (simplify_subreg_regno): Likewise.
16543 * sel-sched.c (init_regs_for_mode): Likewise.
16544 * varasm.c (make_decl_rtl): Likewise.
16545 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
16546 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
16547 HARD_REGNO_MODE_OK.
16548 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
16549 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
16550 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16551 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
16552 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
16553 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16554 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
16555 (arc_mode_class): Delete.
16556 (HARD_REGNO_MODE_OK): Delete.
16557 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16558 (arc_hard_regno_mode_ok): Rename old array to...
16559 (arc_hard_regno_mode_ok_modes): ...this.
16560 (arc_conditional_register_usage): Update accordingly.
16561 (arc_mode_class): Make static.
16562 (arc_hard_regno_mode_ok): New function.
16563 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
16564 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
16565 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16566 (arm_hard_regno_mode_ok): Make static.
16567 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
16568 HARD_REGNO_MODE_OK.
16569 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
16570 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
16571 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
16573 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16574 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
16575 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
16576 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
16577 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
16578 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16579 * config/bfin/predicates.md (valid_reg_operand): Use
16580 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
16581 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
16582 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
16583 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16584 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
16585 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
16586 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16587 (cr16_hard_regno_mode_ok): Make static and return a bool.
16588 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
16589 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16590 (cris_hard_regno_mode_ok): New function.
16591 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
16592 (epiphany_mode_class): Delete.
16593 (HARD_REGNO_MODE_OK): Delete.
16594 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
16595 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16596 (hard_regno_mode_ok): Rename to...
16597 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
16598 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
16599 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
16600 HARD_REGNO_MODE_OK.
16601 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
16602 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
16603 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16604 (frv_hard_regno_mode_ok): Make static and return a bool.
16605 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
16606 HARD_REGNO_MODE_OK.
16607 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
16608 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
16609 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
16610 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
16612 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16613 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
16614 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
16615 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
16617 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16618 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
16619 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16620 (ia64_hard_regno_mode_ok): New function.
16621 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
16622 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16623 (iq2000_hard_regno_mode_ok): New function.
16624 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
16625 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16626 (lm32_hard_regno_mode_ok): New function.
16627 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
16628 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
16629 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
16630 instead of HARD_REGNO_MODE_OK.
16631 (m32c_hard_regno_ok): Rename to...
16632 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
16633 (m32c_cannot_change_mode_class): Update accordingly.
16634 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16635 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
16636 (m32r_mode_class): Delete.
16637 (HARD_REGNO_MODE_OK): Delete.
16638 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16639 (m32r_hard_regno_mode_ok): Rename to...
16640 (m32r_hard_regno_modes): ...this.
16641 (m32r_mode_class): Make static.
16642 (m32r_hard_regno_mode_ok): New function.
16643 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
16644 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
16645 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16646 (m68k_hard_regno_mode_ok): Make static.
16647 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
16648 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16649 (mcore_hard_regno_mode_ok): New function.
16650 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
16651 (HARD_REGNO_MODE_OK): Delete.
16652 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
16654 (microblaze_hard_regno_mode_ok_p): ...this and make static.
16655 (microblaze_hard_regno_mode_ok): New function.
16656 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16657 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
16658 (mips_hard_regno_mode_ok): Delete.
16659 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
16660 (mips_hard_regno_mode_ok_p): ...this and make static.
16661 (mips_hard_regno_mode_ok_p): Rename to...
16662 (mips_hard_regno_mode_ok_uncached): ...this.
16663 (mips_hard_regno_mode_ok): New function.
16664 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
16665 of HARD_REGNO_MODE_OK.
16666 (mips_option_override): Update after above name changes.
16667 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16668 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
16669 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
16670 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
16671 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
16672 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16673 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
16674 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
16675 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
16676 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16677 (msp430_hard_regno_mode_ok): Make static and return a bool.
16678 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
16679 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
16680 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
16682 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16683 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
16684 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
16685 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
16686 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
16687 (PA_HARD_REGNO_MODE_OK): ...this
16688 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
16689 (PA_HARD_REGNO_MODE_OK): ...this.
16690 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16691 (pa_hard_regno_mode_ok): New function.
16692 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
16693 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16694 (pdp11_hard_regno_mode_ok): New function.
16695 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
16696 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
16698 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
16700 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16701 (rs6000_hard_regno_mode_ok): Rename to...
16702 (rs6000_hard_regno_mode_ok_uncached): ...this.
16703 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16704 (rs6000_hard_regno_mode_ok): New function.
16705 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
16706 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
16707 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
16708 (riscv_hard_regno_mode_ok): ...this and make static.
16709 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16710 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
16711 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
16712 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16713 (rl78_hard_regno_mode_ok): Make static and return bool.
16714 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
16715 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
16717 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
16718 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16719 (rs6000_hard_regno_mode_ok): Rename to...
16720 (rs6000_hard_regno_mode_ok_uncached): ...this.
16721 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16722 (rs6000_hard_regno_mode_ok): New function.
16723 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
16724 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
16725 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16726 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
16727 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
16728 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
16729 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16730 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
16731 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
16732 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16733 (sh_hard_regno_mode_ok): Make static.
16734 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
16735 instead of HARD_REGNO_MODE_OK.
16736 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
16737 (sparc_mode_class): Delete.
16738 (HARD_REGNO_MODE_OK): Delete.
16739 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16740 (hard_regno_mode_classes): Make static.
16741 (sparc_mode_class): Likewise.
16742 (sparc_hard_regno_mode_ok): New function.
16743 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
16744 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
16745 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
16747 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16748 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
16749 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
16750 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
16751 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
16752 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16753 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
16754 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
16755 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16756 (visium_hard_regno_mode_ok): New function.
16757 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
16758 instead of HARD_REGNO_MODE_OK.
16759 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
16760 (HARD_REGNO_MODE_OK): Delete.
16761 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
16762 (xtensa_hard_regno_mode_ok_p): ...this and make static.
16763 (xtensa_option_override): Update accordingly.
16764 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16765 (xtensa_hard_regno_mode_ok): New function.
16766 * system.h (HARD_REGNO_MODE_OK): Poison.
16768 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16769 Alan Hayward <alan.hayward@arm.com>
16770 David Sherwood <david.sherwood@arm.com>
16772 * target.def (hard_regno_call_part_clobbered): New hook.
16773 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
16774 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
16775 * doc/tm.texi: Regenerate.
16776 * hooks.h (hook_bool_uint_mode_false): Declare.
16777 * hooks.c (hook_bool_uint_mode_false): New function.
16778 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16779 * cselib.c (cselib_process_insn): Use
16780 targetm.hard_regno_call_part_clobbered instead of
16781 HARD_REGNO_CALL_PART_CLOBBERED.
16782 * ira-conflicts.c (ira_build_conflicts): Likewise.
16783 * ira-costs.c (ira_tune_allocno_costs): Likewise.
16784 * lra-constraints.c (need_for_call_save_p): Likewise.
16785 * lra-lives.c: Include target.h.
16786 (check_pseudos_live_through_calls): Use
16787 targetm.hard_regno_call_part_clobbered instead of
16788 HARD_REGNO_CALL_PART_CLOBBERED.
16789 * regcprop.c: Include target.h.
16790 (copyprop_hardreg_forward_1): Use
16791 targetm.hard_regno_call_part_clobbered instead of
16792 HARD_REGNO_CALL_PART_CLOBBERED.
16793 * reginfo.c (choose_hard_reg_mode): Likewise.
16794 * regrename.c (check_new_reg_p): Likewise.
16795 * reload.c (find_equiv_reg): Likewise.
16796 * reload1.c (emit_reload_insns): Likewise.
16797 * sched-deps.c (deps_analyze_insn): Likewise.
16798 * sel-sched.c (init_regs_for_mode): Likewise.
16799 (mark_unavailable_hard_regs): Likewise.
16800 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
16801 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16802 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
16804 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16805 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16806 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
16808 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
16810 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16811 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16812 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
16814 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16815 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16816 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
16818 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16819 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
16821 * config/powerpcspe/powerpcspe.c
16822 (rs6000_hard_regno_call_part_clobbered): New function.
16823 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16824 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16825 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
16827 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16828 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16829 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
16831 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16832 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16833 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
16835 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16836 Alan Hayward <alan.hayward@arm.com>
16837 David Sherwood <david.sherwood@arm.com>
16839 * rtl.h (subreg_memory_offset): Declare.
16840 * emit-rtl.c (subreg_memory_offset): New function.
16841 * expmed.c (store_bit_field_1): Use it.
16842 * expr.c (undefined_operand_subword_p): Likewise.
16843 * simplify-rtx.c (simplify_subreg): Likewise.
16845 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
16847 PR rtl-optimization/57448
16850 * optabs.c (expand_atomic_load): Place compiler memory barriers if
16851 using atomic_load pattern.
16852 (expand_atomic_store): Likewise.
16854 2017-09-04 Jakub Jelinek <jakub@redhat.com>
16857 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
16858 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
16859 handling. Use replace_call_with_value with NULL instead of
16860 gsi_replace, unlink_stmt_vdef and release_defs.
16862 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
16865 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
16867 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16870 * rtl.h (emit_library_call_value_1): Declare.
16871 (emit_library_call): Replace declaration with a series of overloads.
16872 Remove the parameter count argument.
16873 (emit_library_call_value): Likewise.
16874 * calls.c (emit_library_call_value_1): Make global. Replace varargs
16875 with an "rtx_mode_t *".
16876 (emit_library_call_value): Delete.
16877 (emit_library_call): Likewise.
16878 * asan.c (asan_emit_stack_protection): Update calls accordingly.
16879 (asan_emit_allocas_unpoison): Likewise.
16880 * builtins.c (expand_builtin_powi): Likewise.
16881 (expand_asan_emit_allocas_unpoison): Likewise.
16882 * cfgexpand.c (expand_main_function): Likewise.
16883 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
16884 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
16885 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
16886 * config/arm/arm.c (arm_trampoline_init): Likewise.
16887 (arm_call_tls_get_addr): Likewise.
16888 (arm_expand_divmod_libfunc): Likewise.
16889 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
16890 (smulsi3_highpart): Likewise.
16891 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
16892 (c6x_expand_compare): Likewise.
16893 (c6x_expand_movmem): Likewise.
16894 * config/frv/frv.c (frv_trampoline_init): Likewise.
16895 * config/i386/i386.c (ix86_trampoline_init): Likewise.
16896 (ix86_expand_divmod_libfunc): Likewise.
16897 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
16898 (ia64_expand_compare): Likewise.
16899 (ia64_profile_hook): Likewise.
16900 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
16901 (nonlocal_goto): Likewise.
16902 (restore_stack_nonlocal): Likewise.
16903 * config/m32r/m32r.c (block_move_call): Likewise.
16904 (m32r_trampoline_init): Likewise.
16905 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
16906 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
16907 (m68k_call_m68k_read_tp): Likewise.
16908 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
16909 (microblaze_expand_divide): Likewise.
16910 * config/mips/mips.h (mips_args): Likewise.
16911 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
16912 (MIPS_ICACHE_SYNC): Likewise.
16913 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
16914 (nios2_trampoline_init): Likewise.
16915 * config/pa/pa.c (hppa_tls_call): Likewise.
16916 (pa_trampoline_init): Likewise.
16917 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
16918 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
16919 (expand_strn_compare): Likewise.
16920 (rs6000_generate_compare): Likewise.
16921 (rs6000_expand_float128_convert): Likewise.
16922 (output_profile_hook): Likewise.
16923 (rs6000_trampoline_init): Likewise.
16924 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
16925 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
16926 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
16927 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
16928 (rs6000_generate_compare): Likewise.
16929 (rs6000_expand_float128_convert): Likewise.
16930 (output_profile_hook): Likewise.
16931 (rs6000_trampoline_init): Likewise.
16932 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
16933 * config/sh/sh.c (sh_trampoline_init): Likewise.
16934 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
16935 (sparc_emit_float_lib_cmp): Likewise.
16936 (sparc32_initialize_trampoline): Likewise.
16937 (sparc64_initialize_trampoline): Likewise.
16938 (sparc_profile_hook): Likewise.
16939 * config/spu/spu.c (ea_load_store): Likewise.
16940 * config/spu/spu.md (floatunssidf2): Likewise.
16941 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
16942 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
16943 * config/visium/visium.c (expand_block_move_4): Likewise.
16944 (expand_block_move_2): Likewise.
16945 (expand_block_move_1): Likewise.
16946 (expand_block_set_4): Likewise.
16947 (expand_block_set_2): Likewise.
16948 (expand_block_set_1): Likewise.
16949 (visium_trampoline_init): Likewise.
16950 (visium_profile_hook): Likewise.
16951 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
16952 (xtensa_setup_frame_addresses): Likewise.
16953 (xtensa_trampoline_init): Likewise.
16954 * except.c (sjlj_emit_function_enter): Likewise.
16955 (sjlj_emit_function_exit): Likewise.
16956 * explow.c (allocate_dynamic_stack_space): Likewise.
16957 (probe_stack_range): Likewise.
16958 * expr.c (convert_mode_scalar): Likewise.
16959 * optabs.c (expand_binop): Likewise.
16960 (expand_twoval_binop_libfunc): Likewise.
16961 (expand_unop): Likewise.
16962 (prepare_cmp_insn): Likewise.
16963 (prepare_float_lib_cmp): Likewise.
16964 (expand_float): Likewise.
16965 (expand_fix): Likewise.
16966 (expand_fixed_convert): Likewise.
16967 (maybe_emit_sync_lock_test_and_set): Likewise.
16968 (expand_atomic_compare_and_swap): Likewise.
16969 (expand_mem_thread_fence): Likewise.
16970 (expand_atomic_fetch_op): Likewise.
16972 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
16974 * doc/generic.texi (OpenACC): Adjust URL.
16975 * doc/invoke.texi (C Dialect Options): Ditto.
16977 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
16979 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
16980 predicate for operand 1. Add (m,<S>) constraint.
16981 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
16982 Prevent memory operand 1 with register operand 2.
16984 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
16986 PR rtl-optimization/82024
16987 * combine.c (try_combine): If the combination result is a PARALLEL,
16988 and we only need to retain the SET in there that would be placed
16989 at I2, check that we can place that at I3 instead, before doing so.
16991 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16994 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
16996 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
16997 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
17000 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
17001 Jeff Law <law@redhat.com>
17003 * varasm.c (bss_initializer_p): Do not put constants into .bss
17004 (categorize_decl_for_section): Handle bss_initializer_p returning
17005 false when DECL_INITIAL is NULL.
17007 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17010 * config/s390/s390.c (s390_can_inline_p): New function.
17012 2017-09-01 Jeff Law <law@redhat.com>
17014 PR tree-optimization/82052
17015 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
17016 Always initialize the returned slot after a hash table miss
17017 when INSERT is true.
17019 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
17021 * config/s390/s390.md (mem_signal_fence): Remove.
17022 * doc/md.texi (mem_signal_fence): Remove.
17023 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
17025 * target-insns.def (mem_signal_fence): Remove.
17027 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17030 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
17033 * asan.c (create_odr_indicator): Strip name encoding from assembler
17034 name before appending it after __odr_asan_.
17036 2017-09-01 Martin Liska <mliska@suse.cz>
17038 PR tree-optimization/82059
17039 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
17040 frequency only when an edge is redirected.
17042 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17044 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
17045 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
17046 (arc_conditional_register_usage): Remove ARC600 lp_count
17048 (arc_file_start): Emit Tag_ARC_CPU_variation.
17049 (arc_can_use_doloop_p): New conditions to use ZOLs.
17050 (hwloop_fail): New function.
17051 (hwloop_optimize): Likewise.
17052 (hwloop_pattern_reg): Likewise.
17053 (arc_doloop_hooks): New struct, to be used with reorg_loops.
17054 (arc_reorg_loops): New function, calls reorg_loops.
17055 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
17056 (arc600_corereg_hazard): Remove ZOL checking, case handled by
17058 (arc_loop_hazard): Remove function, functionality moved into
17060 (arc_hazard): Remove arc_loop_hazard call.
17061 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
17062 into hwloop_optimize.
17063 (arc_label_align): Remove ZOL handling.
17064 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
17065 * config/arc/arc.md (doloop_begin): Remove pattern.
17066 (doloop_begin_i): Likewise.
17067 (doloop_end_i): Likewise.
17068 (doloop_fallback): Likewise.
17069 (doloop_fallback_m): Likewise.
17070 (doloop_end): Reimplement expand.
17071 (arc_lp): New pattern for LP instruction.
17072 (loop_end): New pattern.
17073 (loop_fail): Likewise.
17074 (decrement_and_branch_until_zero): Likewise.
17075 * config/arc/arc.opt (mlpc-width): New option.
17076 * doc/invoke.texi (mlpc-width): Document option.
17078 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17080 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
17081 (arc_ccfsm_advance): Fix checking for delay slots.
17082 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
17084 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17086 * config/arc/arc.md (movqi_insn): Add stores to save constant long
17088 (movhi_insn): Update store instruction constraint which are saving
17089 6-bit short immediates.
17090 (movsi_insn): Consider also short scaled load operations.
17091 (zero_extendhisi2_i): Use Usd constraint instead of T.
17092 (extendhisi2_i): Add q constraint.
17093 (arc_clzsi2): Add type and length attributes.
17094 (arc_ctzsi2): Likewise.
17095 * config/arc/constraints.md (Usc): Update constraint, the
17096 assembler can parse two relocations for a single instruction.
17098 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17100 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
17101 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
17103 2017-08-31 Olivier Hainque <hainque@adacore.com>
17105 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
17106 match as powerpc-wrs-vxworks*.
17108 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
17110 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
17111 register constraint for by-element operand.
17112 (aarch64_mls_elt_merge<mode>): Likewise.
17114 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17116 * config/arc/arc.c (arc_can_follow_jump): Check for short
17119 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17121 * config.gcc: Use g.opt for arc.
17122 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
17123 functionality moved to ...
17124 (legitimate_scaled_address_p): New function, ...here.
17125 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
17126 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
17127 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
17129 (arc_override_options): Handle G option.
17130 (arc_output_pic_addr_const): Correct function definition.
17131 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
17132 (arc_decl_anon_ns_mem_p): Delete.
17133 (arc_in_small_data_p): Overhaul this function to take into
17134 consideration the value given via G option.
17135 (arc_rewrite_small_data_1): Renamed and corrected old
17136 arc_rewrite_small_data function.
17137 (arc_rewrite_small_data): New function.
17138 (small_data_pattern): Don't use pic_offset_table_rtx.
17139 * config/arc/arc.h (CC1_SPEC): Recognize G option.
17140 * config/arc/simdext.md (movmisalignv2hi): Use
17141 prepare_move_operands function.
17143 (movmisalign*): Likewise.
17144 * doc/invoke.texi (ARC options): Document -G option.
17146 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17148 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
17150 * config/arc/arc.c (arc_print_operand): Output scalled address for
17151 sdata whenever is possible.
17152 (arc_in_small_data_p): Allow sdata for 64bit datum when double
17153 load/stores are available.
17154 (compact_sda_memory_operand): Check for the alignment required by
17155 code density instructions.
17156 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
17158 * config/arc/constraints.md (Usd): Update constraint.
17159 (Us0): New constraint.
17160 (Usc): Update constraint.
17162 2017-08-31 Richard Biener <rguenther@suse.de>
17164 PR middle-end/82054
17165 * dwarf2out.c (dwarf2out_early_global_decl): Process each
17166 function only once.
17168 2017-08-31 Tamar Christina <tamar.christina@arm.com>
17170 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
17171 Resize type_signature.
17173 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17174 Alan Hayward <alan.hayward@arm.com>
17175 David Sherwood <david.sherwood@arm.com>
17177 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
17178 subregs whose inner modes can be stored in GPRs.
17179 (aarch64_classify_index): Likewise.
17181 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17182 Alan Hayward <alan.hayward@arm.com>
17183 David Sherwood <david.sherwood@arm.com>
17185 * config/aarch64/iterators.md (V_cmp_result): Rename to...
17186 (V_INT_EQUIV): ...this.
17187 (v_cmp_result): Rename to...
17188 (v_int_equiv): ...this.
17189 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
17190 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
17191 (copysign<mode>3): Likewise.
17192 (aarch64_simd_bsl<mode>_internal): Likewise.
17193 (aarch64_simd_bsl<mode>): Likewise.
17194 (vec_cmp<mode><mode>): Likewise.
17195 (vcond<mode><mode>): Likewise.
17196 (vcond<v_cmp_mixed><mode>): Likewise.
17197 (vcondu<mode><v_cmp_mixed>): Likewise.
17198 (aarch64_cm<optab><mode>): Likewise.
17199 (aarch64_cmtst<mode>): Likewise.
17200 (aarch64_fac<optab><mode>): Likewise.
17201 (vec_perm_const<mode>): Likewise.
17202 (vcond_mask_<mode><v_cmp_result>): Rename to...
17203 (vcond_mask_<mode><v_int_equiv>): ...this.
17204 (vec_cmp<mode><v_cmp_result>): Rename to...
17205 (vec_cmp<mode><v_int_equiv>): ...this.
17207 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17208 Alan Hayward <alan.hayward@arm.com>
17209 David Sherwood <david.sherwood@arm.com>
17211 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
17213 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
17214 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
17215 (UNSPEC_LD4_DREG): New unspecs.
17216 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
17217 (aarch64_ld2<mode>_dreg_be): Replace with...
17218 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
17220 (aarch64_ld3<mode>_dreg_le)
17221 (aarch64_ld3<mode>_dreg_be): Replace with...
17222 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
17224 (aarch64_ld4<mode>_dreg_le)
17225 (aarch64_ld4<mode>_dreg_be): Replace with...
17226 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
17229 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17231 PR tree-optimization/81987
17232 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
17233 insert an initializer in a location not dominated by the stride
17236 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17238 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
17239 on the entire header of the finally block in the fallthru case.
17241 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17243 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
17245 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
17247 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
17248 rs6000_emit_move_from_cr and call renamed function.
17249 (rs6000_emit_prologue): Call renamed functions.
17250 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
17251 movesi_from_cr, remove volatile CRs.
17253 2017-08-30 Jon Beniston <jon@beniston.com>
17254 Richard Biener <rguenther@suse.de>
17256 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
17257 of VECTOR_MODE_P check.
17258 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
17259 element vector types.
17261 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17263 * df.h (df_read_modify_subreg_p): Remove in favor of...
17264 * rtl.h (read_modify_subreg_p): ...this new function. Take a
17265 const_rtx instead of an rtx.
17266 * cprop.c (local_cprop_find_used_regs): Update accordingly.
17267 * df-problems.c (df_word_lr_mark_ref): Likewise.
17268 * ira-lives.c (mark_pseudo_reg_live): Likewise.
17269 (mark_pseudo_reg_dead): Likewise.
17270 (mark_ref_dead): Likewise.
17271 * reginfo.c (init_subregs_of_mode): Likewise.
17272 * sched-deps.c (sched_analyze_1): Likewise.
17273 * df-scan.c (df_def_record_1): Likewise.
17274 (df_uses_record): Likewise.
17275 (df_read_modify_subreg_p): Remove in favor of...
17276 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
17277 const_rtx instead of an rtx.
17279 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17280 Alan Hayward <alan.hayward@arm.com>
17281 David Sherwood <david.sherwood@arm.com>
17283 * rtl.h (partial_subreg_p): New function.
17284 * caller-save.c (save_call_clobbered_regs): Use it.
17285 * calls.c (expand_call): Likewise.
17286 * combine.c (combinable_i3pat): Likewise.
17287 (simplify_set): Likewise.
17288 (make_extraction): Likewise.
17289 (make_compound_operation_int): Likewise.
17290 (gen_lowpart_or_truncate): Likewise.
17291 (force_to_mode): Likewise.
17292 (make_field_assignment): Likewise.
17293 (reg_truncated_to_mode): Likewise.
17294 (record_truncated_value): Likewise.
17295 (move_deaths): Likewise.
17296 * cse.c (record_jump_cond): Likewise.
17297 (cse_insn): Likewise.
17298 * cselib.c (cselib_lookup_1): Likewise.
17299 * expmed.c (extract_bit_field_using_extv): Likewise.
17300 * function.c (assign_parm_setup_reg): Likewise.
17301 * ifcvt.c (noce_convert_multiple_sets): Likewise.
17302 * ira-build.c (create_insn_allocnos): Likewise.
17303 * lra-coalesce.c (merge_pseudos): Likewise.
17304 * lra-constraints.c (match_reload): Likewise.
17305 (simplify_operand_subreg): Likewise.
17306 (curr_insn_transform): Likewise.
17307 * lra-lives.c (process_bb_lives): Likewise.
17308 * lra.c (new_insn_reg): Likewise.
17309 (lra_substitute_pseudo): Likewise.
17310 * regcprop.c (mode_change_ok): Likewise.
17311 (maybe_mode_change): Likewise.
17312 (copyprop_hardreg_forward_1): Likewise.
17313 * reload.c (push_reload): Likewise.
17314 (find_reloads): Likewise.
17315 (find_reloads_subreg_address): Likewise.
17316 * reload1.c (alter_reg): Likewise.
17317 (eliminate_regs_1): Likewise.
17318 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17320 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
17322 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
17323 back to if statements, including unpack.
17325 2017-08-30 Martin Liska <mliska@suse.cz>
17327 PR inline-asm/82001
17328 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
17330 (func_checker::compare_asm_inputs_outputs): ... this function.
17331 (func_checker::compare_gimple_asm): Use the function to compare
17332 also ASM constrains.
17333 * ipa-icf-gimple.h: Rename the function.
17335 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17336 Alan Hayward <alan.hayward@arm.com>
17337 David Sherwood <david.sherwood@arm.com>
17339 * coretypes.h (complex_mode): New type.
17340 * gdbhooks.py (build_pretty_printer): Handle it.
17341 * machmode.h (complex_mode): New class.
17342 (complex_mode::includes_p): New function.
17343 (is_complex_int_mode): Likewise.
17344 (is_complex_float_mode): Likewise.
17345 * genmodes.c (get_mode_class): Handle complex mode classes.
17346 * function.c (expand_function_end): Use is_complex_int_mode.
17348 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17349 Alan Hayward <alan.hayward@arm.com>
17350 David Sherwood <david.sherwood@arm.com>
17352 * coretypes.h (scalar_mode_pod): New typedef.
17353 * gdbhooks.py (build_pretty_printer): Handle it.
17354 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
17355 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
17356 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
17357 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
17358 as_a <scalar_mode>.
17360 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17361 Alan Hayward <alan.hayward@arm.com>
17362 David Sherwood <david.sherwood@arm.com>
17364 * machmode.h (mode_for_vector): Take a scalar_mode instead
17366 * stor-layout.c (mode_for_vector): Likewise.
17367 * explow.c (promote_mode): Use as_a <scalar_mode>.
17368 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
17370 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17371 Alan Hayward <alan.hayward@arm.com>
17372 David Sherwood <david.sherwood@arm.com>
17374 * target.def (preferred_simd_mode): Take a scalar_mode
17375 instead of a machine_mode.
17376 * targhooks.h (default_preferred_simd_mode): Likewise.
17377 * targhooks.c (default_preferred_simd_mode): Likewise.
17378 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
17379 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
17380 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
17381 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
17382 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
17383 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
17384 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
17385 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
17387 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
17388 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
17389 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
17390 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
17391 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
17392 * doc/tm.texi: Regenerate.
17393 * optabs-query.c (can_vec_mask_load_store_p): Return false for
17396 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17397 Alan Hayward <alan.hayward@arm.com>
17398 David Sherwood <david.sherwood@arm.com>
17400 * target.def (scalar_mode_supported_p): Take a scalar_mode
17401 instead of a machine_mode.
17402 * targhooks.h (default_scalar_mode_supported_p): Likewise.
17403 * targhooks.c (default_scalar_mode_supported_p): Likewise.
17404 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
17405 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17406 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
17407 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
17408 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
17409 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
17410 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
17411 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
17412 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
17413 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
17414 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
17415 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
17417 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
17418 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
17419 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17420 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17421 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
17423 * doc/tm.texi: Regenerate.
17425 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17426 Alan Hayward <alan.hayward@arm.com>
17427 David Sherwood <david.sherwood@arm.com>
17429 * coretypes.h (opt_scalar_mode): New typedef.
17430 * gdbhooks.py (build_pretty_printers): Handle it.
17431 * machmode.h (mode_iterator::get_2xwider): Add overload for
17433 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
17435 * expr.c (convert_mode_scalar): Likewise.
17436 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17437 * optabs.c (expand_float): Likewise.
17438 (expand_fix): Likewise.
17439 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17441 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17442 Alan Hayward <alan.hayward@arm.com>
17443 David Sherwood <david.sherwood@arm.com>
17445 * optabs.c (expand_float): Explicitly check for scalars before
17446 using a branching expansion.
17447 (expand_fix): Likewise.
17449 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17450 Alan Hayward <alan.hayward@arm.com>
17451 David Sherwood <david.sherwood@arm.com>
17453 * expr.c (convert_mode): Split scalar handling out into...
17454 (convert_mode_scalar): ...this new function. Treat the modes
17457 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17458 Alan Hayward <alan.hayward@arm.com>
17459 David Sherwood <david.sherwood@arm.com>
17461 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
17463 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
17465 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17466 Alan Hayward <alan.hayward@arm.com>
17467 David Sherwood <david.sherwood@arm.com>
17469 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
17470 rather than a machine_mode.
17471 (fixed_from_string): Likewise.
17472 (fixed_convert): Likewise.
17473 (fixed_convert_from_int): Likewise.
17474 (fixed_convert_from_real): Likewise.
17475 (real_convert_from_fixed): Likewise.
17476 * fixed-value.c (fixed_from_double_int): Likewise.
17477 (fixed_from_string): Likewise.
17478 (fixed_convert): Likewise.
17479 (fixed_convert_from_int): Likewise.
17480 (fixed_convert_from_real): Likewise.
17481 (real_convert_from_fixed): Likewise.
17482 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
17484 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17485 Alan Hayward <alan.hayward@arm.com>
17486 David Sherwood <david.sherwood@arm.com>
17488 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
17489 of separate mode class checks. Do not allow vector modes here.
17490 (immed_wide_int_const): Use as_a <scalar_mode>.
17491 * explow.c (trunc_int_for_mode): Likewise.
17492 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
17493 (wi::shwi): Likewise.
17494 (wi::min_value): Likewise.
17495 (wi::max_value): Likewise.
17496 * dwarf2out.c (loc_descriptor): Likewise.
17497 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
17498 for CONST_WIDE_INT.
17500 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17501 Alan Hayward <alan.hayward@arm.com>
17502 David Sherwood <david.sherwood@arm.com>
17504 * tree.h (SCALAR_TYPE_MODE): New macro.
17505 * expr.c (expand_expr_addr_expr_1): Use it.
17506 (expand_expr_real_2): Likewise.
17507 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
17508 (fold_convert_const_fixed_from_int): Likewise.
17509 (fold_convert_const_fixed_from_real): Likewise.
17510 (native_encode_fixed): Likewise
17511 (native_encode_complex): Likewise
17512 (native_encode_vector): Likewise.
17513 (native_interpret_fixed): Likewise.
17514 (native_interpret_real): Likewise.
17515 (native_interpret_complex): Likewise.
17516 (native_interpret_vector): Likewise.
17517 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
17518 (simd_clone_adjust_argument_types): Likewise.
17519 (simd_clone_init_simd_arrays): Likewise.
17520 (simd_clone_adjust): Likewise.
17521 * stor-layout.c (layout_type): Likewise.
17522 * tree.c (build_minus_one_cst): Likewise.
17523 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17524 * tree-inline.c (estimate_move_cost): Likewise.
17525 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
17526 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
17527 (vectorizable_reduction): Likewise.
17528 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
17529 (vect_recog_mixed_size_cond_pattern): Likewise.
17530 (check_bool_pattern): Likewise.
17531 (adjust_bool_pattern): Likewise.
17532 (search_type_for_mask_1): Likewise.
17533 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
17534 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17535 (vectorizable_load): Likewise.
17536 (vectorizable_store): Likewise.
17537 * ubsan.c (ubsan_encode_value): Likewise.
17538 * varasm.c (output_constant): Likewise.
17540 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17541 Alan Hayward <alan.hayward@arm.com>
17542 David Sherwood <david.sherwood@arm.com>
17544 * coretypes.h (scalar_mode): New class.
17545 * machmode.h (scalar_mode): Likewise.
17546 (scalar_mode::includes_p): New function.
17547 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
17548 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
17549 * genmodes.c (get_mode_class): Handle remaining scalar modes.
17550 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
17551 * expmed.c (store_bit_field_1): Likewise.
17552 (extract_bit_field_1): Likewise.
17553 * expr.c (write_complex_part): Likewise.
17554 (read_complex_part): Likewise.
17555 (emit_move_complex_push): Likewise.
17556 (expand_expr_real_2): Likewise.
17557 * function.c (assign_parm_setup_reg): Likewise.
17558 (assign_parms_unsplit_complex): Likewise.
17559 * optabs.c (expand_binop): Likewise.
17560 * rtlanal.c (subreg_get_info): Likewise.
17561 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17562 * varasm.c (output_constant_pool_2): Likewise.
17564 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17565 Alan Hayward <alan.hayward@arm.com>
17566 David Sherwood <david.sherwood@arm.com>
17568 * expmed.c (extract_high_half): Use scalar_int_mode and remove
17570 (expmed_mult_highpart_optab): Likewise.
17571 (expmed_mult_highpart): 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 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
17578 instead of a machine_mode.
17579 (builtin_memset_read_str): Likewise.
17580 * builtins.c (c_readstr): Likewise.
17581 (builtin_memcpy_read_str): Likewise.
17582 (builtin_strncpy_read_str): Likewise.
17583 (builtin_memset_read_str): Likewise.
17584 (builtin_memset_gen_str): Likewise.
17585 (expand_builtin_signbit): Use scalar_int_mode for local variables.
17586 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
17587 instead of a machine_mode.
17588 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
17590 (make_extraction): Likewise.
17591 (try_widen_shift_mode): Take and return scalar_int_modes instead
17593 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
17594 a scalar_int_mode instead of a machine_mode.
17595 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
17596 (avr_addr_space_pointer_mode): Likewise.
17597 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
17598 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
17599 (msp430_unwind_word_mode): Likewise.
17600 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
17601 (spu_addr_space_pointer_mode): Likewise.
17602 (spu_addr_space_address_mode): Likewise.
17603 (spu_libgcc_cmp_return_mode): Likewise.
17604 (spu_libgcc_shift_count_mode): Likewise.
17605 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
17606 (rl78_addr_space_pointer_mode): Likewise.
17607 (fl78_unwind_word_mode): Likewise.
17608 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
17610 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
17611 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
17612 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
17613 (mips_valid_pointer_mode): Likewise.
17614 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
17615 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
17616 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
17618 (ft32_addr_space_address_mode): Likewise.
17619 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
17620 scalar_int_mode instead of a machine_mode.
17621 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
17623 (m32c_addr_space_address_mode): Likewise.
17624 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
17625 (rs6000_eh_return_filter_mode): Likewise.
17626 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
17627 (rs6000_eh_return_filter_mode): Likewise.
17628 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
17629 (s390_libgcc_shift_count_mode): Likewise.
17630 (s390_unwind_word_mode): Likewise.
17631 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
17633 * target.def (mode_rep_extended): Likewise.
17634 (valid_pointer_mode): Likewise.
17635 (addr_space.valid_pointer_mode): Likewise.
17636 (eh_return_filter_mode): Return a scalar_int_mode rather than
17638 (libgcc_cmp_return_mode): Likewise.
17639 (libgcc_shift_count_mode): Likewise.
17640 (unwind_word_mode): Likewise.
17641 (addr_space.pointer_mode): Likewise.
17642 (addr_space.address_mode): Likewise.
17643 * doc/tm.texi: Regenerate.
17644 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
17646 (do_jump): Use scalar_int_mode for local variables.
17647 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
17648 rather than a machine_mode.
17649 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
17650 (scompare_loc_descriptor_wide): Likewise.
17651 (scompare_loc_descriptor_narrow): Likewise.
17652 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
17654 * except.c (sjlj_emit_dispatch_table): Likewise.
17655 (expand_builtin_eh_copy_values): Likewise.
17656 * explow.c (convert_memory_address_addr_space_1): Likewise.
17657 Take a scalar_int_mode rather than a machine_mode.
17658 (convert_memory_address_addr_space): Take a scalar_int_mode rather
17659 than a machine_mode.
17660 (memory_address_addr_space): Use scalar_int_mode for local variables.
17661 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
17662 rather than a machine_mode.
17663 * expmed.c (mask_rtx): Likewise.
17664 (init_expmed_one_conv): Likewise.
17665 (expand_mult_highpart_adjust): Likewise.
17666 (extract_high_half): Likewise.
17667 (expmed_mult_highpart_optab): Likewise.
17668 (expmed_mult_highpart): Likewise.
17669 (expand_smod_pow2): Likewise.
17670 (expand_sdiv_pow2): Likewise.
17671 (emit_store_flag_int): Likewise.
17672 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
17674 (extract_low_bits): Likewise.
17675 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
17677 * expr.c (pieces_addr::adjust): Likewise.
17678 (can_store_by_pieces): Likewise.
17679 (store_by_pieces): Likewise.
17680 (clear_by_pieces_1): Likewise.
17681 (expand_expr_addr_expr_1): Likewise.
17682 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
17683 (expand_expr_real_1): Likewise.
17684 (try_casesi): Likewise.
17685 * final.c (shorten_branches): Likewise.
17686 * fold-const.c (fold_convert_const_int_from_fixed): Change the
17687 type of "mode" to machine_mode.
17688 * internal-fn.c (expand_arith_overflow_result_store): Take a
17689 scalar_int_mode rather than a machine_mode.
17690 (expand_mul_overflow): Use scalar_int_mode for local variables.
17691 * loop-doloop.c (doloop_modify): Likewise.
17692 (doloop_optimize): Likewise.
17693 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
17694 than a machine_mode.
17695 (expand_doubleword_shift_condmove): Likewise.
17696 (expand_doubleword_shift): Likewise.
17697 (expand_doubleword_clz): Likewise.
17698 (expand_doubleword_popcount): Likewise.
17699 (expand_doubleword_parity): Likewise.
17700 (expand_absneg_bit): Use scalar_int_mode for local variables.
17701 (prepare_float_lib_cmp): Likewise.
17702 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
17703 rather than a machine_mode.
17704 (convert_memory_address_addr_space): Likewise.
17705 (get_mode_bounds): Likewise.
17706 (get_address_mode): Return a scalar_int_mode rather than a
17708 * rtlanal.c (get_address_mode): Likewise.
17709 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
17710 than a machine_mode.
17711 * targhooks.c (default_mode_rep_extended): Likewise.
17712 (default_valid_pointer_mode): Likewise.
17713 (default_addr_space_valid_pointer_mode): Likewise.
17714 (default_eh_return_filter_mode): Return a scalar_int_mode rather
17715 than a machine_mode.
17716 (default_libgcc_cmp_return_mode): Likewise.
17717 (default_libgcc_shift_count_mode): Likewise.
17718 (default_unwind_word_mode): Likewise.
17719 (default_addr_space_pointer_mode): Likewise.
17720 (default_addr_space_address_mode): Likewise.
17721 * targhooks.h (default_eh_return_filter_mode): Likewise.
17722 (default_libgcc_cmp_return_mode): Likewise.
17723 (default_libgcc_shift_count_mode): Likewise.
17724 (default_unwind_word_mode): Likewise.
17725 (default_addr_space_pointer_mode): Likewise.
17726 (default_addr_space_address_mode): Likewise.
17727 (default_mode_rep_extended): Take a scalar_int_mode rather than
17729 (default_valid_pointer_mode): Likewise.
17730 (default_addr_space_valid_pointer_mode): Likewise.
17731 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
17733 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
17734 rather than a machine_mode.
17735 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
17736 for local variables.
17737 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17738 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
17739 than a machine_mode.
17741 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17742 Alan Hayward <alan.hayward@arm.com>
17743 David Sherwood <david.sherwood@arm.com>
17745 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
17746 the mode argument to scalar_int_mode.
17747 (do_jump_by_parts_zero_rtx): Likewise.
17748 (do_jump_by_parts_equality_rtx): Likewise.
17749 (do_jump_by_parts_greater): Take a mode argument.
17750 (do_jump_by_parts_equality): Likewise.
17751 (do_jump_1): Update calls accordingly.
17753 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17754 Alan Hayward <alan.hayward@arm.com>
17755 David Sherwood <david.sherwood@arm.com>
17757 * is-a.h (safe_dyn_cast): New function.
17758 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
17759 (jump_table_for_label): Likewise.
17760 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
17761 instead of an rtx_insn *.
17762 (shorten_branches): Use dyn_cast instead of LABEL_P and
17763 JUMP_TABLE_DATA_P. Use jump_table_for_label and
17764 rtx_jump_table_data::get_data_mode.
17765 (final_scan_insn): Likewise.
17767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17768 Alan Hayward <alan.hayward@arm.com>
17769 David Sherwood <david.sherwood@arm.com>
17771 * combine.c (try_combine): Use is_a <scalar_int_mode> when
17772 trying to combine a full-register integer set with a subreg
17775 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17776 Alan Hayward <alan.hayward@arm.com>
17777 David Sherwood <david.sherwood@arm.com>
17779 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
17780 that is always either address_mode or pointer_mode.
17782 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17783 Alan Hayward <alan.hayward@arm.com>
17784 David Sherwood <david.sherwood@arm.com>
17786 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
17787 when the two are known to be equal.
17789 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17790 Alan Hayward <alan.hayward@arm.com>
17791 David Sherwood <david.sherwood@arm.com>
17793 * simplify-rtx.c (simplify_const_unary_operation): Use
17794 is_a <scalar_int_mode> instead of checking for a nonzero
17795 precision. Forcibly convert op_mode to a scalar_int_mode
17796 in that case. More clearly differentiate the operand and
17797 result modes and use the former when deciding what the value
17798 of a count-bits operation should be. Use is_int_mode instead
17799 of checking for a MODE_INT. Remove redundant check for whether
17800 this mode has a zero precision.
17802 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17803 Alan Hayward <alan.hayward@arm.com>
17804 David Sherwood <david.sherwood@arm.com>
17806 * optabs.c (widen_leading): Change the type of the mode argument
17807 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
17808 (widen_bswap): Likewise.
17809 (expand_parity): Likewise.
17810 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
17811 (expand_ffs): Likewise.
17812 (epand_unop): Check for scalar integer modes before calling the
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 (const_scalar_mask_from_tree): Add a mode argument.
17821 (expand_expr_real_1): Update call accordingly.
17823 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17824 Alan Hayward <alan.hayward@arm.com>
17825 David Sherwood <david.sherwood@arm.com>
17827 * expmed.c (store_bit_field_using_insv): Add op0_mode and
17828 value_mode arguments. Use scalar_int_mode internally.
17829 (store_bit_field_1): Rename the new integer mode from imode
17830 to op0_mode and use it instead of GET_MODE (op0). Update calls
17831 to store_split_bit_field, store_bit_field_using_insv and
17832 store_fixed_bit_field.
17833 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
17834 Use scalar_int_mode internally. Use a bit count rather than a mode
17835 when calculating the largest bit size for get_best_mode.
17836 Update calls to store_split_bit_field and store_fixed_bit_field_1.
17837 (store_fixed_bit_field_1): Add mode and value_mode arguments.
17838 Remove assertion that OP0 has a scalar integer mode.
17839 (store_split_bit_field): Add op0_mode and value_mode arguments.
17840 Update calls to extract_fixed_bit_field.
17841 (extract_bit_field_using_extv): Add an op0_mode argument.
17842 Use scalar_int_mode internally.
17843 (extract_bit_field_1): Rename the new integer mode from imode to
17844 op0_mode and use it instead of GET_MODE (op0). Update calls to
17845 extract_split_bit_field, extract_bit_field_using_extv and
17846 extract_fixed_bit_field.
17847 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
17848 to extract_split_bit_field and extract_fixed_bit_field_1.
17849 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
17850 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
17851 on the target mode.
17852 (extract_split_bit_field): Add an op0_mode argument. Update call
17853 to extract_fixed_bit_field.
17855 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17856 Alan Hayward <alan.hayward@arm.com>
17857 David Sherwood <david.sherwood@arm.com>
17859 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
17860 * explow.c (hard_function_value): Likewise.
17861 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
17862 convert_to_mode call outside the loop.
17863 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
17864 for the mode iterator. Require the mode specified by max_pieces
17866 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
17868 (copy_blkmode_to_reg): Likewise.
17869 (set_storage_via_setmem): Likewise.
17870 * optabs.c (prepare_cmp_insn): Likewise.
17871 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17872 * stor-layout.c (finish_bitfield_representative): Likewise.
17874 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17875 Alan Hayward <alan.hayward@arm.com>
17876 David Sherwood <david.sherwood@arm.com>
17878 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
17879 * expr.c (convert_move): Use them.
17880 (convert_modes): Likewise.
17881 (store_expr_with_bounds): Likewise.
17883 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17884 Alan Hayward <alan.hayward@arm.com>
17885 David Sherwood <david.sherwood@arm.com>
17887 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
17888 parameter for the mode of "x". Remove the "known_x", "known_mode"
17889 and "known_ret" arguments. Change the type of the mode argument
17890 to scalar_int_mode.
17891 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
17892 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
17893 (reg_num_sign_bit_copies_for_combine): Likewise.
17894 * rtlanal.c (nonzero_bits1): Likewise.
17895 (num_sign_bit_copies1): Likewise.
17896 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
17897 (reg_num_sign_bit_copies_general): Likewise.
17898 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
17899 (reg_nonzero_bits_general): Likewise.
17901 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17902 Alan Hayward <alan.hayward@arm.com>
17903 David Sherwood <david.sherwood@arm.com>
17905 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
17906 than in subroutines. Return 1 for non-integer modes.
17907 (cached_num_sign_bit_copies): Change the type of the mode parameter
17908 to scalar_int_mode.
17909 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
17910 classes. Handle CONST_INT_P first and then check whether X also
17911 has a scalar integer mode. Check the same thing for inner registers
17912 of a SUBREG and for values that are being extended or truncated.
17914 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17915 Alan Hayward <alan.hayward@arm.com>
17916 David Sherwood <david.sherwood@arm.com>
17918 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
17919 in subroutines. Return the mode mask for non-integer modes.
17920 (cached_nonzero_bits): Change the type of the mode parameter
17921 to scalar_int_mode.
17922 (nonzero_bits1): Likewise. Remove early exit for other mode
17923 classes. Handle CONST_INT_P first and then check whether X
17924 also has a scalar integer mode.
17926 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17927 Alan Hayward <alan.hayward@arm.com>
17928 David Sherwood <david.sherwood@arm.com>
17930 * expr.c (widest_int_mode_for_size): Make the comment match the code.
17931 Return a scalar_int_mode and assert that the size is greater than
17933 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
17934 (op_by_pieces_d::op_by_pieces_d): Likewise.
17935 (op_by_pieces_d::run): Likewise.
17936 (can_store_by_pieces): Likewise.
17938 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17939 Alan Hayward <alan.hayward@arm.com>
17940 David Sherwood <david.sherwood@arm.com>
17942 * combine.c (extract_left_shift): Add a mode argument and update
17944 (make_compound_operation_int): Change the type of the mode parameter
17945 to scalar_int_mode and update the call to extract_left_shift.
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 * combine.c (simplify_and_const_int): Change the type of the mode
17952 parameter to scalar_int_mode.
17953 (simplify_and_const_int_1): Likewise. Update recursive call.
17955 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17956 Alan Hayward <alan.hayward@arm.com>
17957 David Sherwood <david.sherwood@arm.com>
17959 * combine.c (simplify_compare_const): Check that the mode is a
17960 scalar_int_mode (rather than VOIDmode) before testing its
17962 (simplify_comparison): Move COMPARISON_P handling out of the
17963 loop and restrict the latter part of the loop to scalar_int_modes.
17964 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
17965 and when considering SUBREG_REGs. Use is_int_mode instead of
17966 checking GET_MODE_CLASS against MODE_INT.
17968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17969 Alan Hayward <alan.hayward@arm.com>
17970 David Sherwood <david.sherwood@arm.com>
17972 * combine.c (try_widen_shift_mode): Move check for equal modes to...
17973 (simplify_shift_const_1): ...here. Use scalar_int_mode for
17974 shift_unit_mode and for modes involved in scalar shifts.
17976 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17977 Alan Hayward <alan.hayward@arm.com>
17978 David Sherwood <david.sherwood@arm.com>
17980 * combine.c (force_int_to_mode): New function, split out from...
17981 (force_to_mode): ...here. Keep xmode up-to-date and use it
17982 instead of GET_MODE (x).
17984 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17985 Alan Hayward <alan.hayward@arm.com>
17986 David Sherwood <david.sherwood@arm.com>
17988 * optabs-query.h (extraction_insn::struct_mode): Change type to
17989 opt_scalar_int_mode and update comment.
17990 (extraction_insn::field_mode): Change type to scalar_int_mode.
17991 (extraction_insn::pos_mode): Likewise.
17992 * combine.c (make_extraction): Update accordingly.
17993 * optabs-query.c (get_traditional_extraction_insn): Likewise.
17994 (get_optab_extraction_insn): Likewise.
17995 * recog.c (simplify_while_replacing): Likewise.
17996 * expmed.c (narrow_bit_field_mem): Change the type of the mode
17997 parameter to opt_scalar_int_mode.
17999 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18000 Alan Hayward <alan.hayward@arm.com>
18001 David Sherwood <david.sherwood@arm.com>
18003 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
18004 to a scalar_int_mode instead of a machine_mode.
18005 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
18006 (get_best_mode): Return a boolean and use a pointer argument to store
18007 the selected mode. Replace the limit mode parameter with a bit limit.
18008 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
18009 for the values returned by bit_field_mode_iterator::next_mode.
18010 (store_bit_field): Update call to get_best_mode.
18011 (store_fixed_bit_field): Likewise.
18012 (extract_fixed_bit_field): Likewise.
18013 * expr.c (optimize_bitfield_assignment_op): Likewise.
18014 * fold-const.c (optimize_bit_field_compare): Likewise.
18015 (fold_truth_andor_1): Likewise.
18016 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
18017 Update for new type of m_mode.
18018 (get_best_mode): As above.
18020 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18021 Alan Hayward <alan.hayward@arm.com>
18022 David Sherwood <david.sherwood@arm.com>
18024 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
18025 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
18026 (store_bit_field): Check is_a <scalar_int_mode> before calling
18027 strict_volatile_bitfield_p.
18028 (extract_bit_field): Likewise.
18030 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18031 Alan Hayward <alan.hayward@arm.com>
18032 David Sherwood <david.sherwood@arm.com>
18034 * target.def (cstore_mode): Return a scalar_int_mode.
18035 * doc/tm.texi: Regenerate.
18036 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
18037 * targhooks.h (default_cstore_mode): Likewise.
18038 * targhooks.c (default_cstore_mode): Likewise, using a forced
18040 * expmed.c (emit_cstore): Expect the target of the cstore to be
18043 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18044 Alan Hayward <alan.hayward@arm.com>
18045 David Sherwood <david.sherwood@arm.com>
18047 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
18049 (niter_desc): Likewise mode.
18050 (iv_analyze): Add a mode parameter.
18052 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
18053 and change its type to scalar_int_mode.
18054 * loop-iv.c: Update commentary at head of file.
18055 (iv_constant): Pass the mode paraeter before the rtx it describes
18056 and change its type to scalar_int_mode. Remove VOIDmode handling.
18057 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
18058 (iv_extend): Likewise.
18059 (shorten_into_mode): Likewise.
18060 (iv_add): Use scalar_int_mode.
18061 (iv_mult): Likewise.
18062 (iv_shift): Likewise.
18063 (canonicalize_iv_subregs): Likewise.
18064 (get_biv_step_1): Pass the outer_mode parameter before the rtx
18065 it describes and change its mode to scalar_int_mode. Also change
18066 the type of the returned inner_mode to scalar_int_mode.
18067 (get_biv_step): Likewise, turning outer_mode from a pointer
18068 into a direct parameter. Update call to get_biv_step_1.
18069 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
18070 iv_constant and get_biv_step.
18071 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
18072 and change its type to scalar_int_mode. Don't initialise iv->mode
18073 to VOIDmode and remove later checks for its still being VOIDmode.
18074 Update calls to iv_analyze_op and iv_analyze_expr. Check
18075 is_a <scalar_int_mode> when changing the mode under consideration.
18076 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
18077 Update call to iv_analyze_expr.
18078 (iv_analyze_op): Add a mode parameter. Reject subregs whose
18079 inner register is not also a scalar_int_mode. Update call to
18081 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
18082 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
18083 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
18084 separate mode class checks. Update calls to iv_analyze. Remove
18085 fix-up of VOIDmodes after iv_analyze_biv.
18086 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
18087 don't have a scalar_int_mode. Update call to biv_p.
18089 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18090 Alan Hayward <alan.hayward@arm.com>
18091 David Sherwood <david.sherwood@arm.com>
18093 * cfgexpand.c (convert_debug_memory_address): Use
18094 as_a <scalar_int_mode>.
18095 * combine.c (expand_compound_operation): Likewise.
18096 (make_extraction): Likewise.
18097 (change_zero_ext): Likewise.
18098 (simplify_comparison): Likewise.
18099 * cse.c (cse_insn): Likewise.
18100 * dwarf2out.c (minmax_loc_descriptor): Likewise.
18101 (mem_loc_descriptor): Likewise.
18102 (loc_descriptor): Likewise.
18103 * expmed.c (init_expmed_one_mode): Likewise.
18104 (synth_mult): Likewise.
18105 (emit_store_flag_1): Likewise.
18106 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
18107 of a comparison with size.
18108 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
18109 (reduce_to_bit_field_precision): Likewise.
18110 * function.c (expand_function_end): Likewise.
18111 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
18112 * loop-doloop.c (doloop_modify): Likewise.
18113 * optabs.c (expand_binop): Likewise.
18114 (expand_unop): Likewise.
18115 (expand_copysign_absneg): Likewise.
18116 (prepare_cmp_insn): Likewise.
18117 (maybe_legitimize_operand): Likewise.
18118 * recog.c (const_scalar_int_operand): Likewise.
18119 * rtlanal.c (get_address_mode): Likewise.
18120 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18121 (simplify_cond_clz_ctz): Likewise.
18122 * tree-nested.c (get_nl_goto_field): Likewise.
18123 * tree.c (build_vector_type_for_mode): Likewise.
18124 * var-tracking.c (use_narrower_mode): Likewise.
18126 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18127 Alan Hayward <alan.hayward@arm.com>
18128 David Sherwood <david.sherwood@arm.com>
18130 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
18131 * builtins.c (expand_builtin_signbit): Use it.
18132 * cfgexpand.c (expand_debug_expr): Likewise.
18133 * dojump.c (do_jump): Likewise.
18134 (do_compare_and_jump): Likewise.
18135 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
18136 * expmed.c (make_tree): Likewise.
18137 * expr.c (expand_expr_real_2): Likewise.
18138 (expand_expr_real_1): Likewise.
18139 (try_casesi): Likewise.
18140 * fold-const-call.c (fold_const_call_ss): Likewise.
18141 * fold-const.c (unextend): Likewise.
18142 (extract_muldiv_1): Likewise.
18143 (fold_single_bit_test): Likewise.
18144 (native_encode_int): Likewise.
18145 (native_encode_string): Likewise.
18146 (native_interpret_int): Likewise.
18147 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
18148 * internal-fn.c (expand_addsub_overflow): Likewise.
18149 (expand_neg_overflow): Likewise.
18150 (expand_mul_overflow): Likewise.
18151 (expand_arith_overflow): Likewise.
18152 * match.pd: Likewise.
18153 * stor-layout.c (layout_type): Likewise.
18154 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18155 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
18156 * tree-ssanames.c (get_range_info): Likewise.
18157 * tree-switch-conversion.c (array_value_type) Likewise.
18158 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
18159 (vect_recog_divmod_pattern): Likewise.
18160 (vect_recog_mixed_size_cond_pattern): Likewise.
18161 * tree-vrp.c (extract_range_basic): Likewise.
18162 (simplify_float_conversion_using_ranges): Likewise.
18163 * tree.c (int_fits_type_p): Likewise.
18164 * ubsan.c (instrument_bool_enum_load): Likewise.
18165 * varasm.c (mergeable_string_section): Likewise.
18166 (narrowing_initializer_constant_valid_p): Likewise.
18167 (output_constant): Likewise.
18169 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18170 Alan Hayward <alan.hayward@arm.com>
18171 David Sherwood <david.sherwood@arm.com>
18173 * machmode.h (NARROWEST_INT_MODE): New macro.
18174 * expr.c (alignment_for_piecewise_move): Use it instead of
18175 GET_CLASS_NARROWEST_MODE (MODE_INT).
18176 (push_block): Likewise.
18177 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
18179 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18181 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18182 Alan Hayward <alan.hayward@arm.com>
18183 David Sherwood <david.sherwood@arm.com>
18185 * postreload.c (move2add_valid_value_p): Change the type of the
18186 mode parameter to scalar_int_mode.
18187 (move2add_use_add2_insn): Add a mode parameter and use it instead
18189 (move2add_use_add3_insn): Likewise.
18190 (reload_cse_move2add): Update accordingly.
18192 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18193 Alan Hayward <alan.hayward@arm.com>
18194 David Sherwood <david.sherwood@arm.com>
18196 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
18198 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
18199 * optabs.c (expand_unop): Likewise.
18201 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18202 Alan Hayward <alan.hayward@arm.com>
18203 David Sherwood <david.sherwood@arm.com>
18205 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
18206 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
18207 (popcount_loc_descriptor): Likewise.
18208 (bswap_loc_descriptor): Likewise.
18209 (rotate_loc_descriptor): Likewise.
18210 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
18211 calling the functions above.
18213 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18214 Alan Hayward <alan.hayward@arm.com>
18215 David Sherwood <david.sherwood@arm.com>
18217 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
18219 (try_combine): Likewise.
18220 (simplify_if_then_else): Likewise.
18221 * cse.c (cse_insn): Likewise.
18222 * dwarf2out.c (mem_loc_descriptor): Likewise.
18223 * emit-rtl.c (gen_lowpart_common): Likewise.
18224 * simplify-rtx.c (simplify_truncation): Likewise.
18225 (simplify_binary_operation_1): Likewise.
18226 (simplify_const_relational_operation): Likewise.
18227 (simplify_ternary_operation): Likewise.
18228 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
18230 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18231 Alan Hayward <alan.hayward@arm.com>
18232 David Sherwood <david.sherwood@arm.com>
18234 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
18235 * reload.c (push_reload): Likewise.
18236 (find_reloads): Likewise.
18238 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18239 Alan Hayward <alan.hayward@arm.com>
18240 David Sherwood <david.sherwood@arm.com>
18242 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
18243 (make_compound_operation_int): Likewise.
18244 (change_zero_ext): Likewise.
18245 * expr.c (convert_move): Likewise.
18246 (convert_modes): Likewise.
18247 * fwprop.c (forward_propagate_subreg): Likewise.
18248 * loop-iv.c (get_biv_step_1): Likewise.
18249 * optabs.c (widen_operand): Likewise.
18250 * postreload.c (move2add_valid_value_p): Likewise.
18251 * recog.c (simplify_while_replacing): Likewise.
18252 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18253 (simplify_binary_operation_1): Likewise. Remove redundant
18254 mode equality check.
18256 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18257 Alan Hayward <alan.hayward@arm.com>
18258 David Sherwood <david.sherwood@arm.com>
18260 * combine.c (combine_simplify_rtx): Add checks for
18261 is_a <scalar_int_mode>.
18262 (simplify_if_then_else): Likewise.
18263 (make_field_assignment): Likewise.
18264 (simplify_comparison): Likewise.
18265 * ifcvt.c (noce_try_bitop): Likewise.
18266 * loop-invariant.c (canonicalize_address_mult): Likewise.
18267 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18269 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18270 Alan Hayward <alan.hayward@arm.com>
18271 David Sherwood <david.sherwood@arm.com>
18273 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
18274 is_a <scalar_int_mode> instead of != BLKmode.
18276 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18277 Alan Hayward <alan.hayward@arm.com>
18278 David Sherwood <david.sherwood@arm.com>
18280 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
18281 instead of != VOIDmode.
18282 * combine.c (if_then_else_cond): Likewise.
18283 (change_zero_ext): Likewise.
18284 * dwarf2out.c (mem_loc_descriptor): Likewise.
18285 (loc_descriptor): Likewise.
18286 * rtlanal.c (canonicalize_condition): Likewise.
18287 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
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 * simplify-rtx.c (simplify_binary_operation_1): Use
18294 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
18296 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18297 Alan Hayward <alan.hayward@arm.com>
18298 David Sherwood <david.sherwood@arm.com>
18300 * wide-int.h (int_traits<unsigned char>) New class.
18301 (int_traits<unsigned short>) Likewise.
18302 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
18303 Use GET_MODE_UNIT_PRECISION and remove redundant test for
18305 * combine.c (set_nonzero_bits_and_sign_copies): Use
18306 is_a <scalar_int_mode>.
18307 (find_split_point): Likewise.
18308 (combine_simplify_rtx): Likewise.
18309 (simplify_logical): Likewise.
18310 (expand_compound_operation): Likewise.
18311 (expand_field_assignment): Likewise.
18312 (make_compound_operation): Likewise.
18313 (extended_count): Likewise.
18314 (change_zero_ext): Likewise.
18315 (simplify_comparison): Likewise.
18316 * dwarf2out.c (scompare_loc_descriptor): Likewise.
18317 (ucompare_loc_descriptor): Likewise.
18318 (minmax_loc_descriptor): Likewise.
18319 (mem_loc_descriptor): Likewise.
18320 (loc_descriptor): Likewise.
18321 * expmed.c (init_expmed_one_mode): Likewise.
18322 * lra-constraints.c (lra_constraint_offset): Likewise.
18323 * optabs.c (prepare_libcall_arg): Likewise.
18324 * postreload.c (move2add_note_store): Likewise.
18325 * reload.c (operands_match_p): Likewise.
18326 * rtl.h (load_extend_op): Likewise.
18327 * rtlhooks.c (gen_lowpart_general): Likewise.
18328 * simplify-rtx.c (simplify_truncation): Likewise.
18329 (simplify_unary_operation_1): Likewise.
18330 (simplify_binary_operation_1): Likewise.
18331 (simplify_const_binary_operation): Likewise.
18332 (simplify_const_relational_operation): Likewise.
18333 (simplify_subreg): Likewise.
18334 * stor-layout.c (bitwise_mode_for_mode): Likewise.
18335 * var-tracking.c (adjust_mems): Likewise.
18336 (prepare_call_arguments): Likewise.
18338 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18339 Alan Hayward <alan.hayward@arm.com>
18340 David Sherwood <david.sherwood@arm.com>
18342 * machmode.h (is_int_mode): New fuction.
18343 * combine.c (find_split_point): Use it.
18344 (combine_simplify_rtx): Likewise.
18345 (simplify_if_then_else): Likewise.
18346 (simplify_set): Likewise.
18347 (simplify_shift_const_1): Likewise.
18348 (simplify_comparison): Likewise.
18349 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
18350 * cse.c (notreg_cost): Likewise.
18351 (cse_insn): Likewise.
18352 * cselib.c (cselib_lookup_1): Likewise.
18353 * dojump.c (do_jump_1): Likewise.
18354 (do_compare_rtx_and_jump): Likewise.
18355 * dse.c (get_call_args): Likewise.
18356 * dwarf2out.c (rtl_for_decl_init): Likewise.
18357 (native_encode_initializer): Likewise.
18358 * expmed.c (emit_store_flag_1): Likewise.
18359 (emit_store_flag): Likewise.
18360 * expr.c (convert_modes): Likewise.
18361 (store_field): Likewise.
18362 (expand_expr_real_1): Likewise.
18363 * fold-const.c (fold_read_from_constant_string): Likewise.
18364 * gimple-ssa-sprintf.c (get_format_string): Likewise.
18365 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
18366 * optabs.c (expand_binop): Likewise.
18367 (expand_unop): Likewise.
18368 (expand_abs_nojump): Likewise.
18369 (expand_one_cmpl_abs_nojump): Likewise.
18370 * simplify-rtx.c (mode_signbit_p): Likewise.
18371 (val_signbit_p): Likewise.
18372 (val_signbit_known_set_p): Likewise.
18373 (val_signbit_known_clear_p): Likewise.
18374 (simplify_relational_operation_1): Likewise.
18375 * tree.c (vector_type_mode): Likewise.
18377 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18378 Alan Hayward <alan.hayward@arm.com>
18379 David Sherwood <david.sherwood@arm.com>
18381 * machmode.h (smallest_mode_for_size): Fix formatting.
18382 (smallest_int_mode_for_size): New function.
18383 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
18384 instead of smallest_mode_for_size.
18385 * combine.c (make_extraction): Likewise.
18386 * config/arc/arc.c (arc_expand_movmem): Likewise.
18387 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
18388 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
18389 * config/s390/s390.c (s390_expand_insv): Likewise.
18390 * config/sparc/sparc.c (assign_int_registers): Likewise.
18391 * config/spu/spu.c (spu_function_value): Likewise.
18392 (spu_function_arg): Likewise.
18393 * coverage.c (get_gcov_type): Likewise.
18394 (get_gcov_unsigned_t): Likewise.
18395 * dse.c (find_shift_sequence): Likewise.
18396 * expmed.c (store_bit_field_1): Likewise.
18397 * expr.c (convert_move): Likewise.
18398 (store_field): Likewise.
18399 * internal-fn.c (expand_arith_overflow): Likewise.
18400 * optabs-query.c (get_best_extraction_insn): Likewise.
18401 * optabs.c (expand_twoval_binop_libfunc): Likewise.
18402 * stor-layout.c (layout_type): Likewise.
18403 (initialize_sizetypes): Likewise.
18404 * targhooks.c (default_get_mask_mode): Likewise.
18405 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
18407 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18408 Alan Hayward <alan.hayward@arm.com>
18409 David Sherwood <david.sherwood@arm.com>
18411 * machmode.h (opt_mode::else_blk): New function.
18412 (int_mode_for_mode): Declare.
18413 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
18414 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
18416 * cfgexpand.c (expand_debug_expr): Likewise.
18417 * combine.c (gen_lowpart_or_truncate): Likewise.
18418 (gen_lowpart_for_combine): Likewise.
18419 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
18420 * config/avr/avr.c (avr_to_int_mode): Likewise.
18421 (avr_out_plus_1): Likewise.
18422 (avr_out_plus): Likewise.
18423 (avr_out_round): Likewise.
18424 * config/i386/i386.c (ix86_split_to_parts): Likewise.
18425 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
18426 (s390_expand_vcond): Likewise.
18427 * config/spu/spu.c (spu_split_immediate): Likewise.
18428 (spu_expand_mov): Likewise.
18429 * dse.c (get_stored_val): Likewise.
18430 * expmed.c (store_bit_field_1): Likewise.
18431 (convert_extracted_bit_field): Use int_mode_for_mode instead of
18433 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
18434 (extract_low_bits): Likewise.
18435 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
18436 handling rather than repeating the check.
18437 (emit_group_store): Likewise.
18438 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
18439 * optabs.c (expand_absneg_bit): Likewise.
18440 (expand_copysign_absneg): Likewise.
18441 (expand_copysign_bit): Likewise.
18442 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
18443 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18444 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18445 * var-tracking.c (prepare_call_arguments): Likewise.
18446 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
18447 int_mode_for_mode instead of mode_for_size.
18448 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
18450 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18451 Alan Hayward <alan.hayward@arm.com>
18452 David Sherwood <david.sherwood@arm.com>
18454 * machmode.h (int_mode_for_size): New function.
18455 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
18456 instead of mode_for_size.
18457 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
18459 * combine.c (expand_field_assignment): Use int_mode_for_size
18460 instead of mode_for_size.
18461 (make_extraction): Likewise.
18462 (simplify_shift_const_1): Likewise.
18463 (simplify_comparison): Likewise.
18464 * dojump.c (do_jump): Likewise.
18465 * dwarf2out.c (mem_loc_descriptor): Likewise.
18466 * emit-rtl.c (init_derived_machine_modes): Likewise.
18467 * expmed.c (flip_storage_order): Likewise.
18468 (convert_extracted_bit_field): Likewise.
18469 * expr.c (copy_blkmode_from_reg): Likewise.
18470 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
18471 * internal-fn.c (expand_mul_overflow): Likewise.
18472 * lower-subreg.c (simple_move): Likewise.
18473 * optabs-libfuncs.c (init_optabs): Likewise.
18474 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18475 * tree.c (vector_type_mode): Likewise.
18476 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18477 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
18478 * tree-vect-generic.c (expand_vector_parallel): Likewise.
18479 * tree-vect-stmts.c (vectorizable_load): Likewise.
18480 (vectorizable_store): Likewise.
18482 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18483 Alan Hayward <alan.hayward@arm.com>
18484 David Sherwood <david.sherwood@arm.com>
18486 * coretypes.h (pod_mode): New type.
18487 (scalar_int_mode_pod): New typedef.
18488 * machmode.h (pod_mode): New class.
18489 (int_n_data_t::m): Change type to scalar_int_mode_pod.
18490 * genmodes.c (emit_mode_int_n): Update accordingly.
18491 * lower-subreg.h (target_lower_subreg): Change type to
18492 scalar_int_mode_pod.
18493 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
18494 scalar_int_mode_pod.
18496 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18497 Alan Hayward <alan.hayward@arm.com>
18498 David Sherwood <david.sherwood@arm.com>
18500 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
18501 machine_mode to scalar_int_mode.
18502 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
18503 (rs6000_option_override_internal): Remove cast to int.
18504 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
18505 machine_mode to scalar_int_mode.
18506 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
18507 (rs6000_option_override_internal): Remove cast to int.
18508 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
18509 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
18511 * config/s390/s390.c (s390_expand_builtin): Likewise.
18512 * coretypes.h (scalar_int_mode): New type.
18513 (opt_scalar_int_mode): New typedef.
18514 * machmode.h (scalar_int_mode): New class.
18515 (scalar_int_mode::includes_p): New function.
18516 (byte_mode): Change type to scalar_int_mode.
18517 (word_mode): Likewise.
18518 (ptr_mode): Likewise.
18519 * emit-rtl.c (byte_mode): Likewise.
18520 (word_mode): Likewise.
18521 (ptr_mode): Likewise.
18522 (init_derived_machine_modes): Update accordingly.
18523 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
18524 and MODE_PARTIAL_INT.
18525 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
18526 opt_scalar_int_mode.
18528 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18529 Alan Hayward <alan.hayward@arm.com>
18530 David Sherwood <david.sherwood@arm.com>
18532 * target.def (libgcc_floating_mode_supported_p): Take a
18534 * doc/tm.texi: Regenerate.
18535 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
18537 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18538 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
18541 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18542 Alan Hayward <alan.hayward@arm.com>
18543 David Sherwood <david.sherwood@arm.com>
18545 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
18546 * doc/tm.texi: Regenerate.
18547 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
18548 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
18549 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
18550 * targhooks.h (default_floatn_mode): Likewise.
18551 * targhooks.c (default_floatn_mode): Likewise.
18552 * tree.c (build_common_tree_nodes): Update accordingly.
18554 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18555 Alan Hayward <alan.hayward@arm.com>
18556 David Sherwood <david.sherwood@arm.com>
18558 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
18559 (mode_iterator::iterate_p): Likewise.
18560 (mode_iterator::get_wider): Likewise.
18561 * expr.c (init_expr_target): Use opt_scalar_float_mode.
18563 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18564 Alan Hayward <alan.hayward@arm.com>
18565 David Sherwood <david.sherwood@arm.com>
18567 * coretypes.h (opt_scalar_float_mode): New typedef.
18568 * machmode.h (float_mode_for_size): New function.
18569 * emit-rtl.c (double_mode): Delete.
18570 (init_emit_once): Use float_mode_for_size.
18571 * stor-layout.c (layout_type): Likewise.
18572 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
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 * output.h (assemble_real): Take a scalar_float_mode.
18579 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
18580 * config/arm/arm.md (consttable_4): Likewise.
18581 (consttable_8): Likewise.
18582 (consttable_16): Likewise.
18583 * config/mips/mips.md (consttable_float): Likewise.
18584 * config/s390/s390.c (s390_output_pool_entry): Likewise.
18585 * varasm.c (assemble_real): Take a scalar_float_mode.
18586 (output_constant_pool_2): Update accordingly.
18587 (output_constant): Likewise.
18589 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18590 Alan Hayward <alan.hayward@arm.com>
18591 David Sherwood <david.sherwood@arm.com>
18593 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
18594 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
18595 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
18596 (native_encode_real): Likewise.
18597 (native_interpret_real): Likewise.
18598 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
18599 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18601 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18602 Alan Hayward <alan.hayward@arm.com>
18603 David Sherwood <david.sherwood@arm.com>
18605 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
18606 <scalar_float_mode>. Simplify.
18607 (gen_extend_conv_libfunc): Likewise.
18609 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18610 Alan Hayward <alan.hayward@arm.com>
18611 David Sherwood <david.sherwood@arm.com>
18613 * coretypes.h (scalar_float_mode): New type.
18614 * machmode.h (mode_traits::from_int): Use machine_mode if
18615 USE_ENUM_MODES is defined.
18616 (is_a): New function.
18618 (dyn_cast): Likewise.
18619 (scalar_float_mode): New class.
18620 (scalar_float_mode::includes_p): New function.
18621 (is_float_mode): Likewise.
18622 * gdbhooks.py (MachineModePrinter): New class.
18623 (build_pretty_printer): Use it for scalar_float_mode.
18624 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
18625 (format_helper::format_helper): Turn into a template.
18626 * genmodes.c (get_mode_class): New function.
18627 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
18628 or machine_mode if none.
18629 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
18630 as_a <scalar_float_mode>.
18631 * dwarf2out.c (mem_loc_descriptor): Likewise.
18632 (insert_float): Likewise.
18633 (add_const_value_attribute): Likewise.
18634 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18635 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
18636 (expand_unop): Update accordingly.
18637 (expand_abs_nojump): Likewise.
18638 (expand_copysign_absneg): Take a scalar_float_mode.
18639 (expand_copysign_bit): Likewise.
18640 (expand_copysign): Update accordingly.
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 (opt_mode): New class.
18647 * machmode.h (opt_mode): Likewise.
18648 (opt_mode::else_void): New function.
18649 (opt_mode::require): Likewise.
18650 (opt_mode::exists): Likewise.
18651 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
18652 (GET_MODE_2XWIDER_MODE): Likewise.
18653 (mode_iterator::get_wider): Update accordingly.
18654 (mode_iterator::get_2xwider): Likewise.
18655 (mode_iterator::get_known_wider): Likewise, turning into a template.
18656 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
18657 forcing a wider mode to exist.
18658 * config/cr16/cr16.h (LONG_REG_P): Likewise.
18659 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18660 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
18661 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
18662 * lower-subreg.c (init_lower_subreg): Likewise.
18663 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
18664 on the final iteration.
18665 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
18666 a wider mode exists before asking for a move pattern.
18667 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
18668 forcing a wider mode to exist.
18669 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
18670 returning false if no such mode exists.
18671 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
18672 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
18673 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
18674 Avoid checking for a MODE_INT if we already know the mode is not a
18676 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
18677 forcing a wider mode to exist.
18678 (expmed_mult_highpart_optab): Likewise.
18679 (expmed_mult_highpart): Likewise.
18680 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
18682 * lto-streamer-in.c (lto_input_mode_table): Likewise.
18683 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
18684 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
18685 * internal-fn.c (expand_mul_overflow): Update use of
18686 GET_MODE_2XWIDER_MODE.
18687 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18688 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
18689 GET_MODE_WIDER_MODE.
18690 (convert_plusminus_to_widen): Likewise.
18691 * tree-switch-conversion.c (array_value_type): Likewise.
18692 * var-tracking.c (emit_note_insn_var_location): Likewise.
18693 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18694 Return false inside rather than outside the loop if no wider mode
18696 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
18697 and GET_MODE_2XWIDER_MODE
18698 (can_compare_p): Use else_void.
18699 * gdbhooks.py (OptMachineModePrinter): New class.
18700 (build_pretty_printer): Use it for opt_mode.
18702 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18703 Alan Hayward <alan.hayward@arm.com>
18704 David Sherwood <david.sherwood@arm.com>
18706 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
18707 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
18709 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18710 Alan Hayward <alan.hayward@arm.com>
18711 David Sherwood <david.sherwood@arm.com>
18713 * machmode.h (mode_traits): New structure.
18714 (get_narrowest_mode): New function.
18715 (mode_iterator::start): Likewise.
18716 (mode_iterator::iterate_p): Likewise.
18717 (mode_iterator::get_wider): Likewise.
18718 (mode_iterator::get_known_wider): Likewise.
18719 (mode_iterator::get_2xwider): Likewise.
18720 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
18721 (FOR_EACH_MODE): Likewise.
18722 (FOR_EACH_MODE_FROM): Likewise.
18723 (FOR_EACH_MODE_UNTIL): Likewise.
18724 (FOR_EACH_WIDER_MODE): Likewise.
18725 (FOR_EACH_2XWIDER_MODE): Likewise.
18726 * builtins.c (expand_builtin_strlen): Use new mode iterators.
18727 * combine.c (simplify_comparison): Likewise
18728 * config/i386/i386.c (type_natural_mode): Likewise.
18729 * cse.c (cse_insn): Likewise.
18730 * dse.c (find_shift_sequence): Likewise.
18731 * emit-rtl.c (init_derived_machine_modes): Likewise.
18732 (init_emit_once): Likewise.
18733 * explow.c (hard_function_value): Likewise.
18734 * expmed.c (extract_fixed_bit_field_1): Likewise.
18735 (extract_bit_field_1): Likewise.
18736 (expand_divmod): Likewise.
18737 (emit_store_flag_1): Likewise.
18738 * expr.c (init_expr_target): Likewise.
18739 (convert_move): Likewise.
18740 (alignment_for_piecewise_move): Likewise.
18741 (widest_int_mode_for_size): Likewise.
18742 (emit_block_move_via_movmem): Likewise.
18743 (copy_blkmode_to_reg): Likewise.
18744 (set_storage_via_setmem): Likewise.
18745 (compress_float_constant): Likewise.
18746 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18747 * optabs-query.c (get_best_extraction_insn): Likewise.
18748 * optabs.c (expand_binop): Likewise.
18749 (expand_twoval_unop): Likewise.
18750 (expand_twoval_binop): Likewise.
18751 (widen_leading): Likewise.
18752 (widen_bswap): Likewise.
18753 (expand_parity): Likewise.
18754 (expand_unop): Likewise.
18755 (prepare_cmp_insn): Likewise.
18756 (prepare_float_lib_cmp): Likewise.
18757 (expand_float): Likewise.
18758 (expand_fix): Likewise.
18759 (expand_sfix_optab): Likewise.
18760 * postreload.c (move2add_use_add2_insn): Likewise.
18761 * reg-stack.c (reg_to_stack): Likewise.
18762 * reginfo.c (choose_hard_reg_mode): Likewise.
18763 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18764 * stor-layout.c (mode_for_size): Likewise.
18765 (smallest_mode_for_size): Likewise.
18766 (mode_for_vector): Likewise.
18767 (finish_bitfield_representative): Likewise.
18768 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
18769 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
18770 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18771 * var-tracking.c (prepare_call_arguments): Likewise.
18773 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18774 Alan Hayward <alan.hayward@arm.com>
18775 David Sherwood <david.sherwood@arm.com>
18777 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
18778 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
18779 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
18780 * machmode.h (mode_size): Move earlier in file.
18781 (mode_precision): Likewise.
18782 (mode_inner): Likewise.
18783 (mode_nunits): Likewise.
18784 (mode_unit_size): Likewise.
18785 (unit_unit_precision): Likewise.
18786 (mode_wider): Likewise.
18787 (mode_2xwider): Likewise.
18788 (machine_mode): New class.
18789 (mode_to_bytes): New function.
18790 (mode_to_bits): Likewise.
18791 (mode_to_precision): Likewise.
18792 (mode_to_inner): Likewise.
18793 (mode_to_unit_size): Likewise.
18794 (mode_to_unit_precision): Likewise.
18795 (mode_to_nunits): Likewise.
18796 (GET_MODE_SIZE): Use mode_to_bytes.
18797 (GET_MODE_BITSIZE): Use mode_to_bits.
18798 (GET_MODE_PRECISION): Use mode_to_precision.
18799 (GET_MODE_INNER): Use mode_to_inner.
18800 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
18801 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
18802 (GET_MODE_NUNITS): Use mode_to_nunits.
18803 * system.h (ALWAYS_INLINE): New macro.
18804 * config/powerpcspe/powerpcspe-c.c
18805 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
18806 int for arg1_mode and arg2_mode.
18808 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18809 Alan Hayward <alan.hayward@arm.com>
18810 David Sherwood <david.sherwood@arm.com>
18812 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18813 Prefix mode names with E_ in case statements.
18814 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18815 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
18816 (aarch64_split_simd_move): Likewise.
18817 (aarch64_gen_storewb_pair): Likewise.
18818 (aarch64_gen_loadwb_pair): Likewise.
18819 (aarch64_gen_store_pair): Likewise.
18820 (aarch64_gen_load_pair): Likewise.
18821 (aarch64_get_condition_code_1): Likewise.
18822 (aarch64_constant_pool_reload_icode): Likewise.
18823 (get_rsqrte_type): Likewise.
18824 (get_rsqrts_type): Likewise.
18825 (get_recpe_type): Likewise.
18826 (get_recps_type): Likewise.
18827 (aarch64_gimplify_va_arg_expr): Likewise.
18828 (aarch64_simd_container_mode): Likewise.
18829 (aarch64_emit_load_exclusive): Likewise.
18830 (aarch64_emit_store_exclusive): Likewise.
18831 (aarch64_expand_compare_and_swap): Likewise.
18832 (aarch64_gen_atomic_cas): Likewise.
18833 (aarch64_emit_bic): Likewise.
18834 (aarch64_emit_atomic_swap): Likewise.
18835 (aarch64_emit_atomic_load_op): Likewise.
18836 (aarch64_evpc_trn): Likewise.
18837 (aarch64_evpc_uzp): Likewise.
18838 (aarch64_evpc_zip): Likewise.
18839 (aarch64_evpc_ext): Likewise.
18840 (aarch64_evpc_rev): Likewise.
18841 (aarch64_evpc_dup): Likewise.
18842 (aarch64_gen_ccmp_first): Likewise.
18843 (aarch64_gen_ccmp_next): Likewise.
18844 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18845 (alpha_emit_xfloating_libcall): Likewise.
18846 (emit_insxl): Likewise.
18847 (alpha_arg_type): Likewise.
18848 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
18849 (arc_preferred_simd_mode): Likewise.
18850 (arc_secondary_reload): Likewise.
18851 (get_arc_condition_code): Likewise.
18852 (arc_print_operand): Likewise.
18853 (arc_legitimate_constant_p): Likewise.
18854 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18855 * config/arc/arc.md (casesi_load): Likewise.
18856 (casesi_compact_jump): Likewise.
18857 * config/arc/predicates.md (proper_comparison_operator): Likewise.
18858 (cc_use_register): Likewise.
18859 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18860 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18861 (arm_init_iwmmxt_builtins): Likewise.
18862 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
18863 (neon_expand_vector_init): Likewise.
18864 (arm_attr_length_move_neon): Likewise.
18865 (maybe_get_arm_condition_code): Likewise.
18866 (arm_emit_vector_const): Likewise.
18867 (arm_preferred_simd_mode): Likewise.
18868 (arm_output_iwmmxt_tinsr): Likewise.
18869 (thumb1_output_casesi): Likewise.
18870 (thumb2_output_casesi): Likewise.
18871 (arm_emit_load_exclusive): Likewise.
18872 (arm_emit_store_exclusive): Likewise.
18873 (arm_expand_compare_and_swap): Likewise.
18874 (arm_evpc_neon_vuzp): Likewise.
18875 (arm_evpc_neon_vzip): Likewise.
18876 (arm_evpc_neon_vrev): Likewise.
18877 (arm_evpc_neon_vtrn): Likewise.
18878 (arm_evpc_neon_vext): Likewise.
18879 (arm_validize_comparison): Likewise.
18880 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
18881 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
18882 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
18883 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
18884 (c6x_preferred_simd_mode): Likewise.
18885 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
18886 (epiphany_rtx_costs): Likewise.
18887 * config/epiphany/predicates.md (proper_comparison_operator):
18889 * config/frv/frv.c (condexec_memory_operand): Likewise.
18890 (frv_emit_move): Likewise.
18891 (output_move_single): Likewise.
18892 (output_condmove_single): Likewise.
18893 (frv_hard_regno_mode_ok): Likewise.
18894 (frv_matching_accg_mode): Likewise.
18895 * config/h8300/h8300.c (split_adds_subs): Likewise.
18896 (h8300_rtx_costs): Likewise.
18897 (h8300_print_operand): Likewise.
18898 (compute_mov_length): Likewise.
18899 (output_logical_op): Likewise.
18900 (compute_logical_op_length): Likewise.
18901 (compute_logical_op_cc): Likewise.
18902 (h8300_shift_needs_scratch_p): Likewise.
18903 (output_a_shift): Likewise.
18904 (compute_a_shift_length): Likewise.
18905 (compute_a_shift_cc): Likewise.
18906 (expand_a_rotate): Likewise.
18907 (output_a_rotate): Likewise.
18908 * config/i386/i386.c (classify_argument): Likewise.
18909 (function_arg_advance_32): Likewise.
18910 (function_arg_32): Likewise.
18911 (function_arg_64): Likewise.
18912 (function_value_64): Likewise.
18913 (ix86_gimplify_va_arg): Likewise.
18914 (ix86_legitimate_constant_p): Likewise.
18915 (put_condition_code): Likewise.
18916 (split_double_mode): Likewise.
18917 (ix86_avx256_split_vector_move_misalign): Likewise.
18918 (ix86_expand_vector_logical_operator): Likewise.
18919 (ix86_split_idivmod): Likewise.
18920 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
18921 (ix86_build_const_vector): Likewise.
18922 (ix86_build_signbit_mask): Likewise.
18923 (ix86_match_ccmode): Likewise.
18924 (ix86_cc_modes_compatible): Likewise.
18925 (ix86_expand_branch): Likewise.
18926 (ix86_expand_sse_cmp): Likewise.
18927 (ix86_expand_sse_movcc): Likewise.
18928 (ix86_expand_int_sse_cmp): Likewise.
18929 (ix86_expand_vec_perm_vpermi2): Likewise.
18930 (ix86_expand_vec_perm): Likewise.
18931 (ix86_expand_sse_unpack): Likewise.
18932 (ix86_expand_int_addcc): Likewise.
18933 (ix86_split_to_parts): Likewise.
18934 (ix86_vectorize_builtin_gather): Likewise.
18935 (ix86_vectorize_builtin_scatter): Likewise.
18936 (avx_vpermilp_parallel): Likewise.
18937 (inline_memory_move_cost): Likewise.
18938 (ix86_tieable_integer_mode_p): Likewise.
18939 (x86_maybe_negate_const_int): Likewise.
18940 (ix86_expand_vector_init_duplicate): Likewise.
18941 (ix86_expand_vector_init_one_nonzero): Likewise.
18942 (ix86_expand_vector_init_one_var): Likewise.
18943 (ix86_expand_vector_init_concat): Likewise.
18944 (ix86_expand_vector_init_interleave): Likewise.
18945 (ix86_expand_vector_init_general): Likewise.
18946 (ix86_expand_vector_set): Likewise.
18947 (ix86_expand_vector_extract): Likewise.
18948 (emit_reduc_half): Likewise.
18949 (ix86_emit_i387_round): Likewise.
18950 (ix86_mangle_type): Likewise.
18951 (ix86_expand_round_sse4): Likewise.
18952 (expand_vec_perm_blend): Likewise.
18953 (canonicalize_vector_int_perm): Likewise.
18954 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
18955 (expand_vec_perm_1): Likewise.
18956 (expand_vec_perm_interleave3): Likewise.
18957 (expand_vec_perm_even_odd_pack): Likewise.
18958 (expand_vec_perm_even_odd_1): Likewise.
18959 (expand_vec_perm_broadcast_1): Likewise.
18960 (ix86_vectorize_vec_perm_const_ok): Likewise.
18961 (ix86_expand_vecop_qihi): Likewise.
18962 (ix86_expand_mul_widen_hilo): Likewise.
18963 (ix86_expand_sse2_abs): Likewise.
18964 (ix86_expand_pextr): Likewise.
18965 (ix86_expand_pinsr): Likewise.
18966 (ix86_preferred_simd_mode): Likewise.
18967 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
18968 * config/i386/sse.md (*andnot<mode>3): Likewise.
18969 (<mask_codefor><code><mode>3<mask_name>): Likewise.
18970 (*<code><mode>3): Likewise.
18971 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
18972 (ia64_expand_atomic_op): Likewise.
18973 (ia64_arg_type): Likewise.
18974 (ia64_mode_to_int): Likewise.
18975 (ia64_scalar_mode_supported_p): Likewise.
18976 (ia64_vector_mode_supported_p): Likewise.
18977 (expand_vec_perm_broadcast): Likewise.
18978 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
18979 (iq2000_function_arg_advance): Likewise.
18980 (iq2000_function_arg): Likewise.
18981 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
18982 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
18983 (m68k_libcall_value): Likewise.
18984 (m68k_function_value): Likewise.
18985 (sched_attr_op_type): Likewise.
18986 * config/mcore/mcore.c (mcore_output_move): Likewise.
18987 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
18989 (microblaze_function_arg): Likewise.
18990 * config/mips/mips.c (mips16_build_call_stub): Likewise.
18991 (mips_print_operand): Likewise.
18992 (mips_mode_ok_for_mov_fmt_p): Likewise.
18993 (mips_vector_mode_supported_p): Likewise.
18994 (mips_preferred_simd_mode): Likewise.
18995 (mips_expand_vpc_loongson_even_odd): Likewise.
18996 (mips_expand_vec_unpack): Likewise.
18997 (mips_expand_vi_broadcast): Likewise.
18998 (mips_expand_vector_init): Likewise.
18999 (mips_expand_vec_reduc): Likewise.
19000 (mips_expand_msa_cmp): Likewise.
19001 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
19002 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
19003 (cc_flags_for_mode): Likewise.
19004 * config/msp430/msp430.c (msp430_print_operand): Likewise.
19005 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
19006 (nds32_output_casesi_pc_relative): Likewise.
19007 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19008 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
19009 (nvptx_gen_unpack): Likewise.
19010 (nvptx_gen_pack): Likewise.
19011 (nvptx_gen_shuffle): Likewise.
19012 (nvptx_gen_wcast): Likewise.
19013 (nvptx_preferred_simd_mode): Likewise.
19014 * config/pa/pa.c (pa_secondary_reload): Likewise.
19015 * config/pa/predicates.md (base14_operand): Likewise.
19016 * config/powerpcspe/powerpcspe-c.c
19017 (altivec_resolve_overloaded_builtin): Likewise.
19018 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
19020 (rs6000_preferred_simd_mode): Likewise.
19021 (output_vec_const_move): Likewise.
19022 (rs6000_expand_vector_extract): Likewise.
19023 (rs6000_split_vec_extract_var): Likewise.
19024 (reg_offset_addressing_ok_p): Likewise.
19025 (rs6000_legitimate_offset_address_p): Likewise.
19026 (rs6000_legitimize_address): Likewise.
19027 (rs6000_emit_set_const): Likewise.
19028 (rs6000_const_vec): Likewise.
19029 (rs6000_emit_move): Likewise.
19030 (spe_build_register_parallel): Likewise.
19031 (rs6000_darwin64_record_arg_recurse): Likewise.
19032 (swap_selector_for_mode): Likewise.
19033 (spe_init_builtins): Likewise.
19034 (paired_init_builtins): Likewise.
19035 (altivec_init_builtins): Likewise.
19036 (do_load_for_compare): Likewise.
19037 (rs6000_generate_compare): Likewise.
19038 (rs6000_expand_float128_convert): Likewise.
19039 (emit_load_locked): Likewise.
19040 (emit_store_conditional): Likewise.
19041 (rs6000_output_function_epilogue): Likewise.
19042 (rs6000_handle_altivec_attribute): Likewise.
19043 (rs6000_function_value): Likewise.
19044 (emit_fusion_gpr_load): Likewise.
19045 (emit_fusion_p9_load): Likewise.
19046 (emit_fusion_p9_store): Likewise.
19047 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
19048 (fusion_gpr_mem_load): Likewise.
19049 (fusion_addis_mem_combo_load): Likewise.
19050 (fusion_addis_mem_combo_store): Likewise.
19051 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
19052 (fusion_gpr_mem_load): Likewise.
19053 (fusion_addis_mem_combo_load): Likewise.
19054 (fusion_addis_mem_combo_store): Likewise.
19055 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19057 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
19058 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19059 (rs6000_preferred_simd_mode): Likewise.
19060 (output_vec_const_move): Likewise.
19061 (rs6000_expand_vector_extract): Likewise.
19062 (rs6000_split_vec_extract_var): Likewise.
19063 (reg_offset_addressing_ok_p): Likewise.
19064 (rs6000_legitimate_offset_address_p): Likewise.
19065 (rs6000_legitimize_address): Likewise.
19066 (rs6000_emit_set_const): Likewise.
19067 (rs6000_const_vec): Likewise.
19068 (rs6000_emit_move): Likewise.
19069 (rs6000_darwin64_record_arg_recurse): Likewise.
19070 (swap_selector_for_mode): Likewise.
19071 (paired_init_builtins): Likewise.
19072 (altivec_init_builtins): Likewise.
19073 (rs6000_expand_float128_convert): Likewise.
19074 (emit_load_locked): Likewise.
19075 (emit_store_conditional): Likewise.
19076 (rs6000_output_function_epilogue): Likewise.
19077 (rs6000_handle_altivec_attribute): Likewise.
19078 (rs6000_function_value): Likewise.
19079 (emit_fusion_gpr_load): Likewise.
19080 (emit_fusion_p9_load): Likewise.
19081 (emit_fusion_p9_store): Likewise.
19082 * config/rx/rx.c (rx_gen_move_template): Likewise.
19083 (flags_from_mode): Likewise.
19084 * config/s390/predicates.md (s390_alc_comparison): Likewise.
19085 (s390_slb_comparison): Likewise.
19086 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
19087 (s390_vector_mode_supported_p): Likewise.
19088 (s390_cc_modes_compatible): Likewise.
19089 (s390_match_ccmode_set): Likewise.
19090 (s390_canonicalize_comparison): Likewise.
19091 (s390_emit_compare_and_swap): Likewise.
19092 (s390_branch_condition_mask): Likewise.
19093 (s390_rtx_costs): Likewise.
19094 (s390_secondary_reload): Likewise.
19095 (__SECONDARY_RELOAD_CASE): Likewise.
19096 (s390_expand_cs): Likewise.
19097 (s390_preferred_simd_mode): Likewise.
19098 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
19099 * config/sh/sh.c (sh_print_operand): Likewise.
19100 (dump_table): Likewise.
19101 (sh_secondary_reload): Likewise.
19102 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19103 * config/sh/sh.md (casesi_worker_1): Likewise.
19104 (casesi_worker_2): Likewise.
19105 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
19106 (fcc_comparison_operator): Likewise.
19107 * config/sparc/sparc.c (sparc_expand_move): Likewise.
19108 (emit_soft_tfmode_cvt): Likewise.
19109 (sparc_preferred_simd_mode): Likewise.
19110 (output_cbranch): Likewise.
19111 (sparc_print_operand): Likewise.
19112 (sparc_expand_vec_perm_bmask): Likewise.
19113 (vector_init_bshuffle): Likewise.
19114 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
19115 (spu_vector_mode_supported_p): Likewise.
19116 (spu_expand_insv): Likewise.
19117 (spu_emit_branch_or_set): Likewise.
19118 (spu_handle_vector_attribute): Likewise.
19119 (spu_builtin_splats): Likewise.
19120 (spu_builtin_extract): Likewise.
19121 (spu_builtin_promote): Likewise.
19122 (spu_expand_sign_extend): Likewise.
19123 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
19124 (tilegx_simd_int): Likewise.
19125 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
19126 (tilepro_simd_int): Likewise.
19127 * config/v850/v850.c (const_double_split): Likewise.
19128 (v850_print_operand): Likewise.
19129 (ep_memory_offset): Likewise.
19130 * config/vax/vax.c (vax_rtx_costs): Likewise.
19131 (vax_output_int_move): Likewise.
19132 (vax_output_int_add): Likewise.
19133 (vax_output_int_subtract): Likewise.
19134 * config/visium/predicates.md (visium_branch_operator): Likewise.
19135 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
19136 (visium_print_operand_address): Likewise.
19137 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19138 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
19139 (xtensa_expand_conditional_branch): Likewise.
19140 (xtensa_copy_incoming_a7): Likewise.
19141 (xtensa_output_literal): Likewise.
19142 * dfp.c (decimal_real_maxval): Likewise.
19143 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19145 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19146 Alan Hayward <alan.hayward@arm.com>
19147 David Sherwood <david.sherwood@arm.com>
19149 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
19150 (mode_nunits_inline): Likewise.
19151 (mode_inner_inline): Likewise.
19152 (mode_unit_size_inline): Likewise.
19153 (mode_unit_precision_inline): Likewise.
19154 (emit_insn_modes_h): Likewise. Also emit a #define of the
19156 (emit_mode_wider): Add an E_ prefix to mode names.
19157 (emit_mode_complex): Likewise.
19158 (emit_mode_inner): Likewise.
19159 (emit_mode_adjustments): Likewise.
19160 (emit_mode_int_n): Likewise.
19161 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
19162 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
19163 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
19164 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
19165 (CRC32_BUILTIN, ENTRY): Likewise.
19166 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
19167 (aarch64_pop_regs): Likewise.
19168 (aarch64_process_components): Likewise.
19169 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
19170 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
19171 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
19172 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
19173 * config/arm/arm.c (arm_init_libfuncs): Likewise.
19174 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
19176 * config/i386/i386-builtin.def (pcmpestr): Likewise.
19177 (pcmpistr): Likewise.
19178 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
19179 * config/mmix/mmix.c (mmix_output_condition): Likewise.
19180 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19182 * config/rl78/rl78.c (mduc_regs): Likewise.
19183 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19184 (htm_expand_builtin): Likewise.
19185 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
19186 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
19187 * config/xtensa/xtensa.c (print_operand): Likewise.
19188 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
19189 (NUM_MODE_VECTOR_INT): Likewise.
19190 * genoutput.c (null_operand): Likewise.
19191 (output_operand_data): Likewise.
19192 * genrecog.c (print_parameter_value): Likewise.
19193 * lra.c (debug_operand_data): Likewise.
19195 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19197 * dwarf2out.c (scompare_loc_descriptor_wide)
19198 (scompare_loc_descriptor_narrow): New functions, split out from...
19199 (scompare_loc_descriptor): ...here.
19200 * expmed.c (emit_store_flag_int): New function, split out from...
19201 (emit_store_flag): ...here.
19203 2017-08-30 Richard Biener <rguenther@suse.de>
19205 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
19206 (dwarf2out_early_finish): Move setting of AT_pubnames from
19207 early debug output to early finish.
19209 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19211 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
19212 and -mdata-region to the assembler.
19214 2017-08-30 Richard Biener <rguenther@suse.de>
19216 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
19218 (gen_subprogram_die): Add DW_AT_object_pointer only early.
19219 (dwarf2out_early_global_decl): Only generate a DIE for the
19220 abstract origin if it doesn't already exist or is a declaration DIE.
19221 (resolve_addr): Do not add the linkage name twice when
19222 generating a stub DIE for the DW_TAG_GNU_call_site target.
19224 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19226 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19227 Use machine_mode rather than int for arg1_mode.
19229 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19232 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
19233 that the second argument of the built-in functions to unpack
19234 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
19235 switch statement instead a lot of if statements.
19236 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
19237 Allow 64-bit values to be in Altivec registers as well as
19238 traditional floating point registers.
19239 (pack<mode>, FMOVE128_VSX iterator): Likewise.
19241 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
19243 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
19244 MAX_REGS_PER_ADDRESS == 1.
19246 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
19248 * config/i386/i386.opt (flag_fentry): Do not init to -1.
19249 * config/i386/i386.c (ix86_option_override_internal): Simplify
19250 setting of opts->x_flag_entry.
19252 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19253 Jakub Jelinek <jakub@redhat.com>
19254 Richard Biener <rguenther@suse.de>
19256 PR tree-optimization/81503
19257 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
19258 folded constant fits in the target type; reorder tests for clarity.
19260 2017-08-29 Martin Liska <mliska@suse.cz>
19262 * passes.def: Include pass_lower_switch.
19263 * stmt.c (dump_case_nodes): Remove and move to
19264 tree-switch-conversion.
19265 (case_values_threshold): Likewise.
19266 (expand_switch_as_decision_tree_p): Likewise.
19267 (emit_case_decision_tree): Likewise.
19268 (expand_case): Likewise.
19269 (balance_case_nodes): Likewise.
19270 (node_has_low_bound): Likewise.
19271 (node_has_high_bound): Likewise.
19272 (node_is_bounded): Likewise.
19273 (emit_case_nodes): Likewise.
19274 (struct simple_case_node): New struct.
19275 (add_case_node): Remove.
19276 (emit_case_dispatch_table): Use vector instead of case_list.
19277 (reset_out_edges_aux): Remove.
19278 (compute_cases_per_edge): Likewise.
19279 (expand_case): Build list of simple_case_node.
19280 (expand_sjlj_dispatch_table): Use it.
19281 * tree-switch-conversion.c (struct case_node): Moved from
19282 stmt.c and adjusted.
19283 (emit_case_nodes): Likewise.
19284 (node_has_low_bound): Likewise.
19285 (node_has_high_bound): Likewise.
19286 (node_is_bounded): Likewise.
19287 (case_values_threshold): Likewise.
19288 (reset_out_edges_aux): Likewise.
19289 (compute_cases_per_edge): Likewise.
19290 (add_case_node): Likewise.
19291 (dump_case_nodes): Likewise.
19292 (balance_case_nodes): Likewise.
19293 (expand_switch_as_decision_tree_p): Likewise.
19294 (emit_jump): Likewise.
19295 (emit_case_decision_tree): Likewise.
19296 (try_switch_expansion): Likewise.
19297 (do_jump_if_equal): Likewise.
19298 (emit_cmp_and_jump_insns): Likewise.
19299 (fix_phi_operands_for_edge): New function.
19300 (record_phi_operand_mapping): Likewise.
19301 (class pass_lower_switch): New pass.
19302 (pass_lower_switch::execute): New function.
19303 (make_pass_lower_switch): Likewise.
19304 (conditional_probability):
19305 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
19306 * tree-pass.h: Add make_pass_lower_switch.
19308 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19311 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
19314 2017-08-29 Richard Biener <rguenther@suse.de>
19316 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
19317 we do not add a DW_AT_inline attribute twice.
19318 (gen_subprogram_die): Remove code setting DW_AT_inline on
19319 DECL_ABSTRACT_P nodes.
19321 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
19323 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
19324 calls to internal functions.
19325 (gimplify_modify_expr): Likewise.
19326 * tree-call-cdce.c (use_internal_fn): Likewise.
19327 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
19328 (convert_to_divmod): Set the nothrow flag.
19329 * tree-if-conv.c (predicate_mem_writes): Likewise.
19330 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19331 (vectorizable_call): Likewise.
19332 (vectorizable_store): Likewise.
19333 (vectorizable_load): Likewise.
19334 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
19335 (vect_recog_mask_conversion_pattern): Likewise.
19337 2017-08-29 Martin Liska <mliska@suse.cz>
19340 * gcc.c (driver_handle_option): Add new argument.
19341 * opts-common.c (handle_option): Pass
19342 target_option_override_hook.
19343 * opts-global.c (lang_handle_option): Add new option.
19344 (set_default_handlers): Add new argument.
19345 (decode_options): Likewise.
19346 * opts.c (target_handle_option): Likewise.
19347 (common_handle_option): Call target_option_override_hook.
19348 * opts.h (struct cl_option_handler_func): Add hook for
19349 target option override.
19350 (struct cl_option_handlers): Likewise.
19351 (set_default_handlers): Add new argument.
19352 (decode_options): Likewise.
19353 (common_handle_option): Likewise.
19354 (target_handle_option): Likewise.
19355 * toplev.c (toplev::main): Pass targetm.target_option.override
19358 2017-08-29 Richard Biener <rguenther@suse.de>
19359 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
19361 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
19362 life to the active subtree.
19364 2017-08-28 Jeff Law <law@redhat.com>
19366 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
19367 derive_equivalences.
19368 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
19369 Code moved into....
19370 (edge_info::derive_equivalences): New private member function
19372 * tree-ssa-dom.c (class edge_info): Changed from a struct
19373 to a class. Add ctor/dtor, methods and data members.
19374 (edge_info::edge_info): Renamed from allocate_edge_info.
19375 Initialize additional members.
19376 (edge_info::~edge_info): New.
19377 (free_dom_edge_info): Delete the edge info.
19378 (record_edge_info): Use new class & associated member functions.
19379 Tighten forms for testing for edge equivalences.
19380 (record_temporary_equivalences): Iterate over the simple
19381 equivalences rather than assuming there's only one per edge.
19382 (cprop_into_successor_phis): Iterate over the simple
19383 equivalences rather than assuming there's only one per edge.
19384 (optimize_stmt): Use operand_equal_p rather than pointer
19385 equality for mini-DSE code.
19387 2017-08-28 Nathan Sidwell <nathan@acm.org>
19389 * gcc.c (execute): Fold SIGPIPE handling into switch
19390 statement. Adjust internal error message.
19392 2017-08-28 Richard Biener <rguenther@suse.de>
19395 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
19396 Do nothing for removed DIEs.
19398 2017-08-28 Richard Biener <rguenther@suse.de>
19400 PR tree-optimization/81977
19401 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
19404 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
19407 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
19408 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
19409 using targetm.gen_mem_thread_fence.
19411 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
19414 * config/i386/i386.md (*<btsc><mode>): Change operand 2
19415 predicate to register_operand. Reorder operands.
19416 (*btr<mode>): Ditto.
19417 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
19418 (*btr<mode>_mask): Ditto.
19420 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
19422 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
19423 * config/rs6000/xmmintrin.h: New file.
19424 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
19426 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19429 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
19430 parameter and_insn and return it.
19431 (recombine_lvx_pattern): Insert a copy to ensure availability of
19432 the base register of the copied masking operation at the point of
19433 the instruction replacement.
19434 (recombine_stvx_pattern): Likewise.
19436 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
19438 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
19439 undocumented switches.
19440 (-mpower9-dform-vector): Likewise.
19441 (-mpower9-dform): Likewise.
19442 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19443 comments to delete references to -mpower9-dform* switches.
19444 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
19445 Delete reference to -mpower9-dform* switches, test for
19446 -mpower9-vector instead.
19447 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
19448 (OTHER_P9_VECTOR_MASKS): Likewise.
19449 (POWERPC_MASKS): Likewise.
19450 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
19451 tests against -mpower9-dform* to -mpower9-vector. Delete code
19452 that checked for -mpower9-dform* consistancy with other options.
19453 Add test for -mpower9-misc to enable other power9 switches.
19454 (rs6000_init_hard_regno_mode_ok): Likewise.
19455 (rs6000_option_override_internal): Likewise.
19456 (rs6000_emit_prologue): Likewise.
19457 (rs6000_emit_epilogue): Likewise.
19458 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
19459 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
19460 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
19462 (emit_fusion_p9_store): Likewise.
19463 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
19464 resetting these macros if the assembler does not support ISA 3.0
19466 (TARGET_P9_DFORM_VECTOR): Likewise.
19467 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
19468 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
19470 2017-08-25 Alan Modra <amodra@gmail.com>
19472 PR rtl-optimization/81747
19473 * cse.c (cse_extended_basic_block): Don't attempt to record
19474 equivalences for degenerate conditional branches that branch
19475 to their fall-through.
19477 2017-08-24 Martin Sebor <msebor@redhat.com>
19479 PR middle-end/81908
19480 * gimple-fold.c (size_must_be_zero_p): New function.
19481 (gimple_fold_builtin_memory_op): Call it.
19483 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
19485 * config/rs6000/mm_malloc.h: New file.
19487 2017-08-24 Bin Cheng <bin.cheng@arm.com>
19489 PR tree-optimization/81913
19490 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
19491 analysis when either IVs in condition can wrap.
19493 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
19495 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
19496 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
19498 2017-08-24 Richard Biener <rguenther@suse.de>
19501 * targhooks.c (default_target_can_inline_p): Properly
19502 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
19503 is present and always compare.
19504 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
19505 infer -mfpmath=sse from TARGET_SSE_P.
19506 (ix86_can_inline_p): Properly use target_option_default_node when
19507 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
19509 2017-08-24 Richard Biener <rguenther@suse.de>
19512 * dwarf2out.c (output_die): Handle flag_generate_offload like
19514 (output_comp_unit): Likewise.
19515 (gen_array_type_die): Likewise.
19516 (dwarf2out_early_finish): Likewise.
19517 (note_variable_value_in_expr): Likewise.
19518 (dwarf2out_finish): Likewise. Adjust assert.
19519 * cgraphunit.c (symbol_table::compile): Move setting of
19520 flag_generate_offload earlier ...
19521 (symbol_table::finalize_compilation_unit): ... here, before
19522 early debug finalization.
19524 2017-08-24 Richard Biener <rguenther@suse.de>
19526 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
19527 and ipa-fnsummary.h.
19528 (ix86_can_inline_p): When ix86_fpmath flags do not match
19529 check whether the callee uses FP math at all.
19531 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
19533 PR middle-end/81931
19534 * tree-ssanames.c (get_nonzero_bits): Use element_precision
19535 instead of TYPE_PRECISION.
19537 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
19538 Alan Hayward <alan.hayward@arm.com>
19539 David Sherwood <david.sherwood@arm.com>
19541 * combine.c (make_extraction): Use subreg_offset_from_lsb.
19543 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
19545 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
19547 (ix86_frame::stack_realign_allocate): New field.
19548 (struct machine_frame_state): Modify comments.
19549 (machine_frame_state::sp_realigned_fp_end): New field.
19550 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
19551 layout calculation.
19552 (sp_valid_at): Add assertion to assure no attempt to access invalid
19553 offset of a realigned stack.
19554 (fp_valid_at): Likewise.
19555 (choose_baseaddr): Modify comments.
19556 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
19557 ix86_expand_prologue.
19558 (ix86_expand_prologue): Modify stack realignment and allocation.
19559 (ix86_expand_epilogue): Modify comments.
19560 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
19561 avx2_runtime, avx512f, and avx512f_runtime.
19563 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
19565 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
19566 (mstackrealign): Do not init to -1.
19567 * config/i386/i386.c (ix86_option_override_internal):
19568 Check opts_set, not opts when setting default value of
19569 opts->x_ix86_force_align_arg_pointer.
19571 2017-08-23 Richard Biener <rguenther@suse.de>
19573 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
19574 lang_hooks.decl_printable_name.
19575 * print-rtl-function.c (print_rtx_function): Likewise.
19576 * tree-pretty-print.c (dump_function_header): Likewise.
19578 2017-08-23 Richard Biener <rguenther@suse.de>
19581 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
19582 -g0 at compile-time.
19584 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19586 PR middle-end/19706
19587 * doc/sourcebuild.texi (Other hardware attributes):
19590 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19592 PR middle-end/19706
19593 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
19594 Fix single-use check.
19596 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
19598 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
19600 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19602 * doc/install.texi: Modify to add more details on running selected
19605 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19607 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
19608 is combined with -mabi=ms.
19609 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
19612 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19614 PR tree-optimization/81488
19615 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
19616 and cached_basis fields.
19617 (MAX_SPREAD): New constant.
19618 (alloc_cand_and_find_basis): Initialize new fields.
19619 (clear_visited): New function.
19620 (create_phi_basis_1): Rename from create_phi_basis, set visited
19621 and cached_basis fields.
19622 (create_phi_basis): New wrapper function.
19623 (phi_add_costs_1): Rename from phi_add_costs, add spread
19624 parameter, set visited field, short-circuit when limits reached.
19625 (phi_add_costs): New wrapper function.
19626 (record_phi_increments_1): Rename from record_phi_increments, set
19628 (record_phi_increments): New wrapper function.
19629 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
19630 (phi_incr_cost): New wrapper function.
19631 (all_phi_incrs_profitable_1): Rename from
19632 all_phi_incrs_profitable, set visited field.
19633 (all_phi_incrs_profitable): New wrapper function.
19635 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19636 Alan Hayward <alan.hayward@arm.com>
19637 David Sherwood <david.sherwood@arm.com>
19639 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
19640 that takes the outer and inner modes.
19641 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
19642 comparison as the canonical test for a paradoxical subreg.
19643 * combine.c (simplify_set): Use paradoxical_subreg_p.
19644 (make_extraction): Likewise.
19645 (force_to_mode): Likewise.
19646 (rtx_equal_for_field_assignment_p): Likewise.
19647 (gen_lowpart_for_combine): Likewise.
19648 (simplify_comparison): Likewise.
19649 * cse.c (equiv_constant): Likewise.
19650 * expmed.c (store_bit_field_1): Likewise.
19651 * final.c (alter_subreg): Likewise.
19652 * fwprop.c (propagate_rtx): Likewise.
19653 (forward_propagate_subreg): Likewise.
19654 * ira-conflicts.c (ira_build_conflicts): Likewise.
19655 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
19656 * lra-constraints.c (curr_insn_transform): Likewise.
19657 (split_reg): Likewise.
19658 * lra-eliminations.c (move_plus_up): Likewise.
19659 (lra_eliminate_regs_1): Likewise.
19660 * recog.c (general_operand): Likewise.
19661 * ree.c (combine_reaching_defs): Likewise.
19662 * reload.c (push_reload): Likewise.
19663 (find_reloads): Likewise.
19664 * reload1.c (elimination_effects): Likewise.
19665 (compute_reload_subreg_offset): Likewise.
19666 (choose_reload_regs): Likewise.
19667 * rtlanal.c (subreg_lsb_1): Likewise.
19668 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19669 (simplify_subreg): Likewise.
19670 * var-tracking.c (track_loc_p): Likewise.
19671 * emit-rtl.c (byte_lowpart_offset): Likewise.
19672 (paradoxical_subreg_p): Delete out-of-line definition.
19674 2017-08-22 Jeff Law <law@redhat.com>
19676 PR tree-optimization/81741
19677 PR tree-optimization/71947
19678 * tree-ssa-dom.c: Include tree-inline.h.
19679 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
19680 equivalences if one is more expensive to compute than the other.
19681 * tree-ssa-scopedtables.h (class const_or_copies): Make
19682 record_const_or_copy_raw method private.
19683 (class avail_exprs_stack): New method simplify_binary_operation.
19684 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
19685 avail_exprs_stack::simplify_binary_operation as needed.
19686 (avail_exprs_stack::simplify_binary_operation): New function.
19688 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
19690 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
19691 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
19692 (DOT_SYMBOLS): Likewise.
19693 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
19694 (RELOCATABLE_NEEDS_FIXUP): Likewise.
19695 (RS6000_ABI_NAME): Likewise.
19696 (TARGET_CMODEL): Likewise.
19697 (TOC_SECTION_ASM_OP): Likewise.
19698 (SET_CMODEL): New macro.
19699 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
19701 2017-08-22 Richard Biener <rguenther@suse.de>
19703 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
19704 to allow for free-lang-data replacements similar to verify_type_variant.
19706 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19707 Alan Hayward <alan.hayward@arm.com>
19708 David Sherwood <david.sherwood@arm.com>
19710 * config/aarch64/aarch64.md (casesi): Use DImode rather than
19711 VOIDmode for the LABEL_REF.
19713 2017-08-22 Richard Biener <rguenther@suse.de>
19715 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
19717 2017-08-22 Richard Biener <rguenther@suse.de>
19719 * common.opt (feliminate-dwarf2-dups): Ignore.
19720 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
19721 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
19722 same_die_p_wrap, compute_section_prefix,
19723 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
19724 (comdat_symbol_id, comdat_symbol_number): Likewise.
19725 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
19727 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
19728 (output_die): Mark unreachable path unreachable.
19729 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
19730 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
19731 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
19732 (dwarf2out_early_finish): Likewise.
19734 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
19736 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
19738 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
19741 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
19742 not VAR_P. Filter attribute warnings with OPT_Wattributes.
19743 (avr_attribute_table) <io, io_low, address>: Initialize
19744 .decl_required with true.
19746 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
19748 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
19749 undocumented debugging options.
19750 (-mvsx-scalar-double): Likewise.
19751 (-mallow-df-permute): Likewise.
19752 (-mvectorize-builtins): Likewise.
19753 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19754 (rs6000_builtin_vectorized_function): Likewise.
19755 (rs6000_builtin_md_vectorized_function): Likewise.
19756 (rs6000_opt_vars): Likewise.
19758 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
19761 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
19762 (*btrq_imm): Rename from *btrq.
19763 (*btcq_imm): Rename from *btcq.
19764 (btsc): New code attribute.
19765 (*<btsc><mode>): New insn pattern.
19766 (*btr<mode>): Ditto.
19767 (*<btsc><mode>_mask): New insn_and_split pattern.
19768 (*btr<mode>_mask): Ditto.
19770 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19771 Alan Hayward <alan.hayward@arm.com>
19772 David Sherwood <david.sherwood@arm.com>
19774 * function.c (pad_below): Simplify padding calculation.
19776 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19777 Alan Hayward <alan.hayward@arm.com>
19778 David Sherwood <david.sherwood@arm.com>
19780 * target.def (function_prologue): Remove frame size argument.
19781 (function_epilogue): Likewise.
19782 * doc/tm.texi: Regenerate.
19783 * final.c (final_start_function): Update call to function_prologue.
19784 (final_end_function): Update call to function_epilogue.
19785 (default_function_pro_epilogue): Remove frame size argument.
19786 * output.h (default_function_pro_epilogue): Likewise.
19787 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
19788 (arm_output_function_prologue): Likewise.
19789 * config/frv/frv.c (frv_function_prologue): Likewise.
19790 (frv_function_epilogue): Likewise.
19791 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
19792 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
19793 (ia64_output_function_epilogue): Likewise.
19794 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
19795 (m32r_output_function_epilogue): Likewise.
19796 * config/microblaze/microblaze.c (microblaze_function_prologue)
19797 (microblaze_function_epilogue): Likewise.
19798 * config/mips/mips.c (mips_output_function_prologue): Likewise.
19799 (mips_output_function_epilogue): Likewise.
19800 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
19801 (mmix_target_asm_function_epilogue): Likewise.
19802 * config/msp430/msp430.c (msp430_start_function): Likewise.
19803 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
19804 (nds32_asm_function_epilogue): Likewise.
19805 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
19806 * config/pa/pa.c (pa_output_function_prologue): Likewise.
19807 (pa_output_function_epilogue): Likewise.
19808 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
19809 (rs6000_output_function_epilogue): Likewise.
19810 * config/rl78/rl78.c (rl78_start_function): Likewise.
19811 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
19812 (rs6000_output_function_epilogue): Likewise.
19813 * config/rx/rx.c (rx_output_function_prologue): Likewise.
19814 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
19815 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
19816 (sparc_asm_function_epilogue): Likewise.
19818 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19820 * tree.h (type_has_mode_precision_p): New function.
19821 * convert.c (convert_to_integer_1): Use it.
19822 * expr.c (expand_expr_real_2): Likewise.
19823 (expand_expr_real_1): Likewise.
19824 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
19825 * match.pd: Likewise.
19826 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
19827 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
19828 * tree-tailcall.c (process_assignment): Likewise.
19829 * tree-vect-loop.c (vectorizable_reduction): Likewise.
19830 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
19831 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
19832 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19833 (vectorizable_assignment): Likewise.
19834 (vectorizable_shift): Likewise.
19835 (vectorizable_operation): Likewise.
19836 * tree-vrp.c (register_edge_assert_for_2): Likewise.
19838 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
19840 * match.pd: Add pow (C, x) simplification.
19842 2017-08-21 Richard Biener <rguenther@suse.de>
19844 PR tree-optimization/81900
19845 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
19846 for blocks with abnormal predecessors.
19847 (compute_antic): Do not set visited flag prematurely.
19849 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
19852 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
19854 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19856 * stor-layout.h (vector_type_mode): Move to...
19857 * tree.h (vector_type_mode): ...here.
19858 * stor-layout.c (vector_type_mode): Move to...
19859 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
19861 2017-08-21 Richard Biener <rguenther@suse.de>
19863 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
19864 register_external_die hooks.
19865 (debug_false_tree_charstarstar_uhwistar): Declare.
19866 (debug_nothing_tree_charstar_uhwi): Likewise.
19867 * debug.c (do_nothing_debug_hooks): Adjust.
19868 (debug_false_tree_charstarstar_uhwistar): New do nothing.
19869 (debug_nothing_tree_charstar_uhwi): Likewise.
19870 * dbxout.c (dbx_debug_hooks): Adjust.
19871 (xcoff_debug_hooks): Likewise.
19872 * sdbout.c (sdb_debug_hooks): Likewise.
19873 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19874 * dwarf2out.c (macinfo_label_base): New global.
19875 (dwarf2out_register_external_die): New function for the
19876 register_external_die hook.
19877 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
19878 (dwarf2_debug_hooks): Use them.
19879 (dwarf2_lineno_debug_hooks): Adjust.
19880 (struct die_struct): Add with_offset flag.
19881 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
19882 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
19883 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
19884 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
19885 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
19886 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
19887 defining section names for the early LTO debug variants.
19888 (reset_indirect_string): New helper.
19889 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
19890 (print_dw_val): Add support for offsetted symbol references.
19891 (get_ultimate_context): Split out from is_cxx.
19892 (is_cxx): Use get_ultimate_context.
19893 (is_fortran): Add decl overload.
19894 (compute_comp_unit_symbol): Split out worker from
19895 compute_section_prefix.
19896 (compute_section_prefix): Call compute_comp_unit_symbol and
19897 set comdat_type_p here.
19898 (output_die): Skip DIE symbol output for the LTO added one.
19899 Handle DIE symbol references with offset.
19900 (output_comp_unit): Guard section name mangling properly.
19901 For LTO debug sections emit a symbol at the section beginning
19902 which we use to refer to its DIEs.
19903 (add_abstract_origin_attribute): For DIEs registered via
19904 dwarf2out_register_external_die directly refer to the early
19905 DIE rather than indirectly through the shadow one we created.
19906 Remove obsolete call to dwarf2out_abstract_function for
19907 non-function/block origins.
19908 (gen_array_type_die): When generating early LTO debug do
19909 not emit DW_AT_string_length.
19910 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
19911 late when in LTO. As suggested place a gcc_unreachable for
19912 the DECL_ABSTRACT_P case.
19913 (gen_subprogram_die): Avoid another specification DIE
19914 for early built declarations/definitions for the late LTO case.
19915 (gen_variable_die): Add type references for late duplicated VLA dies
19917 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
19918 we have the abstract instance already.
19919 (process_scope_var): Adjust decl DIE contexts in LTO which
19920 first puts them in limbo.
19921 (gen_decl_die): Do not generate type DIEs late apart from
19922 types for VLAs or for decls we do not yet have a DIE. Do not
19923 call dwarf2out_abstract_function late.
19924 (dwarf2out_early_global_decl): Make sure to create DIEs
19925 for abstract instances of a decl first.
19926 (dwarf2out_late_global_decl): Adjust comment.
19927 (output_macinfo_op): With multiple macro sections use
19928 macinfo_label_base to distinguish labels.
19929 (output_macinfo): Likewise. Update macinfo_label_base.
19930 Pass in the line info label.
19931 (note_variable_value_in_expr): When generating LTO resolve
19932 all variable values here by generating DIEs as needed.
19933 (init_sections_and_labels): Add early LTO debug flag parameter
19934 and generate different sections and names if set. Add generation
19935 counter for the labels so we can have multiple of them.
19936 (reset_dies): Helper to allow DIEs to be output multiple times.
19937 (dwarf2out_finish): When outputting DIEs to the fat part of an
19938 LTO object first reset DIEs.
19939 (dwarf2out_early_finish): Output early DIEs when generating LTO.
19940 (modified_type_die): Check for decl_ultimate_origin being self
19942 (gen_type_die_with_usage): Likewise.
19943 (gen_typedef_die): Allow decl_ultimate_origin being self.
19944 (set_decl_abstract_flags): Remove.
19945 (set_block_abstract_flags): Likewise.
19946 (dwarf2out_abstract_function): Treat the early generated DIEs
19947 as the abstract copy and only add DW_AT_inline and
19948 DW_AT_artificial here and call set_decl_origin_self.
19949 If the DIE has an abstract origin don't do anything.
19950 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
19951 if we have none yet (Go fails to build one, PR78628).
19952 (variably_modified_type_p): Prevent endless recursion for Ada
19953 cyclic pointer types.
19954 * lto-streamer-in.c: Include debug.h.
19955 (dref_queue): New global.
19956 (lto_read_tree_1): Stream in DIE references.
19957 (lto_input_tree): Register DIE references.
19958 (input_function): Stream DECL_DEBUG_ARGS.
19959 * lto-streamer-out.c: Include debug.h.
19960 (lto_write_tree_1): Output DIE references.
19961 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
19962 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
19963 (output_function): Stream DECL_DEBUG_ARGS.
19964 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
19965 Stream DECL_ABSTRACT_ORIGIN.
19966 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
19967 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
19968 DECL_CONTEXT for file-scope decls.
19969 * lto-streamer.h (struct dref_entry): Declare.
19970 (dref_queue): Likewise.
19971 * cfgexpand.c (pass_expand::execute): Do not call the
19972 outlining_inline_function hook here.
19973 * lto-wrapper.c (debug_obj): New global.
19974 (tool_cleanup): Unlink it if required.
19975 (debug_objcopy): New function.
19976 (run_gcc): Handle early debug sections in the IL files by
19977 extracting them to separate files, partially linkin them and
19978 feeding the result back as result to the linker.
19979 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
19980 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
19981 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
19982 sections into a separate segment.
19983 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
19985 (darwin_asm_dwarf_section): Likewise.
19986 (darwin_asm_output_dwarf_offset): Likewise.
19987 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
19989 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19990 Alan Hayward <alan.hayward@arm.com>
19991 David Sherwood <david.sherwood@arm.com>
19993 * read-md.h (md_reader::record_potential_iterator_use): Replace
19994 pointer argument with an rtx and an index.
19995 * read-rtl.c (iterator_group::apply_iterator): Likewise.
19996 (apply_mode_iterator): Likewise.
19997 (apply_code_iterator): Likewise.
19998 (apply_int_iterator): Likewise.
19999 (apply_subst_iterator): Likewise.
20000 (record_iterator_use): Likewise.
20001 (record_attribute_use): Likewise.
20002 (md_reader::record_potential_iterator_use): Likewise. Update calls
20003 to record_iterator_use and apply_iterator.
20004 (iterator_use): Replace ptr with x and index.
20005 (attribute_use): Likewise.
20006 (apply_attribute_uses): Update calls to apply_iterator.
20007 (apply_iterators): Likewise. Update initialization of iterator_use.
20008 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
20009 and record_potential_iterator_use.
20010 (rtx_reader::read_rtx_operand): Likewise.
20012 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20013 Alan Hayward <alan.hayward@arm.com>
20014 David Sherwood <david.sherwood@arm.com>
20016 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
20019 2017-08-21 Richard Biener <rguenther@suse.de>
20021 PR middle-end/81884
20022 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
20023 at struct end conservatively when comparing common bases.
20025 2017-08-21 Richard Biener <rguenther@suse.de>
20027 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
20028 (mem_ref_in_stmt): Remove.
20029 (determine_max_movement): Use ref index to get at the reference.
20030 (invariantness_dom_walker::before_dom_children): Deal with
20031 lim data already initialized.
20032 (gather_mem_refs_stmt): Initialize lim data and record ref index.
20034 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20036 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
20037 (TARGET_ISA_ROUND): Ditto.
20038 (TARGET_ROUND): Ditto.
20039 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
20040 * config/i386/i386.md: Ditto.
20041 * config/i386/sse.md: Ditto.
20042 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
20043 with OPTION_MASK_ISA_SSE4_1.
20045 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20048 * doc/extend.texi (x86 Built-in Functions): Correct the name of
20049 __builtin_ia32_lzcnt_u16.
20051 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
20054 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
20055 (rs6000_set_current_function): Rewrite function to use it.
20057 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
20060 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
20061 and TYPE_WARN_IF_NOT_ALIGN.
20062 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
20063 (handle_warn_if_not_align): New.
20064 (place_union_field): Call handle_warn_if_not_align.
20065 (place_field): Call handle_warn_if_not_align.
20066 Copy TYPE_WARN_IF_NOT_ALIGN.
20067 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
20068 (layout_type): Likewise.
20069 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
20071 (tree_decl_common): Add warn_if_not_align.
20072 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
20073 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
20074 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
20075 (DECL_WARN_IF_NOT_ALIGN): Likewise.
20076 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
20077 * doc/extend.texi: Document warn_if_not_aligned attribute.
20078 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
20080 2017-08-17 Martin Liska <mliska@suse.cz>
20083 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
20084 (get_data_dependence): Use it as pointer type.
20085 (distribute_loop): Likewise.
20087 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20089 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
20090 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
20091 (p8_vmrgow_<mode>_direct): New define_insn.
20092 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
20093 handle endianness for vmrgew and vmrgow permute patterns.
20095 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20097 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
20098 * config/rs6000/rs6000-cpus.def: Remove comment.
20099 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
20100 (POWERPC_MASKS): Likewise.
20101 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
20102 use of TARGET_VSX_TIMODE.
20103 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
20104 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
20105 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
20106 (rs6000_option_override_internal): Remove dead code.
20107 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
20108 (rs6000_legitimize_reload_address): Likewise.
20109 (rs6000_legitimate_address_p): Likewise.
20110 (rs6000_opt_masks): Delete "vsx-timode".
20111 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
20112 from function comment.
20113 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
20114 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
20115 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
20117 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
20118 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
20119 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
20120 (VSX_TI): Likewise.
20122 (define_peephole2): Likewise.
20124 2017-08-17 Martin Sebor <msebor@redhat.com>
20127 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
20128 past the end of an array.
20129 (test_pp_format): Add test cases.
20131 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
20133 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
20134 missing ECF_NOTHROW flags.
20136 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20139 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
20140 operands residing in integer registers.
20141 (*vsx_le_perm_load_<mode>): Likewise.
20142 (*vsx_le_perm_store_<mode>): Likewise.
20143 (define_peephole2): Add peepholes to optimize the above.
20145 2017-08-17 Marek Polacek <polacek@redhat.com>
20147 PR middle-end/81814
20148 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
20149 to mimic what shorten_compare did. Change the return type to bool.
20150 (fold_cond_expr_with_comparison): Update call to
20151 operand_equal_for_comparison_p.
20152 (fold_ternary_loc): Likewise.
20154 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
20156 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
20158 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
20160 2017-08-17 Richard Biener <rguenther@suse.de>
20162 * tree-ssa-structalias.c (solve_graph): When propagating
20163 to successors update the graphs succ edges and avoid duplicate work.
20165 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
20168 * config/i386/i386.c (ix86_option_override_internal): Save target
20169 specific options after ix86_stack_protector_guard_reg was changed.
20171 2017-08-17 Richard Biener <rguenther@suse.de>
20173 PR tree-optimization/81827
20174 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
20175 (new_var_info): Initialize it conservatively.
20176 (get_call_vi): Mark register vars.
20177 (new_scalar_tmp_constraint_exp): Likewise.
20178 (handle_rhs_call): Likewise.
20179 (handle_const_call): Likewise.
20180 (create_function_info_for): Likewise.
20181 (solve_constraints): Sort varinfos to separate register from
20182 non-register vars to pack points-to solution bitmaps during
20185 2017-08-17 Marek Polacek <polacek@redhat.com>
20187 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
20189 2017-08-17 Richard Biener <rguenther@suse.de>
20191 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
20192 to true when overflow is undefined and we saturated the result.
20194 2017-08-17 Alan Modra <amodra@gmail.com>
20197 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
20198 Don't use store multiple if only one reg needs saving.
20199 (interesting_frame_related_regno): New function.
20200 (rs6000_frame_related): Don't emit frame info for regs that
20202 (rs6000_emit_epilogue): Likewise.
20204 2017-08-16 Nathan Sidwell <nathan@acm.org>
20206 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
20207 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
20208 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
20209 * tree.c (free_lang_data_in_type): Use else-if chain. Always
20210 clear TYPE_LANG_1. Remove obsolete member-function stripping.
20211 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
20212 (verify_type): Adjust for TYPE_BINFO move.
20213 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
20214 process TYPE_BINFO directly.
20215 (hash_tree): Likewise.
20216 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20218 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20221 2017-08-16 David Malcolm <dmalcolm@redhat.com>
20223 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
20225 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
20228 * config/i386/i386.md (*anddi_1_btr): Change predicates of
20229 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
20230 Add ix86_binary_operator_ok to insn constraint.
20231 (*iordi_1_bts): Ditto.
20232 (*xordi_1_btc): Ditto.
20233 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
20234 Update corresponding peephole2 pattern.
20238 2017-08-16 Bin Cheng <bin.cheng@arm.com>
20240 PR tree-optimization/81832
20241 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
20242 copy loop header which has IFN_LOOP_DIST_ALIAS call.
20244 2017-08-16 Marek Polacek <polacek@redhat.com>
20247 * fold-const.c (fold_indirect_ref_1): Restore original behavior
20248 regarding size_zero_node.
20250 2017-08-16 Martin Liska <mliska@suse.cz>
20253 * config.gcc: Respect previously set extra_objs in case
20256 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
20258 PR tree-optimization/81835
20259 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
20260 the phi SSA_NAME. Check that the condition in a COND_EXPR does
20261 not depend on the phi.
20263 2017-08-16 Alan Modra <amodra@gmail.com>
20265 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
20268 2017-08-16 Alan Modra <amodra@gmail.com>
20270 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
20271 (save_reg_p): ..into this. Update all callers.
20272 (first_reg_to_save): Simplify.
20274 2017-08-16 Alan Modra <amodra@gmail.com>
20276 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
20279 2017-08-15 Joseph Myers <joseph@codesourcery.com>
20283 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
20284 constant count if that count is less than 32.
20286 2017-08-15 Nathan Sidwell <nathan@acm.org>
20288 * gcc.c (execute): Emit friendlier message if inferior is killed
20289 by an external cause.
20291 2017-08-15 Richard Biener <rguenther@suse.de>
20293 PR tree-optimization/81790
20294 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
20295 CONSTRUCTORs from simplifying and VN.
20297 2017-08-14 Martin Sebor <msebor@redhat.com>
20299 * builtin-attrs.def: Add comments.
20301 2017-08-14 Martin Sebor <msebor@redhat.com>
20304 * doc/extend.texi (attribute nonstring): Document new attribute.
20306 2017-08-14 Martin Sebor <msebor@redhat.com>
20309 * tree-diagnostic.c (default_tree_printer): Handle %G.
20310 * gimple-pretty-print.h (percent_G_format): Declare new function.
20311 * gimple-pretty-print.c (percent_G_format): Define.
20312 * tree-pretty-print.c (percent_K_format): Add argument.
20314 2017-08-14 Martin Sebor <msebor@redhat.com>
20316 PR translation/79998
20317 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
20318 Remove a stray space.
20320 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
20323 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
20324 (*iordi_1_bts): Ditto.
20325 (*xordi_1_btc): Ditto.
20327 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20330 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
20331 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20333 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
20334 quoted strings, and make more translator-friendly.
20335 (darwin_rs6000_override_options): Likewise.
20336 (rs6000_option_override_internal): Likewise.
20337 (rs6000_return_in_memory): Fix overlong line.
20338 (init_cmulative_args): Use quoted strings, and make more
20339 translator-friendly.
20340 (rs6000_pass_by_reference): Fix overlong line.
20341 (def_builtin): Use quoted strings.
20342 (altivec_expand_predicate_builtin): Use quoted strings, and make
20343 more translator-friendly.
20344 (htm_expand_builtin): Use quoted strings.
20345 (cpu_expand_builtin): Use quoted strings, and make more
20346 translator-friendly.
20347 (altivec_expand_builtin): Likewise.
20348 (paired_expand_predicate_builtin): Likewise.
20349 (rs6000_invalid_builtin): Likewise.
20350 (builtin_function_type): Use quoted strings.
20351 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
20352 more translator-friendly.
20353 (rs6000_trampoline_init): Likewise.
20354 (rs6000_handle_altivec_attribute): Likewise.
20355 (rs6000_inner_target_options): Use quoted strings.
20356 (rs6000_disable_incompatible_switches): Likewise.
20357 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
20358 strings, and make more translator-friendly.
20359 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
20361 2017-08-14 Bin Cheng <bin.cheng@arm.com>
20363 PR tree-optimization/81799
20364 * tree-loop-distribution.c (version_loop_by_alias_check): Force
20365 cond_expr to simple gimple operand.
20367 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
20369 PR middle-end/46932
20370 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
20372 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
20376 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
20377 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
20378 TARGET_GASISR_PROLOGUES.
20379 * config/avr/avr.c (avr_option_override): Same.
20380 (avr_pass_pre_proep::execute): Same.
20382 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
20385 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
20386 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
20388 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
20390 * config/i386/i386.md (*load_tp_<mode>): Redefine as
20391 define_insn_and_split. Split to a memory load from 0 in
20392 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
20393 using PTR mode iterator.
20394 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
20395 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
20396 (*add_tp_<mode>): Redefine as define_insn_and_split.
20397 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
20398 address space. Merge with *add_tp_x32 using PTR mode iterator.
20399 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
20400 Split to an add with a memory load from 0 in
20401 DEFAULT_TLS_SEG_REG address space.
20403 2017-08-12 Andrew Pinski <apinski@cavium.com>
20405 * config/aarch64/aarch64-option-extensions.def (rdma):
20406 Fix feature string to what Linux prints out in /proc/cpuinfo.
20408 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
20411 * dwarf2out.c (modified_type_die): For C typedef types that have
20412 an ultimate origin, process the ultimate origin instead of the
20414 (gen_typedef_die): Assert that input DECLs have no ultimate
20416 (gen_type_die_with_usage): For typedef variants that have an
20417 ultimate origin, just call gen_decl_die on the original DECL.
20418 (process_scope_var): Avoid creating DIEs for local typedefs and
20419 concrete static variables.
20421 2017-08-12 Alan Modra <amodra@gmail.com>
20425 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
20426 match gnu-user.h startfile.
20427 (ENDFILE_LINUX_SPEC): Similarly.
20429 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
20432 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
20434 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
20436 2017-08-11 Tamar Christina <tamar.christina@arm.com>
20437 * config/aarch64/aarch64.md (mov<mode>): Change.
20438 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20439 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
20440 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
20442 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
20444 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
20445 for storage order barriers.
20447 2017-08-11 Martin Liska <mliska@suse.cz>
20450 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
20451 variables of void type.
20453 2017-08-11 Martin Liska <mliska@suse.cz>
20455 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
20456 TARGET_SUPPORTS_ALIASES.
20457 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
20458 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
20459 (optimize_weakref): Likewise.
20460 * symtab.c (symtab_node::noninterposable_alias): Likewise.
20461 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
20462 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
20464 2017-08-11 Martin Liska <mliska@suse.cz>
20467 * config/i386/i386.c (make_resolver_func): Do complete
20468 refactoring of the function.
20470 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20473 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
20474 * config/i386/i386.c (ix86_stack_protect_guard): Use
20475 ix86_stack_protect_guard_symbol_str to generate varible declaration.
20476 * doc/invoke.texi (x86 Options): Document
20477 -mstack-protector-guard-symbol= option.
20479 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20481 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
20482 * config/i386/i386.c (ix86_split_stack_guard): New function.
20483 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
20484 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
20485 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
20486 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
20487 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
20488 (split_stack_space_check): Call ix86_split_stack_guard.
20490 2017-08-10 Martin Sebor <msebor@redhat.com>
20492 * print-tree.c (print_node): Print location using the established
20494 Replace spaces with colons.
20495 (debug_raw, debug): Ditto.
20497 2017-08-10 Martin Sebor <msebor@redhat.com>
20500 * pretty-print.c (pp_format): Correct the handling of %s precision.
20502 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
20505 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20507 (ix86_finalize_stack_frame_flags): This. Also clear
20508 frame_pointer_needed if -fno-omit-frame-pointer is used without
20510 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20511 with ix86_finalize_stack_frame_flags.
20512 (ix86_expand_epilogue): Likewise.
20513 (ix86_expand_split_stack_prologue): Likewise.
20514 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
20516 2017-08-10 Martin Liska <mliska@suse.cz>
20519 * c-attribs.c (handle_target_attribute):
20520 Report warning for an empty string argument of target attribute.
20522 2017-08-09 Jakub Jelinek <jakub@redhat.com>
20525 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
20527 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
20528 or DECL_NONLOCAL labels.
20529 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
20530 or DECL_NONLOCAL labels here.
20532 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
20534 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
20535 to indicate when early gimple folding has been disabled.
20536 (rs6000_gimple_fold_builtin): Add debug content.
20537 (rs6000_invalid_builtin): Fix whitespace.
20538 (rs6000_expand_builtin): Fix whitespace.
20539 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
20541 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
20544 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
20545 SAVE_MULTIPLE if not all the registers that saves, should be saved.
20547 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
20549 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
20550 (qdf24xx): Likewise.
20551 * config/aarch64/aarch64.md: Include falkor.md.
20552 * config/aarch64/falkor.md: New.
20554 2017-08-09 Marek Polacek <polacek@redhat.com>
20557 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
20558 * diagnostic.c (emit_diagnostic): Add a comment.
20559 (emit_diagnostic_valist): New function.
20561 2017-08-09 Marek Polacek <polacek@redhat.com>
20564 * input.c (make_location): New overload.
20565 * input.h (make_location): Declare.
20567 2017-08-08 Alan Modra <amodra@gmail.com>
20568 H.J. Lu <hongjiu.lu@intel.com>
20571 * gcc.c (NO_PIE_SPEC): Delete.
20572 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
20574 (LINK_PIE_SPEC): ..to here.
20575 (LINK_COMMAND_SPEC): Support -no-pie.
20576 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
20577 chain of crtbegin*.o selection, update for PIE_SPEC changes and
20579 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
20580 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
20581 (ENDFILE_CRTEND_SPEC): Similarly.
20583 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
20586 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
20587 (mstack-protector-guard-offset=): Ditto.
20588 * config/i386/i386.c (ix86_option_override): Handle
20589 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
20591 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
20592 ix86_stack_protect_guard_offset variables.
20593 (TARGET_STACK_PROTECT_GUARD): Always define.
20594 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
20595 and -mstack-protector-guard-offset= options.
20597 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20599 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
20600 boundary case for the last candidate.
20602 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20604 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
20605 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
20607 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20609 PR middle-end/19706
20610 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
20611 * config/aarch64/aarch64-builtins.c
20612 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
20613 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
20614 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
20616 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20617 Andrew Pinski <pinskia@gmail.com>
20619 PR middle-end/19706
20620 * internal-fn.def (XORSIGN): New.
20621 * optabs.def (xorsign_optab): New.
20622 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
20623 (convert_expand_mult_copysign): New.
20624 (pass_optimize_widening_mul::execute): Call
20625 convert_expand_mult_copysign.
20627 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20629 PR tree-optimization/81354
20630 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
20631 Insert on edges rather than explicitly creating landing pads.
20632 (analyze_candidates_and_replace): Commit edge inserts.
20634 2017-08-08 Richard Biener <rguenther@suse.de>
20636 PR middle-end/81719
20637 * tree-ssa-loop-niter.c: Include tree-dfa.h.
20638 (expand_simple_operations): Also look through ADDR_EXPRs with
20639 MEM_REF bases treating them as POINTER_PLUS_EXPR.
20641 2017-08-08 Richard Biener <rguenther@suse.de>
20643 PR tree-optimization/81723
20644 * tree-vect-slp.c (struct bst_traits): New hash traits.
20645 (bst_fail): New global.
20646 (vect_build_slp_tree_2): New worker, split out from ...
20647 (vect_build_slp_tree): ... this now wrapping it with using
20648 bst_fail set to cache SLP tree build fails. Properly handle
20650 (vect_analyze_slp_instance): Allocate and free bst_fail.
20652 2017-08-08 Martin Liska <mliska@suse.cz>
20655 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
20656 LABEL_DECLs that can be from a different function.
20658 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20660 PR tree-optimization/81744
20661 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
20662 loop's number of iterations.
20664 2017-08-08 Martin Liska <mliska@suse.cz>
20666 * asan.c: Include header files.
20667 * attribs.c (build_decl_attribute_variant): New function moved
20669 (build_type_attribute_qual_variant): Likewise.
20670 (cmp_attrib_identifiers): Likewise.
20671 (simple_cst_list_equal): Likewise.
20672 (omp_declare_simd_clauses_equal): Likewise.
20673 (attribute_value_equal): Likewise.
20674 (comp_type_attributes): Likewise.
20675 (build_type_attribute_variant): Likewise.
20676 (lookup_ident_attribute): Likewise.
20677 (remove_attribute): Likewise.
20678 (merge_attributes): Likewise.
20679 (merge_type_attributes): Likewise.
20680 (merge_decl_attributes): Likewise.
20681 (merge_dllimport_decl_attributes): Likewise.
20682 (handle_dll_attribute): Likewise.
20683 (attribute_list_equal): Likewise.
20684 (attribute_list_contained): Likewise.
20685 * attribs.h (lookup_attribute): New function moved from tree.[ch].
20686 (lookup_attribute_by_prefix): Likewise.
20687 * bb-reorder.c: Include header files.
20688 * builtins.c: Likewise.
20689 * calls.c: Likewise.
20690 * cfgexpand.c: Likewise.
20691 * cgraph.c: Likewise.
20692 * cgraphunit.c: Likewise.
20693 * convert.c: Likewise.
20694 * dwarf2out.c: Likewise.
20695 * final.c: Likewise.
20696 * fold-const.c: Likewise.
20697 * function.c: Likewise.
20698 * gimple-expr.c: Likewise.
20699 * gimple-fold.c: Likewise.
20700 * gimple-pretty-print.c: Likewise.
20701 * gimple.c: Likewise.
20702 * gimplify.c: Likewise.
20703 * hsa-common.c: Likewise.
20704 * hsa-gen.c: Likewise.
20705 * internal-fn.c: Likewise.
20706 * ipa-chkp.c: Likewise.
20707 * ipa-cp.c: Likewise.
20708 * ipa-devirt.c: Likewise.
20709 * ipa-fnsummary.c: Likewise.
20710 * ipa-inline.c: Likewise.
20711 * ipa-visibility.c: Likewise.
20713 * lto-cgraph.c: Likewise.
20714 * omp-expand.c: Likewise.
20715 * omp-general.c: Likewise.
20716 * omp-low.c: Likewise.
20717 * omp-offload.c: Likewise.
20718 * omp-simd-clone.c: Likewise.
20719 * opts-global.c: Likewise.
20720 * passes.c: Likewise.
20721 * predict.c: Likewise.
20722 * sancov.c: Likewise.
20723 * sanopt.c: Likewise.
20724 * symtab.c: Likewise.
20725 * toplev.c: Likewise.
20726 * trans-mem.c: Likewise.
20727 * tree-chkp.c: Likewise.
20728 * tree-eh.c: Likewise.
20729 * tree-into-ssa.c: Likewise.
20730 * tree-object-size.c: Likewise.
20731 * tree-parloops.c: Likewise.
20732 * tree-profile.c: Likewise.
20733 * tree-ssa-ccp.c: Likewise.
20734 * tree-ssa-live.c: Likewise.
20735 * tree-ssa-loop.c: Likewise.
20736 * tree-ssa-sccvn.c: Likewise.
20737 * tree-ssa-structalias.c: Likewise.
20738 * tree-ssa.c: Likewise.
20739 * tree-streamer-in.c: Likewise.
20740 * tree-vectorizer.c: Likewise.
20741 * tree-vrp.c: Likewise.
20742 * tsan.c: Likewise.
20743 * ubsan.c: Likewise.
20744 * varasm.c: Likewise.
20745 * varpool.c: Likewise.
20746 * tree.c: Remove functions moved to attribs.[ch].
20747 * tree.h: Likewise.
20748 * config/aarch64/aarch64.c: Add attrs.h header file.
20749 * config/alpha/alpha.c: Likewise.
20750 * config/arc/arc.c: Likewise.
20751 * config/arm/arm.c: Likewise.
20752 * config/avr/avr.c: Likewise.
20753 * config/bfin/bfin.c: Likewise.
20754 * config/c6x/c6x.c: Likewise.
20755 * config/cr16/cr16.c: Likewise.
20756 * config/cris/cris.c: Likewise.
20757 * config/darwin.c: Likewise.
20758 * config/epiphany/epiphany.c: Likewise.
20759 * config/fr30/fr30.c: Likewise.
20760 * config/frv/frv.c: Likewise.
20761 * config/ft32/ft32.c: Likewise.
20762 * config/h8300/h8300.c: Likewise.
20763 * config/i386/winnt.c: Likewise.
20764 * config/ia64/ia64.c: Likewise.
20765 * config/iq2000/iq2000.c: Likewise.
20766 * config/lm32/lm32.c: Likewise.
20767 * config/m32c/m32c.c: Likewise.
20768 * config/m32r/m32r.c: Likewise.
20769 * config/m68k/m68k.c: Likewise.
20770 * config/mcore/mcore.c: Likewise.
20771 * config/microblaze/microblaze.c: Likewise.
20772 * config/mips/mips.c: Likewise.
20773 * config/mmix/mmix.c: Likewise.
20774 * config/mn10300/mn10300.c: Likewise.
20775 * config/moxie/moxie.c: Likewise.
20776 * config/msp430/msp430.c: Likewise.
20777 * config/nds32/nds32-isr.c: Likewise.
20778 * config/nds32/nds32.c: Likewise.
20779 * config/nios2/nios2.c: Likewise.
20780 * config/nvptx/nvptx.c: Likewise.
20781 * config/pa/pa.c: Likewise.
20782 * config/pdp11/pdp11.c: Likewise.
20783 * config/powerpcspe/powerpcspe.c: Likewise.
20784 * config/riscv/riscv.c: Likewise.
20785 * config/rl78/rl78.c: Likewise.
20786 * config/rx/rx.c: Likewise.
20787 * config/s390/s390.c: Likewise.
20788 * config/sh/sh.c: Likewise.
20789 * config/sol2.c: Likewise.
20790 * config/sparc/sparc.c: Likewise.
20791 * config/spu/spu.c: Likewise.
20792 * config/stormy16/stormy16.c: Likewise.
20793 * config/tilegx/tilegx.c: Likewise.
20794 * config/tilepro/tilepro.c: Likewise.
20795 * config/v850/v850.c: Likewise.
20796 * config/vax/vax.c: Likewise.
20797 * config/visium/visium.c: Likewise.
20798 * config/xtensa/xtensa.c: Likewise.
20800 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20803 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
20804 constraints since the -mupper-regs-* switches have been
20806 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
20807 into a vector from a double word element that was extracted from
20808 another vector, and eliminate extra XXPERMDI instructions.
20809 (vsx_concat_<mode>_2): Likewise.
20810 (vsx_concat_<mode>_3): Likewise.
20811 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
20812 concat to allow optimizing inserts from previous extracts.
20814 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
20816 * config/i386/i386.c (ix86_stack_protect_guard): Generate
20817 memory reference to a SSP offset in TLS address space.
20818 (ix86_print_operand) <case '@'>: Remove.
20819 (ix86_print_operand_punct_valid_p): Remove '@' code.
20820 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
20821 UNSPEC_SP_TLS_TEST.
20822 (stack_tls_protect_set_<mode>): Remove.
20823 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
20824 (stack_tls_protect_test_<mode>): Remove.
20825 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
20827 2017-08-07 Olivier Hainque <hainque@adacore.com>
20830 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
20832 2017-08-07 Douglas Rupp <rupp@adacore.com>
20834 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
20835 variable was referenced as multidir in command.
20837 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20840 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
20843 2017-08-07 Martin Liska <mliska@suse.cz>
20845 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
20846 * config/rl78/rl78.c: Add include of attribs.h.
20847 * config/sh/sh.c: Likewise.
20848 * config/v850/v850.c: Likewise.
20850 2017-08-07 Tom de Vries <tom@codesourcery.com>
20852 PR middle-end/78266
20853 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
20855 2017-08-07 Martin Liska <mliska@suse.cz>
20857 * config/mips/mips.c: Include attribs.h.
20859 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
20862 * doc/invoke.texi: Document change in behvaior for -Ofast for
20865 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
20867 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
20869 (aarch64_pop_regs): Likewise.
20870 (aarch64_gen_load_pair): Likewise.
20871 (aarch64_save_callee_saves): Likewise.
20872 (aarch64_restore_callee_saves): Likewise.
20874 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20876 * config/i386/i386.c: Revert the last change.
20878 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20881 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20883 (ix86_finalize_stack_frame_flags): This. Also clear
20884 frame_pointer_needed if -fno-omit-frame-pointer is used without
20886 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20887 with ix86_finalize_stack_frame_flags.
20888 (ix86_expand_epilogue): Likewise.
20889 (ix86_expand_split_stack_prologue): Likewise.
20891 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20894 * config/i386/i386.c (get_builtin_code_for_version): Set priority
20895 to P_AES for Westmere.
20897 2017-08-07 Jonathan Yong <10walls@gmail.com>
20899 * config/i386/mingw.opt (fset-stack-executable): Removed.
20900 * config/i386/cygming.opt (fset-stack-executable): Moved
20902 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
20904 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
20906 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
20908 2017-08-07 Marek Polacek <polacek@redhat.com>
20910 PR middle-end/81737
20911 * fold-const.c (fold_indirect_ref_1): Check type_domain.
20913 2017-08-07 Martin Liska <mliska@suse.cz>
20915 * attribs.h (canonicalize_attr_name): New function.
20916 (cmp_attribs): Move from c-format.c and adjusted.
20917 (is_attribute_p): Moved from tree.h.
20918 * tree-inline.c: Add new includes.
20919 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
20920 (private_is_attribute_p): Remove.
20921 (private_lookup_attribute): Likewise.
20922 (private_lookup_attribute_by_prefix): Simplify.
20923 (remove_attribute): Use is_attribute_p.
20924 * tree.h: Remove removed declarations.
20926 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20928 PR middle-end/81698
20929 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
20930 instead of computing it in the function. Formatting fix.
20931 (expand_case): Don't rely on default_edge being the first edge,
20932 clear it if removing it, pass default_edge to
20933 emit_case_dispatch_table.
20934 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
20937 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
20939 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
20940 insn in the function, emit NOP after the insn.
20942 2017-08-06 Tom de Vries <tom@codesourcery.com>
20944 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
20947 2017-08-06 Tom de Vries <tom@codesourcery.com>
20949 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
20952 2017-08-04 Yury Gribov <tetra2005@gmail.com>
20954 PR tree-optimization/57371
20955 * match.pd: New pattern.
20957 2017-08-04 Marek Polacek <polacek@redhat.com>
20959 PR middle-end/81695
20960 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
20961 perform the computation in offset_int.
20963 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20965 PR tree-optimization/81136
20966 * tree-vectorizer.h: Include tree-hash-traits.h.
20967 (vec_base_alignments): New typedef.
20968 (vec_info): Add a base_alignments field.
20969 (vect_record_base_alignments): Declare.
20970 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
20972 (DR_IS_CONDITIONAL_IN_STMT): New macro.
20973 (create_data_ref): Add an is_conditional_in_stmt argument.
20974 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
20975 the is_conditional_in_stmt field.
20976 (data_ref_loc): Add an is_conditional_in_stmt field.
20977 (get_references_in_stmt): Set the is_conditional_in_stmt field.
20978 (find_data_references_in_stmt): Update call to create_data_ref.
20979 (graphite_find_data_references_in_stmt): Likewise.
20980 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
20981 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
20982 (vect_record_base_alignment): New function.
20983 (vect_record_base_alignments): Likewise.
20984 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
20985 for nested statements even if we fail to compute a misalignment.
20986 Use pooled base alignments for unconditional references.
20987 (vect_find_same_alignment_drs): Compare base addresses instead
20989 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
20990 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
20992 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20994 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
20995 Add an explicit name for the enum. Use auto_vec for slp_instances
20996 and grouped_stores.
20997 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
20999 (_bb_vec_info): Add a constructor and destructor.
21000 (vinfo_for_stmt): Return NULL for uids of -1 as well.
21001 (destroy_loop_vec_info): Delete.
21002 (vect_destroy_datarefs): Likewise.
21003 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
21004 (vec_info::vec_info): New function.
21005 (vec_info::~vec_info): Likewise.
21006 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
21007 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
21008 destroy_loop_vec_info.
21009 * tree-vect-loop.c (new_loop_vec_info): Replace with...
21010 (_loop_vec_info::_loop_vec_info): ...this.
21011 (destroy_loop_vec_info): Replace with...
21012 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
21013 the stmt_vec_infos. Leave handling of vec_info information to its
21014 destructor. Remove explicit vector releases.
21015 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
21016 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
21017 * tree-vect-slp.c (new_bb_vec_info): Replace with...
21018 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
21019 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
21020 (destroy_bb_vec_info): Replace with...
21021 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
21022 information to its destructor.
21023 (vect_slp_analyze_bb_1): Use new and delete instead of
21024 new_bb_vec_info and destroy_bb_vec_info.
21025 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
21028 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21030 * tree-data-ref.h (subscript): Add access_fn field.
21031 (data_dependence_relation): Add could_be_independent_p.
21032 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
21033 (same_access_functions): Move to tree-data-ref.c.
21034 * tree-data-ref.c (ref_contains_union_access_p): New function.
21035 (access_fn_component_p): Likewise.
21036 (access_fn_components_comparable_p): Likewise.
21037 (dr_analyze_indices): Add a reference to access_fn_component_p.
21038 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
21040 (constant_access_functions): Likewise.
21041 (add_other_self_distances): Likewise.
21042 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
21043 (initialize_data_dependence_relation): Use XCNEW and remove
21044 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
21045 of access functions that have the same type. Allow the
21046 subsequence to end with different bases in some circumstances.
21047 Record the chosen access functions in SUB_ACCESS_FN.
21048 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
21049 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
21050 (subscript_dependence_tester_1): Likewise dra and drb.
21051 (build_classic_dist_vector): Update calls accordingly.
21052 (subscript_dependence_tester): Likewise.
21053 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
21054 DDR_COULD_BE_INDEPENDENT_P.
21055 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
21056 comp_alias_ddrs instead of may_alias_ddrs.
21057 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
21059 (vect_analyze_data_ref_dependence): Use it if
21060 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
21061 distance vectors if that fails.
21062 (dependence_distance_ge_vf): New function.
21063 (vect_prune_runtime_alias_test_list): Use it. Don't clear
21064 LOOP_VINFO_MAY_ALIAS_DDRS.
21066 2017-08-04 Richard Biener <rguenther@suse.de>
21068 PR middle-end/81705
21069 * fold-const.c (fold_binary_loc): Properly restrict
21070 minus_var0 && minus_var1 case when associating undefined overflow
21073 2017-08-04 Tom de Vries <tom@codesourcery.com>
21075 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
21077 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21079 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21080 Don't start diagnostic messages with a capital letter.
21081 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21083 (rs6000_invalid_builtin): Likewise.
21084 (rs6000_trampoline_init): Likewise.
21086 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21089 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
21090 after setting changeable df flags.
21092 2017-08-03 Richard Biener <rguenther@suse.de>
21094 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
21095 up if the use is in USE - X.
21097 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
21099 * toplev.c (dumpfile.h): New include.
21100 (internal_error_reentered): New static function. Use it...
21101 (internal_error_function): ...here to handle reentered internal_error.
21103 2017-08-03 Richard Biener <rguenther@suse.de>
21105 PR middle-end/81148
21106 * fold-const.c (split_tree): Add minus_var and minus_con
21107 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
21108 here but always use minus_*.
21109 (associate_trees): Assert we never associate with MINUS_EXPR
21110 and NULL first operand. Do not recurse for PLUS_EXPR operands
21111 when associating as MINUS_EXPR either.
21112 (fold_binary_loc): Track minus_var and minus_con.
21114 2017-08-03 Tom de Vries <tom@codesourcery.com>
21117 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
21118 ACCEL_COMPILER, apply finish_options on
21119 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
21121 2017-08-03 Tom de Vries <tom@codesourcery.com>
21124 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
21125 function_entry_patch_area_size > 0.
21127 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21130 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
21131 instead of 10??LU, perform unit multiplication in wide_int,
21132 don't change alloc_object_size_limit if the limit is larger
21135 PR tree-optimization/81655
21136 PR tree-optimization/81588
21137 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
21138 the case when ranges[i].low and high are 1 for unsigned type with
21141 PR middle-end/81052
21142 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
21143 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
21145 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21147 * tree-vrp.h: Add include guard.
21149 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
21152 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
21153 (ud2): New insn pattern.
21154 * config/i386/i386.c (ix86_expand_epilogue):
21155 For naked functions, generate ud2 instead of trap insn.
21157 2017-08-02 Marek Polacek <polacek@redhat.com>
21160 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
21162 2017-08-02 Tom de Vries <tom@codesourcery.com>
21163 Cesar Philippidis <cesar@codesourcery.com>
21165 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
21166 Add missing edge probabilities.
21168 2017-08-02 Tamar Christina <tamar.christina@arm.com>
21170 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
21171 Correct endianness.
21173 2017-08-02 Jakub Jelinek <jakub@redhat.com>
21175 PR middle-end/79499
21176 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
21177 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
21178 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
21179 prologue_seq sequences - if any.
21181 2017-08-02 Richard Biener <rguenther@suse.de>
21183 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
21184 via vectors if supported, integer extracts via punning if supported
21185 or otherwise vector extracts.
21187 2017-08-02 Richard Biener <rguenther@suse.de>
21189 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
21191 (bitmap_insert_into_set): ... this.
21193 2017-08-02 Richard Biener <rguenther@suse.de>
21195 PR tree-optimization/81633
21197 2015-08-17 Alan Hayward <alan.hayward@arm.com>
21199 PR tree-optimization/71752
21200 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
21202 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
21204 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
21205 (machine_function::call_ms2sysv_pad_out): Remove field.
21206 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
21207 (ix86_compute_frame_layout): Likewise.
21209 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
21212 * config/i386/i386.c (ix86_set_func_type): Disallow naked
21213 attribute with interrupt attribute.
21215 2017-08-01 Andrew Pinski <apinski@cavium.com>
21217 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
21218 BIT_INSERT_EXPR's operand 1
21219 to see if the types precision matches.
21221 2017-08-01 Martin Liska <mliska@suse.cz>
21223 PR middle-end/70140
21224 * builtins.c (expand_builtin_memcpy_args): Remove.
21225 (expand_builtin_memcpy): Call newly added function
21226 expand_builtin_memory_copy_args.
21227 (expand_builtin_memcpy_with_bounds): Likewise.
21228 (expand_builtin_mempcpy): Remove last argument.
21229 (expand_builtin_mempcpy_with_bounds): Likewise.
21230 (expand_builtin_memory_copy_args): New function created from
21231 expand_builtin_mempcpy_args with small modifications.
21232 (expand_builtin_mempcpy_args): Remove.
21233 (expand_builtin_stpcpy): Remove unused argument.
21234 (expand_builtin): Likewise.
21235 (expand_builtin_with_bounds): Likewise.
21237 2017-08-01 Martin Liska <mliska@suse.cz>
21240 Make mempcpy more optimal (PR middle-end/70140).
21242 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21245 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
21246 __builtin_vec_cmpne verify both arguments are compatible vectors
21247 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
21248 verify arg1_type is a pointer or array type. For __builtin_vec_st,
21249 move computation of aligned to after checking the argument types.
21253 * config/rs6000/vsx.md (vextract_fp_from_shorth,
21254 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
21257 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
21258 Doug Rupp <rupp@adacore.com>
21259 Olivier Hainque <hainque@adacore.com>
21261 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
21262 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
21264 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
21265 for TARGET_OS_CPP_BUILTIN.
21266 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
21267 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
21269 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
21270 passing required abi options to the assembler for EABI configurations.
21271 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
21272 of .text.hot and .text.unlikely sections for kernel modules when
21273 using ARM style exceptions.
21274 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
21275 options. Add EXTRA_CC1_SPEC.
21276 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
21278 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
21280 (ARM_TARGET2_DWARF_FORMAT): Define.
21281 * config/arm/t-vxworks: Adjust multilib control to removal of the
21282 Diab command line options.
21284 2017-08-01 Martin Liska <mliska@suse.cz>
21286 PR gcov-profile/81561
21287 * gcov.c (unblock): Make unblocking safe as we need to preserve
21288 index correspondence of blocks and block_lists.
21290 2017-08-01 Richard Biener <rguenther@suse.de>
21292 PR tree-optimization/81181
21293 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
21294 (compute_antic): ... end of iteration here.
21296 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
21298 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
21299 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
21300 (ftree-slp-vectorize): Likewise.
21301 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
21302 can no longer be set independent of flag_tree_loop_vectorize.
21303 * omp-general.c (emp_max_vf): Likewise.
21304 * opts.c (enable_fdo_optimizations): Remove references to
21305 flag_tree_vectorize, these are now implicit.
21306 (common_handle_option): Remove handling for OPT_ftree_vectorize,
21307 and leave it for the options machinery.
21309 2017-08-01 Martin Liska <mliska@suse.cz>
21311 PR middle-end/70140
21312 * builtins.c (expand_builtin_memcpy_args): Remove.
21313 (expand_builtin_memcpy): Call newly added function
21314 expand_builtin_memory_copy_args.
21315 (expand_builtin_memcpy_with_bounds): Likewise.
21316 (expand_builtin_mempcpy): Remove last argument.
21317 (expand_builtin_mempcpy_with_bounds): Likewise.
21318 (expand_builtin_memory_copy_args): New function created from
21319 expand_builtin_mempcpy_args with small modifications.
21320 (expand_builtin_mempcpy_args): Remove.
21321 (expand_builtin_stpcpy): Remove unused argument.
21322 (expand_builtin): Likewise.
21323 (expand_builtin_with_bounds): Likewise.
21325 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21328 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
21329 print "ds:" only for immediates in generic address space.
21331 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21334 * config/i386/i386.c (ix86_funciton_naked): New prototype.
21335 (ix86_function_ok_for_sibcall): Return false for naked functions.
21337 2017-08-01 Richard Biener <rguenther@suse.de>
21339 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
21340 (compute_antic): Seed worklist with exit block predecessors.
21341 * cfganal.c (dfs_find_deadend): For a cycle return the source
21342 of the edge closing it.
21344 2017-08-01 Tamar Christina <tamar.christina@arm.com>
21346 * config/aarch64/aarch64.c
21347 (aarch64_can_const_movi_rtx_p): Move 0 check.
21349 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21351 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
21352 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
21354 * match.pd: Ditto in address comparison pattern.
21356 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21358 PR tree-optimization/81627
21359 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
21360 closed ssa form for store-store chain.
21362 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21364 PR tree-optimization/81620
21365 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
21366 for store-store chain.
21368 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21370 PR tree-optimization/81588
21371 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
21372 ranges[i].in_p, invert comparison code ccode. For >/>=,
21373 swap rhs1 and rhs2 and comparison code unconditionally,
21374 for </<= don't do that. Don't swap rhs1/rhs2 again if
21375 ranges[i].in_p, instead invert comparison code ccode if
21376 opcode or oe->rank is BIT_IOR_EXPR.
21379 * optabs.def (vec_extract_optab, vec_init_optab): Change from
21380 a direct optab to conversion optab.
21381 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
21382 with GET_MODE_INNER as last argument instead of optab_handler.
21383 * expmed.c (extract_bit_field_1): Likewise. Use vector from
21384 vector extraction if possible and optab is available.
21385 * expr.c (store_constructor): Use convert_optab_handler instead
21386 of optab_handler. Use vector initialization from smaller
21387 vectors if possible and optab is available.
21388 * tree-vect-stmts.c (vectorizable_load): Likewise.
21389 * doc/md.texi (vec_extract, vec_init): Document that the optabs
21390 now have two modes.
21391 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
21392 of vec_init from half-sized vectors with the same element mode.
21393 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
21394 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
21395 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
21396 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
21397 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
21398 after mode in gen_vec_extract* calls.
21399 (vec_extract<mode>): Renamed to ...
21400 (vec_extract<mode><ssescalarmodelower>): ... this.
21401 (vec_extract<mode><ssehalfvecmodelower>): New expander.
21402 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
21403 element mode after mode in gen_vec_init* calls.
21404 (VEC_INIT_HALF_MODE): New mode iterator.
21405 (vec_init<mode>): Renamed to ...
21406 (vec_init<mode><ssescalarmodelower>): ... this.
21407 (vec_init<mode><ssehalfvecmodelower>): New expander.
21408 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
21409 (vec_extractv2sfsf): ... this.
21410 (vec_initv2sf): Renamed to ...
21411 (vec_initv2sfsf): ... this.
21412 (vec_extractv2si): Renamed to ...
21413 (vec_extractv2sisi): ... this.
21414 (vec_initv2si): Renamed to ...
21415 (vec_initv2sisi): ... this.
21416 (vec_extractv4hi): Renamed to ...
21417 (vec_extractv4hihi): ... this.
21418 (vec_initv4hi): Renamed to ...
21419 (vec_initv4hihi): ... this.
21420 (vec_extractv8qi): Renamed to ...
21421 (vec_extractv8qiqi): ... this.
21422 (vec_initv8qi): Renamed to ...
21423 (vec_initv8qiqi): ... this.
21424 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
21425 (vec_init<mode>): Renamed to ...
21426 (vec_init<mode><VEC_base_l>): ... this.
21427 (vec_extract<mode>): Renamed to ...
21428 (vec_extract<mode><VEC_base_l>): ... this.
21429 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
21430 (vec_initv2sfsf): ... this.
21431 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
21432 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21433 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
21434 element mode after mode in gen_vec_init* calls.
21435 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
21436 (vec_init<mode><Vel>): ... this.
21437 (vec_extract<mode>): Renamed to ...
21438 (vec_extract<mode><Vel>): ... this.
21439 * config/aarch64/iterators.md (Vel): New mode attribute.
21440 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
21441 Add element mode after mode in gen_vec_extract* calls.
21442 * config/s390/vector.md (non_vec_l): New mode attribute.
21443 (vec_extract<mode>): Renamed to ...
21444 (vec_extract<mode><non_vec_l>): ... this.
21445 (vec_init<mode>): Renamed to ...
21446 (vec_init<mode><non_vec_l>): ... this.
21447 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
21448 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
21450 * config/arm/iterators.md (V_elem_l): New mode attribute.
21451 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
21452 (vec_extract<mode><V_elem_l>): ... this.
21453 (vec_extractv2di): Renamed to ...
21454 (vec_extractv2didi): ... this.
21455 (vec_init<mode>): Renamed to ...
21456 (vec_init<mode><V_elem_l>): ... this.
21457 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
21458 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
21459 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
21460 Add element mode after gen_vec_extract* calls.
21461 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
21462 (vec_init<mode><unitmode>): ... this.
21463 (vec_extract<mode>): Renamed to ...
21464 (vec_extract<mode><unitmode>): ... this.
21465 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
21466 (vec_init<mode><unitmode>): ... this.
21467 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
21468 (vec_initv2sfsf): ... this.
21469 (vec_extractv2sf): Renamed to ...
21470 (vec_extractv2sfsf): ... this.
21471 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
21472 Add element mode after gen_vec_extract* calls.
21473 * config/mips/mips.md (unitmode): New mode iterator.
21474 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
21475 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
21476 * config/spu/spu.md (inner_l): New mode attribute.
21477 (vec_init<mode>): Renamed to ...
21478 (vec_init<mode><inner_l>): ... this.
21479 (vec_extract<mode>): Renamed to ...
21480 (vec_extract<mode><inner_l>): ... this.
21481 * config/sparc/sparc.md (veltmode): New mode iterator.
21482 (vec_init<VMALL:mode>): Renamed to ...
21483 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
21484 * config/ia64/vect.md (vec_initv2si): Renamed to ...
21485 (vec_initv2sisi): ... this.
21486 (vec_initv2sf): Renamed to ...
21487 (vec_initv2sfsf): ... this.
21488 (vec_extractv2sf): Renamed to ...
21489 (vec_extractv2sfsf): ... this.
21490 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
21491 (vec_init<mode>): Renamed to ...
21492 (vec_init<mode><VEC_base_l>): ... this.
21493 (vec_extract<mode>): Renamed to ...
21494 (vec_extract<mode><VEC_base_l>): ... this.
21495 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
21496 (vec_initv2sfsf): ... this.
21497 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
21498 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21499 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
21500 gen_vec_init* calls.
21502 2017-08-01 Richard Biener <rguenther@suse.de>
21504 PR tree-optimization/81297
21505 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
21506 TREE_OVERFLOW from INTEGER_CSTs.
21508 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
21510 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
21512 2017-07-31 Carl Love <cel@us.ibm.com>
21514 * config/rs6000/rs6000-c: Add support for built-in functions
21515 vector signed char vec_xl_be (signed long long, signed char *);
21516 vector unsigned char vec_xl_be (signed long long, unsigned char *);
21517 vector signed int vec_xl_be (signed long long, signed int *);
21518 vector unsigned int vec_xl_be (signed long long, unsigned int *);
21519 vector signed long long vec_xl_be (signed long long, signed long long *);
21520 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
21521 vector signed short vec_xl_be (signed long long, signed short *);
21522 vector unsigned short vec_xl_be (signed long long, unsigned short *);
21523 vector double vec_xl_be (signed long long, double *);
21524 vector float vec_xl_be (signed long long, float *);
21525 * config/rs6000/altivec.h (vec_xl_be): Add #define.
21526 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
21527 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
21529 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
21530 (altivec_expand_builtin): Add switch statement to call
21531 altivec_expand_xl_be for each builtin.
21532 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
21533 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
21534 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
21535 __builtin_vsx_le_be_v16qi.
21536 * doc/extend.texi: Update the built-in documentation file for the
21537 new built-in functions.
21539 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21542 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
21544 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21546 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21548 * config.gcc: Add z14.
21549 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
21550 CPU model numbers for z13s and z14.
21551 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
21553 * config/s390/s390-opts.h (enum processor_type): Rename
21554 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21555 * config/s390/s390.c (processor_table): Add field for CPU name to
21556 be passed to Binutils.
21557 (s390_asm_output_machine_for_arch): Use the new field in
21558 processor_table for Binutils.
21559 (s390_expand_builtin): Replace arch12 with z14.
21560 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21561 (s390_get_sched_attrmask): Likewise.
21562 (s390_get_unit_mask): Likewise.
21563 * config/s390/s390.opt: Add z14 to processor_type enum.
21565 2017-07-31 Martin Jambor <mjambor@suse.cz>
21568 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
21569 regardless of optimization level.
21571 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
21572 Martin Liska <mliska@suse.cz>
21574 * predict.def: Remove old comment and adjust probability.
21575 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
21576 PREDICT statements.
21578 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21581 * config/i386/i386.c (ix86_function_naked): New function.
21582 (ix86_can_use_return_insn_p): Return false for naked functions.
21583 (ix86_expand_prologue): Skip prologue for naked functions.
21584 (ix86_expand_epilogue): Skip epilogue for naked functions
21585 and emit trap instruction.
21586 (ix86_warn_func_return): New function.
21587 (ix86_attribute_table): Add "naked" attribute specification.
21588 (TARGET_WARN_FUNC_RETURN): Define.
21589 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
21591 2017-07-31 Martin Liska <mliska@suse.cz>
21593 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
21594 (dump_gimple_bb_header): Always dump BB info.
21595 (pp_cfg_jump): Do not append info about BB when dumping a jump.
21597 2017-07-31 Martin Liska <mliska@suse.cz>
21600 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
21601 also with current_function_decl non-null equality.
21603 2017-07-31 Jakub Jelinek <jakub@redhat.com>
21606 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
21607 change type to the element type, instead add eltype variable and
21608 use it where we are interested in the element type.
21610 PR tree-optimization/81603
21611 * ipa-polymorphic-call.c
21612 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
21613 offset arithmetic in offset_int, bail out if the resulting bit offset
21614 doesn't fit into shwi.
21616 2017-07-31 Martin Liska <mliska@suse.cz>
21618 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
21619 (gimplify_save_expr): Fix comment.
21621 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
21624 * config/i386/i386.c (ix86_function_arg): Update arguments for
21626 (ix86_compute_frame_layout): Set the initial stack offset to
21627 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
21628 INCOMING_FRAME_SP_OFFSET.
21629 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
21630 stack before exception handler returns.
21631 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
21632 the 'ERROR_CODE' for exception handler.
21634 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
21636 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
21637 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
21638 (ASM_OUTPUT_REG_POP): Ditto.
21639 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
21640 instead of asm_fprintf to output pure string.
21642 2017-07-29 Jakub Jelinek <jakub@redhat.com>
21644 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
21645 to imported_module_or_decl hook.
21646 (debug_nothing_tree_tree_tree_bool): Remove.
21647 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
21648 * debug.c (do_nothing_debug_hooks): Use
21649 debug_nothing_tree_tree_tree_bool_bool instead of
21650 debug_nothing_tree_tree_tree_bool.
21651 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21652 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
21653 * sdbout.c (sdb_debug_hooks): Likewise.
21654 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
21655 (gen_namespace_die): Add DW_AT_export_symbols attribute if
21657 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
21658 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
21659 attribute, don't add anything.
21661 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21663 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
21664 (fold_build2_stat_loc): Likewise.
21665 (fold_build3_stat_loc): Likewise.
21666 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
21667 (fold_build1_loc): Remove macro.
21668 (fold_build2_loc): Likewise.
21669 (fold_build3_loc): Likewise.
21671 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21673 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
21674 (gimple_build_debug_bind_source_stat): Likewise.
21675 * gimple.h (gimple_build_debug_bind): Remove macro.
21676 (gimple_build_debug_bind_source): Likewise.
21678 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21680 * bitmap.c (bitmap_alloc): Adjust.
21681 (bitmap_gc_alloc): Likewise.
21682 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
21684 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21686 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
21687 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
21688 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
21689 (bitmap_gc_alloc_stat): Likewise.
21690 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
21692 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21694 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
21695 * rtl.h (shallow_copy_rtx): Remove macro.
21697 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21699 * emit-rtl.c (gen_raw_REG): Adjust.
21700 * gengenrtl.c (gendef): Likewise.
21701 * rtl.c (rtx_alloc_stat): Remove _stat from name.
21702 * rtl.h (rtx_alloc): Remove macro.
21704 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21706 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
21707 (build_tree_list_stat): Likewise.
21708 * tree.h (build_tree_list): Remove macro.
21709 (build_tree_list_vec): Likewise.
21711 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21713 * tree.c (make_vector_stat): Remove _stat from name.
21714 (build_vector_stat): Likewise.
21715 * tree.h (make_vector_stat): Remove macro.
21716 (build_vector_stat): Likewise.
21718 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21720 * tree.h (build_var_debug_value): Remove prototype.
21722 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21724 * tree.c (tree_cons_stat): Remove _stat from name.
21725 * tree.h (tree_cons): Remove macro.
21727 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21729 * tree.c (build_vl_exp_stat): Remove _stat from name.
21730 * tree.h (build_vl_exp): Remove macro.
21732 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21734 * tree.c (build_decl_stat): Remove _stat from name.
21735 * tree.h (build_decl): Remove macro.
21737 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21739 * gimple.c (gimple_build_with_ops_stat): Adjust.
21740 (gimple_alloc_stat): Remove _stat from name.
21741 * gimple.h (gimple_alloc): Remove macro.
21743 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21745 * tree.c (make_tree_vec_stat): Remove _stat from name.
21746 (grow_tree_vec_stat): Likewise.
21747 * tree.h (make_tree_vec_stat): Adjust prototype.
21748 (grow_tree_vec_stat): Likewise.
21749 (make_tree_vec): Remove macro.
21750 (grow_tree_vec): Likewise.
21752 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21754 * fold-const.c (fold_build1_stat_loc): Adjust.
21755 (fold_build2_stat_loc): Likewise.
21756 (fold_build3_stat_loc): Likewise.
21757 * tree.c (build0_stat): Remove _stat from name.
21758 (build1_stat): Likewise.
21759 (build2_stat): Likewise.
21760 (build3_stat): Likewise.
21761 (build4_stat): Likewise.
21762 (build5_stat): Likewise.
21763 * tree.h (build1_loc): Remove macro, and rename _stat function
21765 (build2_loc): Likewise.
21766 (build3_loc): Likewise.
21767 (build4_loc): Likewise.
21768 (build5_loc): Likewise.
21770 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21772 * tree.c (make_int_cst_stat): Remove _stat from name.
21773 * tree.h (make_int_cst_stat): Adjust prototype.
21774 (make_int_cst): Remove macro.
21776 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21778 * tree.c (make_tre_binfo_stat): Remove _stat from name.
21779 * tree.h (make_tree_binfo_stat): Adjust prototype.
21780 (make_tree_binfo): Remove.
21782 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21784 * tree.c (copy_node_stat): Rename to copy_node.
21785 (build_distinct_type_copy): Adjust.
21786 * tree.h (copy_node_stat): Adjust prototype.
21787 (copy_node): Remove macro.
21789 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21791 * tree.c (make_node_stat): rename to make_node.
21792 (build_tree_list_stat): Adjust.
21793 (build0_stat): Likewise.
21794 (build2_stat): Likewise.
21795 (build3_stat): Likewise.
21796 (build4_stat): Likewise.
21797 (build5_stat): Likewise.
21798 (build_decl_stat): Likewise.
21799 * tree.h (make_node_stat): Adjust prototype.
21800 (make_node): remove macro.
21802 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
21804 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
21805 (PPC_FEATURE2_SCV): Likewise.
21806 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
21808 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21810 * config/aarch64/aarch64.c
21811 (aarch64_internal_mov_immediate): Add new special pattern.
21812 * config/aarch64/aarch64.md (*movdi_aarch64):
21813 Add reg/32bit const mov case.
21815 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21816 Richard Sandiford <richard.sandiford@linaro.org>
21818 * config/aarch64/aarch64.md (mov<mode>): Generalize.
21819 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21820 Add integer and movi cases.
21821 (movi-split-hf-df-sf split, fp16): New.
21822 (enabled): Added TARGET_FP_F16INST.
21823 * config/aarch64/iterators.md (GPF_HF): New.
21824 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
21826 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21828 * config/aarch64/aarch64.c
21829 (aarch64_simd_container_mode): Add prototype.
21830 (aarch64_expand_mov_immediate): Add HI support.
21831 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
21832 (aarch64_can_const_movi_rtx_p): New.
21833 (aarch64_preferred_reload_class):
21834 Remove restrictions of using FP registers for certain SIMD operations.
21835 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
21836 (aarch64_valid_floating_const): Add integer move validation.
21837 (aarch64_simd_imm_scalar_p): Remove.
21838 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
21839 (aarch64_legitimate_constant_p): Expand list of supported cases.
21840 * config/aarch64/aarch64-protos.h
21841 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
21842 (aarch64_reinterpret_float_as_int): New.
21843 (aarch64_simd_imm_scalar_p): Remove.
21844 * config/aarch64/constraints.md (Uvi): New.
21845 (Dd): Split into Ds and new Dd.
21846 * config/aarch64/aarch64.md (*movsi_aarch64):
21848 (*movdi_aarch64): Add SIMD mov case.
21850 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21852 * tree-predcom.c: (struct chain): Handle store-store chain in which
21853 stores for elimination only store loop invariant values.
21854 (execute_pred_commoning_chain): Ditto.
21855 (prepare_initializers_chain_store_elim): Ditto.
21856 (prepare_finalizers): Ditto.
21857 (is_inv_store_elimination_chain): New function.
21858 (initialize_root_vars_store_elim_1): New function.
21860 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21862 * tree-predcom.c: Revise general description of the pass.
21863 (enum chain_type): New enum type for store elimination.
21864 (struct chain): New field supporting store elimination.
21865 (struct component): Ditto.
21866 (dump_chain): Dump store-stores chain.
21867 (release_chain): Release resources.
21868 (split_data_refs_to_components): Compute and create component
21869 contains only stores for elimination.
21870 (get_chain_last_ref_at): New function.
21871 (make_invariant_chain): Initialization.
21872 (make_rooted_chain): Specify chain type in parameter and record it.
21873 (add_looparound_copies): Skip for store-stores chain.
21874 (determine_roots_comp): Compute type of chain and pass it to
21876 (initialize_root_vars_store_elim_2): New function.
21877 (finalize_eliminated_stores): New function.
21878 (remove_stmt): Handle store for elimination.
21879 (execute_pred_commoning_chain): Execute predictive commoning on
21880 store-store chains.
21881 (determine_unroll_factor): Skip unroll for store-stores chain.
21882 (prepare_initializers_chain_store_elim): New function.
21883 (prepare_initializers_chain): Hanlde store-store chain.
21884 (prepare_finalizers_chain, prepare_finalizers): New function.
21885 (tree_predictive_commoning_loop): Return integer value indicating
21886 if loop is unrolled or lcssa form is corrupted.
21887 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
21889 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21891 * tree-predcom.c (initialize_root): Delete.
21892 (execute_pred_commoning_chain): Initialize root vars and replace
21893 reference of non-combined chain directly, rather than call above
21896 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21898 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
21899 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
21901 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21903 * tree-predcom.c (struct chain): New field init_seq.
21904 (release_chain): Release init_seq.
21905 (prepare_initializers_chain): Record intialization stmts in above
21907 (insert_init_seqs): New function.
21908 (tree_predictive_commoning_loop): Call insert_init_seqs.
21910 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21912 * tree-predcom.c (determine_roots_comp): Skip trivial components.
21914 2017-07-28 Richard Biener <rguenther@suse.de>
21916 * match.pd: Remove superfluous :c.
21917 * genmatch.c (simplify::id): Add member.
21918 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
21920 (current_id): New global.
21921 (dt_node::parent): Move from ...
21922 (dt_operand::parent): ... here. Add for_id member.
21923 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
21924 (decision_tree::find_node): Relax order requirement when
21925 merging DT_TRUE nodes to ones inbetween the current simplify
21926 and the one we try to merge with. Add diagnostic whenever
21927 we need to enforce pattern order by not merging.
21928 (decision_tree::insert): Set current_id.
21929 (decision_tree::print_node): Dump parent node and for_id.
21930 (parser::last_id): Add member.
21931 (parser::push_simplify): Assign unique id.
21932 (parser::parser): Initialize last_id.
21934 2017-07-28 Martin Liska <mliska@suse.cz>
21937 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
21938 gimple_build_debug_bind.
21940 2017-07-28 Richard Biener <rguenther@suse.de>
21942 PR tree-optimization/81502
21943 * match.pd: Add pattern combining BIT_INSERT_EXPR with
21945 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
21947 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
21948 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
21949 for BIT_FIELD_REF args.
21950 * fold-const.c (make_bit_field_ref): Likewise.
21951 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
21953 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21956 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
21957 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
21958 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
21959 Or it into SANITIZER_UNDEFINED.
21960 * ubsan.c: Include gimple-fold.h and varasm.h.
21961 (ubsan_expand_ptr_ifn): New function.
21962 (instrument_pointer_overflow): New function.
21963 (maybe_instrument_pointer_overflow): New function.
21964 (instrument_object_size): Formatting fix.
21965 (pass_ubsan::execute): Call instrument_pointer_overflow
21966 and maybe_instrument_pointer_overflow.
21967 * internal-fn.c (expand_UBSAN_PTR): New function.
21968 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
21969 * sanitizer.def (__ubsan_handle_pointer_overflow,
21970 __ubsan_handle_pointer_overflow_abort): New builtins.
21971 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
21972 * internal-fn.def (UBSAN_PTR): New internal function.
21973 * opts.c (sanitizer_opts): Add pointer-overflow.
21974 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
21975 * fold-const.c (build_range_check): Compute pointer range check in
21976 integral type if pointer arithmetics would be needed. Formatting
21979 2017-07-28 Martin Liska <mliska@suse.cz>
21982 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
21983 parameters that are of a variable-length.
21985 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
21987 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
21989 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
21990 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
21991 (CRT_CALL_STATIC_FUNCTION): Likewise.
21992 (ASM_DEFAULT_SPEC): New define.
21993 (ASM_SPEC32): Likewise.
21994 (ASM_SPEC64): Likewise.
21995 (ASM_SPEC_COMMON): Likewise.
21996 (ASM_SPEC): Likewise.
21997 (INVALID_64BIT): Likewise.
21998 (LINK_OS_DEFAULT_SPEC): Likewise.
21999 (LINK_OS_SPEC32): Likewise.
22000 (LINK_OS_SPEC64): Likewise.
22001 (POWERPC_LINUX): Likewise.
22002 (PTRDIFF_TYPE): Likewise.
22003 (RESTORE_FP_PREFIX): Likewise.
22004 (RESTORE_FP_SUFFIX): Likewise.
22005 (SAVE_FP_PREFIX): Likewise.
22006 (SAVE_FP_SUFFIX): Likewise.
22007 (SIZE_TYPE): Likewise.
22008 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
22009 (TARGET_64BIT): Likewise.
22010 (TARGET_64BIT): Likewise.
22011 (TARGET_AIX): Likewise.
22012 (WCHAR_TYPE_SIZE): Likewise.
22013 (WCHAR_TYPE): Undefine.
22014 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
22015 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
22016 (CPP_OS_RTEMS_SPEC): Delete.
22017 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
22018 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
22020 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
22022 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22024 PR tree-optimization/81578
22025 * tree-parloops.c (build_new_reduction): Bail out if
22026 reduction_code isn't one of the standard OpenMP reductions.
22027 Move the details printing after that decision.
22029 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22031 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
22032 related to reload_in_progress.
22033 (splat_input_operand): Likewise.
22034 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
22036 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
22038 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
22039 (TARGET_INSTANTIATE_DECLS): Likewise.
22040 (legitimate_indexed_address_p): Delete reload_in_progress code.
22041 (rs6000_debug_legitimate_address_p): Likewise.
22042 (rs6000_eliminate_indexed_memrefs): Likewise.
22043 (rs6000_emit_le_vsx_store): Likewise.
22044 (rs6000_emit_move_si_sf_subreg): Likewise.
22045 (rs6000_emit_move): Likewise.
22046 (register_to_reg_type): Likewise.
22047 (rs6000_pre_atomic_barrier): Likewise.
22048 (rs6000_machopic_legitimize_pic_address): Likewise.
22049 (rs6000_allocate_stack_temp): Likewise.
22050 (rs6000_address_for_fpconvert): Likewise.
22051 (rs6000_address_for_altivec): Likewise.
22052 (rs6000_secondary_memory_needed_rtx): Delete function.
22053 (rs6000_check_sdmode): Likewise.
22054 (rs6000_alloc_sdmode_stack_slot): Likewise.
22055 (rs6000_instantiate_decls): Likewise.
22056 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
22057 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
22058 Delete reload_in_progress.
22059 (*vec_reload_and_plus_<mptrsize>): Likewise.
22060 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
22061 (vsx_div_v2di): Likewise.
22062 (vsx_udiv_v2di): Likewise.
22064 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22066 * config/rs6000/rs6000.opt (mlra): Replace with stub.
22067 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
22068 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
22069 (rs6000_debug_reg_global): Delete print of LRA status.
22070 (rs6000_option_override_internal): Delete dead LRA related code.
22071 (rs6000_lra_p): Delete function.
22072 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
22074 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
22076 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
22077 * config/riscv/rtems.h: New file.
22079 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22080 Sudakshina Das <sudi.das@arm.com>
22082 * config/aarch64/aarch64.md
22083 (define_split for and<mode>3nr_compare): Move
22084 non aarch64_logical_operand to a register.
22085 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
22086 register immediate operand to a register.
22087 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
22089 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22091 PR middle-end/81564
22092 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
22094 2017-07-27 Richard Biener <rguenther@suse.de>
22096 PR tree-optimization/81573
22097 PR tree-optimization/81494
22098 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
22099 multi defuse cycle case.
22101 2017-07-27 Richard Biener <rguenther@suse.de>
22103 PR tree-optimization/81571
22104 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
22107 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
22109 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
22110 earlier and only if MASK_FPU is set. Adjust formatting.
22112 2017-07-27 Martin Liska <mliska@suse.cz>
22114 * opt-functions.awk: Add validation of value of Init.
22115 * optc-gen.awk: Pass new argument.
22117 2017-07-27 Martin Liska <mliska@suse.cz>
22119 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
22120 Fix wrong condition.
22122 2017-07-27 Martin Liska <mliska@suse.cz>
22124 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
22125 BBs and edges seen by autoFDO.
22127 2017-07-27 Richard Biener <rguenther@suse.de>
22129 PR tree-optimization/81502
22130 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
22131 with incompatible but same sized type.
22132 (execute_update_addresses_taken): Likewise.
22134 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
22136 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
22137 flag_tree_loop_vectorize rather than flag_tree_vectorize.
22139 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22142 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
22143 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
22144 Change s_operand to memory_operand.
22146 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
22148 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
22149 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
22150 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
22151 Emit instructions rather than returning an expression. Handle TFmode
22152 and KFmode by casting to TImode.
22153 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
22154 (rs6000_emit_le_vsx_store): Likewise.
22155 * config/rs6000/vsx.md (VSX_TI): New iterator.
22156 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
22157 (*vsx_le_undo_permute_<mode>): Likewise.
22158 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
22159 emit the split sequence.
22160 (*vsx_le_perm_store_<mode>): Likewise.
22162 2017-07-27 Jakub Jelinek <jakub@redhat.com>
22164 PR tree-optimization/81555
22165 PR tree-optimization/81556
22166 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
22167 if true, force CHANGED for the recursive invocation.
22168 (reassociate_bb): Remember original length of ops array, pass
22169 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
22171 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
22172 attributes for noipa attribute. For naked attribute use
22173 lookup_attribute first before lookup_attribute_spec.
22174 * final.c (rest_of_handle_final): Disable IPA RA for functions with
22176 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
22177 for functions with noipa attribute.
22178 (cgraph_externally_visible_p): Return true for functions with noipa
22180 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
22181 for functions with noipa attribute.
22182 * doc/extend.texi: Document noipa function attribute.
22183 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
22184 also for functions with noipa attribute.
22185 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
22187 2017-07-26 Andrew Pinski <apinski@cavium.com>
22189 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
22190 vec_unalign_load_cost and vec_unalign_store_cost.
22192 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
22194 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
22195 -mvsx-small-integer option.
22196 (ISA_3_0_MASKS_IEEE): Likewise.
22197 (OTHER_VSX_VECTOR_MASKS): Likewise.
22198 (POWERPC_MASKS): Likewise.
22199 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
22200 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
22201 code, only testing for DImode being allowed in non-VSX floating
22203 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
22204 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
22206 (rs6000_option_override_internal): Delete -mvsx-small-integer.
22207 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
22208 TARGET_P8_VECTOR test.
22209 (rs6000_secondary_reload_simple_move): Likewise.
22210 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
22211 since TARGET_P9_VECTOR was already tested.
22212 (rs6000_opt_masks): Remove -mvsx-small-integer.
22213 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
22214 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22216 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
22217 test for TARGET_VEXTRACTUB was used, and that uses
22219 (p9 extract splitter): Likewise.
22220 (vsx_extract_<mode>_di_p9): Likewise.
22221 (vsx_extract_<mode>_store_p9): Likewise.
22222 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
22223 for TARGET_P9_VECTOR was used. Delete code that is now dead with
22224 the elimination of TARGET_VSX_SMALL_INTEGER.
22225 (vsx_extract_<mode>_p8): Likewise.
22226 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
22227 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
22228 (vsx_set_<mode>_p9): Likewise.
22229 (vsx_set_v4sf_p9): Likewise.
22230 (vsx_set_v4sf_p9_zero): Likewise.
22231 (vsx_insert_extract_v4sf_p9): Likewise.
22232 (vsx_insert_extract_v4sf_p9_2): Likewise.
22233 * config/rs6000/rs6000.md (sign extend splitter): Change
22234 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
22235 (floatsi<mode>2_lfiwax_mem): Likewise.
22236 (floatunssi<mode>2_lfiwzx_mem): Likewise.
22237 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
22238 since a test for TARGET_P9_VECTOR was used.
22239 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22240 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
22241 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22242 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22243 TARGET_P8_VECTOR test.
22244 (fix_trunc<mode>si2_stfiwx): Likewise.
22245 (fix_trunc<mode>si2_internal): Likewise.
22246 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
22247 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22249 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22250 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22251 TARGET_P8_VECTOR test.
22252 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22253 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
22254 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22256 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22257 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
22258 since a test for TARGET_P9_VECTOR was used.
22259 (splitter for loading small constants): Likewise.
22261 2017-07-26 Andrew Pinski <apinski@cavium.com>
22263 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
22266 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
22269 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
22270 (fp_valid_at): Likewise.
22272 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22274 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
22275 (qdf24xx_addrcost_table): Likewise.
22276 (cortexa57_tunings): Update to use generic_branch_cost.
22277 (cortexa72_tunings): Likewise.
22278 (cortexa73_tunings): Likewise.
22279 (qdf24xx_tunings): Likewise.
22281 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22283 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
22284 (thunderx2t99_branch_cost): Likewise.
22285 (cortexa35_tunings): Update to use generic_branch_cost.
22286 (cortexa53_tunings): Likewise.
22287 (cortexa57_tunings): Likewise.
22288 (cortexa72_tunings): Likewise.
22289 (cortexa73_tunings): Likewise.
22290 (thunderx2t99_tunings): Likewise.
22292 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22294 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
22295 (sparc_option_override): Honour MASK_FSMULD.
22296 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
22297 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
22298 * config/sparc/sparc.opt (mfsmuld): New option.
22299 * doc/invoke.texi (mfsmuld): Document option.
22301 2017-07-26 Marek Polacek <polacek@redhat.com>
22303 PR middle-end/70992
22304 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
22306 2017-07-26 Richard Biener <rguenther@suse.de>
22308 * gimple-match-head.c (do_valueize): Return OP if valueize
22310 (get_def): New helper to get at the def stmt of a SSA name
22311 if valueize allows.
22312 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
22313 do_valueize to get at the def stmt.
22314 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
22316 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
22318 PR middle-end/46932
22319 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
22321 2017-07-26 Martin Liska <mliska@suse.cz>
22324 * function.c (expand_function_start): Make expansion of
22325 nonlocal_goto_save_area after parm_birth_insn.
22327 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22329 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
22330 from all CPU target flags enable members.
22332 2017-07-26 Richard Biener <rguenther@suse.de>
22334 * genmatch.c (dt_simplify::gen): Make iterator vars const.
22335 (decision_tree::gen): Make 'type' const.
22336 (write_predicate): Likewise.
22338 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22340 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
22341 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
22342 (rs6000_option_override_internal): Likewise.
22343 (rs6000_expand_vector_set): Likewise.
22344 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
22345 (TARGET_UPPER_REGS_SF): Likewise.
22346 (TARGET_UPPER_REGS_DI): Likewise.
22347 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
22348 (TARGET_DIRECT_MOVE_64BIT): Likewise.
22349 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
22350 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22351 (Splitters for DI constants in Altivec registers): Likewise.
22352 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
22353 (vsx_set_v4sf_p9): Likewise.
22354 (vsx_set_v4sf_p9_zero): Likewise.
22355 (vsx_insert_extract_v4sf_p9): Likewise.
22356 (vsx_insert_extract_v4sf_p9_2): Likewise.
22358 2017-07-25 Carl Love <cel@us.ibm.com>
22360 * doc/extend.texi: Update the built-in documentation file for the
22361 existing built-in functions
22362 vector signed char vec_cnttz (vector signed char);
22363 vector unsigned char vec_cnttz (vector unsigned char);
22364 vector signed short vec_cnttz (vector signed short);
22365 vector unsigned short vec_cnttz (vector unsigned short);
22366 vector signed int vec_cnttz (vector signed int);
22367 vector unsigned int vec_cnttz (vector unsigned int);
22368 vector signed long long vec_cnttz (vector signed long long);
22369 vector unsigned long long vec_cnttz (vector unsigned long long);
22371 2017-07-25 Andrew Pinski <apinski@cavium.com>
22373 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
22374 accesses where the use is for the first operand of a BIT_INSERT.
22376 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22379 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22380 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22382 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22384 * config/i386/gstabs.h: Delete.
22385 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
22387 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
22389 * config/i386/i386.c (ix86_decompose_address): Do not check for
22390 register RTX when looking at index_reg or base_reg.
22391 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
22393 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
22395 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
22396 to update EH info here.
22398 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22400 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
22402 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22404 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
22406 2017-07-25 Torsten Duwe <duwe@suse.de>
22408 * common.opt: Introduce -fpatchable-function-entry
22409 command line option, and its variables function_entry_patch_area_size
22410 and function_entry_patch_area_start.
22411 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
22412 including a two-value parser.
22413 * target.def (print_patchable_function_entry): New target hook.
22414 * targhooks.h (default_print_patchable_function_entry): New function.
22415 * targhooks.c (default_print_patchable_function_entry): Likewise.
22416 * toplev.c (process_options): Switch off IPA-RA if
22417 patchable function entries are being generated.
22418 * varasm.c (assemble_start_function): Look at the
22419 patchable-function-entry command line switch and current
22420 function attributes and maybe generate NOP instructions by
22421 calling the print_patchable_function_entry hook.
22422 * doc/extend.texi: Document patchable_function_entry attribute.
22423 * doc/invoke.texi: Document -fpatchable_function_entry
22424 command line option.
22425 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
22427 * doc/tm.texi: Re-generate.
22429 2017-07-25 Jakub Jelinek <jakub@redhat.com>
22432 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
22433 TARGET_AVX512DQ rather than TARGET_AVX512BW.
22435 2017-07-25 Tamar Christina <tamar.christina@arm.com>
22437 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
22439 2017-07-25 Richard Biener <rguenther@suse.de>
22441 PR tree-optimization/81455
22442 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
22443 not walk in cycles when looking for guards.
22445 2017-07-25 Richard Biener <rguenther@suse.de>
22447 PR tree-optimization/81529
22448 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
22449 when optimizing backedge uses.
22451 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
22453 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
22455 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
22456 to dl_section_ref. On AIX, append an expression to subtract
22457 the size of the section length to dl_section_ref.
22459 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
22461 * configure.ac: If any of the config.* scripts fail, exit 1.
22462 * configure: Regenerate.
22464 2017-07-25 Richard Biener <rguenther@suse.de>
22466 PR middle-end/81546
22467 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
22468 of immediate uses, be more verbose on errors.
22470 2017-07-25 Richard Biener <rguenther@suse.de>
22472 PR tree-optimization/81510
22473 * tree-vect-loop.c (vect_is_simple_reduction): When the
22474 reduction stmt is not inside the loop bail out.
22476 2017-07-25 Richard Biener <rguenther@suse.de>
22478 PR tree-optimization/81303
22479 * tree-vect-loop-manip.c (vect_loop_versioning): Build
22480 profitability check against LOOP_VINFO_NITERSM1.
22482 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22484 * domwalk.c (cmp_bb_postorder): Simplify.
22485 (sort_bbs_postorder): New function. Use it...
22486 (dom_walker::walk): ...here to optimize common cases.
22488 2017-07-25 Martin Liska <mliska@suse.cz>
22491 * ipa-visibility.c (function_and_variable_visibility): Make the
22492 redirection just on target that supports aliasing.
22493 Fix GNU coding style.
22495 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
22498 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
22500 * config/arm/rtems.h: Likewise.
22501 * config/bfin/rtems.h: Likewise.
22502 * config/i386/rtemself.h: Likewise.
22503 * config/lm32/rtems.h: Likewise.
22504 * config/m32c/rtems.h: Likewise.
22505 * config/m68k/rtemself.h: Likewise.
22506 * config/microblaze/rtems.h: Likewise.
22507 * config/mips/rtems.h: Likewise.
22508 * config/moxie/rtems.h: Likewise.
22509 * config/nios2/rtems.h: Likewise.
22510 * config/powerpcspe/rtems.h: Likewise.
22511 * config/rs6000/rtems.h: Likewise.
22512 * config/rtems.h: Likewise.
22513 * config/sh/rtems.h: Likewise.
22514 * config/sh/rtemself.h: Likewise.
22515 * config/sparc/rtemself.h: Likewise.
22517 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22520 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
22521 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
22522 * tree-ssa-structalias.c (alias_get_name): Same.
22524 2017-07-25 Bin Cheng <bin.cheng@arm.com>
22527 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
22528 instructions if no du chain is found.
22530 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22532 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
22534 2017-07-25 Richard Biener <rguenther@suse.de>
22536 PR middle-end/81505
22537 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
22540 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22542 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
22543 upper-regs options.
22544 (ISA_2_7_MASKS_SERVER): Likewise.
22545 (ISA_3_0_MASKS_IEEE): Likewise.
22546 (OTHER_P8_VECTOR_MASKS): Likewise.
22547 (OTHER_VSX_VECTOR_MASKS): Likewise.
22548 (POWERPC_MASKS): Likewise.
22549 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
22550 duplicate list of options.
22551 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
22552 explicit -mupper-regs options.
22553 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
22554 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
22555 alias for -mupper-regs-df.
22556 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
22557 (rs6000_init_hard_regno_mode_ok): Likewise.
22558 (rs6000_option_override_internal): Likewise.
22559 (rs6000_opt_masks): Likewise.
22560 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
22561 options in terms of whether -mvsx or -mpower8-vector was used.
22562 (TARGET_UPPER_REGS_DI): Likewise.
22563 (TARGET_UPPER_REGS_SF): Likewise.
22564 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
22565 -mupper-regs-* options.
22567 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22569 * passes.c (emergency_dump_function): Print some empty lines and a
22570 header before the RTL dump.
22572 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22574 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
22576 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
22579 * config/aarch64/aarch64.c (aarch64_classify_symbol):
22580 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
22582 2017-07-24 Carl Love <cel@us.ibm.com>
22584 * config/rs6000/rs6000-c.c: Add support for built-in functions
22585 vector float vec_extract_fp32_from_shorth (vector unsigned short);
22586 vector float vec_extract_fp32_from_shortl (vector unsigned short);
22587 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
22588 vec_extract_fp_from_shortl): Add defines for the two builtins.
22589 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
22590 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
22592 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
22593 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
22594 * doc/extend.texi: Update the built-in documentation file for the
22595 new built-in function.
22597 2017-07-24 Jakub Jelinek <jakub@redhat.com>
22600 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
22602 * doc/generic.texi: Likewise.
22603 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22604 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22606 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
22608 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
22609 (aarch64_mls_elt_merge<mode>): Likewise.
22611 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
22613 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22614 having __cxa_atexit.
22616 2017-07-23 Michael Collison <michael.collison@arm.com>
22618 * config/arm/arm.c (arm_option_override): Deprecate
22619 use of -mstructure-size-boundary.
22620 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
22621 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
22623 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22626 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22627 Reduce cost estimate for direct moves.
22629 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
22632 * config/i386/i386.c (ix86_option_override_internal): Disable
22633 BMI, BMI2 and TBM instructions for -m16.
22635 2017-07-21 Carl Love <cel@us.ibm.com>
22637 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22638 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22639 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22640 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22641 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22642 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22643 VMULOSW): New enum "unspec" values.
22644 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22645 altivec_vmulosw): New patterns.
22646 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22647 VMULOSW): Add definitions.
22649 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
22651 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
22652 (qdf24xx): Likewise.
22653 * config/aarch64/aarch64-options-extensions.def (rdma); New.
22654 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
22655 (AARCH64_FL_V8_1): Renumber.
22656 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
22657 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
22658 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
22659 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
22660 rdma to feature modifiers list.
22662 2017-07-21 Yury Gribov <tetra2005@gmail.com>
22664 PR middle-end/56727
22665 * ipa-visibility (function_and_variable_visibility): Convert
22666 recursive PLT call to direct call if appropriate.
22668 2017-07-21 Andrew Pinski <apinski@cavium.com>
22670 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
22671 operand 1 to see if the types precision matches.
22672 * fold-const.c (operand_equal_p): Likewise.
22674 2017-07-21 Richard Biener <rguenther@suse.de>
22676 PR tree-optimization/81303
22677 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
22678 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
22679 (vect_peeling_hash_get_lowest_cost): Adjust.
22680 (vect_enhance_data_refs_alignment): Likewise. Use
22681 vect_get_peeling_costs_all_drs to compute the penalty for no
22682 peeling to match up costs.
22684 2017-07-21 Richard Biener <rguenther@suse.de>
22686 PR tree-optimization/81500
22687 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
22688 we didn't identify a reduction path.
22690 2017-07-21 Tom de Vries <tom@codesourcery.com>
22691 Cesar Philippidis <cesar@codesourcery.com>
22693 PR gcov-profile/81442
22694 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
22697 2017-07-21 Tom de Vries <tom@codesourcery.com>
22700 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
22702 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
22703 nvptx_override_options_after_change.
22705 2017-07-21 Ulrich Drepper <drepper@redhat.com>
22707 * dwarf2out.c (output_file_names): Avoid double testing for
22708 dwarf_version >= 5.
22710 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
22712 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
22714 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22716 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
22718 (try_crossjump_to_edge): Do not punt on partitioned functions.
22720 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22722 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
22723 Put all BBs reachable only via paths crossing cold region to cold
22725 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
22727 2016-07-21 Richard Biener <rguenther@suse.de>
22729 PR tree-optimization/81303
22730 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
22731 into account prologue and epilogue iterations when raising
22732 min_profitable_iters to sth at least covering one vector iteration.
22734 2017-07-21 Tamar Christina <tamar.christina@arm.com>
22736 * config/arm/arm.c (arm_test_cpu_arch_dat):
22739 2017-07-20 Nathan Sidwell <nathan@acm.org>
22741 Remove TYPE_METHODS.
22742 * tree.h (TYPE_METHODS): Delete.
22743 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
22744 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
22745 (dbxout_type_methods): Scan TYPE_FIELDS.
22746 (dbxout_type): Don't check TYPE_METHODS here.
22747 * function.c (use_register_for_decl): Always ignore register for
22748 class types when not optimizing.
22749 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
22750 * tree.c (free_lang_data_in_type): Stitch out member functions and
22751 templates from TYPE_FIELDS.
22752 (build_distinct_type_copy, verify_type_variant,
22753 verify_type): Member fns are on TYPE_FIELDS.
22754 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
22755 * tree-pretty-print.c (dump_generic_node): Likewise.
22757 2017-07-20 Jakub Jelinek <jakub@redhat.com>
22760 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
22761 V2TImode and V4TImode.
22762 (ix86_expand_vector_extract): Likewise.
22763 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
22764 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
22765 (ssescalarmode): Handle V4TImode and V2TImode.
22766 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
22767 (*vec_extractv2ti, *vec_extractv4ti): New insns.
22768 (VEXTRACTI128_MODE): New mode iterator.
22769 (splitter for *vec_extractv?ti first element): New.
22770 (VEC_INIT_MODE): New mode iterator.
22771 (vec_init<mode>): Consolidate 3 expanders into one using
22772 VEC_INIT_MODE mode iterator.
22774 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22776 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
22777 non_spilled_static_chain_regno_p.
22779 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22781 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
22783 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
22785 * bb-reorder.c (connect_traces): Allow copying of blocks within
22788 2017-07-20 Richard Biener <rguenther@suse.de>
22790 * gimple.h (gimple_phi_result): Add gphi * overload.
22791 (gimple_phi_result_ptr): Likewise.
22792 (gimple_phi_arg): Likewise. Adjust index assert to only
22793 allow actual argument accesses rather than all slots available
22795 (gimple_phi_arg_def): Add gphi * overload.
22796 * tree-phinodes.c (make_phi_node): Initialize only actual
22798 (resize_phi_node): Clear memory not covered by old node,
22799 do not initialize excess argument slots.
22800 (reserve_phi_args_for_new_edge): Initialize new argument slot
22803 2017-07-20 Bin Cheng <bin.cheng@arm.com>
22805 PR tree-optimization/81388
22807 2016-07-21 Bin Cheng <bin.cheng@arm.com>
22809 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
22810 by removing computation of may_be_zero.
22812 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22813 Tom de Vries <tom@codesourcery.com>
22815 PR middle-end/81030
22816 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
22817 when gimple level profile disagrees with what RTL expander did.
22819 2017-07-20 Richard Biener <rguenther@suse.de>
22821 PR tree-optimization/61171
22822 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
22823 (vect_analyze_stmt): Add slp instance parameter.
22824 (vectorizable_reduction): Likewise.
22825 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
22826 (vect_is_simple_reduction): Deal with chains not detected
22827 as SLP reduction chain, specifically not properly associated
22828 chains containing a mix of plus/minus.
22829 (get_reduction_op): Remove.
22830 (get_initial_defs_for_reduction): Simplify, pass in whether
22831 this is a reduction chain, pass in the SLP node for the PHIs.
22832 (vect_create_epilog_for_reduction): Get the SLP instance as
22834 (vectorizable_reduction): Get the SLP instance as arg.
22835 During analysis remember the SLP node with the PHIs in the
22836 instance. Simplify getting at the vectorized reduction PHIs.
22837 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
22838 through SLP instance.
22839 (vect_slp_analyze_operations): Likewise.
22840 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
22841 (vect_transform_stmt): Likewise.
22843 2017-07-20 Tom de Vries <tom@codesourcery.com>
22845 PR tree-optimization/81489
22846 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
22847 read of phi arg location to before loop that modifies phi.
22849 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22851 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
22854 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22856 PR middle-end/81331
22857 * except.c (execute): Fix ordering issue.
22859 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22861 PR rtl-optimization/81423
22862 * combine.c (make_compound_operation_int): Don't try to optimize
22863 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
22865 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22867 PR rtl-optimization/81423
22868 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
22869 with a constant that is -1 in the truncated to mode.
22871 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22873 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
22874 (determine_unlikely_bbs): ... here.
22875 * predict.h (propagate_unlikely_bbs_forward): Declare.
22876 * cfgexpand.c (pass_expand::execute): Use it.
22877 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
22879 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
22880 propagate_unlikely_bbs_forward.
22882 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22884 PR middle-end/81331
22885 * except.c (maybe_add_nop_after_section_switch): New function.
22888 2017-07-19 Tom de Vries <tom@codesourcery.com>
22890 * gimple.h (gimple_phi_set_arg): Make assert more strict.
22892 2017-07-19 Tom de Vries <tom@codesourcery.com>
22894 * gimple.h (gimple_phi_arg): Make assert more strict.
22896 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
22898 * config.gcc (powerpc*-*-*): Add mmintrin.h.
22899 * config/rs6000/mmintrin.h: New file.
22900 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
22902 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22904 PR tree-optimization/81346
22905 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
22907 2017-07-19 Tom de Vries <tom@codesourcery.com>
22909 * config/nvptx/nvptx.md (VECIM): Add V2DI.
22911 2017-07-19 Tom de Vries <tom@codesourcery.com>
22913 * config/nvptx/nvptx-modes.def: Add V2DImode.
22914 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
22915 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
22916 (nvptx_output_mov_insn): Handle lack of mov.b128.
22917 (nvptx_print_operand): Handle 'H' and 'L' codes.
22918 (nvptx_vector_mode_supported): Allow V2DImode.
22919 (nvptx_preferred_simd_mode): New function.
22920 (nvptx_data_alignment): New function.
22921 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
22922 nvptx_preferred_simd_mode.
22923 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
22925 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
22927 2017-07-19 Tom de Vries <tom@codesourcery.com>
22929 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
22930 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
22931 (nvptx_vector_mode_supported): New function. Allow V2SImode.
22932 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
22933 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
22934 (mov<VECIM>_insn): New define_insn.
22935 (define_expand "mov<VECIM>): New define_expand.
22937 2017-07-19 Tom de Vries <tom@codesourcery.com>
22939 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
22941 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22943 PR tree-optimization/81346
22944 * fold-const.h (fold_div_compare, range_check_type): Declare.
22945 * fold-const.c (range_check_type): New function.
22946 (build_range_check): Use range_check_type.
22947 (fold_div_compare): No longer static, rewritten into
22948 a match.pd helper function.
22949 (fold_comparison): Don't call fold_div_compare here.
22950 * match.pd (X / C1 op C2): New optimization using fold_div_compare
22951 as helper function.
22953 2017-07-19 Nathan Sidwell <nathan@acm.org>
22955 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
22956 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
22957 * tree.c (find_decls_types_r, verify_type): Use
22958 TYPE_{MIN,MAX}_VALUE_RAW.
22959 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
22960 (hash_tree): Likewise.
22961 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
22963 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
22966 2017-07-18 Tom de Vries <tom@codesourcery.com>
22968 PR middle-end/81464
22969 * omp-expand.c (expand_omp_for_static_chunk): Handle
22970 equal-argument loop exit phi.
22972 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
22975 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
22976 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
22977 operand 2 predicate.
22978 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
22979 operand 2 predicate.
22980 (ror,rol -> rorx splitters): Use const_int_operand as
22981 operand 2 predicate.
22983 2017-06-18 Richard Biener <rguenther@suse.de>
22985 PR tree-optimization/81410
22986 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
22987 the gap in the ! slp_perm SLP case after each group.
22989 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22991 PR middle-end/81463
22992 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
22995 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22997 PR middle-end/81462
22998 * predict.c (set_even_probabilities): Cleanup; do not affect
22999 probabilities that are already known.
23000 (combine_predictions_for_bb): Call even when count is set.
23002 2017-07-18 Nathan Sidwell <nathan@acm.org>
23004 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
23007 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23010 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
23011 optimization for loop niter analysis.
23013 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
23016 * config/avr/avr.c (avr_optimize_casesi): Don't use
23017 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
23019 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
23021 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
23022 body_cost_vec from _vect_peel_extended_info.
23023 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
23024 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
23027 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23029 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
23031 2017-07-18 Richard Biener <rguenther@suse.de>
23033 PR tree-optimization/80620
23034 PR tree-optimization/81403
23035 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
23036 info when re-using a VN table entry.
23038 2017-07-18 Richard Biener <rguenther@suse.de>
23040 PR tree-optimization/81418
23041 * tree-vect-loop.c (vectorizable_reduction): Properly compute
23042 vectype_in. Verify that with lane-reducing reduction operations
23043 we have a single def-use cycle.
23045 2017-07-17 Carl Love <cel@us.ibm.com>
23047 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
23049 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23050 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23051 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23052 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23053 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23054 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23055 VMULOSW): New enum "unspec" values.
23056 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23057 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23058 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23059 altivec_vmulosw): New patterns.
23060 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23061 VMULOSW): Add definitions.
23063 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
23065 * config/alpha/alpha.c: Include predict.h.
23067 2017-07-17 Yury Gribov <tetra2005@gmail.com>
23069 * tree-vrp.c (compare_assert_loc): Fix comparison function
23070 to return predictable results.
23072 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23074 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
23076 (subdi3): Likewise.
23077 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
23078 * doc/invoke.texi (mexpand-adddi): Update text.
23080 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23082 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
23083 that also clobbers the CC register. The old expand code is moved
23085 (*arc_clzsi2): ... here.
23086 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
23087 the CC register. The old expand code is moved to ...
23088 (arc_ctzsi2): ... here.
23090 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23092 * config/arc/arc.opt (mindexed-loads): Use initial value
23093 TARGET_INDEXED_LOADS_DEFAULT.
23094 (mauto-modify-reg): Use initial value
23095 TARGET_AUTO_MODIFY_REG_DEFAULT.
23096 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23097 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23098 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23099 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23101 2017-07-17 Martin Liska <mliska@suse.cz>
23104 * opts.c (finish_options): Do not allow -fgnu-tm
23105 w/ -fsanitize={kernel-,}address. Say sorry.
23107 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23110 * tree-loop-distribution.c (classify_partition): Only assert on
23111 numer of iterations.
23112 (merge_dep_scc_partitions): Delete prameter. Update function call.
23113 (distribute_loop): Remove code handling loop with unknown niters.
23114 (pass_loop_distribution::execute): Skip loop with unknown niters.
23116 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23119 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
23120 function sort_partitions_by_post_order.
23122 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23124 PR tree-optimization/81374
23125 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
23126 the max index of basic blocks, rather than number of basic blocks.
23128 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23130 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
23132 (arc_legitimate_pic_operand_p): Likewise.
23133 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
23135 (arc_needs_pcl_p): Likewise.
23136 (arc_legitimate_pc_offset_p): Likewise.
23137 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
23138 function is also used in constrains.md.
23139 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
23140 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
23141 PLUS. Only return true/false in known cases, otherwise assert.
23142 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
23143 is already called in arc_legitimate_constant_p.
23144 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
23146 (LEGITIMATE_PIC_OPERAND_P): Use
23147 arc_raw_symbolic_reference_mentioned_p function.
23148 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
23153 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23154 Andrew Burgess <andrew.burgess@embecosm.com>
23156 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
23157 (arc_return_address_register): New function.
23158 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
23159 (arc_handle_fndecl_attribute): Add naked attribute.
23160 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
23161 (TARGET_WARN_FUNC_RETURN): Likewise.
23162 (arc_allocate_stack_slots_for_args): New function.
23163 (arc_warn_func_return): Likewise.
23164 (machine_function): Change type fn_type.
23165 (arc_compute_function_type): Consider new naked function type,
23166 change function return type.
23167 (arc_must_save_register): Adapt to handle new
23168 arc_compute_function_type's return type.
23169 (arc_expand_prologue): Likewise.
23170 (arc_expand_epilogue): Likewise.
23171 (arc_return_address_regs): Delete.
23172 (arc_return_address_register): New function.
23173 (arc_epilogue_uses): Use above function.
23174 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
23175 (arc_function_type): Change encoding, add naked type.
23176 (ARC_INTERRUPT_P): Change to handle the new encoding.
23177 (ARC_FAST_INTERRUPT_P): Likewise.
23178 (ARC_NORMAL_P): Define.
23179 (ARC_NAKED_P): Likewise.
23180 (arc_compute_function_type): Delete prototype.
23181 * config/arc/arc.md (in_ret_delay_slot): Use
23182 arc_return_address_register function.
23183 (simple_return): Likewise.
23184 (p_return_i): Likewise.
23186 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23188 PR tree-optimization/81428
23189 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
23190 can't be built for those types.
23192 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23194 Remove stuff dead since r239246.
23196 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
23197 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
23198 (avr_inform_devices): Remove dead stuff.
23200 2017-07-17 Tamar Christina <tamar.christina@arm.com>
23202 * config/arm/arm_neon.h: Fix softp typo.
23204 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23206 PR tree-optimization/81365
23207 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
23208 aggregate moves onto bb predecessor edges, make sure there are no
23209 loads that could alias the lhs in between the start of bb and the
23212 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23215 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
23216 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
23217 [LSHIFTRT, outer_code = TRUNCATE]: Same.
23219 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23221 PR tree-optimization/81396
23222 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
23223 (init_symbolic_number): Initialize it to 1.
23224 (perform_symbolic_merge): Add n_ops from both operands into the new
23226 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
23227 without base_addr as useless if they need more than one operation.
23228 (bswap_replace): Handle !bswap case for NULL base_addr.
23230 2017-07-17 Tom de Vries <tom@codesourcery.com>
23233 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
23236 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
23238 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
23239 conditional builtin define __FIX_LEON3FT_B2BST.
23241 2017-07-17 Daniel Cederman <cederman@gaisler.com>
23243 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
23244 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
23247 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23249 PR rtl-optimization/81424
23250 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
23251 to remove potential trapping from operands if -fnon-call-exceptions.
23253 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23255 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
23256 profile_proability for scalling.
23257 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
23259 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23261 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
23263 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23265 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
23266 fixpoint arithmetics.
23268 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23270 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
23271 fixpoint arithmetics.
23273 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23275 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
23276 fixpoint arithmetics.
23278 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23280 * profile-count.h (profile_probability::from_reg_br_prob_note,
23281 profile_probability::to_reg_br_prob_note): New functions.
23282 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
23283 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
23284 * predict.c (probability_reliable_p): Update.
23285 (edge_probability_reliable_p): Update.
23286 (br_prob_note_reliable_p): Update.
23287 (invert_br_probabilities): Update.
23288 (add_reg_br_prob_note): New function.
23289 (combine_predictions_for_insn): Update.
23290 * asan.c (asan_clear_shadow): Update.
23291 * cfgbuild.c (compute_outgoing_frequencies): Update.
23292 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23293 (update_br_prob_note): Update.
23294 (rtl_verify_edges): Update.
23295 (purge_dead_edges): Update.
23296 (fixup_reorder_chain): Update.
23297 * emit-rtl.c (try_split): Update.
23298 * ifcvt.c (cond_exec_process_insns): Update.
23299 (cond_exec_process_if_block): Update.
23300 (dead_or_predicable): Update.
23301 * internal-fn.c (expand_addsub_overflow): Update.
23302 (expand_neg_overflow): Update.
23303 (expand_mul_overflow): Update.
23304 * loop-doloop.c (doloop_modify): Update.
23305 * loop-unroll.c (compare_and_jump_seq): Update.
23306 * optabs.c (emit_cmp_and_jump_insn_1): Update.
23307 * predict.h: Update.
23308 * reorg.c (mostly_true_jump): Update.
23310 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
23311 * config/alpha/alpha.c (emit_unlikely_jump): Update.
23312 * config/arc/arc.c: (emit_unlikely_jump): Update.
23313 * config/arm/arm.c: (emit_unlikely_jump): Update.
23314 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
23315 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
23316 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
23317 (ix86_print_operand): Update.
23318 (ix86_split_fp_branch): Update.
23319 (predict_jump): Update.
23320 * config/ia64/ia64.c (ia64_print_operand): Update.
23321 * config/mmix/mmix.c (mmix_print_operand): Update.
23322 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
23323 (rs6000_expand_split_stack_prologue): Update.
23324 * config/rs6000/rs6000.c: Update.
23325 * config/s390/s390.c (s390_expand_vec_strlen): Update.
23326 (s390_expand_vec_movstr): Update.
23327 (s390_expand_cs_tdsi): Update.
23328 (s390_expand_split_stack_prologue): Update.
23329 * config/sh/sh.c (sh_print_operand): Update.
23330 (expand_cbranchsi4): Update.
23331 (expand_cbranchdi4): Update.
23332 * config/sparc/sparc.c (output_v9branch): Update.
23333 * config/spu/spu.c (get_branch_target): Update.
23334 (ea_load_store_inline): Update.
23335 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
23336 * config/tilepro/tilepro.c: Update.
23338 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23340 * gimplify.c (mostly_copy_tree_r): Revert latest change.
23341 (gimplify_save_expr): Likewise.
23343 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23345 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
23347 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23349 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
23351 * timevar.def (TV_IPA_FNSUMMARY): Define.
23353 2017-07-16 Daniel Cederman <cederman@gaisler.com>
23355 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
23356 to back store errata sensitive sequence from being generated.
23357 (sqrtdf2_fix): Likewise.
23359 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23361 * tree-ssa-threadupdate.c (compute_path_counts,
23362 update_joiner_offpath_counts): Use profile_probability.
23364 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
23367 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23369 * config/arm/arm-c.c (arm_cpu_builtins): Define
23370 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23372 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23374 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23375 array entries to represent __ieee128 versions of the
23376 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
23377 scalar_extract_sig, and scalar_insert_exp built-in functions.
23378 (altivec_resolve_overloaded_builtin): Add special case handling
23379 for the __builtin_scalar_insert_exp function, as represented by
23380 the P9V_BUILTIN_VEC_VSIEDP constant.
23381 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
23382 exponent support for __ieee128 argument.
23383 (VSESQP): Add scalar extract signature support for __ieee128
23385 (VSTDCNQP): Add scalar test negative support for __ieee128
23387 (VSIEQP): Add scalar insert exponent support for __int128 argument
23388 with __ieee128 result.
23389 (VSIEQPF): Add scalar insert exponent support for __ieee128
23390 argument with __ieee128 result.
23391 (VSTDCQP): Add scalar test data class support for __ieee128
23393 (VSTDCNQP): Add overload support for scalar test negative with
23394 __ieee128 argument.
23395 (VSTDCQP): Add overload support for scalar test data class
23396 __ieee128 argument.
23397 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
23398 UNSPEC_VSX_SXSIGDP.
23399 (UNSPEC_VSX_SIEXPQP): New constant.
23400 (xsxexpqp): New insn for VSX scalar extract exponent quad
23402 (xsxsigqp): New insn for VSX scalar extract significand quad
23404 (xsiexpqpf): New insn for VSX scalar insert exponent quad
23405 precision with floating point argument.
23406 (xststdcqp): New expand for VSX scalar test data class quad
23408 (xststdcnegqp): New expand for VSX scalar test negative quad
23410 (xststdcqp): New insn to match expansions for VSX scalar test data
23411 class quad precision and VSX scalar test negative quad precision.
23412 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
23413 special case operand checking to enforce that second operand of
23414 VSX scalar test data class with quad precision argument is a 7-bit
23416 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23417 prototypes and descriptions of __ieee128 versions of
23418 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
23419 scalar_test_data_class, and scalar_test_neg built-in functions.
23421 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23423 PR tree-optimization/81162
23424 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
23425 replace a negate with an add.
23427 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
23429 * doc/invoke.texi (arm/-mcpu): Document +crypto.
23431 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23433 * config/arm/arm-c.c (arm_cpu_builtins): Define
23434 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23436 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23438 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
23439 (armv8-r): Set ARM Cortex-R52 as default CPU.
23440 * config/arm/arm-tables.opt: Regenerate.
23441 * config/arm/arm-tune.md: Regenerate.
23442 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23444 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
23445 extension for -mcpu=cortex-r52.
23447 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23449 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
23450 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
23451 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
23452 (fp-armv8): Define it as FP_ARMv8 only.
23453 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
23454 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
23456 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
23457 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
23458 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
23459 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
23460 than TARGET_FPU_ARMV8.
23461 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
23462 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
23463 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
23465 * config/arm/neon.md (neon_vrint): Likewise.
23466 (neon_vcvt): Likewise.
23467 (neon_<fmaxmin_op><mode>): Likewise.
23468 (<fmaxmin><mode>3): Likewise.
23469 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
23470 * config/arm/predicates.md (arm_cond_move_operator): Check against
23471 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
23473 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
23475 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
23476 to top of function.
23478 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23480 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
23481 loop in comment with memset.
23483 2017-07-14 Martin Liska <mliska@suse.cz>
23485 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
23486 * dwarf2out.c (is_java): Remove the function.
23487 (output_pubname): Remove usage of the function.
23488 (lower_bound_default): Remove usage of DW_LANG_Java.
23489 (gen_compile_unit_die): Likewise.
23490 * gcc.c: Remove compiler defaults for .java and .zip files.
23491 * gimple-expr.c (remove_suffix): Change as there's no longer
23492 extension than 4-letter one.
23493 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
23494 (gimplify_save_expr): Likewise.
23495 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
23496 as it's possible even for other languages than Java.
23497 * langhooks.h (struct lang_hooks): Remove Java from a comment.
23498 * lto-opts.c (lto_write_options): Remove reference to Java.
23499 * opts.c (strip_off_ending): Update file extension handling.
23500 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
23501 * tree-eh.c (lower_resx): Likewise.
23502 * tree.c (free_lang_data_in_type): Remove dead code.
23503 (find_decls_types_r): Likewise.
23504 (build_common_builtin_nodes): Remove Java from a comment.
23505 (verify_type): Remove dead code.
23506 * varasm.c (assemble_external): Remove Java from a comment.
23508 2017-07-14 Martin Liska <mliska@suse.cz>
23510 * opts.c (finish_options): Add quotes.
23511 (common_handle_option): Likewise.
23513 2017-07-14 Martin Liska <mliska@suse.cz>
23515 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
23516 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
23517 Remove N_SO_PASCAL.
23518 * dwarf2out.c (lower_bound_default): Do not handle
23520 (gen_compile_unit_die): Likewise.
23521 * gcc.c: Remove default extension binding for GNU Pascal.
23522 * stmt.c: Remove Pascal language from a comment.
23523 * xcoffout.c: Likewise.
23525 2017-07-13 David Malcolm <dmalcolm@redhat.com>
23528 * diagnostic-show-locus.c (fixit_cmp): New function.
23529 (layout::layout): Sort m_fixit_hints.
23530 (column_range::column_range): Assert that the values are valid.
23531 (struct char_span): New struct.
23532 (correction::overwrite): New method.
23533 (struct source_line): New struct.
23534 (line_corrections::add_hint): Add assertions. Reimplement memcpy
23535 calls in terms of classes source_line and char_span, and
23536 correction::overwrite.
23537 (selftest::test_overlapped_fixit_printing_2): New function.
23538 (selftest::diagnostic_show_locus_c_tests): Call it.
23540 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
23542 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
23543 early if there is no lhs.
23545 2017-07-13 Martin Liska <mliska@suse.cz>
23547 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
23548 (gen_reference_type_die): Likewise.
23549 * stor-layout.c: Remove Pascal-related comment.
23551 2017-07-13 Martin Liska <mliska@suse.cz>
23553 * opts.c (finish_options): Add quotes to error messages.
23554 (parse_sanitizer_options): Likewise.
23556 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
23558 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
23560 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
23562 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
23564 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23566 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
23568 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
23570 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
23573 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
23574 provides the hardware capability bits, define the macro
23575 __BUILTIN_CPU_SUPPORTS__.
23576 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
23577 if GLIBC does not provide the hardware capability bits. Add a
23578 gcc_unreachable call if the built-in cpu function is neither
23579 __builtin_cpu_is nor __builtin_cpu_supports.
23580 (rs6000_get_function_versions_dispatcher): Change the warning
23581 that an old GLIBC is used which does not export the capability
23582 bits to be an error.
23583 * doc/extend.texi (target_clones attribute): Document the
23584 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
23585 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
23586 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
23587 the macros defined by GCC if the newer GLIBC is available.
23589 2017-07-12 Jeff Law <law@redhat.com>
23591 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
23592 remaining includes slightly.
23593 * config/riscv/riscv-builtins.c: Include profile-count.h.
23595 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23598 * config/avr/avr.c (avr_set_current_function): In diagnostic
23599 messages: Quote keywords and (parts of) identifiers.
23600 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
23603 2017-07-12 Carl Love <cel@us.ibm.com>
23605 * config/rs6000/rs6000-c.c: Add support for built-in functions
23606 vector bool char vec_revb (vector bool char);
23607 vector bool short vec_revb (vector short char);
23608 vector bool int vec_revb (vector bool int);
23609 vector bool long long vec_revb (vector bool long long);
23610 * doc/extend.texi: Update the built-in documentation file for the
23611 new built-in functions.
23613 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23615 * config/s390/s390.md: Remove movcc splitter.
23617 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23619 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
23620 load/store on condition.
23622 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23625 * config/avr/avr.c (avr_encode_section_info)
23626 [progmem && !TREE_READONLY]: Error if progmem object needs
23629 2017-07-11 Michael Collison <michael.collison@arm.com>
23631 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
23634 2017-07-11 Carl Love <cel@us.ibm.com>
23636 * config/rs6000/rs6000-c.c: Add support for builtins
23637 vector unsigned int vec_parity_lsbb (vector signed int);
23638 vector unsigned int vec_parity_lsbb (vector unsigned int);
23639 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
23640 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
23641 vector unsigned long long vec_parity_lsbb (vector signed long long);
23642 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
23643 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
23644 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
23645 * doc/extend.texi: Update the built-in documentation file for the
23646 new built-in functions.
23648 2017-07-11 David Malcolm <dmalcolm@redhat.com>
23650 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
23651 (layout::m_primary_loc): New field.
23652 (layout::layout): Initialize new field. Move location filtering
23653 logic from here to...
23654 (layout::maybe_add_location_range): ...this new method. Add
23655 support for filtering to just the lines already specified by other
23657 (layout::will_show_line_p): New method.
23658 (gcc_rich_location::add_location_if_nearby): New method.
23659 (selftest::test_add_location_if_nearby): New test function.
23660 (selftest::diagnostic_show_locus_c_tests): Call it.
23661 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
23664 2017-07-11 Tom de Vries <tom@codesourcery.com>
23666 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
23667 (bb_first_real_insn): New function.
23668 (nvptx_single): Add extra initialization of broadcasted condition
23671 2017-07-11 Nathan Sidwell <nathan@acm.org>
23673 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
23675 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
23677 * doc/extend.texi (AVR Function Attributes): Remove weblink to
23678 Binutils doc as TEXI will mess them up.
23679 * doc/invoke.texi (AVR Options): Same here.
23681 2017-07-11 Daniel Cederman <cederman@gaisler.com>
23683 * config/sparc/sparc.opt (mfix-ut700): New option.
23684 (mfix-gr712rc): Likewise.
23685 (sparc_fix_b2bst): New variable.
23686 * doc/invoke.texi (SPARC options): Document them.
23687 (ARM options): Fix warnings.
23688 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
23689 instructions to prevent sequences that can trigger the store-store
23690 errata for certain LEON3FT processors.
23691 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
23692 (sparc_option_override): Set sparc_fix_b2bst appropriately.
23693 * config/sparc/sparc.md (fix_b2bst): New attribute.
23694 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
23696 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
23699 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
23701 (*rsqrtsf2_sse): Ditto.
23703 (div<mode>3): Macroize insn from divdf3 and divsf3
23704 using MODEF mode iterator.
23706 2017-07-10 Martin Sebor <msebor@redhat.com>
23708 PR tree-optimization/80397
23709 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
23710 instead of testing for equality to INTEGER_TYPE.
23712 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
23714 * config.gcc: Remove uclibc from arc target spec.
23716 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
23718 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
23720 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23723 * lto-wrapper.c (remove_option): New function.
23724 (merge_and_complain): Merge PIC/PIE options more realistically.
23726 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23728 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
23732 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
23733 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
23734 * config.in: Regenerate.
23735 * configure: Regenerate.
23736 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
23737 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
23738 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
23739 (TARGET_GASISR_PROLOGUES): ...target mask.
23740 * common/config/avr/avr-common.c
23741 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
23742 Set -mgas-isr-prologues.
23743 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
23744 INSERT_PASS_BEFORE for it.
23745 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
23746 * config/avr/avr.c (avr_option_override)
23747 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
23748 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
23749 (avr_attribute_table) <no_gccisr>: Add new function attribute.
23750 (avr_set_current_function) <is_no_gccisr>: Init machine field.
23751 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
23753 (make_avr_pass_pre_proep): New function.
23754 (emit_push_sfr) <treg>: Add argument to function and use it
23755 instead of TMP_REG.
23756 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
23757 and set machine->gasisr.yes.
23758 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
23759 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
23760 __gcc_isr.n_pushed to .L__stack_usage.
23761 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
23762 (avr_asm_final_postscan_insn): ...this new static function.
23763 * config/avr/avr.h (machine_function)
23764 <is_no_gccisr, use_L__stack_usage>: New fields.
23765 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
23766 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
23767 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
23768 (gasisr, *gasisr): New expander and insn.
23769 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23770 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
23771 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
23773 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
23775 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
23778 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23780 Move jump-tables out of .text again.
23783 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
23784 (ASM_OUTPUT_ADDR_VEC): New function.
23785 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
23786 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
23787 INSN_ADDRESSes as asm comment.
23788 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
23789 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
23790 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
23791 * config/avr/avr.md (*tablejump): Adjust comment.
23792 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
23793 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
23795 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
23796 (avr_output_addr_vec): New proto.
23797 (avr_log_t) <insn_addresses>: New field.
23799 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
23802 * config/i386/i386.c (ix86_function_arg_advance): Set
23803 outgoing_args_on_stack to true if there are outgoing arguments
23805 (ix86_function_arg): Likewise.
23806 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
23807 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
23808 * config/i386/i386.h (machine_function): Add
23809 outgoing_args_on_stack.
23811 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
23813 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23814 supporting pthreds.
23815 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
23817 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
23819 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
23820 (REAL_H): Remove $(MACHMODE_H).
23821 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
23823 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
23824 $(MACHMODE_H) and double-int.h.
23825 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
23827 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
23830 2017-07-07 Andrew Pinski <apinski@cavium.com>
23832 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
23833 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
23835 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23837 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
23838 Add warning if GCC was not configured to link against a GLIBC that
23839 exports the hardware capability bits.
23840 (make_resolver_func): Make resolver function private and not a
23841 COMDAT function. Create the name with clone_function_name instead
23842 of make_unique_name.
23845 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
23846 correct operand in doing the split.
23848 2017-07-07 Carl Love <cel@us.ibm.com>
23850 * config/rs6000/rs6000-c: Add support for built-in function
23851 vector unsigned short vec_pack_to_short_fp32 (vector float,
23853 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
23854 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
23855 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
23856 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
23857 (convert_4f32_8i16): Add define_expand.
23858 * doc/extend.texi: Update the built-in documentation file for the
23859 new built-in function.
23861 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23863 * config/sparc/m8.md: New file.
23864 * config/sparc/sparc.md: Include m8.md.
23866 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23868 * config/sparc/sparc.opt: New option -mvis4b.
23869 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
23870 (sparc_option_override): Handle VIS4B.
23871 (enum sparc_builtins): Define
23872 SPARC_BUILTIN_DICTUNPACK{8,16,32},
23873 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
23874 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
23875 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
23876 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
23877 (check_constant_argument): New function.
23878 (sparc_vis_init_builtins): Define builtins
23879 __builtin_vis_dictunpack{8,16,32},
23880 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
23881 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
23882 __builtin_vis_fpcmpde{8,16,32}shl and
23883 __builtin_vis_fpcmpur{8,16,32}shl.
23884 (sparc_expand_builtin): Check that the constant operands to
23885 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
23886 constant and in range.
23887 * config/sparc/sparc-c.c (sparc_target_macros): Handle
23889 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
23890 (SPARC_IMM5_P): Likewise.
23891 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
23892 (enabled): Handle vis4b.
23893 (UNSPEC_DICTUNPACK): New unspec.
23894 (UNSPEC_FPCMPSHL): Likewise.
23895 (UNSPEC_FPUCMPSHL): Likewise.
23896 (UNSPEC_FPCMPDESHL): Likewise.
23897 (UNSPEC_FPCMPURSHL): Likewise.
23898 (cpu_feature): New CPU feature `vis4b'.
23899 (dictunpack{8,16,32}): New insns.
23900 (FPCSMODE): New mode iterator.
23901 (fpcscond): New code iterator.
23902 (fpcsucond): Likewise.
23903 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
23904 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
23905 (fpcmpde{8,16,32}{si,di}shl): Likewise.
23906 (fpcmpur{8,16,32}{si,di}shl): Likewise.
23907 * config/sparc/constraints.md: Define constraints `q' for unsigned
23908 2-bit integer constants and `t' for unsigned 5-bit integer
23910 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
23912 (imm5_operand_dictunpack16): Likewise.
23913 (imm5_operand_dictunpack32): Likewise.
23914 (imm2_operand): Likewise.
23915 * doc/invoke.texi (SPARC Options): Document -mvis4b.
23916 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
23917 ditunpack* and fpcmp*shl builtins.
23919 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23921 * config.gcc: Handle m8 in --with-{cpu,tune} options.
23922 * config.in: Add HAVE_AS_SPARC6 define.
23923 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
23925 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
23927 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
23928 (CPP_CPU_SPEC): Handle m8.
23929 (ASM_CPU_SPEC): Likewise.
23930 * config/sparc/sparc-opts.h (enum processor_type): Add
23932 * config/sparc/sparc.c (m8_costs): New struct.
23933 (sparc_option_override): Handle TARGET_CPU_m8.
23934 (sparc32_initialize_trampoline): Likewise.
23935 (sparc64_initialize_trampoline): Likewise.
23936 (sparc_issue_rate): Likewise.
23937 (sparc_register_move_cost): Likewise.
23938 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
23939 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
23940 (ASM_CPU64_DEFAULT_SPEC): Likewise.
23941 (CPP_CPU_SPEC): Handle M8.
23942 (ASM_CPU_SPEC): Likewise.
23943 (AS_M8_FLAG): Define.
23944 * config/sparc/sparc.md: Add m8 to the cpu attribute.
23945 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
23946 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
23948 * configure: Regenerate.
23949 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
23952 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23954 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
23956 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
23957 ("*movdi_insn_sp32"): Do not set v3pipe.
23958 ("*movsi_insn"): Likewise.
23959 ("*movdi_insn_sp64"): Likewise.
23960 ("*movsf_insn"): Likewise.
23961 ("*movdf_insn_sp32"): Likewise.
23962 ("*movdf_insn_sp64"): Likewise.
23963 ("*zero_extendsidi2_insn_sp64"): Likewise.
23964 ("*sign_extendsidi2_insn"): Likewise.
23965 ("*mov<VM32:mode>_insn"): Likewise.
23966 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23967 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23968 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23969 ("<vlop:code><VL:mode>3"): Likewise.
23970 ("*not_<vlop:code><VL:mode>3"): Likewise.
23971 ("*nand<VL:mode>_vis"): Likewise.
23972 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
23973 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
23974 ("one_cmpl<VL:mode>2"): Likewise.
23975 ("faligndata<VM64:mode>_vis"): Likewise.
23976 ("alignaddrsi_vis"): Likewise.
23977 ("alignaddrdi_vis"): Likweise.
23978 ("alignaddrlsi_vis"): Likewise.
23979 ("alignaddrldi_vis"): Likewise.
23980 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23981 ("bmaskdi_vis"): Likewise.
23982 ("bmasksi_vis"): Likewise.
23983 ("bshuffle<VM64:mode>_vis"): Likewise.
23984 ("cmask8<P:mode>_vis"): Likewise.
23985 ("cmask16<P:mode>_vis"): Likewise.
23986 ("cmask32<P:mode>_vis"): Likewise.
23987 ("pdistn<P:mode>_vis"): Likewise.
23988 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23990 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23992 * config/sparc/sparc.md ("subtype"): New insn attribute.
23993 ("*wrgsr_sp64"): Set insn subtype.
23994 ("*rdgsr_sp64"): Likewise.
23995 ("alignaddrsi_vis"): Likewise.
23996 ("alignaddrdi_vis"): Likewise.
23997 ("alignaddrlsi_vis"): Likewise.
23998 ("alignaddrldi_vis"): Likewise.
23999 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24000 ("fexpand_vis"): Likewise.
24001 ("fpmerge_vis"): Likewise.
24002 ("faligndata<VM64:mode>_vis"): Likewise.
24003 ("bshuffle<VM64:mode>_vis"): Likewise.
24004 ("cmask8<P:mode>_vis"): Likewise.
24005 ("cmask16<P:mode>_vis"): Likewise.
24006 ("cmask32<P:mode>_vis"): Likewise.
24007 ("fchksm16_vis"): Likewise.
24008 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
24009 ("fmean16_vis"): Likewise.
24010 ("fp<plusminus_insn>64_vis"): Likewise.
24011 ("<plusminus_insn>v8qi3"): Likewise.
24012 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24013 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
24014 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
24015 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
24016 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
24017 ("*movqi_insn"): Likewise.
24018 ("*movhi_insn"): Likewise.
24019 ("*movsi_insn"): Likewise.
24020 ("movsi_pic_gotdata_op"): Likewise.
24021 ("*movdi_insn_sp32"): Likewise.
24022 ("*movdi_insn_sp64"): Likewise.
24023 ("movdi_pic_gotdata_op"): Likewise.
24024 ("*movsf_insn"): Likewise.
24025 ("*movdf_insn_sp32"): Likewise.
24026 ("*movdf_insn_sp64"): Likewise.
24027 ("*zero_extendhisi2_insn"): Likewise.
24028 ("*zero_extendqihi2_insn"): Likewise.
24029 ("*zero_extendqisi2_insn"): Likewise.
24030 ("*zero_extendqidi2_insn"): Likewise.
24031 ("*zero_extendhidi2_insn"): Likewise.
24032 ("*zero_extendsidi2_insn_sp64"): Likewise.
24033 ("ldfsr"): Likewise.
24034 ("prefetch_64"): Likewise.
24035 ("prefetch_32"): Likewise.
24036 ("tie_ld32"): Likewise.
24037 ("tie_ld64"): Likewise.
24038 ("*tldo_ldub_sp32"): Likewise.
24039 ("*tldo_ldub1_sp32"): Likewise.
24040 ("*tldo_ldub2_sp32"): Likewise.
24041 ("*tldo_ldub_sp64"): Likewise.
24042 ("*tldo_ldub1_sp64"): Likewise.
24043 ("*tldo_ldub2_sp64"): Likewise.
24044 ("*tldo_ldub3_sp64"): Likewise.
24045 ("*tldo_lduh_sp32"): Likewise.
24046 ("*tldo_lduh1_sp32"): Likewise.
24047 ("*tldo_lduh_sp64"): Likewise.
24048 ("*tldo_lduh1_sp64"): Likewise.
24049 ("*tldo_lduh2_sp64"): Likewise.
24050 ("*tldo_lduw_sp32"): Likewise.
24051 ("*tldo_lduw_sp64"): Likewise.
24052 ("*tldo_lduw1_sp64"): Likewise.
24053 ("*tldo_ldx_sp64"): Likewise.
24054 ("*mov<VM32:mode>_insn"): Likewise.
24055 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24056 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24058 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24060 * config/sparc/sparc.md ("type"): New insn type viscmp.
24061 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
24063 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
24064 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
24065 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24066 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
24068 ("n7_vis_logical_11cycle"): Likewise.
24069 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
24070 * config/sparc/niagara2.md ("niag3_vis": Likewise.
24071 * config/sparc/niagara.md ("niag_vis"): Likewise.
24072 * config/sparc/ultra3.md ("us3_fga"): Likewise.
24073 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
24075 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24077 * config/sparc/sparc.md: New instruction type `bmask'.
24078 (bmaskdi_vis): Use the `bmask' type.
24079 (bmasksi_vis): Likewise.
24080 * config/sparc/ultra3.md (us3_array): Likewise.
24081 * config/sparc/niagara7.md (n7_array): Likewise.
24082 * config/sparc/niagara4.md (n4_array): Likewise.
24083 * config/sparc/niagara2.md (niag2_vis): Likewise.
24084 (niag3_vis): Likewise.
24085 * config/sparc/niagara.md (niag_vis): Likewise.
24087 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24089 * ipa-comdats.c: Remove optimize check from gate.
24090 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
24091 for functions not optimized.
24092 (ipa_fn_summary_read): Skip optimize check.
24093 (ipa_fn_summary_write): Likewise.
24094 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
24096 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
24098 (can_inline_edge_p): Check flag_pcc_struct_return for match.
24099 (check_callers): Give up on caller which is not optimized.
24100 (inline_small_functions): Likewise.
24101 (ipa_inline): Do not give up when not optimizing.
24102 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
24103 away unoptimizes cdtors.
24104 (whole_program_function_and_variable_visibility): Do
24105 ipa_discover_readonly_nonaddressable_vars in LTO mode.
24106 * ipa.c (process_references): Do not check optimize.
24107 (symbol_table::remove_unreachable_nodes): Update optimize check.
24108 (set_writeonly_bit): Update optimize check.
24109 (pass_ipa_cdtor_merge::gate): Do not check optimize.
24110 (pass_ipa_single_use::gate): Remove.
24112 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24114 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
24115 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
24116 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
24117 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
24118 permute_load, permute_store, adjust_extract, adjust_splat,
24119 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
24120 replace_swap_with_copy, dump_swap_insn_table,
24121 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
24122 recombine_lvx_pattern, recombine_stvx_pattern,
24123 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
24124 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
24125 to file rs6000-p8swap.c.
24126 * config/rs6000/rs6000-p8swap.c: New file.
24127 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
24128 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
24129 and rs6000*-*-* targets.
24131 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24133 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
24135 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24137 * lto-wrapper.c (merge_and_complain): Do not merge
24138 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
24139 fsigned_zeros, ftrapping_math, fwrapv.
24140 (append_compiler_options): Do not track these options.
24141 (append_linker_options): Likewie
24143 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24145 * cgraphunit.c (cgraph_node::finalize_function): When
24146 !flag_toplevel_reorde set no_reorder flag.
24147 (varpool_node::finalize_decl): Likewise.
24148 (symbol_table::compile): Drop no toplevel reorder path.
24150 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24152 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
24153 edges; zero probability is not better than uninitialized.
24155 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24157 * asan.h (asan_sanitize_allocas_p): Declare.
24158 * asan.c (asan_sanitize_allocas_p): New function.
24159 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
24160 (handle_builtin_alloca): Likewise.
24161 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
24162 if !asan_sanitize_allocas_p.
24163 * params.def (asan-instrument-allocas): Add new option.
24164 * params.h (ASAN_PROTECT_ALLOCAS): Define.
24165 * opts.c (common_handle_option): Disable allocas sanitization for
24168 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24170 * asan.c: Include gimple-fold.h.
24171 (get_last_alloca_addr): New function.
24172 (handle_builtin_stackrestore): Likewise.
24173 (handle_builtin_alloca): Likewise.
24174 (asan_emit_allocas_unpoison): Likewise.
24175 (get_mem_refs_of_builtin_call): Add new parameter, remove const
24176 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
24177 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
24178 (instrument_builtin_call): Pass gimple iterator to
24179 get_mem_refs_of_builtin_call.
24180 (last_alloca_addr): New global.
24181 * asan.h (asan_emit_allocas_unpoison): Declare.
24182 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
24183 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
24184 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
24185 if function calls alloca.
24186 * gimple-fold.c (replace_call_with_value): Remove static keyword.
24187 * gimple-fold.h (replace_call_with_value): Declare.
24188 * internal-fn.c: Include asan.h.
24189 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
24190 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
24192 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24194 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
24195 (C_SELFTEST_FLAGS): New.
24196 (CPP_SELFTEST_FLAGS): New.
24197 (SELFTEST_DEPS): New, from deps of s-selftest.
24198 (C_SELFTEST_DEPS): New, from deps of s-selftest.
24199 (CPP_SELFTEST_DEPS): New.
24200 (selftest): Add dependency on s-selftest-c++.
24201 (s-selftest): Rename to...
24202 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
24203 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
24204 than SELFTEST_FLAGS.
24205 (selftest-gdb): Rename to...
24206 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
24208 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
24209 (selftest-valgrind): Rename to...
24210 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
24212 (selftest-valgrind): Reintroduce as an alias for
24213 selftest-c-valgrind.
24214 (s-selftest-c++): New.
24215 (selftest-c++-gdb): New.
24216 (selftest-c++-valgrind): New.
24218 2017-07-06 Olivier Hainque <hainque@adacore.com>
24220 * gcc.c (process_command): When deciding if undefined variables
24221 should be ignored when processing specs, accept "gcc -v" as well.
24223 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24225 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
24226 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
24228 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24230 * config/arm/arm-cpus.in (armv8-r): Add new entry.
24231 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
24232 * config/arm/arm-tables.opt: Regenerate.
24233 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
24235 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
24237 2017-07-06 Carl Love <cel@us.ibm.com>
24239 * ChangeLog: Clean up from mid air collision
24241 2017-07-06 Carl Love <cel@us.ibm.com>
24243 * config/rs6000/rs6000-c.c: Add support for built-in functions
24244 vector signed int vec_subc (vector signed int, vector signed int);
24245 vector signed __int128 vec_subc (vector signed __int128,
24246 vector signed __int128);
24247 vector unsigned __int128 vec_subc (vector unsigned __int128,
24248 vector unsigned __int128);
24249 vector signed int vec_sube (vector signed int, vector signed int,
24250 vector signed int);
24251 vector unsigned int vec_sube (vector unsigned int,
24252 vector unsigned int,
24253 vector unsigned int);
24254 vector signed __int128 vec_sube (vector signed __int128,
24255 vector signed __int128,
24256 vector signed__int128);
24257 vector unsigned __int128 vec_sube (vector unsigned __int128,
24258 vector unsigned __int128,
24259 vector unsigned __int128);
24260 vector signed int vec_subec (vector signed int, vector signed int,
24261 vector signed int);
24262 vector unsigned int vec_subec (vector unsigned int,
24263 vector unsigned int,
24264 vector unsigned int);
24265 vector signed __int128 vec_subec (vector signed __int128,
24266 vector signed __int128,
24267 vector signed__int128);
24268 vector unsigned __int128 vec_subec (vector unsigned __int128,
24269 vector unsigned __int128,
24270 vector unsigned __int128);
24271 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
24272 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
24273 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
24274 BU_ALTIVEC_OVERLOAD_X definitions.
24275 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
24276 * doc/extend.texi: Update the built-in documentation file for the new
24277 built-in functions.
24279 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24282 * diagnostic-show-locus.c (layout::layout): Use start and finish
24283 spelling location for the start and finish of each range.
24284 * genmatch.c (linemap_client_expand_location_to_spelling_point):
24285 Add unused aspect param.
24286 * input.c (expand_location_1): Add "aspect" param, and use it
24287 to access the correct part of the location.
24288 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
24290 (expand_location_to_spelling_point): Likewise.
24291 (linemap_client_expand_location_to_spelling_point): Add "aspect"
24292 param, and pass it to expand_location_1.
24294 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
24296 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
24297 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
24298 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
24299 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
24300 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
24301 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
24302 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
24303 _mm_maskz_getmant_ss): New intrinsics.
24304 (__builtin_ia32_getexpss128_mask): Changed to ...
24305 __builtin_ia32_getexpss128_round ... this.
24306 (__builtin_ia32_getexpsd128_mask): Changed to ...
24307 __builtin_ia32_getexpsd128_round ... this.
24308 * config/i386/i386-builtin-types.def
24309 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
24310 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
24311 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
24312 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
24313 __builtin_ia32_getmantss_mask_round): New builtins.
24314 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
24315 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
24316 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
24317 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
24318 * config/i386/sse.md
24319 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
24320 avx512f_sgetexp<mode><mask_scalar_name>
24321 <round_saeonly_scalar_name> ... this.
24322 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
24323 %0, %1, %2<round_saeonly_op3>}): Changed to ...
24324 vgetexp<ssescalarmodesuffix>
24325 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24326 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
24327 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
24328 avx512f_vgetmant<mode><mask_scalar_name>
24329 <round_saeonly_scalar_name> ... this.
24330 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24331 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
24332 vgetmant<ssescalarmodesuffix>
24333 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
24334 %0<mask_scalar_operand4>, %1, %2
24335 <round_saeonly_scalar_mask_op4>, %3} ... this.
24336 * config/i386/subst.md (mask_scalar_operand4,
24337 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
24338 round_saeonly_scalar_nimm_predicate): New subst attributes.
24340 2017-07-06 Julia Koval <julia.koval@intel.com>
24342 * config/i386/i386.c (ix86_erase_embedded_rounding):
24343 Remove code for old rounding pattern.
24345 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
24347 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
24349 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
24351 * doc/sourcebuild.texi (Test Directives, Variants of
24352 dg-require-support): Add documentation for dg-require-stack-check.
24354 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
24356 * config/i386/subst.md (mask_scalar, round_scalar,
24357 round_saeonly_scalar): New meta-templates.
24358 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
24359 round_scalar_mask_operand3, round_scalar_mask_op3,
24360 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
24361 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
24362 round_saeonly_scalar_constraint,
24363 round_saeonly_scalar_prefix): New subst attribute.
24364 * config/i386/sse.md
24365 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
24366 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
24367 <round_scalar_name> ... this.
24368 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
24369 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
24370 <round_scalar_name> ... this.
24371 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
24372 <sse>_vm<code><mode>3<mask_scalar_name>
24373 <round_saeonly_scalar_name> ... this.
24374 (v<plusminus_mnemonic><ssescalarmodesuffix>
24375 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24376 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24377 v<plusminus_mnemonic><ssescalarmodesuffix>
24378 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24379 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24380 (v<multdiv_mnemonic><ssescalarmodesuffix>
24381 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24382 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24383 v<multdiv_mnemonic><ssescalarmodesuffix>
24384 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24385 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24386 (v<maxmin_float><ssescalarmodesuffix>
24387 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
24388 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
24389 v<maxmin_float><ssescalarmodesuffix>
24390 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24391 %0<mask_scalar_operand3>, %1, %<iptr>2
24392 <round_saeonly_scalar_mask_op3>} ... this.
24394 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
24396 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
24397 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
24399 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
24400 Alan Hayward <alan.hayward@arm.com>
24401 David Sherwood <david.sherwood@arm.com>
24403 * combine.c (simplify_if_then_else): Remove "enum" before
24405 * compare-elim.c (can_eliminate_compare): Likewise.
24406 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
24408 (aarch64_lookup_simd_builtin_type): Likewise.
24409 (aarch64_simd_builtin_type): Likewise.
24410 (aarch64_init_simd_builtin_types): Likewise.
24411 (aarch64_simd_expand_args): Likewise.
24412 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
24414 (aarch64_reverse_mask): Likewise.
24415 (aarch64_simd_emit_reg_reg_move): Likewise.
24416 (aarch64_gen_adjusted_ldpstp): Likewise.
24417 (aarch64_ccmp_mode_to_code): Likewise.
24418 (aarch64_operands_ok_for_ldpstp): Likewise.
24419 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24420 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
24422 (aarch64_min_divisions_for_recip_mul): Likewise.
24423 (aarch64_reassociation_width): Likewise.
24424 (aarch64_get_condition_code_1): Likewise.
24425 (aarch64_simd_emit_reg_reg_move): Likewise.
24426 (aarch64_simd_attr_length_rglist): Likewise.
24427 (aarch64_reverse_mask): Likewise.
24428 (aarch64_operands_ok_for_ldpstp): Likewise.
24429 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24430 (aarch64_gen_adjusted_ldpstp): Likewise.
24431 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
24433 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
24434 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
24435 (arm_lookup_simd_builtin_type): Likewise.
24436 (arm_simd_builtin_type): Likewise.
24437 (arm_init_simd_builtin_types): Likewise.
24438 (arm_expand_builtin_args): Likewise.
24439 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
24440 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
24441 (ft32_setup_incoming_varargs): Likewise.
24442 (ft32_function_arg): Likewise.
24443 (ft32_function_arg_advance): Likewise.
24444 (ft32_pass_by_reference): Likewise.
24445 (ft32_arg_partial_bytes): Likewise.
24446 (ft32_valid_pointer_mode): Likewise.
24447 (ft32_addr_space_pointer_mode): Likewise.
24448 (ft32_addr_space_legitimate_address_p): Likewise.
24449 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
24451 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
24452 (ix86_emit_outlined_ms2sysv_restore): Likewise.
24453 (iamcu_alignment): Likewise.
24454 (canonicalize_vector_int_perm): Likewise.
24455 (ix86_noce_conversion_profitable_p): Likewise.
24456 (ix86_mpx_bound_mode): Likewise.
24457 (ix86_operands_ok_for_move_multiple): Likewise.
24458 * config/microblaze/microblaze-protos.h
24459 (microblaze_expand_conditional_branch_reg): Likewise.
24460 * config/microblaze/microblaze.c
24461 (microblaze_expand_conditional_branch_reg): Likewise.
24462 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
24464 (rs6000_reassociation_width): Likewise.
24465 (rs6000_invalid_binary_op): Likewise.
24466 (fusion_p9_p): Likewise.
24467 (emit_fusion_p9_load): Likewise.
24468 (emit_fusion_p9_store): Likewise.
24469 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
24471 (riscv_hard_regno_mode_ok_p): Likewise.
24472 (riscv_address_insns): Likewise.
24473 (riscv_split_symbol): Likewise.
24474 (riscv_legitimize_move): Likewise.
24475 (riscv_function_value): Likewise.
24476 (riscv_hard_regno_nregs): Likewise.
24477 (riscv_expand_builtin): Likewise.
24478 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
24479 (riscv_build_integer): Likewise.
24480 (riscv_split_integer): Likewise.
24481 (riscv_legitimate_constant_p): Likewise.
24482 (riscv_cannot_force_const_mem): Likewise.
24483 (riscv_regno_mode_ok_for_base_p): Likewise.
24484 (riscv_valid_base_register_p): Likewise.
24485 (riscv_valid_offset_p): Likewise.
24486 (riscv_valid_lo_sum_p): Likewise.
24487 (riscv_classify_address): Likewise.
24488 (riscv_legitimate_address_p): Likewise.
24489 (riscv_address_insns): Likewise.
24490 (riscv_load_store_insns): Likewise.
24491 (riscv_force_binary): Likewise.
24492 (riscv_split_symbol): Likewise.
24493 (riscv_force_address): Likewise.
24494 (riscv_legitimize_address): Likewise.
24495 (riscv_move_integer): Likewise.
24496 (riscv_legitimize_const_move): Likewise.
24497 (riscv_legitimize_move): Likewise.
24498 (riscv_address_cost): Likewise.
24499 (riscv_subword): Likewise.
24500 (riscv_output_move): Likewise.
24501 (riscv_canonicalize_int_order_test): Likewise.
24502 (riscv_emit_int_order_test): Likewise.
24503 (riscv_function_arg_boundary): Likewise.
24504 (riscv_pass_mode_in_fpr_p): Likewise.
24505 (riscv_pass_fpr_single): Likewise.
24506 (riscv_pass_fpr_pair): Likewise.
24507 (riscv_get_arg_info): Likewise.
24508 (riscv_function_arg): Likewise.
24509 (riscv_function_arg_advance): Likewise.
24510 (riscv_arg_partial_bytes): Likewise.
24511 (riscv_function_value): Likewise.
24512 (riscv_pass_by_reference): Likewise.
24513 (riscv_setup_incoming_varargs): Likewise.
24514 (riscv_print_operand): Likewise.
24515 (riscv_elf_select_rtx_section): Likewise.
24516 (riscv_save_restore_reg): Likewise.
24517 (riscv_for_each_saved_reg): Likewise.
24518 (riscv_register_move_cost): Likewise.
24519 (riscv_hard_regno_mode_ok_p): Likewise.
24520 (riscv_hard_regno_nregs): Likewise.
24521 (riscv_class_max_nregs): Likewise.
24522 (riscv_memory_move_cost): Likewise.
24523 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
24524 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
24525 (rl78_addr_space_address_mode): Likewise.
24526 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
24528 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
24529 (rs6000_reassociation_width): Likewise.
24530 (rs6000_invalid_binary_op): Likewise.
24531 (fusion_p9_p): Likewise.
24532 (emit_fusion_p9_load): Likewise.
24533 (emit_fusion_p9_store): Likewise.
24534 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
24535 (ok_for_simple_move_operands): Likewise.
24536 (ok_for_simple_move_strict_operands): Likewise.
24537 (ok_for_simple_arith_logic_operands): Likewise.
24538 (visium_legitimize_reload_address): Likewise.
24539 (visium_select_cc_mode): Likewise.
24540 (output_cbranch): Likewise.
24541 (visium_split_double_move): Likewise.
24542 (visium_expand_copysign): Likewise.
24543 (visium_expand_int_cstore): Likewise.
24544 (visium_expand_fp_cstore): Likewise.
24545 * config/visium/visium.c (visium_pass_by_reference): Likewise.
24546 (visium_function_arg): Likewise.
24547 (visium_function_arg_advance): Likewise.
24548 (visium_libcall_value): Likewise.
24549 (visium_setup_incoming_varargs): Likewise.
24550 (visium_legitimate_constant_p): Likewise.
24551 (visium_legitimate_address_p): Likewise.
24552 (visium_legitimize_address): Likewise.
24553 (visium_secondary_reload): Likewise.
24554 (visium_register_move_cost): Likewise.
24555 (visium_memory_move_cost): Likewise.
24556 (prepare_move_operands): Likewise.
24557 (ok_for_simple_move_operands): Likewise.
24558 (ok_for_simple_move_strict_operands): Likewise.
24559 (ok_for_simple_arith_logic_operands): Likewise.
24560 (visium_function_value_1): Likewise.
24561 (rtx_ok_for_offset_p): Likewise.
24562 (visium_legitimize_reload_address): Likewise.
24563 (visium_split_double_move): Likewise.
24564 (visium_expand_copysign): Likewise.
24565 (visium_expand_int_cstore): Likewise.
24566 (visium_expand_fp_cstore): Likewise.
24567 (visium_split_cstore): Likewise.
24568 (visium_select_cc_mode): Likewise.
24569 (visium_split_cbranch): Likewise.
24570 (output_cbranch): Likewise.
24571 (visium_print_operand_address): Likewise.
24572 * expmed.c (flip_storage_order): Likewise.
24573 * expmed.h (emit_cstore): Likewise.
24574 (flip_storage_order): Likewise.
24575 * genrecog.c (validate_pattern): Likewise.
24576 * hsa-gen.c (gen_hsa_addr): Likewise.
24577 * internal-fn.c (expand_arith_overflow): Likewise.
24578 * ira-color.c (allocno_copy_cost_saving): Likewise.
24579 * lra-assigns.c (find_hard_regno_for_1): Likewise.
24580 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
24581 (process_invariant_for_inheritance): Likewise.
24582 * lra-eliminations.c (move_plus_up): Likewise.
24583 * omp-low.c (lower_oacc_reductions): Likewise.
24584 * simplify-rtx.c (simplify_subreg): Likewise.
24585 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
24586 (TARGET_CHKP_BOUND_MODE): Likewise..
24587 * targhooks.c (default_chkp_bound_mode): Likewise.
24588 (default_setup_incoming_vararg_bounds): Likewise.
24589 * targhooks.h (default_chkp_bound_mode): Likewise.
24590 (default_setup_incoming_vararg_bounds): Likewise.
24591 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
24592 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
24593 (have_whole_vector_shift): Likewise.
24594 * tree-vect-stmts.c (vectorizable_load): Likewise.
24595 * doc/tm.texi: Regenerate.
24597 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24599 Graceful degrade if Binutils PR21472 is not available.
24602 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
24603 .rodata in flash test fails.
24604 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
24605 * confgure: Regenerate.
24606 * config.in: Regenerate.
24607 * config/avr/avr.c (avr_asm_named_section)
24608 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
24609 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
24610 (avr_asm_init_sections): Same.
24612 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24614 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
24615 (fma<VH:mode>4_intrinsic): Likewise.
24616 (*fmsub<VCVTF:mode>4): Likewise.
24617 (*fmsub<VH:mode>4_intrinsic): Likewise.
24619 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24622 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
24623 Don't depend on "optimize > 0".
24624 (out_movhi_r_mr, out_movqi_mr_r): Same.
24625 (out_movhi_mr_r, out_movqi_r_mr): Same.
24626 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
24627 io_address_operand on "optimize > 0".
24629 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24631 * tree-loop-distribution.c: Add general explanantion on the pass.
24632 (generate_loops_for_partition): Mark distributed loop.
24633 (pg_add_dependence_edges): New parameter. Handle alias data
24634 dependence specially and record it in the parameter if asked.
24635 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
24636 (init_partition_graph_vertices, add_partition_graph_edge): New.
24637 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
24638 (free_partition_graph_vdata, build_partition_graph): New.
24639 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
24640 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
24641 (data_ref_segment_size, latch_dominated_by_data_ref): New.
24642 (compute_alias_check_pairs, version_loop_by_alias_check): New.
24643 (version_for_distribution_p, finalize_partitions): New.
24644 (distribute_loop): Handle alias data dependence specially. Factor
24645 out loop fusion code as functions and call these functions.
24647 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24649 * tree-loop-distribution.c (classify_partition): New parameter and
24650 better handle reduction statement.
24651 (rdg_build_partitions): Revise comment.
24652 (distribute_loop): Compute statements in all partitions and pass it
24653 to classify_partition.
24655 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24657 * tree-loop-distribution.c (enum partition_type): New.
24658 (struct partition): New field type.
24659 (partition_merge_into): Add parameter. Update partition type.
24660 (data_dep_in_cycle_p, update_type_for_merge): New functions.
24661 (build_rdg_partition_for_vertex): Compute partition type.
24662 (rdg_build_partitions): Dump partition type.
24663 (distribute_loop): Update calls to partition_merge_into.
24665 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24667 * tree-loop-distribution.c (struct ddr_hasher): New.
24668 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
24670 (classify_partition): Call get_data_dependence.
24671 (pg_add_dependence_edges): Ditto.
24672 (distribute_loop): Release data dependence hash table.
24674 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24676 * tree-loop-distribution.c (ref_base_address): Delete.
24677 (similar_memory_accesses): Rename ...
24678 (share_memory_accesses): ... to this. Check if partitions access
24679 the same memory reference.
24680 (distribute_loop): Call share_memory_accesses.
24682 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24684 * tree-loop-distribution.c (struct partition): New field recording
24685 its data reference.
24686 (partition_alloc, partition_free): Init and release data refs.
24687 (partition_merge_into): Merge data refs.
24688 (build_rdg_partition_for_vertex): Collect data refs for partition.
24689 (pg_add_dependence_edges): Change parameters from vector to bitmap.
24691 (distribute_loop): Remve data refs from vertice data of partition
24694 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24696 * tree-loop-distribution.c (params.h): Include header file.
24697 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
24698 (datarefs_vec): New global var.
24699 (create_rdg_vertices): Use datarefs_vec directly.
24700 (free_rdg): Don't free data references.
24701 (build_rdg): Update use. Don't free data references.
24702 (distribute_loop): Compute global variable for data references.
24703 Bail out if there are too many data references.
24705 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24707 * tree-loop-distribution.c (loop_nest): New global var.
24708 (build_rdg): Use loop directly, rather than loop nest.
24709 (pg_add_dependence_edges): Remove loop nest parameter. Use global
24711 (distribute_loop): Compute global variable loop nest. Update use.
24713 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24715 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
24716 (partition_merge_into): New parameter. Dump reason for fusion.
24717 (distribute_loop): Update use of partition_merge_into.
24719 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24721 * tree-loop-distribution.c (bb_top_order_index): New.
24722 (bb_top_order_index_size, bb_top_order_cmp): New.
24723 (stmts_from_loop): Use topological order.
24724 (pass_loop_distribution::execute): Compute and release topological
24725 order for basic blocks.
24727 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24729 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
24732 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24734 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
24735 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
24736 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
24737 * internal-fn.def (LOOP_DIST_ALIAS): New.
24738 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
24739 (fold_loop_internal_call): ... this.
24740 (vect_loop_dist_alias_call): New function.
24741 (set_uid_loop_bbs): Call fold_loop_internal_call.
24742 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
24745 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24748 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
24749 Require dead FLAGS_REG at the beginning of a peephole.
24751 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24754 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
24755 arguments in the call to __builtin_ia32_sbb_u32.
24756 (_subborrow_u64): Swap _X and _Y arguments in the call to
24757 __builtin_ia32_sbb_u64.
24759 2017-07-04 Jakub Jelinek <jakub@redhat.com>
24762 * tree-vrp.c (compare_assert_loc): Turn into a function template
24763 with stable template parameter. Only test if a->e is NULL,
24764 !a->e == !b->e has been verified already. Use e == NULL or
24765 e != NULL instead of e or ! e tests. If stable is true, don't use
24766 iterative_hash_expr, on the other side allow a or b or both NULL
24767 and sort the NULLs last.
24768 (process_assert_insertions): Sort using compare_assert_loc<false>
24769 instead of compare_assert_loc, later sort using
24770 compare_assert_loc<true> before calling process_assert_insertions_for
24771 in a loop. Use break instead of continue once seen NULL pointer.
24773 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24775 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24776 Cortex-R7 and Cortex-R8 processors.
24778 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24780 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
24781 uninitialized while src is not.
24783 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
24785 * common/config/arm/arm-common.c: Adjust include path for
24787 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
24788 (arm-cpu.h): Create in build directory. Adjust dependency rules.
24789 (arm-cpu-data.h): Likewise.
24790 (arm-cpu-cdata.h): Likewise.
24791 * config/arm/arm-cpu.h: Delete.
24792 * config/arm/arm-cpu-cdata.h: Delete.
24793 * config/arm/arm-cpu-data.h: Delete.
24795 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
24797 * config/arm/arm-cpus.in (cortex-a55): New.
24798 (cortex-a75): Likewise.
24799 (cortex-a75.cortex-a55): Likewise.
24800 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
24802 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
24803 * config/arm/arm-cpu-cdata.h: Regenerate.
24804 * config/arm/arm-cpu-data.h: Regenerate.
24805 * config/arm/arm-cpu.h: Regenerate.
24806 * config/arm/arm-tables.opt: Regenerate.
24807 * config/arm/arm-tune.md: Regenerate.
24809 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24811 * haifa-sched.c (sched_create_recovery_edges): Update profile.
24813 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24815 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
24818 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
24820 PR tree-optimization/81292
24821 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
24822 full_string_p, also call adjust_related_strinfos if the adjustment
24823 is simple, otherwise invalidate related strinfos.
24825 2017-07-04 Martin Liska <mliska@suse.cz>
24828 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
24829 newly created variable as DECL_IGNORED_P.
24831 2017-07-04 Martin Liska <mliska@suse.cz>
24834 * ipa-inline.c (inline_small_functions):
24835 Use xstrdup_for_dump.
24837 2017-07-04 Tom de Vries <tom@codesourcery.com>
24839 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
24841 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24844 * config/darwin.c (darwin_function_switched_text_sections):
24847 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
24849 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
24851 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24853 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
24855 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24857 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
24858 min_profitable_iters, and th as inclusive lower bounds.
24859 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
24860 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
24861 for min_profitable_iters and min_profitable_estimate.
24862 (vect_transform_loop): Treat th as an inclusive lower bound.
24863 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24865 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24868 * config/darwin.c (darwin_function_switched_text_sections):
24869 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
24870 in two pieces, and suppress the use of buf.
24872 2017-07-03 Nathan Sidwell <nathan@acm.org>
24874 * hash-table.h (hash_table_mod1): Fix indentation.
24876 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24878 PR middle-end/81290
24879 * predict.c (force_edge_cold): Be more careful about propagation
24881 * profile-count.h (profile_probability::guessed,
24882 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
24884 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
24886 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
24888 * doc/invoke.texi (rcpc architecture extension): Document it.
24890 2017-07-03 Richard Biener <rguenther@suse.de>
24892 PR tree-optimization/60510
24893 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
24894 the scalar reduction PHI and use it.
24895 (vectorizable_reduction): Properly guard the single_defuse_cycle
24896 path for non-SLP reduction chains where we cannot use it.
24897 Rework reduc_def/index and vector type deduction. Rework
24898 vector operand gathering during reduction op code-gen.
24899 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
24900 chains dissolve the chain and leave it to non-SLP reduction
24903 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24905 * tree-data-ref.h (dr_alignment): Declare.
24906 * tree-data-ref.c (dr_alignment): New function.
24907 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
24908 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
24910 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
24912 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24914 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
24915 and base_misalignment fields.
24916 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
24917 * tree-data-ref.c: Include builtins.h.
24918 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
24919 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
24920 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
24921 * tree-vect-data-refs.c: Include tree-cfg.h.
24922 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
24923 fields instead of calculating an alignment here.
24924 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
24925 innermost_loop_behavior fields.
24927 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24929 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
24931 (DR_STEP_ALIGNMENT): New macro.
24932 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
24933 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
24934 (create_data_ref): Print it.
24935 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
24936 to tell whether the step preserves vector (mis)alignment.
24937 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24938 Move the check for an integer step and generalise to all INTEGER_CST.
24939 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
24940 Print the outer step alignment.
24942 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24944 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
24945 with offset_alignment.
24946 (DR_ALIGNED_TO): Delete.
24947 (DR_OFFSET_ALIGNMENT): New macro.
24948 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
24949 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
24950 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
24951 (create_data_ref): Likewise.
24952 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24953 (vect_analyze_data_refs): Likewise.
24954 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
24955 creating dummy innermost behavior.
24957 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24959 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
24960 with a "innermost_loop_behavior *" and refeence tree.
24961 * tree-data-ref.c (dr_analyze_innermost): Likewise.
24962 (create_data_ref): Update call accordingly.
24963 * tree-predcom.c (find_looparound_phi): Likewise.
24965 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24967 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
24968 fields with dr_wrt_vec_loop.
24969 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
24970 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
24971 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
24972 (vect_dr_behavior): New function.
24973 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24974 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
24975 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
24976 track whether the step preserves the misalignment.
24977 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24978 Use vect_dr_behavior.
24979 (vect_setup_realignment): Update call accordingly.
24980 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
24981 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
24982 call to vect_create_addr_base_for_vector_ref.
24983 (vect_create_cond_for_align_checks): Likewise.
24984 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
24985 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
24986 (vect_recog_mask_conversion_pattern): Likewise.
24987 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
24988 (new_stmt_vec_info): Remove redundant zeroing.
24990 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24992 * common/config/arm/arm-common.c (arm_be8_option): New function.
24993 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
24994 (ISA_ARMv6): Add isa_bit_be8.
24995 * config/arm/arm.h (arm_be8_option): Add prototype.
24996 (BE8_SPEC_FUNCTION): New define.
24997 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
24998 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
24999 (mlittle-endian): Similarly.
25000 (mbe8, mbe32): New options.
25001 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
25002 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
25004 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25006 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
25008 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25010 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
25011 (cleanup_tree_cfg_bb): Use it.
25012 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
25014 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
25016 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25019 * loop-doloop.c (add_test): Update profile.
25021 2017-07-03 Martin Liska <mliska@suse.cz>
25024 * sanopt.c (rewrite_usage_of_param): New function.
25025 (sanitize_rewrite_addressable_params): Likewise.
25026 (pass_sanopt::execute): Call rewrite_usage_of_param.
25028 2017-07-03 Richard Biener <rguenther@suse.de>
25030 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
25031 back to using VIEW_CONVERT_EXPR.
25033 2017-07-03 Martin Liska <mliska@suse.cz>
25036 * doc/extend.texi: Document when a resolver function is
25037 generated for target_clones.
25039 2017-07-03 Martin Liska <mliska@suse.cz>
25041 * asan.c (asan_emit_stack_protection): Unpoison just red zones
25042 and shadow memory of auto variables which are subject of
25043 use-after-scope sanitization.
25044 (asan_expand_mark_ifn): Add do set only when is_poison.
25046 2016-07-03 Richard Biener <rguenther@suse.de>
25048 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
25050 (vect_force_simple_reduction): Record reduction def -> phi mapping.
25051 (vectorizable_reduction): Perform reduction PHI creation when
25052 visiting a reduction PHI and adjust and simplify code generation
25053 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
25054 (vect_transform_loop): Visit reduction PHIs.
25055 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
25056 defs into the SLP tree.
25057 (vect_build_slp_tree): Reduction defs terminate the recursion.
25058 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
25060 (vect_get_vec_defs_for_stmt_copy): Export.
25061 (vect_get_vec_defs): Likewise.
25062 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
25064 (vect_get_vec_defs_for_stmt_copy): Declare.
25065 (vect_get_vec_defs): Likewise.
25067 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25069 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
25070 parameter with a "loop" parameter and use it instead of the
25071 loop containing DR_STMT. Don't check simple_iv when doing
25072 BB analysis. Describe the two analysis modes in the comment.
25074 2017-07-03 Tom de Vries <tom@codesourcery.com>
25076 PR tree-optimization/69468
25077 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
25078 (find_same_succ_bb): Handle ignore_edge_flags.
25080 2017-07-03 Tom de Vries <tom@codesourcery.com>
25082 PR tree-optimization/81192
25083 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
25085 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
25087 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
25089 2017-07-03 Tom de Vries <tom@codesourcery.com>
25091 PR tree-optimization/81192
25092 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
25093 BB_SAME_SUCC (bb) == NULL.
25095 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25097 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
25100 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25102 * dumpfile.c: Include profile-count.h
25103 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
25105 (insert_cond_bb): Update profile.
25106 * tree-cfg.h (insert_cond_bb): Update prototype.
25107 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
25108 * tree-dump.c: Do not include tree-cfg.
25110 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25112 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
25114 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25116 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
25119 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25121 * tree-complex.c (expand_complex_div_wide): update profile.
25123 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25124 Alan Hayward <alan.hayward@arm.com>
25125 David Sherwood <david.sherwood@arm.com>
25127 * Makefile.in (MACHMODE_H): Remove insn-modes.h
25128 (CORETYPES_H): New define.
25129 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
25130 (insn-modes-inline.h, s-modes-inline-h): New rules.
25131 (generated_files): Add insn-modes-inline.h.
25132 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
25133 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
25134 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
25135 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
25136 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
25137 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
25138 (build/gencodes.o, build/genconditions.o): Likewise.
25139 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
25140 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
25141 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
25142 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
25143 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
25144 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
25145 * coretypes.h: Include everything up to real.h for generators.
25146 Include insn-modes.h first. Include wide-int-print.h after
25147 wide-int.h. Include insn-modes-inline.h and then machmode.h.
25148 * machmode.h: Don't include insn-modes.h here.
25149 * function-tests.c: Remove includes of signop.h, machmode.h,
25150 double-int.h and wide-int.h.
25152 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
25154 * optc-save-gen.awk: Likewise.
25155 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
25156 * godump.c: Remove include of wide-int-print.h.
25157 * pretty-print.h: Likewise.
25158 * wide-int-print.cc: Likewise.
25159 * wide-int.cc: Likewise.
25160 * hash-map-tests.c: Remove include of signop.h.
25161 * hash-set-tests.c: Likewise.
25162 * rtl-tests.c: Likewise.
25163 * mkconfig.sh: Remove include of machmode.h.
25164 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
25166 (emit_insn_modes_inline_h): ...this new function. Emit the code
25167 into an insn-modes-inline.h header file, adding appropriate
25168 include guards and end comments.
25169 (emit_insn_modes_c_header): Remove include of machmode.h.
25170 (emit_min_insn_modes_c_header): Include coretypes.h rather than
25172 (main): Handle -i flag and call emit_insn_modes_inline_h when
25175 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25177 * tree-ssa-strlen.c (strinfo): Rename the length field to
25178 nonzero_chars. Add a full_string_p field.
25179 (compare_nonzero_chars, zero_length_string_p): New functions.
25180 (get_addr_stridx): Add an offset_out parameter.
25181 Use compare_nonzero_chars.
25182 (get_stridx): Update accordingly. Use compare_nonzero_chars.
25183 (new_strinfo): Update after above changes to strinfo.
25184 (set_endptr_and_length): Set full_string_p.
25185 (get_string_length): Update after above changes to strinfo.
25186 (unshare_strinfo): Update call to new_strinfo.
25187 (maybe_invalidate): Likewise.
25188 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
25189 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
25190 as a uhwi instead of an shwi. Update after above changes to
25191 strinfo and new_strinfo.
25192 (zero_length_string): Assert that chainsi contains full strings.
25193 Use zero_length_string_p. Update call to new_strinfo.
25194 (adjust_related_strinfos): Update after above changes to strinfo.
25195 Copy full_string_p from origsi.
25196 (adjust_last_stmt): Use zero_length_string_p.
25197 (handle_builtin_strlen): Update after above changes to strinfo and
25198 new_strinfo. Install the lhs as the string length if the previous
25199 entry didn't describe a full string.
25200 (handle_builtin_strchr): Update after above changes to strinfo
25202 (handle_builtin_strcpy): Likewise.
25203 (handle_builtin_strcat): Likewise.
25204 (handle_builtin_malloc): Likewise.
25205 (handle_pointer_plus): Likewise.
25206 (handle_builtin_memcpy): Likewise. Track nonzero characters
25207 that aren't necessarily followed by a nul terminator.
25208 (handle_char_store): Likewise.
25210 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25212 PR tree-optimization/80769
25213 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
25214 for malloc and calloc. Document the new invariant that all related
25215 strinfos have delayed lengths or none do.
25216 (verify_related_strinfos): Move earlier in file.
25217 (set_endptr_and_length): New function, split out from...
25218 (get_string_length): ...here. Also set the lengths of related
25220 (zero_length_string): Assert that chainsi has known (rather than
25222 (adjust_related_strinfos): Likewise.
25224 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25226 PR tree-optimization/81136
25227 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
25228 assert that two references with the same misalignment have the same
25229 compile-time misalignment if those compile-time misalignments
25232 2017-07-01 Andi Kleen <ak@linux.intel.com>
25234 * print-tree.c (print_node): Print all attributes.
25236 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25238 * cfg.c (scale_bbs_frequencies): New function.
25239 * cfg.h (scale_bbs_frequencies): Declare it.
25240 * cfgloopanal.c (single_likely_exit): Cleanup.
25241 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
25243 (scale_loop_profile): Likewise.
25244 (loop_version): Likewise.
25245 (create_empty_loop_on_edge): Update.
25246 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
25247 scale_loop_frequencies, scale_loop_profile, loopify,
25248 loop_version): Update prototypes.
25249 * modulo-sched.c (sms_schedule): Update.
25250 * predict.c (unlikely_executed_edge_p): Also check probability.
25251 (probably_never_executed_edge_p): Fix typo.
25252 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25253 * tree-parloops.c (gen_parallel_loop): Update.
25254 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
25255 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25256 * tree-ssa-loop-split.c (split_loop): Update.
25257 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25258 * tree-vect-loop-manip.c (vect_do_peeling): Update.
25259 (vect_loop_versioning): Update.
25260 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25262 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25264 * trans-mem.c (split_bb_make_tm_edge): Update profile.
25266 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25268 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
25269 to keep profile consistent.
25271 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25273 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
25274 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
25275 * profile-count.h (max_safe_multiplier): Make unsigned.
25276 (profile_count::guessed_zero): New.
25278 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25280 * bb-reorder.c (fix_up_crossing_landing_pad,
25281 fix_crossing_conditional_branches): Use make_single_succ_edge
25282 to keep profile consistent.
25284 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25286 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
25289 2017-07-01 Jakub Jelinek <jakub@redhat.com>
25292 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
25293 the right scopes, make sure cond_jump isn't preserved between multiple
25294 iterations. Search for fallthru edge whenever there are 3+ edges and
25295 use find_fallthru_edge for it.
25297 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25299 Patch by Alexander Monakov <amonakov@ispras.ru>
25300 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
25301 probabilities consistently.
25303 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25305 * pa.c (pa_expand_compare_and_swap_loop): Update call of
25306 emit_cmp_and_jump_insns.
25308 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25311 * tree-inline.c (expand_call_inline): Combine profile statuses.
25313 2017-06-30 Andrew Pinski <apinski@cavium.com>
25315 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
25316 fold_stmt returned true.
25318 2017-06-30 Nathan Sidwell <nathan@acm.org>
25320 * ggc.h (empty_string): Delete.
25321 * cfgexpand.c (expand_asm_stmt): Use plain "".
25322 * optabs.c (expand_asm_memory_barrier): Likewise.
25323 * stringpool.c (empty_string): Delete.
25324 (digit_vector, digit_string): Delete.
25325 (ggc_alloc_string): Use plain "", don't optimize single digit
25326 strings. Use ggc_alloc_atomic.
25328 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
25330 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
25331 comparison set and one other set, use the cost of the non-comparison
25334 2017-06-30 Nathan Sidwell <nathan@acm.org>
25336 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
25339 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
25341 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
25342 loops. Remove now unneeded calls to gimple_switch_set_label() that
25343 just set removed labels to NULL_TREE.
25345 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
25347 * tree-ssanames.c (set_range_info_raw): Abstract from ...
25348 (set_range_info): ...here. Only call set_range_info_raw if domain
25350 (set_nonzero_bits): Call set_range_info_raw.
25351 * tree-ssanames.h (set_range_info_raw): New.
25353 2017-06-30 Jakub Jelinek <jakub@redhat.com>
25356 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
25357 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
25358 of nonimmediate_operand and <store_mask_constraint> instead of m
25359 for the input operand. For V8FI iterator, always split if input
25360 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
25361 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
25362 <store_mask_predicate> instead of register_operand and
25363 <store_mask_constraint> instead of v for the input operand. Make
25364 sure both operands aren't MEMs for if not <mask_applied>.
25366 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
25368 * lto-wrapper.c (copy_file) Close both file descriptors before
25371 2017-06-30 Martin Liska <mliska@suse.cz>
25374 * multiple_target.c (create_dispatcher_calls): Make ifunc
25375 also for function that don't have calls or are not referenced.
25377 2017-06-30 Richard Biener <rguenther@suse.de>
25379 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
25380 analyze the first scalar stmt. Move vector type computation
25381 for the BB case here from ...
25382 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
25383 live operation processing in the SLP case properly.
25385 2017-06-30 Richard Biener <rguenther@suse.de>
25387 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
25389 2017-06-30 Martin Liska <mliska@suse.cz>
25392 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
25393 before BUILT_IN_UNWIND_RESUME when ASAN is used.
25395 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
25397 * doc/invoke.texi (AArch64): Add missing options and remove redundant
25400 2017-06-30 Richard Biener <rguenther@suse.de>
25402 PR tree-optimization/81249
25403 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
25404 condition reduction result to original scalar type.
25406 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25408 * profile-count.h (enum profile_quality): Fix typos and whitespace
25411 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25413 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
25414 type for branch probabilities.
25416 2017-06-29 Julian Brown <julian@codesourcery.com>
25417 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25419 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
25420 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
25421 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
25422 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
25424 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25426 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
25427 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
25428 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
25429 CC usage from generic code to here.
25430 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
25431 CC usage into the target macros.
25433 2017-06-29 Maya Rashish <coypu@sdf.org>
25435 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
25438 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25440 * arm/arm-builtins.c: Include profile-count.h
25441 * except.c (sjlj_emit_function_enter): Use
25442 profile_probability::unlikely.
25444 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25446 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
25447 and tocrel_offset be pointer args rather than implicitly using
25449 (legitimate_constant_pool_address_p, rs6000_emit_move,
25450 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
25451 tocrel_offset and use in toc_relative_expr_p call.
25452 (print_operand, print_operand_address): Use static tocrel_base_oac
25453 and tocrel_offset_oac.
25454 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
25457 2017-06-29 Maya Rashish <coypu@sdf.org>
25459 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
25461 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
25463 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
25464 objects, take into account only the alignment of 'op0' and 'mode1' if
25467 2017-06-29 Steve Ellcey <sellcey@cavium.com>
25469 * ccmp.c (ccmp_tree_comparison_p): New function.
25470 (ccmp_candidate_p): Update to use above function.
25471 (get_compare_parts): New function.
25472 (expand_ccmp_next): Update to use new functions.
25473 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
25475 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
25476 take mode as argument.
25477 * ccmp.h (expand_ccmp_expr): Add mode as argument.
25478 * expr.c (expand_expr_real_1): Pass mode as argument.
25480 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
25482 * combine.c (combine_instructions): Print insns to dump_file, together
25485 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25487 * asan.c (asan_emit_stack_protection): Update.
25488 (create_cond_insert_point): Update.
25489 * auto-profile.c (afdo_propagate_circuit): Update.
25490 * basic-block.h (struct edge_def): Turn probability to
25491 profile_probability.
25492 (EDGE_FREQUENCY): Update.
25493 * bb-reorder.c (find_traces_1_round): Update.
25494 (better_edge_p): Update.
25495 (sanitize_hot_paths): Update.
25496 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
25497 (make_single_succ_edge): Update.
25498 (check_bb_profile): Update.
25499 (dump_edge_info): Update.
25500 (update_bb_profile_for_threading): Update.
25501 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
25502 probabilitycount to 0.
25503 * cfgbuild.c (compute_outgoing_frequencies): Update.
25504 * cfgcleanup.c (try_forward_edges): Update.
25505 (outgoing_edges_match): Update.
25506 (try_crossjump_to_edge): Update.
25507 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
25508 (expand_gimple_tailcall): Update.
25509 (construct_init_block): Use make_single_succ_edge.
25510 (construct_exit_block): Use make_single_succ_edge.
25511 * cfghooks.c (verify_flow_info): Update.
25512 (redirect_edge_succ_nodup): Update.
25513 (split_edge): Update.
25514 (account_profile_record): Update.
25515 * cfgloopanal.c (single_likely_exit): Update.
25516 * cfgloopmanip.c (scale_loop_profile): Update.
25517 (set_zero_probability): Remove.
25518 (duplicate_loop_to_header_edge): Update.
25519 * cfgloopmanip.h (loop_version): Update prototype.
25520 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
25521 (force_nonfallthru_and_redirect): Update.
25522 (update_br_prob_note): Update.
25523 (rtl_verify_edges): Update.
25524 (purge_dead_edges): Update.
25525 (rtl_lv_add_condition_to_bb): Update.
25526 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
25527 * cgraphunit.c (init_lowered_empty_function): Update.
25528 (cgraph_node::expand_thunk): Update.
25529 * cilk-common.c: Include profile-count.h
25530 * dojump.c (inv): Remove.
25531 (jumpifnot): Update.
25532 (jumpifnot_1): Update.
25533 (do_jump_1): Update.
25535 (do_jump_by_parts_greater_rtx): Update.
25536 (do_compare_rtx_and_jump): Update.
25537 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
25538 do_jump_1. do_compare_rtx_and_jump): Update prototype.
25539 * dwarf2cfi.c: Include profile-count.h
25540 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
25541 (sjlj_emit_dispatch_table): Likewise.
25542 * explow.c: Include profile-count.h
25543 * expmed.c (emit_store_flag_force): Update.
25544 (do_cmp_and_jump): Update.
25545 * expr.c (compare_by_pieces_d::generate): Update.
25546 (compare_by_pieces_d::finish_mode): Update.
25547 (emit_block_move_via_loop): Update.
25548 (store_expr_with_bounds): Update.
25549 (store_constructor): Update.
25550 (expand_expr_real_2): Update.
25551 (expand_expr_real_1): Update.
25552 * expr.h (try_casesi, try_tablejump): Update prototypes.
25553 * gimple-pretty-print.c (dump_probability): Update.
25554 (dump_profile): New.
25555 (dump_gimple_label): Update.
25556 (dump_gimple_bb_header): Update.
25557 * graph.c (draw_cfg_node_succ_edges): Update.
25558 * hsa-gen.c (convert_switch_statements): Update.
25559 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
25560 (find_if_case_1): Update.
25561 (find_if_case_2): Update.
25562 * internal-fn.c (expand_arith_overflow_result_store): Update.
25563 (expand_addsub_overflow): Update.
25564 (expand_neg_overflow): Update.
25565 (expand_mul_overflow): Update.
25566 (expand_vector_ubsan_overflow): Update.
25567 * ipa-cp.c (good_cloning_opportunity_p): Update.
25568 * ipa-split.c (split_function): Use make_single_succ_edge.
25569 * ipa-utils.c (ipa_merge_profiles): Update.
25570 * loop-doloop.c (add_test): Update.
25571 (doloop_modify): Update.
25572 * loop-unroll.c (compare_and_jump_seq): Update.
25573 (unroll_loop_runtime_iterations): Update.
25574 * lra-constraints.c (lra_inheritance): Update.
25575 * lto-streamer-in.c (input_cfg): Update.
25576 * lto-streamer-out.c (output_cfg): Update.
25577 * mcf.c (adjust_cfg_counts): Update.
25578 * modulo-sched.c (sms_schedule): Update.
25579 * omp-expand.c (expand_omp_for_init_counts): Update.
25580 (extract_omp_for_update_vars): Update.
25581 (expand_omp_ordered_sink): Update.
25582 (expand_omp_for_ordered_loops): Update.
25583 (expand_omp_for_generic): Update.
25584 (expand_omp_for_static_nochunk): Update.
25585 (expand_omp_for_static_chunk): Update.
25586 (expand_cilk_for): Update.
25587 (expand_omp_simd): Update.
25588 (expand_omp_taskloop_for_outer): Update.
25589 (expand_omp_taskloop_for_inner): Update.
25590 * omp-simd-clone.c (simd_clone_adjust): Update.
25591 * optabs.c (expand_doubleword_shift): Update.
25592 (expand_abs): Update.
25593 (emit_cmp_and_jump_insn_1): Update.
25594 (expand_compare_and_swap_loop): Update.
25595 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
25596 * predict.c (predictable_edge_p): Update.
25597 (edge_probability_reliable_p): Update.
25598 (set_even_probabilities): Update.
25599 (combine_predictions_for_insn): Update.
25600 (combine_predictions_for_bb): Update.
25601 (propagate_freq): Update.
25602 (estimate_bb_frequencies): Update.
25603 (force_edge_cold): Update.
25604 * profile-count.c (profile_count::dump): Add missing space into dump.
25605 (profile_count::debug): Add newline.
25606 (profile_count::differs_from_p): Explicitly convert to unsigned.
25607 (profile_count::stream_in): Update.
25608 (profile_probability::dump): New member function.
25609 (profile_probability::debug): New member function.
25610 (profile_probability::differs_from_p): New member function.
25611 (profile_probability::differs_lot_from_p): New member function.
25612 (profile_probability::stream_in): New member function.
25613 (profile_probability::stream_out): New member function.
25614 * profile-count.h (profile_count_quality): Rename to ...
25615 (profile_quality): ... this one.
25616 (profile_probability): New.
25617 (profile_count): Update.
25618 * profile.c (compute_branch_probabilities): Update.
25619 * recog.c (peep2_attempt): Update.
25620 * sched-ebb.c (schedule_ebbs): Update.
25621 * sched-rgn.c (find_single_block_region): Update.
25622 (compute_dom_prob_ps): Update.
25623 (schedule_region): Update.
25624 * sel-sched-ir.c (compute_succs_info): Update.
25625 * stmt.c (struct case_node): Update.
25626 (do_jump_if_equal): Update.
25627 (get_outgoing_edge_probs): Update.
25628 (conditional_probability): Update.
25629 (emit_case_dispatch_table): Update.
25630 (expand_case): Update.
25631 (expand_sjlj_dispatch_table): Update.
25632 (emit_case_nodes): Update.
25633 * targhooks.c: Update.
25634 * tracer.c (better_p): Update.
25635 (find_best_successor): Update.
25636 * trans-mem.c (expand_transaction): Update.
25637 * tree-call-cdce.c: Update.
25638 * tree-cfg.c (gimple_split_edge): Upate.
25639 (move_sese_region_to_fn): Upate.
25640 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
25641 * tree-eh.c (lower_resx): Upate.
25642 (cleanup_empty_eh_move_lp): Upate.
25643 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25644 * tree-inline.c (copy_edges_for_bb): Update.
25645 (copy_cfg_body): Update.
25646 * tree-parloops.c (gen_parallel_loop): Update.
25647 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
25648 (gimple_gen_time_profiler): Update.
25649 * tree-ssa-dce.c (remove_dead_stmt): Update.
25650 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25651 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
25652 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
25653 (unloop_loops): Update.
25654 (try_peel_loop): Update.
25655 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25656 * tree-ssa-loop-split.c (connect_loops): Update.
25657 (split_loop): Update.
25658 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25659 (hoist_guard): Update.
25660 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
25661 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
25662 (value_replacement): Update.
25663 * tree-ssa-reassoc.c (branch_fixup): Update.
25664 * tree-ssa-tail-merge.c (replace_block_by): Update.
25665 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
25666 (create_edge_and_update_destination_phis): Update.
25667 (compute_path_counts): Update.
25668 (recompute_probabilities): Update.
25669 (update_joiner_offpath_counts): Update.
25670 (freqs_to_counts_path): Update.
25671 (duplicate_thread_path): Update.
25672 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
25673 (struct switch_conv_info): Update.
25674 (gen_inbound_check): Update.
25675 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25676 (vect_do_peeling): Update.
25677 (vect_loop_versioning): Update.
25678 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25679 (optimize_mask_stores): Update.
25680 * ubsan.c (ubsan_expand_null_ifn): Update.
25681 * value-prof.c (gimple_divmod_fixed_value): Update.
25682 (gimple_divmod_fixed_value_transform): Update.
25683 (gimple_mod_pow2): Update.
25684 (gimple_mod_pow2_value_transform): Update.
25685 (gimple_mod_subtract): Update.
25686 (gimple_mod_subtract_transform): Update.
25687 (gimple_ic): Update.
25688 (gimple_stringop_fixed_value): Update.
25689 (gimple_stringops_transform): Update.
25690 * value-prof.h: Update.
25692 2017-06-29 Carl Love <cel@us.ibm.com>
25694 * config/rs6000/rs6000-c.c: Add support for built-in functions
25695 vector signed int vec_signed (vector float);
25696 vector signed long long vec_signed (vector double);
25697 vector signed int vec_signed2 (vector double, vector double);
25698 vector signed int vec_signede (vector double);
25699 vector signed int vec_signedo (vector double);
25700 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
25701 instruction generator.
25702 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25703 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
25704 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
25706 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
25707 vunsignede_v2df): Add define_expands.
25708 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
25709 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
25710 VEC_UNSIGNEDO): Add definitions.
25711 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25712 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
25713 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
25714 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
25715 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
25716 * config/rs6000/altivec.h (vec_signed, vec_signed2,
25717 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
25718 vec_unsignede, vec_unsignedo): Add builtin defines.
25719 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
25721 * doc/extend.texi: Update the built-in documentation file for the
25722 new built-in functions.
25724 2017-06-29 Richard Biener <rguenther@suse.de>
25726 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
25727 reduction chains to LOOP_VINFO_REDUCTIONS.
25728 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
25729 SLP reductions after processing reduction chains.
25731 2017-06-29 Nathan Sidwell <nathan@acm.org>
25733 * builtins.c (fold_builtin_FUNCTION): Use
25734 lang_hooks.decl_printable_name.
25736 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
25738 PR middle-end/81194
25739 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
25740 with only one label.
25741 * stmt.c (expand_case): Assert NCASES is greater than one.
25743 2017-06-29 Richard Biener <rguenther@suse.de>
25745 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
25747 (group_case_labels): Likewise.
25748 (find_taken_edge): Push sanity checking on val to workers...
25749 (find_taken_edge_cond_expr): ... here
25750 (find_taken_edge_switch_expr): ... and here, handle cases
25751 with just a default label.
25752 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
25753 (group_case_labels): Likewise.
25754 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
25755 group_case_labels does anything cleanup the CFG again.
25757 2017-06-29 Bin Cheng <bin.cheng@arm.com>
25759 PR tree-optimization/81196
25760 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
25761 exit condition comparing two IVs.
25763 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
25765 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
25766 profile to the dummy entry at the end of the list of architectures.
25767 * config/arm/arm-cpu-cdata.h: Regenerated.
25769 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25770 Michael Collison <michael.collison@arm.com>
25773 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
25775 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
25776 (*aarch64_reg_<mode>3_minus_mask): New pattern.
25777 (*aarch64_<optab>_reg_di3_mask2): New pattern.
25778 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
25779 of shift when the shift amount is masked with constant equal to
25780 the size of the mode.
25781 * config/aarch64/predicates.md (subreg_lowpart_operator): New
25784 2017-06-29 Martin Liska <mliska@suse.cz>
25786 * config/i386/i386.opt: Change range from [1,5] to [0,5].
25788 2017-06-29 Yury Gribov <tetra2005@gmail.com>
25791 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
25794 (edge_growth_cache_entry::edge_growth_cache_entry): New
25796 (reset_edge_growth_cache): Update to use constructor.
25798 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25800 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
25801 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25802 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25804 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
25806 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
25807 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
25809 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
25811 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
25812 (*-linux-uclibc*): Add t-uclibc tmake_file.
25813 * config/t-musl: New.
25814 * config/t-uclibc: New.
25816 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
25818 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
25820 (gen_comm_data): Emit architectural setting of arch_prof.
25821 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
25823 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
25824 (armv8-m.base, armv8-m.main): Likewise.
25825 * arm-protos.h (arm_build_target): Add profile field.
25826 (arch_option): Likewise.
25827 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
25829 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
25830 arm_active_target.profile.
25832 2017-06-28 Richard Biener <rguenther@suse.de>
25834 PR middle-end/81227
25835 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
25836 TYPE_OVERFLOW_WRAPS.
25837 * match.pd (negate_expr_p): Likewise.
25838 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
25839 fold_build2, not fold_binary.
25841 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25843 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
25844 Convert memory address to Pmode.
25845 (aarch64_print_operand): Assert MEM operands are always Pmode.
25847 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25850 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
25851 Remove redundant if.
25852 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
25853 * config/arm/aarch-common-protos.h
25854 (aarch_forward_to_shift_is_not_shifted_re): Remove.
25855 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
25857 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
25860 * multiple_target.c (create_dispatcher_calls): Set the default
25861 clone to be static, not public.
25863 2017-06-28 Richard Biener <rguenther@suse.de>
25865 * tree-vect-loop.c (vectorizable_reduction): Move special
25866 cond reduction IV var creation ...
25867 (vect_create_epilog_for_reduction): ... here. Remove induction_index
25868 parameter. Use STMT_VINFO_VECTYPE.
25869 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
25872 2017-06-28 Martin Liska <mliska@suse.cz>
25875 * ipa-visibility.c (non_local_p): Handle visibility.
25877 2017-06-28 Martin Liska <mliska@suse.cz>
25880 * common.opt: Add IntegerRange to various options.
25881 * opt-functions.awk (integer_range_info): New function.
25882 * optc-gen.awk: Add integer_range_info to cl_options struct.
25883 * opts-common.c (decode_cmdline_option): Handle
25884 CL_ERR_INT_RANGE_ARG.
25885 (cmdline_handle_error): Likewise.
25886 * opts.c (print_filtered_help): Show valid interval in
25887 when --help is provided.
25888 * opts.h (struct cl_option): Add range_min and range_max fields.
25889 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
25891 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
25893 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
25894 (x * C EQ/NE y * C): New transformation.
25896 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
25898 * genmultilib (combination_space): Accept '+' in option names.
25900 2017-06-28 Martin Liska <mliska@suse.cz>
25903 * asan.c (instrument_derefs): Bail out inner references
25904 that are hard register variables.
25906 2017-06-28 Jakub Jelinek <jakub@redhat.com>
25909 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
25910 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
25912 2017-06-28 Richard Biener <rguenther@suse.de>
25914 * tree-vectorizer.h (vect_get_vec_defs): Remove.
25915 (vect_get_slp_defs): Adjust.
25916 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
25918 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
25920 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
25921 get_initial_defs_for_reduction instead of vect_get_vec_defs.
25922 (vectorizable_reduction): Adjust.
25923 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
25925 (vect_get_slp_defs): Likewise.
25926 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
25927 (vectorizable_bswap): Adjust.
25928 (vectorizable_call): Likewise.
25929 (vectorizable_conversion): Likewise.
25930 (vectorizable_assignment): Likewise.
25931 (vectorizable_shift): Likewise.
25932 (vectorizable_operation): Likewise.
25933 (vectorizable_store): Likewise.
25934 (vectorizable_condition): Likewise.
25935 (vectorizable_comparison): Likewise.
25937 2017-06-28 Michael Collison <michael.collison@arm.com>
25940 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
25942 (arm_gen_movmemqi): Removed unused variable 'i'.
25943 Convert 'for' loop into 'while' loop.
25944 (arm_expand_prologue): Remove last unnecessary set of insn.
25945 (thumb_pop): Remove unused variable 'pushed_words'.
25946 (thumb_exit): Remove last unnecessary set of regs_to_pop.
25948 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25950 * config/s390/predicates.md: Use s390_rel_address_ok_p.
25951 * config/s390/s390-protos.h: Add prototype of
25952 s390_rel_address_ok_p.
25953 * config/s390/s390.c (s390_got_symbol): New function.
25954 (s390_rel_address_ok_p): New function.
25955 (legitimize_pic_address): Use s390_rel_address_ok_p.
25956 (s390_load_got): Use s390_got_symbol.
25957 (s390_option_override): Issue error if
25958 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
25959 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
25961 * config/s390/s390.opt: New option mpic-data-is-text-relative.
25963 2017-06-27 Andrew Pinski <apinski@cavium.com>
25965 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
25966 (X * copysign (1.0, X)): New pattern.
25967 (X * copysign (1.0, -X)): New pattern.
25968 (copysign (-1.0, CST)): New pattern.
25970 2017-06-27 Joseph Myers <joseph@codesourcery.com>
25972 * genmultilib (combination_space): Remove variable.
25973 Validate reuse rules against regular expression for any sequence
25974 of multilib options in any order.
25976 2017-06-27 Michael Collison <michael.collison@arm.com>
25978 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
25979 call aarch64_split_simd_combine.
25980 * (aarch64_combine_internal<mode>): Delete pattern.
25981 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
25982 Allow register and subreg operands.
25984 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25986 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
25987 specific need, just fallback on defaults.
25988 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
25990 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25991 Olivier Hainque <hainque@adacore.com>
25993 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
25995 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
25996 targets. Pick a default if no particular attempt applied.
25997 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
26000 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26002 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
26003 (x86_64-wrs-vxworks7): Likewise.
26005 2017-06-27 Marek Polacek <polacek@redhat.com>
26008 * ubsan.c (instrument_null): Check get_base_address's result for null.
26010 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26012 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
26014 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26016 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
26017 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
26018 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
26019 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
26020 New function types.
26021 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
26022 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
26023 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
26024 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
26025 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
26026 BUILT_IN_FEUPDATEENV): New builtins.
26027 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
26028 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
26029 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
26030 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
26032 (builtin_structptr_types): Adjust size.
26033 * tree.c (builtin_structptr_types): Add four entries.
26035 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26036 Olivier Hainque <hainque@adacore.com>
26038 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
26039 (TLS_SYM): New local macro, forcing reference to __tls__ on
26040 link command lines for VxWorks 7 RTPs, triggering initialization
26042 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
26043 OS features TLS support, true for RTPs on VxWorks 7.
26044 * config/vxworks.c (vxworks_override_options): Setup emutls
26047 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26049 * predict.c (test_prediction_value_range): Use -1U instead of -1
26050 to avoid narrowing conversion warning.
26051 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
26052 to avoid narrowing conversion warning.
26053 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
26055 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
26057 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26059 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
26060 64bit configurations.
26061 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
26062 (SIZE_TYPE): Likewise.
26063 * config/vxworks.c (vxworks_emutls_var_fields): Use
26064 long_unsigned_type_node instead of unsigned_type_node as the offset
26065 field type, which is "pointer" mode in emutls.c.
26067 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26070 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
26072 PR middle-end/81207
26073 * gimple-fold.c (replace_call_with_call_and_fold): Handle
26074 gimple_vuse copying separately from gimple_vdef copying.
26076 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26078 * value-prof.c (free_hist): Remove call to memset and the enclosing if
26081 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26082 Olivier Hainque <hainque@adacore.com>
26084 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
26085 for all vxworks7 targets.
26086 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
26087 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
26088 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
26089 variations for VX6/VX7 and 32/64bits later on in ...
26090 (VXWORKS_LIB_SPEC): Leverage new macros.
26091 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
26092 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
26094 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26096 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
26097 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
26099 2017-06-26 Carl Love <cel@us.ibm.com>
26101 * config/rs6000/rs6000-c.c: Add support for built-in functions
26102 vector bool char vec_reve (vector bool char);
26103 vector signed char vec_reve (vector signed char);
26104 vector unsigned char vec_reve (vector unsigned char);
26105 vector bool int vec_reve (vector bool int);
26106 vector signed int vec_reve (vector signed int);
26107 vector unsigned int vec_reve (vector unsigned int);
26108 vector bool long long vec_reve (vector bool long long);
26109 vector signed long long vec_reve (vector signed long long);
26110 vector unsigned long long vec_reve (vector unsigned long long);
26111 vector bool short vec_reve (vector bool short);
26112 vector signed short vec_reve (vector signed short);
26113 vector double vec_reve (vector double);
26114 vector float vec_reve (vector float);
26115 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
26116 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
26117 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
26118 (altivec_vreve): New pattern.
26119 * config/rs6000/altivec.h (vec_reve): New define.
26120 * doc/extend.texi (vec_rev): Update the built-in documentation file
26121 for the new built-in functions.
26123 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26125 PR tree-optimization/71815
26126 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
26128 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
26130 (slsr_process_phi): Likewise.
26131 (replace_uncond_cands_and_profitable_phis): Don't replace a
26132 multiply candidate with a stride of 1 (copy or cast).
26133 (phi_incr_cost): Call uses_consumed_by_stmt rather than
26135 (lowest_cost_path): Likewise.
26136 (total_savings): Likewise.
26138 2017-06-26 Richard Biener <rguenther@suse.de>
26141 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
26142 Use def_builtin_pure for all gather builtins.
26144 2017-06-26 Richard Biener <rguenther@suse.de>
26146 PR tree-optimization/81203
26147 * tree-tailcall.c (find_tail_calls): Do not move stmts into
26148 non-dominating BBs.
26150 2017-06-26 Marek Polacek <polacek@redhat.com>
26153 * doc/invoke.texi: Document -Wmultistatement-macros.
26155 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
26157 * doc/sourcebuild.texi (ARM-specific attributes): Document new
26158 arm_neon_ok_no_float_abi effective target.
26160 2017-06-26 Richard Biener <rguenther@suse.de>
26162 PR tree-optimization/80928
26163 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
26164 (copy_bbs): Set BB_DUPLICATED flag early.
26165 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
26167 (execute_on_shrinking_pred): Likewise.
26168 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
26169 BB_DUPLICATED blocks.
26170 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
26171 iterate over all PHIs considering removal of *gsi.
26173 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
26175 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
26178 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26180 * config/rs6000/rs6000-string.c: (expand_block_clear,
26181 do_load_for_compare, select_block_compare_mode,
26182 compute_current_alignment, expand_block_compare,
26183 expand_strncmp_align_check, expand_strn_compare,
26184 expand_block_move, rs6000_output_load_multiple)
26185 Move functions related to string/block move/compare
26186 to a separate file.
26187 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
26188 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
26189 for this function which is now used in two files.
26190 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
26191 * config.gcc: Add rs6000-string.o to extra_objs for
26192 targets powerpc*-*-* and rs6000*-*-*.
26194 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
26197 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
26198 32-bit, since indexed is not valid for DImode.
26199 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
26200 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
26201 (define_peephole2 for Altivec d-form load): Add 32-bit support.
26202 (define_peephole2 for Altivec d-form store): Likewise.
26205 * multiple_target.c (create_dispatcher_calls): Only create the
26206 dispatcher call if the function is the default clone of a
26207 versioned function.
26209 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
26211 PR middle-end/80902
26212 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
26213 a call, force the call to not be a tail call.
26215 2017-06-23 Jeff Law <law@redhat.com>
26217 * doc/contrib.texi: Add entry for Steven Pemberton's work on
26220 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
26222 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
26223 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
26224 handling for early expansion of vector shifts (sl,sr,sra,rl).
26225 (builtin_function_type): Add vector shift right instructions
26226 to the unsigned argument list.
26228 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
26230 rtl-optimizatoin/79286
26231 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
26232 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
26233 trap. PIC register plus a const unspec without offset can never trap.
26235 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
26237 * tree.h (builtin_structptr_type): New type.
26238 (builtin_structptr_types): Declare new array.
26239 * tree.c (builtin_structptr_types): New array.
26240 (free_lang_data, build_common_tree_nodes): Use it.
26242 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
26245 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
26248 2017-06-22 Matt Turner <mattst88@gmail.com>
26250 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
26251 Lake models to skylake case. Assume skylake for unknown
26252 models with clflushopt.
26254 2017-06-22 Jeff Law <law@redhat.com>
26256 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
26257 frame sizes that do not satisfy aarch64_uimm12_shift.
26259 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
26261 * profile-count.h (apply_probability,
26262 apply_scale, probability_in): Fix checks for zero.
26264 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26266 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
26267 * doc/cppdiropts.texi (-I @var{dir}): Document it.
26269 2016-06-22 Richard Biener <rguenther@suse.de>
26271 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26272 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
26273 REDUC_MAX_EXPR support.
26274 (vectorizable_reduction): Likewise.
26275 (vect_create_epilog_for_reduction): Likewise.
26277 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
26279 * match.pd (A / (1 << B) -> A >> B): New.
26280 * generic-match-head.c: Include optabs-tree.h.
26281 * gimple-match-head.c: Likewise.
26282 * optabs-tree.h (target_supports_op_p): New.
26283 * optabs-tree.c (target_supports_op_p): New.
26285 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26287 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
26288 $gcc_cv_ld --help output.
26289 (gcc_cv_ld_demangle): Likewise.
26290 (gcc_cv_ld_eh_frame_hdr): Likewise.
26291 (gcc_cv_ld_pie): Likewise.
26292 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
26293 (gcc_cv_ld_buildid): Likewise.
26294 (gcc_cv_ld_sysroot): Likewise.
26295 (ld_bndplt_support): Likewise.
26296 (ld_pushpopstate_support): Likewise.
26297 * configure: Regenerate.
26298 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
26300 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26303 * config/i386/sse.md (round<mode>2): Renumber match_dup and
26304 operands indexes to avoid gap between operands and match_dups.
26306 2017-06-21 Andrew Pinski <apinski@cavium.com>
26308 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
26309 Increment Arith_shift and Arith_shift_reg by 1.
26310 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
26312 * config/aarch64/aarch64.c (thunderx_tunings): Enable
26313 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
26314 (aarch64_strip_extend): Add new argument and test for it.
26315 (aarch64_cheap_mult_shift_p): New function.
26316 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
26317 add a cost if it is true.
26318 Update calls to aarch64_strip_extend.
26319 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
26321 2017-06-21 Andrew Pinski <apinski@cavium.com>
26323 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
26325 (thunderxt88): Likewise.
26326 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
26327 (thunderx_prefetch_tune): New variable.
26328 (thunderx2t99_prefetch_tune): Update for the correct values.
26329 (thunderxt88_tunings): New variable.
26330 (thunderx_tunings): Use thunderx_prefetch_tune instead of
26331 generic_prefetch_tune.
26332 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
26334 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26336 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
26337 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
26338 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
26339 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
26340 (aarch64_atomic_cas<mode>, GPI): Likewise.
26342 2017-06-21 Martin Liska <mliska@suse.cz>
26344 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
26345 statements on cold and hot labels.
26346 * predict.c (tree_estimate_probability_bb): Remove the
26347 prediction from this place.
26349 2017-06-21 Martin Liska <mliska@suse.cz>
26351 PR tree-optimization/79489
26352 * gimplify.c (maybe_add_early_return_predict_stmt): New
26354 (gimplify_return_expr): Call the function.
26355 * predict.c (tree_estimate_probability_bb): Remove handling
26357 * predict.def: Update comment about early return predictor.
26358 * gimple-predict.h (is_gimple_predict): New function.
26359 * predict.def: Change default value of early return to 66.
26360 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
26362 * passes.def: Put pass_strip_predict_hints to the beginning of
26365 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
26367 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
26368 FUNCTION_DECL declarations.
26369 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
26371 (dwaf2out_decl): Likewise.
26372 * godump.c (go_early_global_decl): Skip call to the real debug hook
26373 for FUNCTION_DECL declarations.
26374 * passes.c (rest_of_decl_compilation): Skip call to the
26375 early_global_decl debug hook for FUNCTION_DECL declarations, unless
26376 -fdump-go-spec is passed.
26378 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26380 * config/i386/i386.c (struct builtin_isa): New field pure_p.
26381 Reorder for compactness.
26382 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
26383 (def_builtin_pure, def_builtin_pure2): New functions.
26384 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
26386 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26388 * match.pd (nop_convert): New predicate.
26389 ((A +- CST1) +- CST2): Allow some NOP conversions.
26391 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26394 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
26395 with ctors/dtors if GOVD_SHARED is set.
26397 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26399 * config/aarch64/aarch64.md (movti_aarch64):
26400 Emit mov rather than orr.
26401 (movtf_aarch64): Likewise.
26402 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
26403 Emit mov rather than orr.
26405 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26407 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
26408 Swap alternatives, make integer dup more expensive.
26410 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26412 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
26413 Return true for non-tls symbols.
26415 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
26417 * config/aarch64/aarch64-cores.def (cortex-a55): New.
26418 (cortex-a75): Likewise.
26419 (cortex-a75.cortex-a55): Likewise.
26420 * config/aarch64/aarch64-tune.md: Regenerate.
26421 * doc/invoke.texi (-mtune): Document new values for -mtune.
26423 2017-06-21 Tom de Vries <tom@codesourcery.com>
26425 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
26426 stack_size feature.
26427 (Effective-Target Keywords, Other attributes): Suggest using
26428 dg-add-options stack_size feature to get stack limit in stack_size
26429 effective target documentation.
26431 2017-06-21 Julian Brown <julian@codesourcery.com>
26432 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26434 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
26435 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
26436 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
26438 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
26439 attribute type list for neon_multiply.
26440 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
26441 attribute type list for neon_multiply.
26442 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
26443 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
26444 attribute type list for neon_multiply.
26445 * config/arm/types.md (crypto_pmull): Add.
26446 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
26447 attribute type list.
26449 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
26451 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
26454 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26456 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
26457 to make sure not to dereference a NULL cost_classes_ptr pointer.
26459 2017-06-20 Carl Love <cel@us.ibm.com>
26461 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26462 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
26463 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
26464 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
26465 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
26466 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
26467 VMULOSW): New enum "unspec" values.
26468 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
26469 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
26470 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
26471 altivec_vmulosw): New patterns.
26472 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
26473 VMULOSW): Add definitions.
26475 2017-06-20 Julia Koval <julia.koval@intel.com>
26477 * config/i386/i386.c: Fix rounding expand for new pattern.
26478 * config/i386/subst.md: Fix pattern (parallel -> unspec).
26480 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26482 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
26483 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
26485 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26487 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
26490 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26492 * config/aarch64/aarch64-cores.def: Rearrange to sort by
26493 architecture, then by implementer ID.
26494 * config/aarch64/aarch64-tune.md: Regenerate.
26496 2017-06-20 Richard Biener <rguenther@suse.de>
26498 PR middle-end/81097
26499 * fold-const.c (split_tree): Fold to type before negating.
26501 2017-06-20 David Malcolm <dmalcolm@redhat.com>
26503 * diagnostic-show-locus.c
26504 (selftest::test_fixit_deletion_affecting_newline): New function.
26505 (selftest::diagnostic_show_locus_c_tests): Call it.
26507 2017-06-20 Andreas Schwab <schwab@suse.de>
26510 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
26513 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26515 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
26516 __ARM_FEATURE_COPROC according to support.
26518 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26520 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26521 Rewritten to avoid overflow for > 32-bit pointers.
26524 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
26525 by removing enum keyword.
26526 (ubsan_type_descriptor): Likewise. Formatting fix.
26529 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
26530 splitter): Require TARGET_SSE2 in the condition.
26532 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
26535 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
26536 for doing vector set of SFmode on ISA 3.0.
26537 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
26538 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
26540 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
26541 SFmode value into a V4SF variable that was extracted from another
26542 V4SF variable without converting the element to double precision
26543 and back to single precision vector format.
26544 (vsx_insert_extract_v4sf_p9_2): Likewise.
26546 2017-06-19 Jakub Jelinek <jakub@redhat.com>
26548 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
26549 in UWHI to avoid undefined overflow.
26552 * ubsan.h (enum ubsan_encode_value_phase): New.
26553 (ubsan_encode_value): Change second argument to
26554 enum ubsan_encode_value_phase with default value of
26555 UBSAN_ENCODE_VALUE_GENERIC.
26556 * ubsan.c (ubsan_encode_value): Change second argument to
26557 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
26558 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
26559 create_tmp_var_raw instead of create_tmp_var and use a
26561 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
26562 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
26563 ubsan_encode_value callers.
26566 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
26567 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
26568 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
26570 2017-06-19 Richard Biener <rguenther@suse.de>
26572 PR middle-end/81118
26573 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
26574 estimates if we changed anything.
26576 2017-06-19 Richard Biener <rguenther@suse.de>
26578 PR tree-optimization/80887
26579 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
26580 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
26581 simplified lookups, then reset mprts_hook.
26582 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
26584 (try_to_simplify): Likewise.
26586 2017-06-19 Martin Liska <mliska@suse.cz>
26589 * gimplify.c (gimplify_switch_expr):
26590 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
26592 2017-06-19 Martin Liska <mliska@suse.cz>
26594 * doc/install.texi: Document that PGO runs in 4 stages.
26596 2017-06-19 Martin Liska <mliska@suse.cz>
26599 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
26600 to dispatcher function name.
26601 * multiple_target.c (replace_function_decl): New function.
26602 (create_dispatcher_calls): Redirect both edges and references.
26604 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
26606 * profile-count.c (profile_count::dump): Dump quality.
26607 (profile_count::differs_from_p): Update for unsigned val.
26608 * profile-count.h (profile_count_quality): New enum.
26609 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
26611 2017-06-19 Richard Biener <rguenther@suse.de>
26613 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
26614 struct function as arg.
26615 (estimate_numbers_of_iterations): Export overload with loop arg.
26616 (free_numbers_of_iterations_estimates_loop): Use an overload of
26617 free_numbers_of_iterations_estimates instead.
26618 * tree-cfg.c (remove_bb): Adjust.
26619 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
26620 * tree-parloops.c (gen_parallel_loop): Likewise.
26621 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
26623 (tree_unroll_loops_completely): Likewise.
26624 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
26625 Use an overload instead and export.
26626 (estimated_loop_iterations): Adjust.
26627 (max_loop_iterations): Likewise.
26628 (likely_max_loop_iterations): Likewise.
26629 (estimate_numbers_of_iterations): Take struct function as arg
26631 (loop_exits_before_overflow): Adjust.
26632 (free_numbers_of_iterations_estimates_loop): Use an overload.
26633 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
26634 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
26636 2017-06-19 Richard Biener <rguenther@suse.de>
26639 * ipa-prop.c (find_constructor_constant_at_offset): Handle
26640 RANGE_EXPR conservatively.
26642 2017-06-16 Carl Love <cel@us.ibm.com>
26644 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26645 definitions for vec_float, vec_float2, vec_floato,
26646 vec_floate built-ins.
26647 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
26648 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
26650 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
26651 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
26652 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
26653 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
26654 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
26655 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
26656 vec_floato): Add builtin defines.
26657 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
26658 Update the built-in documentation file for the new built-in
26661 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26663 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
26664 (mthumb): Mark as the negative of -marm.
26666 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26668 * doc/invoke.texi (ARM Options, -mcpu): Document supported
26670 (ARM Options, -mtune): Document that this accepts the same
26671 extension options as -mcpu.
26672 (ARM Options, -mfpu): Document addition of -mfpu=auto.
26674 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26676 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
26677 permitted extensions.
26679 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26681 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
26682 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
26683 (armv8-m.main): Add option +nodsp.
26684 * config/arm/arm-cpu-cdata.h: Regenerated.
26686 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26688 * config/arm/t-fuchsia: New file.
26689 * config.gcc (arm*-*-fuchsia*): Use it.
26691 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26693 * config/arm/t-symbian: Rewrite for new option infrastructure.
26695 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26697 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
26698 (MULTILIB_REQUIRED): Likewise.
26700 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26702 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
26703 (MULTILIB_RESUE): Likewise.
26704 (MULTILIB_MATCHES): Likewise.
26705 (MULTLIB_REQUIRED): Likewise.
26707 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26709 * config/arm/t-rtems: Rewrite for new option framework.
26711 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26713 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
26714 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
26715 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
26716 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
26717 * config/arm/t-multilib: ... here.
26718 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
26719 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
26720 armv7-a and armv8*-a when A-profile libraries have not been built.
26721 * config/arm/t-rmprofile: Rewrite.
26723 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26725 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
26726 with a backslash. Remove the backslash after substituting unescaped
26728 * doc/fragments.texi (MULTILIB_REUSE): Document it.
26730 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26732 * config.gcc: (arm*-*-*): When building a-profile libraries, force
26733 the driver to pass through the default setting of -mfloat-abi.
26734 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
26736 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
26737 (all_feat_combs): New rule.
26738 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
26740 * config/arm/t-aprofile: Rewrite.
26742 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26744 * config/arm/arm.h (FPUTYPE_AUTO): Define.
26745 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
26746 fpu is not specified by the user/command-line.
26747 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
26748 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
26749 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
26750 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
26751 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
26752 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
26754 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26756 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
26757 * config/arm/t-arm-elf: Rewritten.
26759 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26761 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
26762 have some floating-point instructions.
26763 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
26764 (TARGET_MAYBE_HARD_FLOAT): New macro.
26765 * config/arm/arm-builtins.c (arm_init_builtins): Use
26766 TARGET_MAYBE_HARD_FLOAT.
26767 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
26769 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26771 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
26772 (configargs.h): Include it.
26773 (arm_print_hint_for_fpu_option): New function.
26774 (arm_parse_fpu_option): New function.
26775 (candidate_extension): New class.
26776 (arm_canon_for_multilib): New function.
26777 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
26778 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
26779 (ARCH_CANONICAL_SPECS): New macro.
26780 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
26782 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26784 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
26785 are set after handling multilib fragments. Set target_cpu_default2
26788 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26790 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
26792 (arm*-*-*): Set target_cpu_default2 to a quoted string.
26793 * config/arm/parsecpu.awk (check_cpu): Validate any extension
26795 (check_arch): Likewise.
26796 * config/arm/arm.c (arm_configure_build_target): Handle
26797 TARGET_CPU_DEFAULT being a string constant. Scan any feature
26798 options in the default.
26800 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26802 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
26803 when an option is an alias of another.
26804 * config/arm/parsecpu.awk (optalias): New parser token.
26805 (gen_comm_data): Mark non-alias options as such. Emit entries
26806 for extension aliases.
26807 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
26808 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
26809 (armv6kz, armv6zk, armv6t2): Likewise.
26810 (armv7): Make vfpv3-d16 an alias.
26811 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
26813 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
26814 Sort in canonical order.
26815 (armv8-a): Sort in canonical order.
26816 (armv8.1-a, armv8.2-a): Likewise.
26817 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
26819 (cortex-a9): Sort in canonical order.
26820 * config/arm/arm.c (selftests.h): Include it.
26821 (arm_test_cpu_arch_data): New function.
26822 (arm_run_self_tests): New function.
26823 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
26824 (targetm): Move declaration to the end of the file.
26825 * arm-cpu-cdata.h: Regenerated.
26827 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26829 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
26830 call to target_mode_check describing the type of option passed.
26831 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
26832 (arm_target_thumb_only): Use arm_parse_arch_option_name or
26833 arm_parse_cpu_option_name to match parameters against list of
26835 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
26836 arm_arch_core_flags data structure.
26837 * config/arm/arm-cpu_cdata.h: Regenerated.
26839 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26841 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
26843 (arm_print_hint_for_cpu_option): Likewise.
26844 (arm_print_hint_for_arch_option): Likewise.
26845 (arm_parse_cpu_option_name): Likewise.
26846 (arm_parse_arch_option_name): Likewise.
26847 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
26848 of entries in the all_fpus list.
26849 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
26850 (arm_parse_cpu_option_name): Declare.
26851 (arm_parse_arch_option_name): Declare.
26852 (arm_parse_option_features): Declare.
26853 (arm_intialize_isa): Declare.
26854 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
26856 (gen_comm_data): ... here. Make definitions non-static.
26857 * config/arm/arm-cpu-data.h: Regenerated.
26858 * config/arm/arm-cpu-cdata.h: Regenerated.
26860 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26862 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
26863 (cpu_arch_extension): New structure.
26864 (cpu_arch_option, arch_option, cpu_option): New structures.
26865 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
26866 architecture types.
26867 (gen_data): Generate new format data tables.
26868 * config/arm/arm.c (cpu_tune): New structure.
26869 (cpu_option, processors): Delete.
26870 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
26871 (arm_print_hint_for_cpu_option): ... this and ...
26872 (arm_print_hint_for_arch_option): ... this.
26873 (arm_parse_arch_cpu_name): Delete. Replace with ...
26874 (arm_parse_cpu_option_name): ... this and ...
26875 (arm_parse_arch_option_name): ... this.
26876 (arm_unrecognized_feature): Change type of target parameter to
26878 (arm_parse_arch_cpu_features): Delete. Replace with ...
26879 (arm_parse_option_features): ... this.
26880 (arm_configure_build_target): Rework to use new configuration data
26882 (arm_print_tune_info): Rework for new configuration data tables.
26883 * config/arm/arm-cpu-data.h: Regenerated.
26884 * config/arm/arm-cpu.h: Regenerated.
26886 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26888 * Makefile.in (OBJS): Move sbitmap.o from here ...
26889 (OBJS-libcommon): ... to here.
26891 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26893 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
26894 (ISA_ALL_CRYPTO): New macro.
26895 (ISA_ALL_SIMD): New macro
26896 (ISA_ALL_FP): New macro.
26897 * config/arm/arm.c (fpu_bitlist): Update initializer.
26898 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
26900 (arm9e): Add fpu. Add option for nofp
26901 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
26902 (arm926ej-s, arm1026ej-s): Likewise.
26903 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
26904 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
26905 neon-fp16, neon-vfpv4, nofp and nosimd.
26906 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
26907 (cortex-a8): Add fpu. Add option for nofp.
26908 (cortex-a9): Add fpu. Add options for nosimd and nofp.
26909 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
26910 (cortex-r4f): Add fpu.
26911 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
26912 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
26914 (cortex-r8): Likewise.
26915 (cortex-m4): Add fpu. Add option for nofp.
26916 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
26917 (cortex-a17.cortex-a7): Likewise.
26918 (cortex-a32): Add fpu. Add options for crypto and nofp.
26919 (cortex-a35, cortex-a53): Likewise.
26920 (cortex-a57): Add fpu. Add option for crypto.
26921 (cortex-a72, cortex-a73): Likewise.
26922 (exynos-m1): Likewise.
26923 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
26924 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
26925 (cortex-m33): Add fpu. Add option for nofp.
26926 * config/arm/arm-cpu-cdata.h: Regenerated
26927 * config/arm/arm-cpu-data.h: Regenerated.
26929 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26931 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
26932 (armv5te, armv5tej): Likewise.
26933 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
26934 (armv7): Add options fp and vfpv3-d16.
26935 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
26936 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
26938 (armv7ve): Likewise.
26939 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
26940 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
26941 (armv8-a): Add nocrypto option.
26942 (armv8.1-a, armv8.2-a): Likewise.
26943 (armv8-m.main): add options fp, fp.dp and nofp.
26945 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26947 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
26949 (armv8-a+crc): Delete.
26950 (armv8.1-a): Add options simd, crypto and nofp.
26951 (armv8.2-a): Add options fp16, simd, crypto and nofp.
26952 (armv8.2-a+fp16): Delete.
26953 (armv8-m.main): Add option dsp.
26954 (armv8-m.main+dsp): Delete.
26955 (cortex-a8): Add fpu. Add nofp option.
26956 (cortex-a9): Add fpu. Add nofp and nosimd options.
26957 * config/arm/parsecpu.awk (gen_data): Generate option tables and
26958 link to main cpu and architecture data structures.
26959 (gen_comm_data): Only put isa attributes from the main architecture
26961 (option): New statement for architecture and CPU entries.
26962 * arm.c (struct cpu_option): New structure.
26963 (struct processors): Add entry for options.
26964 (arm_unrecognized_feature): New function.
26965 (arm_parse_arch_cpu_name): Ignore any characters after the first
26967 (arm_parse_arch_cpu_feature): New function.
26968 (arm_configure_build_target): Separate out any CPU and architecture
26969 features and parse separately. Don't error out if -mfpu=auto is
26970 used with only an architecture string.
26971 (arm_print_asm_arch_directives): New function.
26972 (arm_file_start): Call it.
26973 * config/arm/arm-cpu-cdata.h: Regenerated.
26974 * config/arm/arm-cpu-data.h: Likewise.
26975 * config/arm/arm-tables.opt: Likewise.
26977 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26979 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
26980 assembler when it is not -mfpu=auto.
26982 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26984 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
26985 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
26986 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
26987 (ASM_CPU_SPEC): Rewrite.
26988 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
26989 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
26990 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
26991 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
26992 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
26993 copied string is NUL-terminated. Also strip any characters prefixed
26995 (arm_rewrite_selected_arch): New function.
26996 (arm_rewrite_march): New function.
26998 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27000 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
27001 (x_arm_cpu_string, x_arm_tune_string): Likewise.
27002 (march, mcpu, mtune): Convert to string-based options.
27003 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
27004 (arm_parse_arch_cpu_name): New function.
27005 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
27006 identify selected architecture or CPU.
27007 (arm_option_save): New function.
27008 (TARGET_OPTION_SAVE): Redefine.
27009 (arm_option_restore): Restore string options.
27010 (arm_option_print): Print string options.
27012 2017-06-16 Martin Sebor <msebor@redhat.com>
27014 PR tree-optimization/80933
27015 PR tree-optimization/80934
27016 * builtins.c (fold_builtin_3): Do not handle bcmp here.
27017 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
27018 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
27019 (gimple_fold_builtin): Call them.
27021 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27023 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
27024 as unlikely; update profile.
27026 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27028 * predict.c (force_edge_cold): Handle declaring edges impossible
27031 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27033 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
27035 (try_unroll_loop_completely): Fix reporting.
27037 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27039 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
27041 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
27044 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
27045 if given a non-constant argument for an intrinsic which requires a
27048 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27050 * profile.c (compare_freqs): New function.
27051 (branch_prob): Sort edge list.
27052 (find_spanning_tree): Assume that the list is priority sorted.
27054 2017-06-16 Richard Biener <rguenther@suse.de>
27056 PR tree-optimization/81090
27057 * passes.def (pass_record_bounds): Remove.
27058 * tree-pass.h (make_pass_record_bounds): Likewise.
27059 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
27060 make_pass_record_bounds): Likewise.
27061 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
27062 not free niter estimates at the beginning but at the end.
27063 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
27065 2017-06-16 Richard Biener <rguenther@suse.de>
27067 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
27068 initializer to workaround ICE in host GCC 4.8.
27070 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27072 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
27074 (clone_inlined_nodes): Update.
27076 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27078 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
27079 prefetch settings, and enable prefetching by default at -O3.
27081 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27083 * config/aarch64/aarch64.c (aarch64_override_options_internal):
27084 Set flag_prefetch_loop_arrays according to tuning data.
27086 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27088 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
27089 New tune structure.
27090 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
27091 [Unrelated to main purpose of the patch] Place the pointer field last
27092 to enable type checking errors when tune structure are wrongly merged.
27093 * config/aarch64/aarch64.c (generic_prefetch_tune,)
27094 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
27095 (thunderx2t99_prefetch_tune): New tune constants.
27096 (tune_params *_tunings): Update all tunings (no functional change).
27097 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
27098 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
27099 from tunings structures.
27101 2017-06-16 Jakub Jelinek <jakub@redhat.com>
27104 * ubsan.c (instrument_null): Add T argument, use it instead
27105 of computing it based on IS_LHS.
27106 (instrument_object_size): Likewise.
27107 (pass_ubsan::execute): Adjust instrument_null and
27108 instrument_object_size callers to pass gimple_get_lhs or
27109 gimple_assign_rhs1 result to it. Use instrument_null instead of
27110 calling get_base_address and instrument_mem_ref. Handle
27111 aggregate call arguments for object-size sanitization.
27113 2017-06-16 Yury Gribov <tetra2005@gmail.com>
27115 PR tree-optimization/81089
27116 * tree-vrp.c (is_masked_range_test): Validate operands of
27119 2017-06-15 Martin Sebor <msebor@redhat.com>
27122 * dumpfile.c (dump_register): Avoid calling memset to initialize
27123 a class with a default ctor.
27124 * gcc.c (struct compiler): Remove const qualification.
27125 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
27126 * hash-table.h: Ditto.
27127 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
27129 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
27130 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
27132 * params.h (struct param_info): Make struct members non-const.
27133 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
27134 with default initialization.
27135 * vec.h (vec_copy_construct, vec_default_construct): New helper
27137 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
27138 with vec_copy_construct.
27139 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
27140 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
27141 * doc/invoke.texi (-Wclass-memaccess): Document.
27143 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27145 * emit-rtl.h (is_leaf): Update comment about local
27146 register allocator.
27148 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27151 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
27152 for a variable to have a section before checking if the section has a
27154 Set section to.persistent if persistent attribute is set.
27155 Warn if .persistent attribute is used on an automatic variable.
27157 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
27159 PR rtl-optimization/80474
27160 * reorg.c (update_block): Do not ignore instructions in a delay slot.
27162 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
27164 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
27167 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
27169 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
27170 (casesi): Emit bounds checking as RTL.
27171 (casesi_internal_mips16_<mode>): Remove bounds checking.
27173 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
27175 * config/xtensa/xtensa.c (xtensa_option_override): Append
27176 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
27177 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
27178 xtensa_doloop_hooks): Define unconditionally.
27179 (xtensa_reorg_loops): Only call reorg_loops in the presence of
27181 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
27182 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
27183 for it in xtensa_option_override.
27184 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
27185 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
27187 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
27189 * doc/cppopts.texi: Document '-' special value to -MF.
27191 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
27193 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
27194 (cortex_a53_fconst): Likewise.
27195 (cortex_a53_fpmul): Likewise.
27196 (cortex_a53_f_load_64): Likewise.
27197 (cortex_a53_f_load_many): Likewise.
27198 (cortex_a53_advsimd_alu): Likewise.
27199 (cortex_a53_advsimd_alu_q): Likewise.
27200 (cortex_a53_advsimd_mul): Likewise.
27201 (cortex_a53_advsimd_mul_q): Likewise.
27202 (fpmac bypass): Add new bypass for fpmac-fpmac case.
27203 Add missing fmul, r2f_cvt and fconst cases.
27205 2017-06-14 Richard Biener <rguenther@suse.de>
27207 PR middle-end/81088
27208 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
27210 (fold_binary_loc): When associating do not treat pre-existing
27211 TREE_OVERFLOW on literal constants as a reason to allow
27212 TREE_OVERFLOW on associated literal constants.
27214 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27216 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
27217 (MASK_FEATURES): New macro.
27218 * config/sparc/sparc.c (sparc_option_override): Remove the special
27219 handling of -mfpu and generalize it to all MASK_FEATURES switches.
27221 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27223 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
27224 a division of 0 if non-call exceptions are enabled.
27226 2017-06-14 Andrew Pinski <apinski@cavium.com>
27227 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27230 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
27231 Improve vector initialization code gen for only variable case.
27233 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27235 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
27237 2017-06-14 Richard Biener <rguenther@suse.de>
27239 PR tree-optimization/81083
27240 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
27243 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27245 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
27246 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
27247 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
27248 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
27249 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
27250 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
27252 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27254 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
27255 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
27257 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27259 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
27261 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27263 * config/rs6000/t-rtems: Don't handle SPE.
27265 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27267 * config/rs6000/t-linux: Don't handle SPE.
27269 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27271 * config/rs6000/eabispe.h: Delete file.
27273 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27275 * config/rs6000/t-spe: Delete file.
27277 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27279 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
27280 (rs6000_legitimate_offset_address_p): Return false for anything in
27281 V2SImode or V2SFmode.
27283 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27285 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
27286 except V2SF and V2SI. Rearrange the vector modes, and add comments.
27287 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
27289 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
27290 (rs6000_legitimate_offset_address_p): Ditto.
27291 (rs6000_emit_move): Ditto.
27292 (rs6000_init_builtins): Remove V4HI_type_node.
27294 2017-06-13 Martin Liska <mliska@suse.cz>
27297 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
27298 (gate_asan): Likewise.
27299 * asan.h (asan_no_sanitize_address_p): Remove the function.
27300 (sanitize_flags_p): New function.
27301 * builtins.def: Fix coding style.
27302 * common.opt: Use renamed enum value.
27303 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
27304 * doc/extend.texi: Document no_sanitize attribute.
27305 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
27306 to SANITIZE_UNDEFINED_NONDEFAULT.
27307 * gcc.c (sanitize_spec_function): Use the renamed enum value.
27308 * gimple-fold.c (optimize_atomic_compare_exchange_p):
27309 Use sanitize_flags_p.
27310 * gimplify.c (gimplify_function_tree): Likewise.
27311 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
27312 * opts.c (parse_no_sanitize_attribute): New function.
27313 (common_handle_option): Use renamed enum value.
27314 * opts.h (parse_no_sanitize_attribute): Declare.
27315 * tree.c (sanitize_flags_p): New function.
27316 * tree.h: Declared here.
27317 * tsan.c: Use sanitize_flags_p.
27318 * ubsan.c (ubsan_expand_null_ifn): Likewise.
27319 (instrument_mem_ref): Likewise.
27320 (instrument_bool_enum_load): Likewise.
27321 (do_ubsan_in_current_function): Remove the function.
27322 (pass_ubsan::execute): Use sanitize_flags_p.
27323 * ubsan.h: Remove do_ubsan_in_current_function
27324 * tree-cfg.c (print_no_sanitize_attr_value): New function.
27325 (dump_function_to_file): Use it here.
27327 2017-06-13 Martin Jambor <mjambor@suse.cz>
27329 PR tree-optimization/80803
27330 PR tree-optimization/81063
27331 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
27332 (propagate_subaccesses_across_link): Enqueue subtree whenever
27333 necessary instead of relying on the caller.
27335 2017-06-13 Martin Jambor <mjambor@suse.cz>
27337 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
27338 that have a first_link.
27339 (sort_and_splice_var_accesses): Do not check first_link before
27341 (subtree_mark_written_and_enqueue): Likewise.
27342 (propagate_all_subaccesses): Likewise and do not stop at first
27343 parent with a first_link.
27345 2017-06-13 Martin Jambor <mjambor@suse.cz>
27347 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
27350 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27352 * match.pd: New pattern.
27354 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27356 * tree-vrp.c (is_masked_range_test): New function.
27357 (register_edge_assert_for): Determine ranges for
27360 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27362 PR tree-optimization/67328
27363 * fold-const.c (maskable_range_p): New function.
27364 (build_range_check): Generate bittests if possible.
27366 2017-06-13 Martin Liska <mliska@suse.cz>
27368 * gimple-pretty-print.c (dump_probability): Add new argument.
27369 (dump_edge_probability): Dump both probability and count.
27370 (dump_gimple_label): Likewise.
27371 (dump_gimple_bb_header): Likewise.
27373 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
27376 * config/avr/avr-devices.c: Fix indentation.
27377 * config/avr/gen-avr-mmcu-specs.c: Dito.
27379 2017-06-13 Richard Biener <rguenther@suse.de>
27381 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
27382 instead get vector type from stmt_info.
27383 (vectorizable_reduction): Adjust. Remove dead code.
27385 2017-06-13 Richard Biener <rguenther@suse.de>
27387 PR middle-end/81065
27388 * fold-const.c (extract_muldiv_1): Remove bogus distribution
27389 case of C * (x * C2 + C3).
27390 (fold_addr_of_array_ref_difference): Properly fold index difference.
27392 2017-06-12 David S. Miller <davem@davemloft.net>
27395 * config/sparc/sparc.md (return expander): Emit frame blockage if
27396 function uses alloca.
27398 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
27400 * combine.c (make_field_assignment): Check len rather than the mode
27401 precision when calling force_to_mode.
27403 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
27405 Support multilibs and devices that see flash in RAM address range.
27408 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
27409 (avr_mcu_t) <flash_pm_offset>: New field.
27410 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
27411 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
27412 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
27413 (AVR_TINY_PM_OFFSET): Remove macro.
27414 * config/avr/avr.opt (-mshort-calls): New option.
27415 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
27416 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
27417 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
27418 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
27419 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
27420 instead of avr_arch->have_jmp_call.
27421 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
27422 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
27423 avr_arch->flash_pm_offset to define.
27424 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
27425 new field flash_pm_offset. Add entry for avrxmega3.
27426 (avr_texinfo): Add entry for avrxmega3.
27427 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
27428 attiny212, attiny214,
27429 attiny412, attiny414, attiny416, attiny417,
27430 attiny814, attiny816, attiny817,
27431 attiny1614, attiny1616, attiny1617,
27432 attiny3214, attiny3216, attiny3217.
27433 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
27434 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
27435 (avr_print_operand_address) [AVR_TINY]: Same.
27436 (avr_asm_init_sections) <readonly_data_section>: Only patch
27437 callback if avr_arch->flash_pm_offset = 0.
27438 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
27439 for rodata if avr_arch->flash_pm_offset != 0.
27440 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
27441 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
27442 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
27443 (m_options): Append opt_rcall.
27444 (m_dirnames): Append dir_rcall.
27445 * config/avr/t-multilib: Regenerate.
27447 * configure.ac [target=avr]: Check whether avrxmega3 default
27448 linker description file works as needed.
27449 * configure: Regenerate.
27450 * doc/avr-mmcu.texi: Regenerate.
27451 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
27452 <__AVR_ARCH__>: Document avrxmega3 and 103.
27453 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
27454 <__AVR_SHORT_CALLS__>: Document it.
27455 <__AVR_PM_BASE_ADDRESS__>: Document it.
27456 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
27457 (AVR Variable Attributes) <progmem>: Document this is
27458 not needed for avrxmega3.
27459 (AVR Named Address Spaces) <__flash>: Dito.
27461 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
27463 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
27465 2017-06-12 Doug Rupp <rupp@adacore.com>
27467 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
27468 Append vxworks-stdint.h to the tm_file list.
27469 * config/vxworks-stdint.h: New file.
27471 2017-06-12 Martin Liska <mliska@suse.cz>
27473 PR tree-optimization/81041
27474 * tree-profile.c (gimple_gen_ic_func_profiler):
27475 Create an extra BB in profile-generate
27476 (gimple_gen_time_profiler): Likewise.
27478 2017-06-12 Jakub Jelinek <jakub@redhat.com>
27480 PR tree-optimization/81003
27481 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
27482 (update_range_test): Use it instead of force_gimple_operand_gsi.
27484 2017-06-12 Richard Biener <rguenther@suse.de>
27486 PR tree-optimization/81053
27487 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
27488 with backedge value not defined in loop. Simplify def stmt
27491 2017-06-11 Tom de Vries <tom@codesourcery.com>
27494 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
27496 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
27497 nvptx_cannot_force_const_mem.
27499 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27501 * opts.c (finish_options): Move test for flag_split_stack after
27502 it has been initialized.
27504 2017-06-11 Jason Merrill <jason@redhat.com>
27506 * tree.h (id_equal): New.
27507 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
27508 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
27509 instead of strcmp of IDENTIFIER_POINTER.
27511 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27513 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
27514 (mark_all_inlined_calls_cdtor): Fix formating.
27515 (inline_transform): Rescale profile before inlining.
27517 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27519 * cgraph.h (cgraph_edge::clone): Update prototype.
27520 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
27521 (cgraph_node::create_clone): Update.
27522 (cgraph_node::create_version_clone): Update.
27523 * tree-inline.c (copy_bb): Update.
27524 (expand_call_inline): Update.
27526 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27528 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
27529 factored out from ...
27530 (rs6000_emit_prologue): ... here.
27532 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27534 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
27535 factored out from ...
27536 (rs6000_emit_prologue): ... here.
27538 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27540 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
27542 (handle_missing_profiles): Fix computation of tp_first_run.
27543 (counts_to_freqs): Do not touch freqs when count is 0.
27545 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27547 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
27550 2017-06-10 Tom de Vries <tom@codesourcery.com>
27552 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27553 attributes): Document signal effective target.
27555 2017-06-10 Tom de Vries <tom@codesourcery.com>
27557 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27558 Document effective target stack_size.
27560 2017-06-09 David Malcolm <dmalcolm@redhat.com>
27562 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
27563 to the edit_context if they can be auto-applied.
27565 2017-06-9 Ian Lance Taylor <iant@golang.org>
27567 * opts.c (finish_options): If -fsplit-stack, disable implicit
27568 -forder-blocks-and-partition.
27569 * doc/invoke.texi (Optimize Options): Document that when using
27570 -fsplit-stack -forder-blocks-and-partition is not implicitly
27573 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27575 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
27576 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
27577 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
27578 * builtins.def (abort, trap, unreachable): Declare cold.
27579 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
27580 * tree-core.h (ECF_COLD): New.
27581 * tree.c (set_call_expr_flags): Handle ECF_COLD.
27582 (build_common_builtin_nodes): Mark unreachable and abort as cold.
27584 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27586 * predict.c (unlikely_executed_stmt_p): Cleanup.
27588 2017-06-09 Richard Biener <rguenther@suse.de>
27590 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
27591 model if the ref is always written to.
27593 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27595 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
27597 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27599 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
27602 2017-06-09 Tom de Vries <tom@codesourcery.com>
27605 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
27606 "target cannot support label values" when encountering LABEL_REF.
27608 2017-06-09 Martin Liska <mliska@suse.cz>
27610 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
27611 (gimple_gen_ic_func_profiler): Emit direct comparison
27612 of __gcov_indirect_call_callee with NULL.
27613 (gimple_gen_time_profiler): Change probability from
27614 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
27616 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27618 * profile.c (edge_gcov_counts): Turn to pointer.
27619 (compute_branch_probabilities, compute_branch_probabilities): Update.
27620 (branch_prob): Do not clear edge_gcov_count.
27621 * profile.h (edge_gcov_counts): Turn to pointer.
27622 (edge_gcov_count): Update.
27624 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27626 * gimple.h (gimple_check_failed): Mark cold.
27628 2017-06-09 Richard Biener <rguenther@suse.de>
27630 PR tree-optimization/66623
27631 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
27632 refactor check_reduction into two parts, properly computing
27633 whether we have to check reduction validity for outer loop
27636 2017-06-09 Richard Biener <rguenther@suse.de>
27638 PR tree-optimization/79483
27639 * graphite-scop-detection.c (order): New global.
27640 (get_order): Compute bb to order mapping that satisfies code
27641 generation constraints.
27642 (cmp_pbbs): New helper.
27643 (build_scops): Start domwalk at entry block, sort generated
27646 2017-06-09 Richard Biener <rguenther@suse.de>
27648 PR middle-end/81007
27649 * ipa-polymorphic-call.c
27650 (ipa_polymorphic_call_context::restrict_to_inner_class):
27651 Skip FIELD_DECLs with error_mark_node type.
27652 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
27655 2017-06-09 Martin Liska <mliska@suse.cz>
27657 * predict.c (struct branch_predictor): New struct.
27658 (test_prediction_value_range): New test.
27659 (predict_c_tests): New function.
27660 * selftest-run-tests.c (selftest::run_tests): Run the function.
27661 * selftest.h: Declare new tests.
27663 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
27666 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
27667 gen_add3_insn did not fail.
27668 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
27669 r0, construct that number in a temporary reg and add that reg to r0.
27670 If asked to put the result in r0 as well, fail.
27672 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
27674 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27675 for early expansion of vec_eqv.
27677 2017-06-08 Jakub Jelinek <jakub@redhat.com>
27679 PR middle-end/81005
27680 * ubsan.c (instrument_null): Avoid pointless code temporary.
27681 (pass_ubsan::execute): Instrument aggregate arguments of calls.
27683 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
27687 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
27690 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
27693 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27695 * predict.c (unlikely_executed_edge_p): Move ahead.
27696 (probably_never_executed_edge_p): Use it.
27698 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27700 PR middle-end/79988
27701 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
27702 gimple_call_builtin_p call.
27704 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27706 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
27707 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
27708 rtl_check_failed_type2, rtl_check_failed_code1,
27709 rtl_check_failed_code2, rtl_check_failed_code_mode,
27710 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
27711 rtvec_check_failed_bounds, rtl_check_failed_flag,
27712 _fatal_insn_not_found, _fatal_insn): Likewise.
27713 * tree.h (tree_contains_struct_check_failed,
27714 tree_check_failed, tree_not_check_failed,
27715 tree_class_check_failed, tree_range_check_failed,
27716 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
27717 tree_vec_elt_check_failed, phi_node_elt_check_failed,
27718 tree_operand_check_failed, omp_clause_check_failed,
27719 omp_clause_operand_check_failed, omp_clause_range_check_failed):
27722 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27724 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
27725 flag_branch_probabilities.
27726 * ipa-inline.c (edge_badness): Likewise.
27727 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
27728 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
27729 * predict.c (maybe_hot_frequency_p): Likewise.
27730 (probably_never_executed): Likewise.
27731 * sched-ebb.c (schedule_ebbs): Likewise.
27732 * sched-rgn.c (find_single_block_region): Likewise.
27733 * tracer.c (tail_duplicate): Likewise.
27735 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27737 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
27738 longer requires x_flag_profile_use.
27740 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27742 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
27743 instead of flag_reorder_blocks_and_partition.
27744 * dbxout.c (dbxout_function_end): Likewise.
27745 * dwarf2out.c (gen_subprogram_die): Likewise.
27746 * haifa-sched.c (sched_create_recovery_edges): Likewise.
27747 * hw-doloop.c (reorg_loops): Likewise.
27748 * varasm.c (assemble_start_function,
27749 assemble_end_function): Likewise.
27750 (decide_function_section): Do not check for
27751 flag_reorder_blocks_and_partition.
27753 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27755 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27757 (chkp_get_hard_register_fake_addr_expr): Ditto.
27758 (chkp_build_addr_expr): Add check for hard reg case.
27759 (chkp_parse_array_and_component_ref): Ditto.
27760 (chkp_find_bounds_1): Ditto.
27761 (chkp_process_stmt): Don't generate bounds store for
27764 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27766 * predict.c (maybe_hot_bb_p): Do not check profile status.
27767 (maybe_hot_edge_p): Likewise.
27768 (probably_never_executed): Check for zero counts even if profile
27770 (unlikely_executed_edge_p): New function.
27771 (unlikely_executed_stmt_p): New function.
27772 (unlikely_executed_bb_p): New function.
27773 (set_even_probabilities): Use unlikely predicates.
27774 (combine_predictions_for_bb): Likewise.
27775 (predict_paths_for_bb): Likewise.
27776 (predict_paths_leading_to_edge): Likewise.
27777 (determine_unlikely_bbs): New function.
27778 (estimate_bb_frequencies): Use it.
27779 (compute_function_frequency): Use zero counts even if profile is
27781 * profile-count.h: Fix typo.
27783 2017-08-08 Julia Koval <julia.koval@intel.com>
27785 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
27786 _mm512_mask_cvtsepi16_storeu_epi8,
27787 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
27788 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
27789 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
27790 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
27791 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
27792 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
27793 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
27794 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
27795 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
27796 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
27797 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
27798 __builtin_ia32_pmovuswb256mem_mask,
27799 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
27800 __builtin_ia32_pmovwb512mem_mask): New builtins.
27802 2017-08-08 Julia Koval <julia.koval@intel.com>
27804 PR target/73350,80862
27805 * config/i386/subst.md (round): Fix round pattern.
27806 * config/i386/i386.c (ix86_erase_embedded_rounding):
27807 Fix erasing rounding for the fixed pattern.
27809 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27811 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
27813 2017-06-08 Martin Liska <mliska@suse.cz>
27815 PR gcov-profile/80911
27816 * gcov.c (block_info::block_info): New constructor.
27818 2017-06-07 Carl Love <cel@us.ibm.com>
27820 * config/rs6000/rs6000-c: The return type of the following
27821 built-in functions was implemented as int not long long. Fix sign
27822 of return value for the unsigned version of vec_mulo and vec_mule.
27823 vector unsigned long long vec_bperm (vector unsigned long long,
27824 vector unsigned char)
27825 vector signed long long vec_mule (vector signed int,
27827 vector unsigned long long vec_mule (vector unsigned int,
27828 vector unsigned int)
27829 vector signed long long vec_mulo (vector signed int,
27831 vector unsigned long long vec_mulo (vector unsigned int,
27832 vector unsigned int)
27833 * doc/extend.texi: Fix the documentation for the built-in
27836 2017-06-07 Carl Love <cel@us.ibm.com>
27839 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
27842 2017-06-07 Carl Love <cel@us.ibm.com>
27844 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
27845 support, Generate doublehv for signed int/float for BE case only.
27847 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
27849 * doc/invoke.texi (mcx16): Rewrite.
27851 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27853 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
27854 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
27855 *mov<mode>_softfloat, and an anonymous splitter): Use
27856 nonimmediate_operand instead of rs6000_nonimmediate_operand.
27858 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27860 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
27862 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27863 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
27864 (rs6000_debug_reg_global): Adjust.
27865 (rs6000_init_hard_regno_mode_ok): Adjust.
27866 (rs6000_dbx_register_number): Adjust.
27867 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
27868 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27869 Remove SPE_ACC and SPEFSCR.
27870 (REG_ALLOC_ORDER): Ditto.
27871 (FRAME_POINTER_REGNUM): Change to 111.
27872 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
27873 (REG_CLASS_NAMES): Ditto.
27874 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
27875 (REGISTER_NAMES): Ditto.
27876 (ADDITIONAL_REG_NAMES): Ditto.
27877 (rs6000_reg_names): Ditto.
27878 * config/rs6000/rs6000.md: Renumber some register number
27881 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27883 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
27885 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27886 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
27888 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
27889 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27890 Delete the SPE high registers.
27891 (REG_ALLOC_ORDER): Ditto.
27892 (enum reg_class): Remove SPE_HIGH_REGS.
27893 (REG_CLASS_NAMES): Ditto.
27894 (REG_CLASS_CONTENTS): Delete the SPE high registers.
27895 (REGISTER_NAMES): Ditto.
27896 (rs6000_reg_names): Ditto.
27897 * doc/tm.texi.in: Remove SPE as example.
27898 * doc/tm.texi: Regenerate.
27900 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27902 * config/rs6000/8540.md (ppc8540_brinc): Delete.
27903 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
27904 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
27905 * config/rs6000/rs6000.md (type): Remove "brinc".
27907 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27909 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
27910 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
27911 * config/rs6000/linuxspe.h: Delete file.
27912 * config/rs6000/rs6000.md: Don't include spe.md.
27913 * config/rs6000/spe.h: Delete file.
27914 * config/rs6000/spe.md: Delete file.
27915 * config/rs6000/t-rs6000: Remove spe.md.
27917 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27919 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
27920 (reg_or_none500mem_operand): Delete.
27921 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
27922 instead of reg_or_none500mem_operand.
27924 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27926 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27927 handling of SPE flags.
27928 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
27930 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27932 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
27934 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
27935 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
27936 paired_divv2sf3): Similar.
27937 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
27938 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
27939 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
27941 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
27942 Rename the paired_* instruction patterns.
27943 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
27945 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
27946 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
27947 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
27948 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
27949 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
27950 PAIRED_VECTOR_MODE.
27951 (struct machine_function): Delete field spe_insn_chain_scanned_p.
27952 (spe_func_has_64bit_regs_p): Delete.
27953 (spe_expand_predicate_builtin): Delete.
27954 (spe_expand_evsel_builtin): Delete.
27955 (TARGET_DWARF_REGISTER_SPAN): Do not define.
27956 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
27957 (invalid_e500_subreg): Delete.
27958 (rs6000_legitimize_address): Always force_reg op2 as well, for
27959 paired single memory accesses.
27960 (rs6000_member_type_forces_blk): Delete.
27961 (rs6000_spe_function_arg): Delete.
27962 (rs6000_expand_unop_builtin): Delete SPE handling.
27963 (rs6000_expand_binop_builtin): Ditto.
27964 (spe_expand_stv_builtin): Delete.
27965 (bdesc_2arg_spe): Delete.
27966 (spe_expand_builtin): Delete.
27967 (spe_expand_predicate_builtin): Delete.
27968 (spe_expand_evsel_builtin): Delete.
27969 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
27970 (spe_init_builtins): Delete.
27971 (spe_func_has_64bit_regs_p): Delete.
27972 (savres_routine_name): Delete "info" parameter. Adjust callers.
27973 (rs6000_emit_stack_reset): Ditto.
27974 (rs6000_dwarf_register_span): Delete.
27975 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
27976 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
27977 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
27979 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
27981 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
27982 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
27983 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
27984 mulv2sf3, divv2sf3): Delete expanders.
27986 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27988 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
27990 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27992 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
27993 * config/rs6000/rs6000.c: Ditto.
27995 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27997 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
27998 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
27999 comparison_operator.
28001 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28003 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
28004 * config/rs6000/rs6000.opt: Ditto.
28005 * config/rs6000/t-rtems: Ditto.
28007 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28009 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
28010 TARGET_E500_SINGLE by 0, simplify.
28011 * config/rs6000/rs6000.c: Ditto.
28012 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
28013 (spe_build_register_parallel): Delete.
28014 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
28015 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
28016 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
28017 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
28018 (E500_CONVERT): Delete.
28019 * config/rs6000/spe.md: Remove many patterns and all define_constants.
28021 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28023 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
28024 * config/rs6000/dfp.md: Ditto.
28025 (negdd2, *negdd2_fpr): Merge.
28026 (absdd2, *absdd2_fpr): Merge.
28027 (negtd2, *negtd2_fpr): Merge.
28028 (abstd2, *abstd2_fpr): Merge.
28029 * config/rs6000/e500.h: Delete file.
28030 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
28031 TARGET_FPRS by 1 and simplify.
28032 * config/rs6000/rs6000-c.c: Ditto.
28033 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
28034 TARGET_DF_SPE by 0.
28035 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
28037 * config/rs6000/rs6000.md: Ditto.
28038 (floatdidf2, *floatdidf2_fpr): Merge.
28039 (move_from_CR_gt_bit): Delete.
28040 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
28041 (E500_CR_IOR_COMPARE): Delete.
28042 (All patterns that require !TARGET_FPRS): Delete.
28043 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
28045 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28047 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
28049 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28051 * graphds.c (add_edge): Intitialize edge's attached data.
28052 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
28053 pointer parameter. Call pointed function on each edge during
28054 graph traversing. Skip traversing the edge when the function
28056 (graphds_dfs, graphds_scc): Ditto.
28057 (for_each_edge): New parameter. Pass the new parameter to callback
28059 * graphds.h (skip_edge_callback): New function pointer type.
28060 (graphds_dfs, graphds_scc): New function pointer parameter.
28061 (graphds_edge_callback, for_each_edge): New parameter.
28063 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28065 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
28066 out code checking if runtime alias check is possible to below ...
28067 Call the new function.
28068 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
28069 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
28071 2017-06-07 Marek Polacek <polacek@redhat.com>
28074 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
28075 TYPE_OVERFLOW_WRAPS check.
28077 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28079 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
28080 if versioning is required.
28081 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
28082 peeling with the check for versioning.
28084 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28086 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
28087 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
28088 Set true to new parameter if new ssa variable is defined.
28089 (vect_gen_vector_loop_niters): Refactor. Set range information
28090 for the new vector loop bound variable.
28091 (vect_do_peeling): Ditto.
28093 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28095 * tree-affine.c (ssa.h): Include header file.
28096 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
28097 has wrapping overflow behavior.
28099 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28101 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
28103 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28105 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
28106 (tree_to_aff_combination): ... here.
28108 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28110 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
28111 reg_pressure model function.
28112 (ivopts_global_cost_for_size): Delete.
28113 (determine_set_costs, iv_ca_recount_cost): Call new model function
28114 ivopts_estimate_reg_pressure.
28116 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28118 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
28119 expensive than udiv. Remove floating point cases from mod.
28121 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28123 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
28124 Increase idiv cost.
28126 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28128 * config/aarch64/aarch64.md
28129 (copysignsf3): Fix mask generation.
28131 2017-06-07 Jakub Jelinek <jakub@redhat.com>
28133 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
28135 (class dump_manager): Add register_dumps method.
28136 * dumpfile.c: Include langhooks.h.
28137 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
28138 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
28139 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
28140 (dump_manager::dump_register): Start with 512 entries instead of 32.
28141 (dump_manager::register_dumps): New method.
28142 * toplev.c (general_init): Instead of invoking register_dumps
28143 langhook, invoke register_dumps method on the dump manager.
28144 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
28147 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
28149 * doc/md.texi: Clarify the restrictions on a define_insn condition.
28150 Say that # requires an associated define_split to exist, and that
28151 the define_split must be suitable for use after register allocation.
28153 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28155 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
28156 (compute_outgoing_frequencies): Also initialize zero counts.
28157 (find_many_sub_basic_blocks): Do not produce uninitialized profile
28158 around loops; preserve more of profile when nothing changes.
28160 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
28162 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
28164 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
28165 * config/arm/arm-cpu-cdata.h: Regenerate.
28166 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
28167 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
28168 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
28169 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
28170 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
28172 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
28174 * config/arm/t-rmprofile: Likewise.
28175 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
28177 2017-06-06 David S. Miller <davem@davemloft.net>
28180 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
28181 blockage if function uses alloca.
28183 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28185 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
28186 New "uid" fields to hold pretty-print IDs of group and ref.
28187 Memory references are now identified as <group_id>:<ref_id>
28188 instead of using [random] addresses.
28189 (dump_mem_details): Simplify, no functional change.
28190 (dump_mem_ref): Simplify and make output more concise.
28191 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
28192 (find_or_create_group): Initialize group uid.
28193 (record_ref): Initialize ref uid. Improve debug output.
28194 (prune_group_by_reuse, should_issue_prefetch_p,)
28195 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
28196 (mark_nontemporal_store, determine_loop_nest_reuse):
28197 Improve debug output.
28199 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28201 * dbgcnt.def (prefetch): New debug counter.
28202 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
28203 (schedule_prefetches): Stop issueing prefetches if debug counter
28206 2017-06-06 Tom de Vries <tom@codesourcery.com>
28208 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
28209 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
28211 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28213 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
28214 Use aarch64_reg_or_zero predicate for operand 4.
28215 (aarch64_compare_and_swap<mode> define_insn_and_split):
28216 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
28217 (aarch64_store_exclusive<mode>): Likewise for operand 2.
28219 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
28221 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
28222 (arm_compute_save_core_reg_mask): This.
28223 (thumb1_compute_save_reg_mask): Rename into ...
28224 (thumb1_compute_save_core_reg_mask): This.
28225 (arm_compute_save_reg0_reg12_mask): Adapt comment.
28226 (arm_compute_frame_layout): Likewise.
28228 2017-06-06 Richard Biener <rguenther@suse.de>
28230 PR tree-optimization/80974
28231 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
28232 keep or clear leaders SSA info.
28234 2017-06-06 Tom de Vries <tom@codesourcery.com>
28236 * config/nvptx/nvptx.c (split_mode_p): New function.
28237 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
28239 2017-06-06 Tom de Vries <tom@codesourcery.com>
28241 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
28243 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28246 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
28249 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28251 * shrink-wrap.c (handle_simple_exit): Update profile.
28252 (try_shrink_wrapping): Upate profile.
28254 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28256 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
28257 (tree_guess_outgoing_edge_probabilities): New.
28258 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
28259 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
28261 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28263 * ipa-split.c (split_function): Initialize return bb profile.
28265 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28267 * profile.c (compute_branch_probabilities): Also initialize
28268 EXIT_BLOCK profile.
28270 2017-06-06 Richard Biener <rguenther@suse.de>
28272 PR tree-optimization/80928
28273 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
28274 (vect_analyze_loop_operations): Properly guard analysis for
28276 (vect_transform_loop): Likewise.
28277 (vect_analyze_loop_2): Also reset SLP type on PHIs.
28278 (vect_model_induction_cost): Do not cost for pure SLP.
28279 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
28280 of induction in inner loop vectorization.
28281 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
28282 (vect_get_and_check_slp_defs): Handle vect_induction_def.
28283 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
28285 (vect_analyze_slp_cost_1): Cost induction.
28286 (vect_detect_hybrid_slp_stmts): Handle PHIs.
28287 (vect_get_slp_vect_defs): Likewise.
28288 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
28289 (vect_transform_stmt): Handle SLP reductions.
28290 * tree-vectorizer.h (vectorizable_induction): Adjust.
28292 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28294 * config/rs6000/rs6000.c (make_resolver_func): Update
28295 init_lowered_empty_function call.
28297 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
28299 * doc/invoke.texi: Document the -fprofile-abs-path option.
28300 * common.opt (fprofile-abs-path): New option.
28301 * gcov-io.h (gcov_write_filename): Declare.
28302 * gcov-io.c (gcov_write_filename): New function.
28303 * coverage.c (coverage_begin_function): Use gcov_write_filename.
28304 * profile.c (output_location): Likewise.
28306 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28308 * shring-wrap.c: Revert accidental commit.
28310 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
28312 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
28314 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28316 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
28318 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
28319 profile in callgraph edge.
28320 * profile-count.h (apply_probability): If THIS is 0, then result is 0
28321 (apply_scale): Likewise.
28322 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
28323 Also scale profile when inlining function with zero profile.
28324 (initialize_cfun): Update exit block profile even when it is zero.
28325 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
28326 when profile is read.
28328 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28330 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
28331 (CLONE_*): New constants to define the processors we can generate
28332 code for with the target_clone attribute.
28333 (rs6000_clone_map): New array to identify which clone processors
28334 the current program is running on.
28335 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
28336 target_clone attribute.
28337 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
28338 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
28339 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
28340 (cpu_expand_builtin): Add support for target_clone attribute.
28341 (rs6000_valid_attribute_p): Allow "default" attribute.
28342 (get_decl_name): New debug function to simplify printing the
28343 current function name in debugging statements.
28344 (rs6000_clone_priority): New functions to support the target_clone
28345 attribute, and be able to generate code to switch between ISA 2.05
28346 through ISA 3.0 (power6 through power9).
28347 (rs6000_compare_version_priority): Likewise.
28348 (rs6000_get_function_versions_dispatcher): Likewise.
28349 (make_resolver_func): Likewise.
28350 (add_condition_to_bb): Likewise.
28351 (dispatch_function_versions): Likewise.
28352 (rs6000_generate_version_dispatcher_body): Likewise.
28353 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
28354 (fusion_gpr_load_p): Fix a spacing issue.
28355 * doc/extend.texi (Common Function Attributes): Document that the
28356 PowerPC supports the target_clone attribute.
28358 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28360 * config/arm/arm.h: explain F symbol found in description of ARM
28361 register allocation in its legend.
28363 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28365 * config/mips/frame-header-opt.c: Include profile-count.h.
28366 * config/riscv/riscv.c: Include profile-count.h
28368 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28370 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
28372 (sm_set_flag_if_changed): Add bbs field.
28373 (execute_sm_if_changed_flag_set): Pass BBS.
28374 (execute_sm): Update.
28376 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28378 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
28381 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28383 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
28384 (peephole2): New peephole2 to emit the above.
28385 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
28387 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28389 * config/aarch64/aarch64.c (define_peephole2 above
28390 *sub_<shift>_<mode>): New peephole.
28392 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28394 * config/i386/i386.c (make_resolver_func): Update.
28395 * Makefile.in: Add profile-count.h and profile-count.o
28396 * auto-profile.c (afdo_indirect_call): Update to new API.
28397 (afdo_set_bb_count): Update.
28398 (afdo_propagate_edge): Update.
28399 (afdo_propagate_circuit): Update.
28400 (afdo_calculate_branch_prob): Update.
28401 (afdo_annotate_cfg): Update.
28402 * basic-block.h: Include profile-count.h
28403 (struct edge_def): Turn count to profile_count.
28404 (struct basic_block_def): Likewie.
28405 (REG_BR_PROB_BASE): Move to profile-count.h
28406 (RDIV): Move to profile-count.h
28407 * bb-reorder.c (max_entry_count): Turn to profile_count.
28408 (find_traces): Update.
28409 (rotate_loop):Update.
28410 (connect_traces):Update.
28411 (sanitize_hot_paths):Update.
28412 * bt-load.c (migrate_btr_defs): Update.
28413 * cfg.c (RDIV): Remove.
28414 (init_flow): Use alloc_block.
28415 (alloc_block): Uninitialize count.
28416 (unchecked_make_edge): Uninitialize count.
28417 (check_bb_profile): Update.
28418 (dump_edge_info): Update.
28419 (dump_bb_info): Update.
28420 (update_bb_profile_for_threading): Update.
28421 (scale_bbs_frequencies_int): Update.
28422 (scale_bbs_frequencies_gcov_type): Update.
28423 (scale_bbs_frequencies_profile_count): New.
28424 * cfg.h (update_bb_profile_for_threading): Update.
28425 (scale_bbs_frequencies_profile_count): Declare.
28426 * cfgbuild.c (compute_outgoing_frequencies): Update.
28427 (find_many_sub_basic_blocks): Update.
28428 * cfgcleanup.c (try_forward_edges): Update.
28429 (try_crossjump_to_edge): Update.
28430 * cfgexpand.c (expand_gimple_tailcall): Update.
28431 (construct_exit_block): Update.
28432 * cfghooks.c (verify_flow_info): Update.
28433 (dump_bb_for_graph): Update.
28434 (split_edge): Update.
28435 (make_forwarder_block): Update.
28436 (duplicate_block): Update.
28437 (account_profile_record): Update.
28438 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
28439 (get_estimated_loop_iterations): Update.
28440 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
28441 (single_likely_exit): Update.
28442 * cfgloopmanip.c (scale_loop_profile): Update.
28444 (set_zero_probability): Update.
28445 (lv_adjust_loop_entry_edge): Update.
28446 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
28447 (purge_dead_edges): Update.
28448 (rtl_account_profile_record): Update.
28449 * cgraph.c (cgraph_node::create): Uninitialize count.
28450 (symbol_table::create_edge): Uninitialize count.
28451 (cgraph_update_edges_for_call_stmt_node): Update.
28452 (cgraph_edge::dump_edge_flags): Update.
28453 (cgraph_node::dump): Update.
28454 (cgraph_edge::maybe_hot_p): Update.
28455 * cgraph.h: Include profile-count.h
28456 (create_clone), create_edge, create_indirect_edge): Update.
28457 (cgraph_node): Turn count to profile_count.
28458 (cgraph_edge0: Likewise.
28459 (make_speculative, clone): Update.
28460 (create_edge): Update.
28461 (init_lowered_empty_function): Update.
28462 * cgraphclones.c (cgraph_edge::clone): Update.
28463 (duplicate_thunk_for_node): Update.
28464 (cgraph_node::create_clone): Update.
28465 * cgraphunit.c (cgraph_node::analyze): Update.
28466 (cgraph_node::expand_thunk): Update.
28467 * final.c (dump_basic_block_info): Update.
28468 * gimple-streamer-in.c (input_bb): Update.
28469 * gimple-streamer-out.c (output_bb): Update.
28470 * graphite.c (print_global_statistics): Update.
28471 (print_graphite_scop_statistics): Update.
28472 * hsa-brig.c: Include basic-block.h.
28473 * hsa-dump.c: Include basic-block.h.
28474 * hsa-gen.c (T sum_slice): Update.
28475 (convert_switch_statements):Update.
28476 * hsa-regalloc.c: Include basic-block.h.
28477 * ipa-chkp.c (chkp_produce_thunks): Update.
28478 * ipa-cp.c (struct caller_statistics): Update.
28479 (init_caller_stats): Update.
28480 (gather_caller_stats): Update.
28481 (ipcp_cloning_candidate_p): Update.
28482 (good_cloning_opportunity_p): Update.
28483 (get_info_about_necessary_edges): Update.
28484 (dump_profile_updates): Update.
28485 (update_profiling_info): Update.
28486 (update_specialized_profile): Update.
28487 (perhaps_add_new_callers): Update.
28488 (decide_about_value): Update.
28489 (ipa_cp_c_finalize): Update.
28490 * ipa-devirt.c (struct odr_type_warn_count): Update.
28491 (struct decl_warn_count): Update.
28492 (struct final_warning_record): Update.
28493 (possible_polymorphic_call_targets): Update.
28494 (ipa_devirt): Update.
28495 * ipa-fnsummary.c (redirect_to_unreachable): Update.
28496 * ipa-icf.c (sem_function::merge): Update.
28497 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
28498 * ipa-inline.c (compute_uninlined_call_time): Update.
28499 (compute_inlined_call_time): Update.
28500 (want_inline_small_function_p): Update.
28501 (want_inline_self_recursive_call_p): Update.
28502 (edge_badness): Update.
28503 (lookup_recursive_calls): Update.
28504 (recursive_inlining): Update.
28505 (inline_small_functions): Update.
28506 (dump_overall_stats): Update.
28507 (dump_inline_stats): Update.
28508 * ipa-profile.c (ipa_profile_generate_summary): Update.
28509 (ipa_propagate_frequency): Update.
28510 (ipa_profile): Update.
28511 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28512 * ipa-utils.c (ipa_merge_profiles): Update.
28513 * loop-doloop.c (doloop_modify): Update.
28514 * loop-unroll.c (report_unroll): Update.
28515 (unroll_loop_runtime_iterations): Update.
28516 * lto-cgraph.c (lto_output_edge): Update.
28517 (lto_output_node): Update.
28518 (input_node): Update.
28519 (input_edge): Update.
28520 (merge_profile_summaries): Update.
28521 * lto-streamer-in.c (input_cfg): Update.
28522 * lto-streamer-out.c (output_cfg): Update.
28523 * mcf.c (create_fixup_graph): Update.
28524 (adjust_cfg_counts): Update.
28525 (sum_edge_counts): Update.
28526 * modulo-sched.c (sms_schedule): Update.
28527 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
28528 * predict.c (maybe_hot_count_p): Update.
28529 (probably_never_executed): Update.
28530 (dump_prediction): Update.
28531 (combine_predictions_for_bb): Update.
28532 (propagate_freq): Update.
28533 (handle_missing_profiles): Update.
28534 (counts_to_freqs): Update.
28535 (rebuild_frequencies): Update.
28536 (force_edge_cold): Update.
28537 * predict.h: Include profile-count.h
28538 (maybe_hot_count_p, counts_to_freqs): UPdate.
28539 * print-rtl-function.c: Do not include cfg.h
28540 * print-rtl.c: Include basic-block.h
28541 * profile-count.c: New file.
28542 * profile-count.h: New file.
28543 * profile.c (is_edge_inconsistent): Update.
28544 (correct_negative_edge_counts): Update.
28545 (is_inconsistent): Update.
28546 (set_bb_counts): Update.
28547 (read_profile_edge_counts): Update.
28548 (compute_frequency_overlap): Update.
28549 (compute_branch_probabilities): Update; Initialize and deinitialize
28551 (branch_prob): Update.
28552 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
28553 (edge_gcov_count): New.
28554 (bb_gcov_count): New.
28555 * shrink-wrap.c (try_shrink_wrapping): Update.
28556 * tracer.c (better_p): Update.
28557 * trans-mem.c (expand_transaction): Update.
28558 (ipa_tm_insert_irr_call): Update.
28559 (ipa_tm_insert_gettmclone_call): Update.
28560 * tree-call-cdce.c: Update.
28561 * tree-cfg.c (gimple_duplicate_sese_region): Update.
28562 (gimple_duplicate_sese_tail): Update.
28563 (gimple_account_profile_record): Update.
28564 (execute_fixup_cfg): Update.
28565 * tree-inline.c (copy_bb): Update.
28566 (copy_edges_for_bb): Update.
28567 (initialize_cfun): Update.
28568 (freqs_to_counts): Update.
28569 (copy_cfg_body): Update.
28570 (expand_call_inline): Update.
28571 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
28572 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
28573 (try_unroll_loop_completely): Update.
28574 (try_peel_loop): Update.
28575 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
28576 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
28577 * tree-ssa-loop-split.c (connect_loops): Update.
28578 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
28579 * tree-ssa-reassoc.c (branch_fixup): Update.
28580 * tree-ssa-tail-merge.c (replace_block_by): Update.
28581 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
28582 (compute_path_counts): Update.
28583 (update_profile): Update.
28584 (recompute_probabilities): Update.
28585 (update_joiner_offpath_counts): Update.
28586 (estimated_freqs_path): Update.
28587 (freqs_to_counts_path): Update.
28588 (clear_counts_path): Update.
28589 (ssa_fix_duplicate_block_edges): Update.
28590 (duplicate_thread_path): Update.
28591 * tree-switch-conversion.c (case_bit_test_cmp): Update.
28592 (struct switch_conv_info): Update.
28593 * tree-tailcall.c (decrease_profile): Update.
28594 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
28595 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
28596 * value-prof.c (check_counter): Update.
28597 (gimple_divmod_fixed_value): Update.
28598 (gimple_mod_pow2): Update.
28599 (gimple_mod_subtract): Update.
28600 (gimple_ic_transform): Update.
28601 (gimple_stringop_fixed_value): Update.
28602 * value-prof.h (gimple_ic): Update.
28604 2017-06-02 Carl Love <cel@us.ibm.com>
28606 * config/rs6000/rs6000-c: Add support for built-in functions
28607 vector double vec_doublee (vector signed int);
28608 vector double vec_doublee (vector unsigned int);
28609 vector double vec_doublee (vector float);
28610 vector double vec_doubleh (vector signed int);
28611 vector double vec_doubleh (vector unsigned int);
28612 vector double vec_doubleh (vector float);
28613 vector double vec_doublel (vector signed int);
28614 vector double vec_doublel (vector unsigned int);
28615 vector double vec_doublel (vector float);
28616 vector double vec_doubleo (vector signed int);
28617 vector double vec_doubleo (vector unsigned int);
28618 vector double vec_doubleo (vector float);.
28619 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
28620 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
28622 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
28623 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
28624 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
28626 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
28627 vec_doublel, vec_doubleh.
28628 * doc/extend.texi: Update the built-in documentation file for the
28629 new built-in functions.
28631 2017-06-02 David Malcolm <dmalcolm@redhat.com>
28634 * ipa-inline-analysis.c (free_growth_caches): Set
28635 edge_removal_hook_holder to NULL after removing it.
28637 2017-06-02 Sudakshina Das <sudi.das@arm.com>
28639 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
28640 comparision with zero.
28642 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28643 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28644 for early expansion of vec_min and vec_max builtins.
28645 (builtin_function_type): Add min/max unsigned variants to those
28646 identified as having unsigned arguments.
28648 2017-06-02 Olivier Hainque <hainque@adacore.com>
28650 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
28652 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28654 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
28655 Use VALL_F16 iterator rather than VALL.
28657 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28659 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
28660 Emit CBNZ inside loop when doing a strong exchange and comparing
28661 against zero. Generate the CC flags after the loop.
28663 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
28665 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
28666 (dl_section_ref): New.
28667 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
28668 On AIX, append an expression to subtract the size of the
28669 section length to dl_section_ref.
28671 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28673 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28674 for early expansion of vector absolute builtins.
28676 2017-06-02 Richard Biener <rguenther@suse.de>
28678 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
28679 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
28681 2017-06-02 Richard Biener <rguenther@suse.de>
28683 PR tree-optimization/80948
28684 * tree-tailcall.c (find_tail_calls): Track stmts to move in
28685 stmt order as well.
28687 2017-06-02 Richard Biener <rguenther@suse.de>
28689 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
28691 * tree-vect-stmts.c (process_use): Do not mark backedge defs
28692 for inductions as relevant.
28694 2017-06-02 Richard Biener <rguenther@suse.de>
28696 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
28697 (vectorizable_induction): ... this. Remove dead code.
28699 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
28701 * builtins. (expand_builtin_alloca): Remove second parameter and
28702 infer its value from the first parameter instead.
28703 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
28705 2017-06-02 Jakub Jelinek <jakub@redhat.com>
28707 PR rtl-optimization/80903
28708 * loop-doloop.c (add_test): Unshare sequence.
28710 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
28712 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
28714 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
28716 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
28718 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
28719 xlogue_layout::get_instance, logue_layout::xlogue_layout,
28720 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
28721 (xlogue_layout::get_stub_rtx): Make static.
28722 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
28723 (xlogue_layout::compute_stub_managed_regs): Rename to...
28724 (xlogue_layout::count_stub_managed_regs): ...this.
28725 (xlogue_layout::is_stub_managed_reg): New function.
28726 (xlogue_layout::m_stub_names): Rename to...
28727 (xlogue_layout::s_stub_names): ...this, make static.
28728 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
28729 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
28730 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
28731 xlogue_layout::s_stub_names): Instantiate statics.
28732 (stub_managed_regs): Remove.
28733 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
28734 (disable_call_ms2sysv_xlogues): Rename to...
28735 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
28736 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
28738 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
28739 change after reload_completed.
28740 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
28742 (ix86_expand_prologue): Likewise.
28743 (ix86_expand_epilogue): Likewise.
28744 (ix86_expand_split_stack_prologue): Likewise.
28745 (ix86_compute_frame_layout): Remove frame parameter ...
28746 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
28747 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
28749 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
28750 (ix86_frame): Move from here ...
28751 * config/i386/i386.h (ix86_frame): ... to here.
28752 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
28753 complete ix86_frame data structure instead. Remove some_ld_name.
28755 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28757 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
28758 symbols that hold a DECL_VALUE_EXPR.
28760 2017-06-01 Martin Jambor <mjambor@suse.cz>
28762 PR tree-optimization/80898
28763 * tree-sra.c (process_subtree_disqualification): Removed.
28764 (disqualify_candidate): Do not acll
28765 process_subtree_disqualification.
28766 (subtree_mark_written_and_enqueue): New function.
28767 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
28768 RHS has been disqualified and re-queue LHS if necessary. Apart
28769 from that, ignore disqualified RHS.
28771 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28773 * config/s390/s390.c (s390_emit_epilogue): Disable early return
28774 address fetch for z10 or later.
28776 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28778 * config/arc/arc.md (tst_movb): Add guard when splitting.
28780 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28782 * config/arc/arc.c (arc_can_eliminate): Test against
28783 arc_frame_pointer_needed.
28785 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28787 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
28788 to prevent store reordering.
28789 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
28790 (type): Add block type.
28791 (stack_tie): Define special instruction to be used in
28794 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28796 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
28797 constraint. It is not valid for the pattern.
28798 (noncommutative_binary_comparison): Likewise.
28800 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28802 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
28805 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28807 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
28808 be used by the reg-alloc.
28810 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28812 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
28813 reg-alloc when having mul64 or mul32x16 instructions.
28814 (mulsidi3): Likewise.
28815 (umulsidi3): Likewise.
28816 (mulsi32x16): New pattern.
28817 (mulsi64): Likewise.
28818 (mulsidi64): Likewise.
28819 (umulsidi64): Likewise.
28820 (MUL32x16_REG): Define.
28821 (mul64_600): Use MUL32x16_REG.
28822 (mac64_600): Likewise.
28823 (umul64_600): Likewise.
28824 (umac64_600): Likewise.
28826 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28828 * config/arc/arc.md (mulsi3_700): Make it commutative.
28830 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
28832 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
28834 (*sign_extendsidi2_insn): Likewise for movstosw.
28836 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28838 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
28839 the type of the input discriminant value. Convert the
28840 discriminant value of signedness vary.
28842 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
28844 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
28845 Add to -Wall section.
28847 2017-06-01 Richard Biener <rguenther@suse.de>
28849 PR middle-end/66313
28850 * fold-const.c (fold_plusminus_mult_expr): If the factored
28851 factor may be zero use a wrapping type for the inner operation.
28852 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
28853 and handle moved defs.
28854 (process_assignment): Properly guard the unary op case. Return a
28855 tri-state indicating that moving the stmt before the call may allow
28856 to continue. Pass through to_move.
28857 (find_tail_calls): Handle moving unrelated defs before
28860 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
28863 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
28864 splitter result in the canonical way.
28866 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
28868 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
28869 also for 32bit target. Update insn attributes.
28870 (zero-extendsidi2 splitter): Allow all registers for operand 1.
28872 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
28874 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
28875 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
28876 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
28877 (_mm_maskz_min_ss): New intrinsics.
28879 2017-05-31 Martin Liska <mliska@suse.cz>
28881 * tree-vect-loop.c (vect_create_epilog_for_reduction):
28882 Change comment style to one we normally use.
28883 (vectorizable_reduction): Likewise.
28884 (vectorizable_induction): Likewise.
28885 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
28886 (vectorizable_call): Likewise.
28887 (vectorizable_simd_clone_call): Likewise.
28888 (vectorizable_conversion): Likewise.
28889 (vectorizable_assignment): Likewise.
28890 (vectorizable_shift): Likewise.
28891 (vectorizable_operation): Likewise.
28892 (vectorizable_store): Likewise.
28893 (vectorizable_load): Likewise.
28894 * tree-vectorizer.h: Likewise.
28896 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
28898 * passes.c (emergency_dump_function): New.
28899 * tree-pass.h (emergency_dump_function): Declare.
28900 * plugin.c (plugins_internal_error_function): Remove.
28901 * plugin.h (plugins_internal_error_function): Remove declaration.
28902 * toplev.c (internal_error_function): New static function. Use it...
28903 (general_init): ...here.
28905 2017-05-31 Graham Markall <graham.markall@embecosm.com>
28907 * config/arc/arc.c (arc_print_operand): Handle constant operands.
28908 (arc_rtx_costs): Add costs for new patterns.
28909 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
28910 * config/arc/predicates.md: Add _1_2_3_operand predicate.
28912 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28914 * tree-ssa-strlen.c (get_next_strinfo): New function.
28915 (get_stridx_plus_constant): Use it.
28916 (zero_length_string): Likewise.
28917 (adjust_related_strinfos): Likewise.
28918 (adjust_last_stmt): Likewise.
28920 2017-05-31 Richard Biener <rguenther@suse.de>
28923 * config/i386/i386.c (ix86_expand_builtin): Remove assert
28924 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
28926 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28928 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
28929 loop_vinfo argument and use of dependence distance vectors.
28930 Check instead whether the two references differ only in their
28931 initial value and assume that they have the same alignment if the
28932 difference is a multiple of the vector alignment.
28933 (vect_analyze_data_refs_alignment): Update call accordingly.
28935 2017-05-31 Martin Liska <mliska@suse.cz>
28938 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
28940 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28942 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
28943 (create_intersect_range_checks): Move from ...
28944 * tree-data-ref.c (create_intersect_range_checks_index)
28945 (create_intersect_range_checks): ... to here.
28946 (create_runtime_alias_checks): New function factored from ...
28947 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
28948 here. Call above function.
28949 * tree-data-ref.h (create_runtime_alias_checks): New function.
28951 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28953 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
28954 segment length for dr_b and compute it in wide_int.
28956 2017-05-31 Richard Biener <rguenther@suse.de>
28958 PR tree-optimization/80906
28959 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
28960 and pass through iv_map.
28961 (copy_bb_and_scalar_dependences): Adjust.
28962 (translate_pending_phi_nodes): Likewise.
28963 (copy_loop_close_phi_args): Handle code-generating IVs instead
28966 2017-05-30 David Malcolm <dmalcolm@redhat.com>
28968 * diagnostic-color.c (color_dict): Add "type-diff".
28969 (parse_gcc_colors): Update comment.
28970 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
28971 -fdiagnostics-show-template-tree and -fno-elide-type.
28972 (GCC_COLORS): Add type-diff to example.
28974 (-fdiagnostics-show-template-tree): New.
28975 (-fno-elide-type): New.
28976 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
28977 the pp_format_decoder callback. Call any m_format_postprocessor's
28979 (pretty_printer::pretty_printer): Initialize
28980 m_format_postprocessor.
28981 (pretty_printer::~pretty_printer): Delete any
28982 m_format_postprocessor.
28983 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
28984 (class format_postprocessor): New class.
28985 (struct pretty_printer::format_decoder): Document the new parameters.
28986 (struct pretty_printer::m_format_postprocessor): New field.
28987 * tree-diagnostic.c (default_tree_printer): Update for new
28988 bool and const char ** params.
28989 * tree-diagnostic.h (default_tree_printer): Likewise.
28991 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
28993 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
28994 (lwa_operand): Delete rs6000_gen_cell_microcode test.
28995 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28996 rs6000_gen_cell_microcode code.
28997 (rs6000_final_prescan_insn): Delete.
28998 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
28999 "warn-cell-microcode" entries.
29000 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
29001 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
29002 throughout. Change cc_reg_not_micro_cr0_operand to
29003 cc_reg_not_cr0_operand throughout.
29004 (*extendhi<mode>2_noload): Delete.
29005 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
29006 (mwarn-cell-microcode): Delete.
29007 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
29008 -mgen-cell-microcode and -mwarn-cell-microcode.
29010 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
29013 * config/i386/constraints.md (Yd): New constraint.
29015 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
29016 and (?Yd, r) alternatives. Update insn attributes.
29017 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
29018 and (?*Yd, r) alternatives. Update insn attributes.
29019 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
29021 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
29023 * gimplify.c (gimplify_modify_expr): Don't create a
29024 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
29027 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
29029 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
29031 2017-05-30 Richard Biener <rguenther@suse.de>
29033 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
29034 and reduc_def fields.
29035 (STMT_VINFO_REDUC_TYPE): New define.
29036 (STMT_VINFO_REDUC_DEF): Likewise.
29037 (vect_force_simple_reduction): Adjust prototype.
29038 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
29039 (vect_is_simple_reduction): Remove check_reduction argument.
29040 (vect_force_simple_reduction): Adjust and set
29041 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29042 (vectorizable_reduction): Do not re-do reduction analysis
29043 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29044 * tree-parloops.c (gather_scalar_reductions): Adjust.
29046 2017-05-30 Richard Biener <rguenther@suse.de>
29048 PR middle-end/80901
29049 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
29052 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
29054 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
29055 Introduce unknown_misalignment parameter and remove vf.
29056 (vect_peeling_hash_get_lowest_cost):
29057 Pass unknown_misalignment parameter.
29058 (vect_enhance_data_refs_alignment):
29059 Fix unsupportable data ref treatment.
29061 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29063 * tree-vect-data-refs.c (vect_get_data_access_cost):
29064 Workaround for SLP handling.
29065 (vect_enhance_data_refs_alignment):
29066 Compute costs for doing no peeling at all, compare to the best
29067 peeling costs so far and avoid peeling if cheaper.
29069 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29071 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
29072 Return peeling info and set costs to zero for unlimited cost
29074 (vect_enhance_data_refs_alignment): Also inspect all datarefs
29075 with unknown misalignment. Compute and costs for unknown
29076 misalignment, compare them to the costs for known misalignment
29077 and choose the cheapest for peeling.
29079 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29081 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
29082 (vect_get_peeling_costs_all_drs): Create function.
29083 (vect_peeling_hash_get_lowest_cost):
29084 Use vect_get_peeling_costs_all_drs.
29085 (vect_peeling_supportable): Create function.
29086 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
29088 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29090 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
29091 DR_HAS_NEGATIVE_STEP.
29092 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
29093 (vect_enhance_data_refs_alignment): Use.
29094 (vect_duplicate_ssa_name_ptr_info): Use.
29095 * tree-vectorizer.h (dr_misalignment): Use.
29096 (known_alignment_for_access_p): Use.
29098 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
29101 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
29103 (has_section_name): New function.
29105 2017-05-30 Martin Liska <mliska@suse.cz>
29108 * auto-profile.c (get_function_decl_from_block): Fix
29111 2017-05-30 Richard Biener <rguenther@suse.de>
29113 PR middle-end/80876
29114 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
29116 2017-05-30 Martin Liska <mliska@suse.cz>
29118 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
29119 * dumpfile.h (struct dump_file_info): Remove ctors.
29121 2017-05-30 Martin Liska <mliska@suse.cz>
29123 * predict.def: Fix GNU coding style.
29125 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29127 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29128 Mark 'to' argument with ATTRIBUTE_UNUSED.
29130 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29132 * config/xtensa/xtensa.c (xtensa_emit_call): Use
29133 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
29134 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
29137 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
29139 * doc/install.texi (Options specification): Restore entry of
29140 --enable-sjlj-exceptions.
29142 2017-05-27 Michael Eager <eager@eagercon.com>
29145 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
29147 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
29149 * config/microblaze/microblaze.h
29150 (FIXED_REGISTERS): Update in macro.
29151 (CALL_USED_REGISTERS): Update in macro.
29153 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
29155 * doc/install.texi: Add links to macOS binary distributions.
29157 2017-05-27 Jakub Jelinek <jakub@redhat.com>
29161 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29163 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29165 2017-05-26 Martin Liska <mliska@suse.cz>
29167 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
29169 2017-05-26 Martin Liska <mliska@suse.cz>
29171 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
29172 always leading ';; '.
29173 (dump_bb_info): Likewise.
29174 (brief_dump_cfg): Likewise.
29175 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
29176 * dumpfile.c: Remove usage of TDF_VERBOSE.
29177 * dumpfile.h (enum dump_kind): Likewise.
29178 (dump_gimple_bb_header): Do not use TDF_COMMENT.
29179 * print-tree.c (debug_verbose): Remove.
29180 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
29181 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
29182 * tree-diagnostic.c (default_tree_printer): Replace
29183 TDF_DIAGNOSTIC with TDF_SLIM.
29185 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29187 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
29188 in parameter loop, rather than loop_vinfo.
29189 (create_intersect_range_checks): Ditto.
29190 (vect_create_cond_for_alias_checks): Update call to above functions.
29192 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29194 PR tree-optimization/80815
29195 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
29196 for merging runtime alias checks. Handle negative DR_STEPs.
29198 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29200 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
29202 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
29203 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
29204 out code pruning runtime alias checks.
29205 * tree-data-ref.c (prune_runtime_alias_test_list): New function
29206 factored out from above.
29207 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29209 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29211 (prune_runtime_alias_test_list): New decalaration.
29213 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29215 * tree-vect-data-refs.c (compare_tree): Rename and move ...
29216 * tree-data-ref.c (data_ref_compare_tree): ... to here.
29217 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
29218 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
29219 (operator==, comp_dr_with_seg_len_pair): Ditto.
29220 (vect_prune_runtime_alias_test_list): Ditto.
29222 2017-05-26 Martin Liska <mliska@suse.cz>
29225 * params.def: Bound partial-inlining-entry-probability param.
29227 2017-05-26 Marek Polacek <polacek@redhat.com>
29230 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
29233 2017-05-26 Richard Biener <rguenther@suse.de>
29235 PR tree-optimization/80842
29236 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
29239 2017-05-26 Richard Biener <rguenther@suse.de>
29241 PR tree-optimization/80844
29242 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
29244 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
29246 * doc/md.texi (Machine Constraints): Update x86 family
29247 machine constraints section to match 'config/i386/constraints.md'.
29249 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
29251 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
29253 2017-05-25 Nathan Sidwell <nathan@acm.org>
29255 * doc/invoke.texi (--enable-languages): Update documentation.
29257 2017-05-25 Martin Liska <mliska@suse.cz>
29259 * dumpfile.c: Add TDF_FOLDING.
29260 * dumpfile.h (enum dump_kind): Likewise.
29261 * genmatch.c (dt_simplify::gen_1): Use it.
29263 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29265 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
29267 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29269 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29270 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
29272 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29274 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
29275 * match.pd (X == C): Rewrite it here.
29276 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
29277 with_certain_nonzero_bits2): New predicates.
29278 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
29280 2017-05-24 Nathan Sidwell <nathan@acm.org>
29282 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
29285 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
29288 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29290 * config/powerpcspe: New port. Files are copied from the rs6000
29291 port, with "rs6000" in filenames replaced by "powerpcspe".
29293 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
29295 PR rtl-optimization/80754
29296 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
29298 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
29300 * config/sparc/sparc.md (length): Return the correct value for -mflat
29301 sibcalls to match output_sibcall.
29303 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29307 * config/rs6000/rs6000.c (struct machine_function): Add new field
29309 (rs6000_get_separate_components): Init that field, use it.
29310 (rs6000_components_for_bb): Use the field.
29312 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
29314 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
29316 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
29318 PR middle-end/80823
29319 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
29321 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29324 * config/s390/s390.c (s390_check_qrst_address): Check incoming
29325 address against address_operand predicate.
29326 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
29328 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
29330 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
29331 parameters passed indirectly.
29333 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
29335 * config/i386/i386.md (*movdi_internal): Remove SSE4
29336 alternative 18 (?r, *v). Update insn attributes.
29337 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
29338 Update insn attributes.
29339 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
29340 Update insn attributes.
29341 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
29342 alternative 1 (r, v). Remove isa attribute.
29343 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
29344 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
29345 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
29347 2017-05-23 Tom de Vries <tom@codesourcery.com>
29349 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
29352 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29354 * cgraphunit.c (symbol_table::process_new_functions): Update.
29355 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
29356 (inline_generate_summary): Rename to ...
29357 (ipa_fn_summary_generate): ... this one.
29358 (inline_read_summary): Rename to ...
29359 (ipa_fn_summary_read): ... this one.
29360 (inline_write_summary): Rename to ...
29361 (ipa_fn_summary_write): ... this one.
29362 (inline_free_summary): Rename to ...
29363 (ipa_free_fn_summary): ... this one.
29364 (pass_data_local_fn_summary, pass_local_fn_summary,
29365 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
29366 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
29367 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
29368 make_pass_ipa_fn_summary): New.
29369 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
29370 inline_write_summary, inline_free_summary): Remove.
29371 (ipa_free_fn_summary) : New.
29372 * ipa-inline.c (ipa_inline): Update.
29373 (pass_ipa_inline): Do not generate summaries.
29374 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
29376 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
29377 and add pass_ipa_fn_summary.
29378 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
29380 (make_pass_inline_parameters): Remove.
29382 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
29384 * omp-low.c (struct omp_context): Remove "default_kind" member.
29387 * omp-offload.c (execute_oacc_device_lower): Remove the
29388 parallelism dimensions function attributes for unparallelized
29389 OpenACC kernels constructs.
29391 2017-05-23 Martin Liska <mliska@suse.cz>
29393 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
29395 (cgraph_edge::make_speculative): Likewise.
29396 (cgraph_edge::resolve_speculation): Likewise.
29397 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29398 (cgraph_node::dump): Likewise.
29399 * cgraph.h: Likewise.
29400 * cgraphunit.c (analyze_functions): Likewise.
29401 (symbol_table::compile): Likewise.
29402 * ipa-cp.c (print_all_lattices): Likewise.
29403 (determine_versionability): Likewise.
29404 (initialize_node_lattices): Likewise.
29405 (ipcp_verify_propagated_values): Likewise.
29406 (estimate_local_effects): Likewise.
29407 (update_profiling_info): Likewise.
29408 (create_specialized_node): Likewise.
29409 (perhaps_add_new_callers): Likewise.
29410 (decide_about_value): Likewise.
29411 (decide_whether_version_node): Likewise.
29412 (identify_dead_nodes): Likewise.
29413 (ipcp_store_bits_results): Likewise.
29414 * ipa-devirt.c (dump_targets): Likewise.
29415 (ipa_devirt): Likewise.
29416 * ipa-icf.c (sem_item::dump): Likewise.
29417 (sem_function::equals): Likewise.
29418 (sem_variable::equals): Likewise.
29419 (sem_item_optimizer::read_section): Likewise.
29420 (sem_item_optimizer::execute): Likewise.
29421 (congruence_class::dump): Likewise.
29422 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
29423 (dump_inline_summary): Likewise.
29424 (estimate_node_size_and_time): Likewise.
29425 (inline_analyze_function): Likewise.
29426 * ipa-inline-transform.c (inline_call): Likewise.
29427 * ipa-inline.c (report_inline_failed_reason): Likewise.
29428 (want_early_inline_function_p): Likewise.
29429 (edge_badness): Likewise.
29430 (update_edge_key): Likewise.
29431 (inline_small_functions): Likewise.
29432 * ipa-profile.c (ipa_profile): Likewise.
29433 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
29434 (ipa_make_edge_direct_to_target): Likewise.
29435 (remove_described_reference): Likewise.
29436 (ipa_impossible_devirt_target): Likewise.
29437 (propagate_controlled_uses): Likewise.
29438 (ipa_print_node_params): Likewise.
29439 (ipcp_transform_function): Likewise.
29440 * ipa-pure-const.c (pure_const_read_summary): Likewise.
29441 (propagate_pure_const): Likewise.
29442 * ipa-reference.c (generate_summary): Likewise.
29443 (read_write_all_from_decl): Likewise.
29444 (propagate): Likewise.
29445 (ipa_reference_read_optimization_summary): Likewise.
29446 * ipa-utils.c (ipa_merge_profiles): Likewise.
29447 * ipa.c (walk_polymorphic_call_targets): Likewise.
29448 (symbol_table::remove_unreachable_nodes): Likewise.
29449 (ipa_single_use): Likewise.
29450 * passes.c (execute_todo): Likewise.
29451 * predict.c (drop_profile): Likewise.
29452 * symtab.c (symtab_node::get_dump_name): New function.
29453 (symtab_node::dump_name): Likewise.
29454 (symtab_node::dump_asm_name): Likewise.
29455 (symtab_node::dump_references): Likewise.
29456 (symtab_node::dump_referring): Likewise.
29457 (symtab_node::dump_base): Likewise.
29458 (symtab_node::debug_symtab): Likewise.
29459 * tree-sra.c (convert_callers_for_node): Likewise.
29460 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
29461 * value-prof.c (init_node_map): Likewise.
29463 2017-05-23 Martin Liska <mliska@suse.cz>
29465 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
29466 and symtab_node::debug_symtab to symbol_table::debug.
29467 * cgraphunit.c (analyze_functions): Use the renamed function.
29468 (symbol_table::compile): Likewise.
29469 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
29470 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
29471 * passes.c (execute_todo): Likewise.
29472 * symtab.c (symbol_table::dump): New function.
29473 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
29475 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29477 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
29478 that nonconst implies exec.
29480 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29482 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29483 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29484 (inline_edge_summary_vec): Turn into ...
29485 (ipa_call_summaries): ... this one.
29486 (redirect_to_unreachable, edge_set_predicate,
29487 evaluate_properties_for_edge, inline_summary_alloc,
29488 reset_ipa_call_summary, reset_inline_summary,
29489 inline_summary_t::duplicate): Update.
29490 (inline_edge_duplication_hook): Turn to ...
29491 (ipa_call_summary_t::duplicate): ... this one.
29492 (inline_edge_removal_hook): Turn to ...
29493 (ipa_call_summary_t::remove): ... this one.
29494 (dump_inline_edge_summary): Turn to ...
29495 (dump_ipa_call_summary): ... this one.
29496 (estimate_function_body_sizes): Update.
29497 (inline_update_callee_summaries): Update.
29498 (remap_edge_change_prob): Update.
29499 (remap_edge_summaries): Update.
29500 (inline_merge_summary): Update.
29501 (do_estimate_edge_time): Update.
29502 (inline_generate_summary): Update.
29503 (inline_read_section): Update.
29504 (inline_read_summary): Update.
29505 (inline_free_summary): Update.
29506 * ipa-inline.c (can_inline_edge_p): Update.
29507 (compute_inlined_call_time): Update.
29508 (want_inline_small_function_p): Update.
29509 (edge_badness): Update.
29510 (early_inliner): Update.
29511 * ipa-inline.h (inline_edge_summary): Turn to ...
29512 (ipa_call_summary): ... this one.
29513 (ipa_call_summary_t): New class.
29514 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29515 (ipa_call_summaries): New.
29516 (inline_edge_summary): Remove.
29517 (estimate_edge_growth): Update.
29518 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29519 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29520 * ipa-split.c (execute_split_functions): Update.
29521 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29523 2017-05-23 Tom de Vries <tom@codesourcery.com>
29525 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29526 attributes): Document rdrand effective target.
29528 2017-05-23 Tom de Vries <tom@codesourcery.com>
29530 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29531 attributes): Sort alphabetically.
29533 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
29535 * config/avr/genmultilib.awk: Use gsub instead of gensub.
29537 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29540 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
29541 V2DF/V2DI splat into two separate patterns, one that handles
29542 registers, and the other that only handles memory. Drop support
29543 for splatting from a GPR on ISA 2.07 and then splitting the
29544 splat into direct move and splat.
29545 (vsx_splat_<mode>_reg): Likewise.
29546 (vsx_splat_<mode>_mem): Likewise.
29548 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
29550 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
29552 2017-05-22 Jakub Jelinek <jakub@redhat.com>
29554 PR middle-end/80809
29555 * omp-low.c (finish_taskreg_remap): New function.
29556 (finish_taskreg_scan): If unit size of ctx->record_type
29557 is non-constant, unshare the size expression and replace
29558 decls in it with possible outer var refs.
29560 PR middle-end/80809
29561 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
29562 GOVD_SHARED rather than GOVD_PRIVATE with it.
29563 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
29564 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
29566 PR middle-end/80853
29567 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
29568 as last argument to build_outer_var_ref for pointer bases of array
29569 section reductions.
29571 2017-05-19 Martin Sebor <msebor@redhat.com>
29573 * print-tree.c (print_node): Print DECL_READ_P flag.
29575 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29577 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
29578 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
29579 * cgraph.c: Likewise.
29580 * cgraphunit.c: Likewise.
29581 * gengtype.c: Likewise.
29582 * ipa-cp.c: Likewise.
29583 * ipa-devirt.c: Likewise.
29584 * ipa-icf.c: Likewise.
29585 * ipa-predicate.c: Likewise.
29586 * ipa-profile.c: Likewise.
29587 * ipa-prop.c: Likewise.
29588 * ipa-split.c: Likewise.
29590 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
29591 edge_predicate_pool, dump_inline_hints,
29592 inline_summary::account_size_time, redirect_to_unreachable,
29593 edge_set_predicate, set_hint_predicate,
29594 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
29595 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
29596 inline_summary_t::remove, remap_hint_predicate_after_duplication,
29597 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
29598 ipa_call_summary_t::remove, initialize_growth_caches,
29599 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
29600 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
29601 mark_modified, unmodified_parm_1, unmodified_parm,
29602 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
29603 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
29604 compute_bb_predicates, will_be_nonconstant_expr_predicate,
29605 will_be_nonconstant_predicate, record_modified_bb_info,
29606 get_minimal_bb, record_modified, param_change_prob,
29607 phi_result_unknown_predicate, predicate_for_phi_result,
29608 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
29609 estimate_function_body_sizes, compute_inline_parameters,
29610 compute_inline_parameters_for_curren, pass_data_inline_parameters,
29611 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
29612 inline_update_callee_summaries, remap_edge_change_prob,
29613 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
29614 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
29615 inline_analyze_function, inline_summary_t::insert,
29616 inline_generate_summary, read_ipa_call_summary, inline_read_section,
29617 inline_read_summary, write_ipa_call_summary, inline_write_summary,
29618 inline_free_summary): Move to ipa-fnsummary.h
29619 (predicate_t): Remove.
29620 * ipa-fnsummary.c: New file.
29621 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
29622 (enum inline_hints_vals, inline_hints, agg_position_info,
29623 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
29624 inline_summaries, ipa_call_summary, ipa_call_summary_t,
29625 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
29626 dump_inline_summary, dump_inline_hints, inline_generate_summary,
29627 inline_read_summary, inline_write_summary, inline_free_summary,
29628 inline_analyze_function, initialize_inline_failed,
29629 inline_merge_summary, inline_update_overall_summary,
29630 compute_inline_parameters): Move to ipa-fnsummary.h
29631 * ipa-fnsummary.h: New file.
29632 * ipa-inline-transform.h: Include ipa-inline.h.
29633 * ipa-inline.c: LIkewise.
29635 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29637 * ipa-inline.c (edge_badness): Use inlined_time instead of
29638 inline_summaries->get.
29640 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29642 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
29644 2017-05-22 Nathan Sidwell <nathan@acm.org>
29646 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
29647 (fdump-lang): Document 'raw' option.
29648 * dumpfile.h (TDI_tu): Delete.
29649 * dumpfile.c (dump_files): Remove translation-unit.
29650 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
29652 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
29654 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
29655 command option from $(AWK) call.
29656 * config/avr/genmultilib.awk: Simplify and rewrite so that it
29657 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
29658 [FORMAT]: Remove handling of variable.
29659 * config/avr/t-multilib: Regenerate.
29661 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29663 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
29665 (dump_inline_summary): Do not print self_time.
29666 (estimate_function_body_sizes): Do not set self_time.
29667 (compute_inline_parameters): Likewise.
29668 (inline_read_section, inline_write_summary): Do not stream self_time.
29669 * ipa-inline.h (inline_summary): Drop self_time.
29671 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29673 * ipa-inline-analysis.c (account_size_time): Rename to ...
29674 (inline_summary::account_size_time): ... this one.
29675 (reset_ipa_call_summary): Turn to ...
29676 (ipa_call_summary::reset): ... this one.
29677 (reset_inline_summary): Turn to ...
29678 (inline_summary::reset): ... this one.
29679 (inline_summary_t::remove): Update.
29680 (inline_summary_t::duplicate): Update.
29681 (ipa_call_summary_t::remove): Update.
29682 (dump_inline_summary): Update.
29683 (estimate_function_body_sizes): Update.
29684 (compute_inline_parameters): Update.
29685 (estimate_node_size_and_time): Update.
29686 (inline_merge_summary): Update.
29687 (inline_update_overall_summary): Update.
29688 (inline_read_section): Update.
29689 (inline_write_summary): Update.
29690 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
29691 add account_size_time and reset member functions.
29692 (ipa_call_summary): Add reset function.
29693 * ipa-predicate.h (predicate::operator &): Constify.
29695 2017-05-22 Richard Biener <rguenther@suse.de>
29697 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
29699 2017-05-19 Jason Merrill <jason@redhat.com>
29701 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
29703 2017-05-19 Marek Polacek <polacek@redhat.com>
29706 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
29707 TYPE_OVERFLOW_WRAPS checks.
29709 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
29711 * tree-core.h (enum omp_clause_default_kind): Add
29712 "OMP_CLAUSE_DEFAULT_PRESENT".
29713 * tree-pretty-print.c (dump_omp_clause): Handle it.
29714 * gimplify.c (enum gimplify_omp_var_data): Add
29715 "GOVD_MAP_FORCE_PRESENT".
29716 (gimplify_adjust_omp_clauses_1): Map it to
29717 "GOMP_MAP_FORCE_PRESENT".
29718 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
29720 * gimplify.c (oacc_default_clause): Clarify.
29722 2017-05-19 Nathan Sidwell <nathan@acm.org>
29724 LANG_HOOK_REGISTER_DUMPS
29725 * toplev.c (general_init): Call register dump lang hook.
29726 * doc/invoke.texi: Document -fdump-lang option family.
29727 * dumpfile.c (dump_files): Remove class dump here.
29728 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
29729 * dumpfile.h (tree_dump_index): Remove TDI_class.
29730 * langhooks-def.h (lhd_register_dumps): Declare.
29731 (LANG_HOOKS_REGISTER_DUMPS): Define.
29732 (LANG_HOOKS_INITIALIZER): Add it.
29733 * langhooks.c (lhd_register_dumps): Define.
29734 * langhooks.h (struct lang_hooks): Add register_dumps.
29736 2017-05-19 Nathan Sidwell <nathan@acm.org>
29738 * context.h (context::set_passes): New.
29739 * context.c (context::context): Do not create pass manager.
29740 * toplev.c (general_init): Create pass manager here.
29742 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
29744 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
29745 use this splitter if two add or or instructions would also work for
29746 the constant we want to generate.
29748 2017-05-19 Richard Biener <rguenther@suse.de>
29751 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
29752 predicate evaluation.
29754 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29756 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
29758 * ipa-inline.c (want_inline_small_function_p): Do not cast to
29761 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29763 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29764 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29765 (inline_edge_summary_vec): Turn into ...
29766 (ipa_call_summaries): ... this one.
29767 (redirect_to_unreachable, edge_set_predicate,
29768 evaluate_properties_for_edge, inline_summary_alloc,
29769 reset_ipa_call_summary, reset_inline_summary,
29770 inline_summary_t::duplicate): Update.
29771 (inline_edge_duplication_hook): Turn to ...
29772 (ipa_call_summary_t::duplicate): ... this one.
29773 (inline_edge_removal_hook): Turn to ...
29774 (ipa_call_summary_t::remove): ... this one.
29775 (dump_inline_edge_summary): Turn to ...
29776 (dump_ipa_call_summary): ... this one.
29777 (estimate_function_body_sizes): Update.
29778 (inline_update_callee_summaries): Update.
29779 (remap_edge_change_prob): Update.
29780 (remap_edge_summaries): Update.
29781 (inline_merge_summary): Update.
29782 (do_estimate_edge_time): Update.
29783 (inline_generate_summary): Update.
29784 (inline_read_section): Update.
29785 (inline_read_summary): Update.
29786 (inline_free_summary): Update.
29787 * ipa-inline.c (can_inline_edge_p): Update.
29788 (compute_inlined_call_time): Update.
29789 (want_inline_small_function_p): Update.
29790 (edge_badness): Update.
29791 (early_inliner): Update.
29792 * ipa-inline.h (inline_edge_summary): Turn to ...
29793 (ipa_call_summary): ... this one.
29794 (ipa_call_summary_t): New class.
29795 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29796 (ipa_call_summaries): New.
29797 (inline_edge_summary): Remove.
29798 (estimate_edge_growth): Update.
29799 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29800 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29801 * ipa-split.c (execute_split_functions): Update.
29802 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29804 2017-05-19 Richard Biener <rguenther@suse.de>
29806 PR middle-end/80764
29807 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
29809 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
29811 * config/rs6000/rs6000.c (struct machine_function): Add field
29812 fpr_is_wrapped_separately.
29813 (rs6000_get_separate_components): Use 64 components. Handle the
29814 new FPR components.
29815 (rs6000_components_for_bb): Handle the FPR components.
29816 (rs6000_emit_prologue_components): Handle the FPR components.
29817 (rs6000_emit_epilogue_components): Handle the FPR components.
29818 (rs6000_set_handled_components): Handle the FPR components.
29819 (rs6000_emit_prologue): Don't output prologue code for those FPRs
29820 that are already separately shrink-wrapped.
29821 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
29822 that are already separately shrink-wrapped.
29824 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29827 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
29830 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
29831 (define_peephole2 for Altivec d-form load): Add peepholes to catch
29832 cases where the register allocator uses a move and an offsettable
29833 memory operation to/from a FPR register on ISA 2.06/2.07.
29834 (define_peephole2 for Altivec d-form store): Likewise.
29836 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
29839 * config/i386/mmx.md (*mov<mode>_internal): Enable
29840 alternatives 11, 12, 13 and 14 also for 32bit targets.
29841 Remove alternatives 15, 16, 17 and 18.
29842 * config/i386/sse.md (vec_concatv2di): Change
29843 alternative (!x, *y) to (x, ?!*Yn).
29845 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
29847 * dumpfile.h (enum dump_kind): Remove stray comma.
29849 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29851 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
29852 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
29853 predicate::num_conditions
29854 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
29855 (CHANGED): turn into predicate::changed.
29856 (agg_position_info): Move to ipa-predicate.h
29857 (add_condition, predicate::add_clause, predicate::operator &=,
29858 predicate::or_with, predicate::evaluate, predicate::probability,
29859 dump_condition, dump_clause, predicate::dump,
29860 predicate::remap_after_duplication, predicate::remap_after_inlining,
29861 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
29862 (evaluate_conditions_for_known_args): Update.
29863 (set_cond_stmt_execution_predicate): Update.
29864 * ipa-inline.h: Include ipa-predicate.h
29865 (condition, inline_param_summary, conditions, agg_position_info,
29866 predicate): Move to ipa-predicate.h
29867 * ipa-predicate.c: New file.
29868 * ipa-predicate.h: New file.
29870 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
29872 * final.c (leaf_function_p): Check we are not in a sequence.
29874 2017-05-18 Martin Liska <mliska@suse.cz>
29876 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
29877 * dumpfile.c (dump_register): Use new enum dump_kind.
29878 (get_dump_file_name): Likewise.
29879 (dump_enable_all): Likewise.
29880 (dump_switch_p_1): Likewise.
29881 (enable_rtl_dump_file): Remove usage of TDF_RTL.
29882 * dumpfile.h (enum dump_kind): New enum type.
29883 (struct dump_file_info): Create constructor and
29884 format fields and comments.
29885 * passes.c (pass_manager::register_one_dump_file):
29887 * statistics.c (statistics_early_init): Likewise.
29888 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
29889 TDF_TREE with TDF_SLIM.
29890 (gather_memory_references_ref): Likewise.
29892 2017-05-18 Martin Liska <mliska@suse.cz>
29894 * vec.h (struct vnull): Use it.
29896 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29898 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
29899 (true_predicate, false_predicate, true_predicate_p,
29900 false_predicate_p): Remove.
29901 (single_cond_predicate, not_inlined_predicate): Turn to member function
29903 (add_condition): Update.
29904 (add_clause): Turn to...
29905 (predicate::add_clause): ... this one; update; allow passing NULL
29907 (and_predicates): Turn to ...
29908 (predicate::operator &=): ... this one.
29909 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
29910 (or_predicates): Turn to ...
29911 (predicate::or_with): ... this one.
29912 (evaluate_predicate): Turn to ...
29913 (predicate::evaluate): ... this one.
29914 (predicate_probability): Turn to ...
29915 (predicate::probability): ... this one.
29916 (dump_condition): Update.
29917 (dump_predicate): Turn to ...
29918 (predicate::dump): ... this one.
29919 (account_size_time): Update.
29920 (edge_set_predicate): Update.
29921 (set_hint_predicate): UPdate.
29922 (evaluate_conditions_for_known_args): Update.
29923 (evaluate_properties_for_edge): Update.
29924 (remap_predicate_after_duplication): Turn to...
29925 (predicate::remap_after_duplication): ... this one.
29926 (remap_hint_predicate_after_duplication): Update.
29927 (inline_summary_t::duplicate): UPdate.
29928 (dump_inline_edge_summary): Update.
29929 (dump_inline_summary): Update.
29930 (set_cond_stmt_execution_predicate): Update.
29931 (set_switch_stmt_execution_predicate): Update.
29932 (compute_bb_predicates): Update.
29933 (will_be_nonconstant_expr_predicate): Update.
29934 (will_be_nonconstant_predicate): Update.
29935 (phi_result_unknown_predicate): Update.
29936 (predicate_for_phi_result): Update.
29937 (array_index_predicate): Update.
29938 (estimate_function_body_sizes): Update.
29939 (estimate_node_size_and_time): Update.
29940 (estimate_ipcp_clone_size_and_time): Update.
29941 (remap_predicate): Rename to ...
29942 (predicate::remap_after_inlining): ... this one.
29943 (remap_hint_predicate): Update.
29944 (inline_merge_summary): Update.
29945 (inline_update_overall_summary): Update.
29946 (estimate_size_after_inlining): Update.
29947 (read_predicate): Rename to ...
29948 (predicate::stream_in): ... this one.
29949 (read_inline_edge_summary): Update.
29950 (write_predicate): Rename to ...
29951 (predicate::stream_out): ... this one.
29952 (write_inline_edge_summary): Update.
29953 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
29954 (clause_t): Turn to uint32_t
29955 (predicate): Turn to class; implement constructor and operators
29957 (size_time_entry): Update.
29958 (inline_summary): Update.
29959 (inline_edge_summary): Update.
29961 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
29963 * fold-const.c (fold_binary_loc): Move transformation...
29964 * match.pd (C - X CMP X): ... here.
29966 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
29968 * config/sparc/sparc.c (sparc_option_override): Set function
29969 alignment for -mcpu=niagara7 to 64 to match the I$ line.
29970 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
29972 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
29974 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
29976 2017-05-18 Marek Polacek <polacek@redhat.com>
29979 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
29980 (pass_ubsan::execute): Call gimple_assign_single_p instead of
29981 gimple_assign_load_p.
29983 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
29985 PR middle-end/80692
29986 * real.c (do_compare): Give decimal_do_compare preference over
29987 comparing just the signs.
29989 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
29991 * doc/md.texi (Canonicalization of Instructions): Describe the
29992 canonical form of instructions that inherently set a condition
29995 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
29997 PR middle-end/80775
29998 * tree-cfg.c: Move deletion of unreachable case statements to after
29999 the merging of consecutive case labels.
30001 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
30003 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
30004 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
30005 restoring of callee-saved registers.
30007 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
30009 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
30010 * config/visium/visium.c (single_set_and_flags): Likewise.
30011 * config/visium/visium.md (Substitutions): Likewise.
30013 2017-05-17 Martin Liska <mliska@suse.cz>
30015 * cfg.c: Introduce dump_flags_t type and
30016 use it instead of int type.
30018 * cfghooks.c: Likewise.
30019 * cfghooks.h (struct cfg_hooks): Likewise.
30020 * cfgrtl.c: Likewise.
30021 * cfgrtl.h: Likewise.
30022 * cgraph.c (cgraph_node::get_body): Likewise.
30023 * coretypes.h: Likewise.
30024 * domwalk.c: Likewise.
30025 * domwalk.h: Likewise.
30026 * dumpfile.c (struct dump_option_value_info): Likewise.
30027 (dump_enable_all): Likewise.
30028 (dump_switch_p_1): Likewise.
30029 (opt_info_switch_p): Likewise.
30030 * dumpfile.h (enum tree_dump_index): Likewise.
30031 (struct dump_file_info): Likewise.
30032 * genemit.c: Likewise.
30033 * generic-match-head.c: Likewise.
30034 * gengtype.c (open_base_files): Likewise.
30035 * gimple-pretty-print.c: Likewise.
30036 * gimple-pretty-print.h: Likewise.
30037 * graph.c (print_graph_cfg): Likewise.
30038 * graphite-scop-detection.c (dot_all_sese): Likewise.
30039 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
30040 * loop-unroll.c (report_unroll): Likewise.
30041 * passes.c (pass_manager::register_one_dump_file): Likewise.
30042 * print-tree.c: Likewise.
30043 * statistics.c: Likewise.
30044 * tree-cfg.c: Likewise.
30045 * tree-cfg.h: Likewise.
30046 * tree-dfa.c: Likewise.
30047 * tree-dfa.h: Likewise.
30048 * tree-dump.c (dump_function): Likewise.
30049 * tree-dump.h (struct dump_info): Likewise.
30050 * tree-pretty-print.c: Likewise.
30051 * tree-pretty-print.h: Likewise.
30052 * tree-ssa-live.c: Likewise.
30053 * tree-ssa-live.h: Likewise.
30054 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
30055 * tree-vect-loop.c: Likewise.
30056 * tree-vect-slp.c: Likewise.
30058 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
30059 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30061 PR tree-optimization/80457
30062 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
30063 of all arguments to a statement as scalar_to_vec operations.
30064 (vectorizable_call): Adjust call to vect_model_simple_cost for
30066 (vectorizable_conversion): Likewise.
30067 (vectorizable_assignment): Likewise.
30068 (vectorizable_shift): Likewise.
30069 (vectorizable_operation): Likewise.
30070 (vectorizable_comparison): Likewise.
30071 (vect_is_simple_cond): Record the def types for operands.
30072 (vectorizable_condition): Likewise, call vect_model_simple_cost.
30073 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
30074 for statement argument count.
30076 2017-05-16 Carl Love <cel@us.ibm.com>
30078 * config/rs6000/rs6000-c: Add support for built-in functions
30079 vector unsigned long long vec_bperm (vector unsigned long long,
30080 vector unsigned char)
30081 vector signed long long vec_mule (vector signed int,
30083 vector unsigned long long vec_mule (vector unsigned int,
30084 vector unsigned int)
30085 vector signed long long vec_mulo (vector signed int,
30087 vector unsigned long long vec_mulo (vector unsigned int,
30088 vector unsigned int)
30089 vector signed char vec_sldw (vector signed char,
30090 vector signed char,
30092 vector unsigned char vec_sldw (vector unsigned char,
30093 vector unsigned char,
30095 vector signed short vec_sldw (vector signed short,
30096 vector signed short,
30098 vector unsigned short vec_sldw (vector unsigned short,
30099 vector unsigned short,
30101 vector signed int vec_sldw (vector signed int,
30104 vector unsigned int vec_sldw (vector unsigned int,
30105 vector unsigned int,
30107 vector signed long long vec_sldw (vector signed long long,
30108 vector signed long long,
30110 vector unsigned long long vec_sldw (vector unsigned long long,
30111 vector unsigned long long,
30113 * config/rs6000/rs6000-c: Add support for built-in functions
30114 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
30115 * config/rs6000/altivec.h: Add defintion for vec_sldw.
30116 * doc/extend.texi: Update the built-in documentation for the
30117 new built-in functions.
30119 2017-05-16 Marek Polacek <polacek@redhat.com>
30123 * tree.c (save_expr): Don't fold the expression.
30125 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
30127 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
30128 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
30129 and (?*y,m). Update insn attributes.
30131 2017-05-16 Martin Liska <mliska@suse.cz>
30133 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
30134 flags argument of print_gimple_stmt, print_gimple_expr,
30135 print_generic_stmt and print_generic_expr.
30136 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
30137 * coretypes.h: Likewise.
30138 * except.c (dump_eh_tree): Likewise.
30139 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
30140 * gimple-pretty-print.h: Likewise.
30141 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
30142 (backprop::push_to_worklist): Likewise.
30143 (backprop::pop_from_worklist): Likewise.
30144 (backprop::process_use): Likewise.
30145 (backprop::intersect_uses): Likewise.
30146 (note_replacement): Likewise.
30147 * gimple-ssa-store-merging.c
30148 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
30149 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
30150 (pass_store_merging::execute): Likewise.
30151 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
30152 (ssa_base_cand_dump_callback): Likewise.
30153 (dump_incr_vec): Likewise.
30154 (replace_refs): Likewise.
30155 (replace_mult_candidate): Likewise.
30156 (create_add_on_incoming_edge): Likewise.
30157 (create_phi_basis): Likewise.
30158 (insert_initializers): Likewise.
30159 (all_phi_incrs_profitable): Likewise.
30160 (introduce_cast_before_cand): Likewise.
30161 (replace_one_candidate): Likewise.
30162 * gimplify.c (gimplify_expr): Likewise.
30163 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
30164 (set_rename): Likewise.
30165 (rename_uses): Likewise.
30166 (copy_loop_phi_nodes): Likewise.
30167 (add_close_phis_to_merge_points): Likewise.
30168 (copy_loop_close_phi_args): Likewise.
30169 (copy_cond_phi_args): Likewise.
30170 (graphite_copy_stmts_from_block): Likewise.
30171 (translate_pending_phi_nodes): Likewise.
30172 * graphite-poly.c (print_pdr): Likewise.
30173 (dump_gbb_cases): Likewise.
30174 (dump_gbb_conditions): Likewise.
30175 (print_scop_params): Likewise.
30176 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
30177 (build_cross_bb_scalars_use): Likewise.
30178 (gather_bbs::before_dom_children): Likewise.
30179 * hsa-dump.c (dump_hsa_immed): Likewise.
30180 * ipa-cp.c (print_ipcp_constant_value): Likewise.
30181 (get_replacement_map): Likewise.
30182 * ipa-inline-analysis.c (dump_condition): Likewise.
30183 (estimate_function_body_sizes): Likewise.
30184 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
30185 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
30186 * ipa-prop.c (ipa_dump_param): Likewise.
30187 (ipa_print_node_jump_functions_for_edge): Likewise.
30188 (ipa_modify_call_arguments): Likewise.
30189 (ipa_modify_expr): Likewise.
30190 (ipa_dump_param_adjustments): Likewise.
30191 (ipa_dump_agg_replacement_values): Likewise.
30192 (ipcp_modif_dom_walker::before_dom_children): Likewise.
30193 * ipa-pure-const.c (check_stmt): Likewise.
30194 (pass_nothrow::execute): Likewise.
30195 * ipa-split.c (execute_split_functions): Likewise.
30196 * omp-offload.c (dump_oacc_loop_part): Likewise.
30197 (dump_oacc_loop): Likewise.
30198 * trans-mem.c (tm_log_emit): Likewise.
30199 (tm_memopt_accumulate_memops): Likewise.
30200 (dump_tm_memopt_set): Likewise.
30201 (dump_tm_memopt_transform): Likewise.
30202 * tree-cfg.c (gimple_verify_flow_info): Likewise.
30203 (print_loop): Likewise.
30204 * tree-chkp-opt.c (chkp_print_addr): Likewise.
30205 (chkp_gather_checks_info): Likewise.
30206 (chkp_get_check_result): Likewise.
30207 (chkp_remove_check_if_pass): Likewise.
30208 (chkp_use_outer_bounds_if_possible): Likewise.
30209 (chkp_reduce_bounds_lifetime): Likewise.
30210 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
30211 (chkp_mark_completed_bounds): Likewise.
30212 (chkp_register_incomplete_bounds): Likewise.
30213 (chkp_mark_invalid_bounds): Likewise.
30214 (chkp_maybe_copy_and_register_bounds): Likewise.
30215 (chkp_build_returned_bound): Likewise.
30216 (chkp_get_bound_for_parm): Likewise.
30217 (chkp_build_bndldx): Likewise.
30218 (chkp_get_bounds_by_definition): Likewise.
30219 (chkp_generate_extern_var_bounds): Likewise.
30220 (chkp_get_bounds_for_decl_addr): Likewise.
30221 * tree-chrec.c (chrec_apply): Likewise.
30222 * tree-data-ref.c (dump_data_reference): Likewise.
30223 (dump_subscript): Likewise.
30224 (dump_data_dependence_relation): Likewise.
30225 (analyze_overlapping_iterations): Likewise.
30226 * tree-inline.c (expand_call_inline): Likewise.
30227 (tree_function_versioning): Likewise.
30228 * tree-into-ssa.c (dump_defs_stack): Likewise.
30229 (dump_currdefs): Likewise.
30230 (dump_names_replaced_by): Likewise.
30231 (dump_update_ssa): Likewise.
30232 (update_ssa): Likewise.
30233 * tree-object-size.c (pass_object_sizes::execute): Likewise.
30234 * tree-parloops.c (build_new_reduction): Likewise.
30235 (try_create_reduction_list): Likewise.
30236 (ref_conflicts_with_region): Likewise.
30237 (oacc_entry_exit_ok_1): Likewise.
30238 (oacc_entry_exit_single_gang): Likewise.
30239 * tree-pretty-print.h: Likewise.
30240 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
30241 (get_scalar_evolution): Likewise.
30242 (add_to_evolution): Likewise.
30243 (get_loop_exit_condition): Likewise.
30244 (analyze_evolution_in_loop): Likewise.
30245 (analyze_initial_condition): Likewise.
30246 (analyze_scalar_evolution): Likewise.
30247 (instantiate_scev): Likewise.
30248 (number_of_latch_executions): Likewise.
30249 (gather_chrec_stats): Likewise.
30250 (final_value_replacement_loop): Likewise.
30251 (scev_const_prop): Likewise.
30252 * tree-sra.c (dump_access): Likewise.
30253 (disqualify_candidate): Likewise.
30254 (create_access): Likewise.
30255 (reject): Likewise.
30256 (maybe_add_sra_candidate): Likewise.
30257 (create_access_replacement): Likewise.
30258 (analyze_access_subtree): Likewise.
30259 (analyze_all_variable_accesses): Likewise.
30260 (sra_modify_assign): Likewise.
30261 (initialize_constant_pool_replacements): Likewise.
30262 (find_param_candidates): Likewise.
30263 (decide_one_param_reduction): Likewise.
30264 (replace_removed_params_ssa_names): Likewise.
30265 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
30266 * tree-ssa-copy.c (dump_copy_of): Likewise.
30267 (copy_prop_visit_cond_stmt): Likewise.
30268 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
30269 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
30270 (record_equivalences_from_stmt): Likewise.
30271 * tree-ssa-dse.c (compute_trims): Likewise.
30272 (delete_dead_call): Likewise.
30273 (delete_dead_assignment): Likewise.
30274 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
30275 (forward_propagate_into_cond): Likewise.
30276 (pass_forwprop::execute): Likewise.
30277 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
30278 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
30280 (move_computations_worker): Likewise.
30281 (execute_sm): Likewise.
30282 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
30283 (remove_exits_and_undefined_stmts): Likewise.
30284 (remove_redundant_iv_tests): Likewise.
30285 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
30286 (adjust_iv_update_pos): Likewise.
30287 * tree-ssa-math-opts.c (bswap_replace): Likewise.
30288 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
30289 (value_replacement): Likewise.
30290 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
30291 * tree-ssa-pre.c (print_pre_expr): Likewise.
30292 (get_representative_for): Likewise.
30293 (create_expression_by_pieces): Likewise.
30294 (insert_into_preds_of_block): Likewise.
30295 (eliminate_insert): Likewise.
30296 (eliminate_dom_walker::before_dom_children): Likewise.
30297 (eliminate): Likewise.
30298 (remove_dead_inserted_code): Likewise.
30299 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
30300 * tree-ssa-reassoc.c (get_rank): Likewise.
30301 (eliminate_duplicate_pair): Likewise.
30302 (eliminate_plus_minus_pair): Likewise.
30303 (eliminate_not_pairs): Likewise.
30304 (undistribute_ops_list): Likewise.
30305 (eliminate_redundant_comparison): Likewise.
30306 (update_range_test): Likewise.
30307 (optimize_range_tests_var_bound): Likewise.
30308 (optimize_vec_cond_expr): Likewise.
30309 (rewrite_expr_tree): Likewise.
30310 (rewrite_expr_tree_parallel): Likewise.
30311 (linearize_expr): Likewise.
30312 (break_up_subtract): Likewise.
30313 (linearize_expr_tree): Likewise.
30314 (attempt_builtin_powi): Likewise.
30315 (attempt_builtin_copysign): Likewise.
30316 (transform_stmt_to_copy): Likewise.
30317 (transform_stmt_to_multiply): Likewise.
30318 (dump_ops_vector): Likewise.
30319 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
30320 (print_scc): Likewise.
30321 (set_ssa_val_to): Likewise.
30322 (visit_reference_op_store): Likewise.
30323 (visit_use): Likewise.
30324 (sccvn_dom_walker::before_dom_children): Likewise.
30325 (run_scc_vn): Likewise.
30326 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
30328 (expr_hash_elt::print): Likewise.
30329 (const_and_copies::pop_to_marker): Likewise.
30330 (const_and_copies::record_const_or_copy_raw): Likewise.
30331 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
30332 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
30333 (dump_predicates): Likewise.
30334 (find_uninit_use): Likewise.
30335 (warn_uninitialized_phi): Likewise.
30336 (pass_late_warn_uninitialized::execute): Likewise.
30337 * tree-ssa.c (verify_vssa): Likewise.
30338 (verify_ssa): Likewise.
30339 (maybe_optimize_var): Likewise.
30340 * tree-vrp.c (dump_value_range): Likewise.
30341 (dump_all_value_ranges): Likewise.
30342 (dump_asserts_for): Likewise.
30343 (register_edge_assert_for_2): Likewise.
30344 (vrp_visit_cond_stmt): Likewise.
30345 (vrp_visit_switch_stmt): Likewise.
30346 (vrp_visit_stmt): Likewise.
30347 (vrp_visit_phi_node): Likewise.
30348 (simplify_cond_using_ranges_1): Likewise.
30349 (fold_predicate_in): Likewise.
30350 (evrp_dom_walker::before_dom_children): Likewise.
30351 (evrp_dom_walker::push_value_range): Likewise.
30352 (evrp_dom_walker::pop_value_range): Likewise.
30353 (execute_early_vrp): Likewise.
30355 2017-05-16 Richard Biener <rguenther@suse.de>
30357 * dwarf2out.c (loc_list_from_tree_1): Do not create
30358 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
30360 2017-05-16 Richard Biener <rguenther@suse.de>
30362 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
30364 (note_variable_value_in_expr): If we resolved the decl ref
30365 do not push to the stack.
30367 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
30369 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
30370 operations in fast-math mode.
30371 (vaddq_f16): Likewise.
30372 (vmul_f16): Likewise.
30373 (vmulq_f16): Likewise.
30374 (vsub_f16): Likewise.
30375 (vsubq_f16): Likewise.
30376 * config/arm/neon.md (add<mode>3): New.
30378 (fma:<VH:mode>3): New. Also remove outdated comment.
30381 2017-05-16 Martin Liska <mliska@suse.cz>
30385 * ipa-devirt.c (warn_types_mismatch): Fix typo.
30386 (odr_types_equivalent_p): Likewise.
30388 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
30390 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
30392 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
30395 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
30396 non-interunit SSE move alternatives with '?'.
30397 (zero-extendsidi peephole2): New peephole to skip intermediate
30398 general register in SSE zero-extend sequence.
30400 2017-05-15 Jeff Law <law@redhat.com>
30402 * reorg.c (relax_delay_slots): Create a new variable to hold
30403 the temporary target rather than clobbering TARGET_LABEL.
30405 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
30406 missing argument to extract_bit_field call.
30407 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
30409 2017-05-15 Martin Liska <mliska@suse.cz>
30412 * gcc.c (process_command): Do not allow empty argument of -o option.
30414 2017-05-15 Renlin Li <renlin.li@arm.com>
30416 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
30417 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
30418 * config/aarch64/constraints.md (Usf): Add long call check.
30419 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
30420 (call_value): Likewise.
30421 (sibcall): Likewise.
30422 (sibcall_value): Likewise.
30424 (call_value_insn): New.
30425 (sibcall_insn): Update rtx pattern.
30426 (sibcall_value_insn): Likewise.
30427 (call_internal): Remove.
30428 (call_value_internal): Likewise.
30429 (sibcall_internal): Likewise.
30430 (sibcall_value_internal): Likewise.
30431 (call_reg): Likewise.
30432 (call_symbol): Likewise.
30433 (call_value_reg): Likewise.
30434 (call_value_symbol): Likewise.
30436 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
30439 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
30441 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
30443 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
30444 compatible with CCGOCmode and with CCZmode.
30446 2017-05-14 Martin Sebor <msebor@redhat.com>
30448 PR middle-end/77671
30449 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
30450 (gimple_fold_builtin_snprintf): Same.
30451 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
30452 (gimple_fold_builtin_snprintf): Same.
30453 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
30454 of character types.
30455 (is_call_safe): New function.
30456 (try_substitute_return_value): Call it.
30457 (try_simplify_call): New function.
30458 (pass_sprintf_length::handle_gimple_call): Call it.
30460 2017-05-14 Martin Sebor <msebor@redhat.com>
30462 PR middle-end/80669
30463 * builtins.c (expand_builtin_stpncpy): Simplify.
30465 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
30467 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
30468 * config/i386/i386.h
30469 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
30470 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
30471 (struct machine_function): Add new members call_ms2sysv,
30472 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
30473 (struct machine_frame_state): New fields sp_realigned and
30474 sp_realigned_offset.
30475 * config/i386/i386.c
30476 (enum xlogue_stub): New enum.
30477 (enum xlogue_stub_sets): New enum.
30478 (class xlogue_layout): New class.
30479 (struct ix86_frame): New fields stack_realign_allocate_offset,
30480 stack_realign_offset and outlined_save_offset. Modify comments to
30481 detail stack layout when using out-of-line stubs.
30482 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
30483 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
30484 -mcall-ms2sysv-xlogues.
30485 (stub_managed_regs): New static variable.
30486 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
30487 registers managed by out-of-line stub.
30488 (disable_call_ms2sysv_xlogues): New function.
30489 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
30490 m->call_ms2sysv when appropriate and compute frame layout for
30492 (sp_valid_at, fp_valid_at): New inline functions.
30493 (choose_basereg): New function.
30494 (choose_baseaddr): Add align parameter, use choose_basereg and modify
30496 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
30497 Use align parameter of choose_baseaddr to generated aligned SSE movs
30499 (pro_epilogue_adjust_stack): Modify to track
30500 machine_frame_state::sp_realigned.
30501 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
30502 (ix86_nsaved_sseregs): Likewise.
30503 (ix86_emit_save_regs): Likewise.
30504 (ix86_emit_save_regs_using_mov): Likewise.
30505 (ix86_emit_save_sse_regs_using_mov): Likewise.
30506 (get_scratch_register_on_entry): Likewise.
30507 (gen_frame_set): New function.
30508 (gen_frame_load): Likewise.
30509 (gen_frame_store): Likewise.
30510 (emit_outlined_ms2sysv_save): Likewise.
30511 (emit_outlined_ms2sysv_restore): Likewise.
30512 (ix86_expand_prologue): Modify stack re-alignment code and call
30513 emit_outlined_ms2sysv_save when appropriate.
30514 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
30515 parameter rtx_insn *insn, which allows the function to be used to only
30516 generate the notes.
30517 (ix86_expand_epilogue): Modify validity checks of frame and stack
30518 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
30519 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
30520 * config/i386/predicates.md
30521 (save_multiple): New predicate.
30522 (restore_multiple): Likewise.
30523 * config/i386/sse.md
30524 (save_multiple<mode>): New pattern.
30525 (save_multiple_realign<mode>): Likewise.
30526 (restore_multiple<mode>): Likewise.
30527 (restore_multiple_and_return<mode>): Likewise.
30528 (restore_multiple_leave_return<mode>): Likewise.
30529 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
30531 2017-05-14 Julia Koval <julia.koval@intel.com>
30533 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
30534 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
30535 (__builtin_ia32_xsetbv): New builtins.
30536 * config/i386/i386.c (ix86_expand_special_args_builtin):
30538 (ix86_expand_builtin): Special expand for new intrinsics.
30539 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
30540 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
30541 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
30543 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30545 * cfganal.c (inverted_post_order_compute): Change argument type
30547 * cfganal.h (inverted_post_order_compute): Adjust prototype.
30548 * df-core.c (rest_of_handle_df_initialize): Adjust.
30549 (rest_of_handle_df_finish): Likewise.
30550 (df_analyze_1): Likewise.
30551 (df_analyze): Likewise.
30552 (loop_inverted_post_order_compute): Change argument to be a vec *.
30553 (df_analyze_loop): Adjust.
30554 (df_get_n_blocks): Likewise.
30555 (df_get_postorder): Likewise.
30556 * df.h (struct df_d): Change field to be a vec.
30557 * lcm.c (compute_laterin): Adjust.
30558 (compute_available): Likewise.
30559 * lra-lives.c (lra_create_live_ranges_1): Likewise.
30560 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
30561 * tree-ssa-pre.c (compute_antic): Likewise.
30563 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30565 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
30566 (depth_first_search::depth_first_search): Change structure init
30567 function to this constructor.
30568 (depth_first_search::add_bb): Rename function to this member.
30569 (depth_first_search::execute): Likewise.
30570 (flow_dfs_compute_reverse_finish): Adjust.
30572 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30574 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
30575 (longest_simple_path): Likewise.
30576 * shrink-wrap.c (spread_components): Likewise.
30577 (disqualify_problematic_components): Likewise.
30578 (emit_common_heads_for_components): Likewise.
30579 (emit_common_tails_for_components): Likewise.
30580 (insert_prologue_epilogue_for_components): Likewise.
30582 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30584 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
30587 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30589 * df-core.c (df_set_blocks): Start using auto_bitmap.
30590 (df_compact_blocks): Likewise.
30591 * df-problems.c (df_rd_confluence_n): Likewise.
30592 * df-scan.c (df_insn_rescan_all): Likewise.
30593 (df_process_deferred_rescans): Likewise.
30594 (df_update_entry_block_defs): Likewise.
30595 (df_update_exit_block_uses): Likewise.
30596 (df_entry_block_bitmap_verify): Likewise.
30597 (df_exit_block_bitmap_verify): Likewise.
30598 (df_scan_verify): Likewise.
30599 * lra-constraints.c (lra_constraints): Likewise.
30600 (undo_optional_reloads): Likewise.
30601 (lra_undo_inheritance): Likewise.
30602 * lra-remat.c (calculate_gen_cands): Likewise.
30603 (do_remat): Likewise.
30604 * lra-spills.c (assign_spill_hard_regs): Likewise.
30605 (spill_pseudos): Likewise.
30606 * tree-ssa-pre.c (bitmap_set_and): Likewise.
30607 (bitmap_set_subtract_values): Likewise.
30609 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30611 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
30612 management with auto_bitmap.
30613 (fix_inter_tick): Likewise.
30614 (fix_recovery_deps): Likewise.
30615 * ira.c (add_store_equivs): Likewise.
30616 (find_moveable_pseudos): Likewise.
30617 (split_live_ranges_for_shrink_wrap): Likewise.
30618 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
30619 (rtx_reuse_manager::seen_def_p): Likewise.
30620 (rtx_reuse_manager::set_seen_def): Likewise.
30621 * print-rtl.h (class rtx_reuse_manager): Likewise.
30623 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30625 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
30627 (migrate_btr_def): Likewise.
30628 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
30629 * df-core.c (loop_post_order_compute): Likewise.
30630 (loop_inverted_post_order_compute): Likewise.
30631 * hsa-common.h: Likewise.
30632 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
30633 * init-regs.c (initialize_uninitialized_regs): Likewise.
30634 * ipa-inline.c (resolve_noninline_speculation): Likewise.
30635 (inline_small_functions): Likewise.
30636 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
30637 * ira.c (combine_and_move_insns): Likewise.
30638 (build_insn_chain): Likewise.
30639 * loop-invariant.c (find_invariants): Likewise.
30640 * lower-subreg.c (propagate_pseudo_copies): Likewise.
30641 * predict.c (tree_predict_by_opcode): Likewise.
30642 (predict_paths_leading_to): Likewise.
30643 (predict_paths_leading_to_edge): Likewise.
30644 (estimate_loops_at_level): Likewise.
30645 (estimate_loops): Likewise.
30646 * shrink-wrap.c (try_shrink_wrapping): Likewise.
30647 (spread_components): Likewise.
30648 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
30649 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
30650 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
30651 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
30652 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
30653 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
30654 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
30655 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
30656 (mark_threaded_blocks): Likewise.
30657 (thread_through_all_blocks): Likewise.
30658 * tree-ssa.c (verify_ssa): Likewise.
30659 (execute_update_addresses_taken): Likewise.
30660 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
30662 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30664 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
30666 (post_order_compute): Likewise.
30667 (inverted_post_order_compute): Likewise.
30668 (pre_and_rev_post_order_compute_fn): Likewise.
30670 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30672 * genrecog.c (int_set::int_set): Explicitly construct our
30673 auto_vec base class.
30674 * vec.h (auto_vec::auto_vec): New constructor.
30676 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30678 * bitmap.h (class auto_bitmap): New constructor taking
30679 bitmap_obstack * argument.
30681 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30683 * bitmap.h (class auto_bitmap): Change type of m_bits to
30684 bitmap_head, and adjust ctor / dtor and member operators.
30686 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
30688 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
30689 when returned register mode doesn't match original mode.
30691 2017-05-12 Jeff Law <law@redhat.com>
30692 Jakub Jelinek <jakub@redhat.com>
30694 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
30695 we look for cc setter after the compare-elim changes.
30696 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
30697 within the vector to match what compare-elim now expects.
30698 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
30699 (xorsi3_flags, one_cmplsi2_flags): Likewise.
30701 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
30702 after the compare-elim changes.
30703 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
30704 the vector to match what compare-elim now expects.
30705 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
30706 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
30707 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
30708 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
30709 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
30711 * config/visium/visium.c (single_set_and_flags): Fix where
30712 we look for cc setter after the compare-elim changes.
30713 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
30714 with the vector to match what compare-elim now expects.
30715 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
30716 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
30717 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
30718 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
30719 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
30720 (neg<mode>2_insn_set_overflow): Likewise.
30722 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
30724 PR middle-end/79794
30725 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
30726 maybe_expand_insn call, set ops[0].target. If still set after call,
30727 set alt_rtl. Add extra arg to recursive calls.
30728 (extract_bit_field): Add alt_rtl argument. Pass to
30730 * expmed.h (extract_bit_field): Fix prototype.
30731 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
30732 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
30733 to extract_bit_field_calls.
30734 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
30735 Pass alt_rtl to extract_bit_field calls.
30736 * calls.c (store_unaligned_arguments_into_psuedos)
30737 load_register_parameters): Pass extra NULL to extract_bit_field calls.
30738 * optabs.c (maybe_legitimize_operand): Clear op->target when call
30740 * optabs.h (struct expand_operand): Add target bitfield.
30742 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30744 * compare-elim.c (try_eliminate_compare): Canonicalize
30745 operation with embedded compare to
30746 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
30747 (set (reg) (operation)].
30749 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
30751 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30754 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
30755 cost of adding a carry flag for ADC instruction.
30756 [case MINUS]: Ignore the cost of subtracting a carry flag
30757 for SBB instruction.
30759 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
30761 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
30763 * config/rs6000/bmiintrin.h: New file.
30764 * config/rs6000/bmi2intrin.h: New file.
30765 * config/rs6000/x86intrin.h: New file.
30767 2017-05-12 Jeff Law <law@redhat.com>
30769 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
30772 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
30774 PR middle-end/80707
30775 * tree-cfg.c: Remove cfg edges of unreachable case statements.
30777 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30779 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30780 early expansion of vector divide builtins.
30781 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
30782 builtins identified as having unsigned arguments.
30784 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30786 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
30787 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
30788 expansion of vector logical operations (and, andc, or, xor,
30791 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30793 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
30794 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
30796 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30798 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30799 early GIMPLE expansion of vector multiplies.
30801 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30803 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
30804 TARGET_HAVE_MOVT conditional.
30805 (movt splitter): Likewise.
30807 2017-05-12 Richard Biener <rguenther@suse.de>
30809 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
30810 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30811 Fold all stmts not inplace.
30813 2017-05-12 Richard Biener <rguenther@suse.de>
30815 PR tree-optimization/80713
30816 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
30817 inserted_exprs bit for not removed stmts.
30819 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
30821 PR middle-end/69921
30822 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
30823 parallelized" attribute for parallelized OpenACC kernels.
30824 * omp-offload.c (execute_oacc_device_lower): Use it.
30826 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
30827 Set "oacc kernels" attribute.
30828 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
30829 parameter. Adjust all users.
30830 (oacc_fn_attrib_kernels_p): Remove function.
30831 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
30832 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
30833 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
30834 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
30835 assert "oacc kernels" attribute is set.
30837 2017-05-11 Carl Love <cel@us.ibm.com>
30839 * config/rs6000/rs6000-c: Add support for built-in functions
30840 vector unsigned char vec_popcnt (vector signed char)
30841 vector unsigned char vec_popcnt (vector unsigned char)
30842 vector unsigned short vec_popcnt (vector signed short)
30843 vector unsigned short vec_popcnt (vector unsigned short)
30844 vector unsigned int vec_popcnt (vector signed int)
30845 vector unsigned int vec_popcnt (vector unsigned int)
30846 vector unsigned long long vec_popcnt (vector signed long long)
30847 vector unsigned long long vec_popcnt (vector unsigned long long)
30848 vector signed long long vec_slo (vector signed long long,
30849 vector signed char)
30850 vector signed long long vec_slo (vector signed long long,
30851 vector unsigned char)
30852 vector unsigned long long vec_slo (vector unsigned long long,
30853 vector signed char)
30854 vector unsigned long long vec_slo (vector unsigned long long,
30855 vector unsigned char)
30856 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
30857 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
30858 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
30859 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
30860 * doc/extend.texi: Update the built-in documentation file for the
30861 new built-in functions.
30863 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
30865 * attribs.h (sorted_attr_string): Move machine independent
30866 functions for target clone support from the i386 port to common
30867 code. Rename ix86_function_versions to common_function_versions.
30868 Rename make_name to make_unique_name.
30869 (common_function_versions): Likewise.
30870 (make_unique_name): Likewise.
30871 (make_dispatcher_decl): Likewise.
30872 (is_function_default_version): Likewise.
30873 * attribs.c (attr_strcmp): Likewise.
30874 (sorted_attr_string): Likewise.
30875 (common_function_versions): Likewise.
30876 (make_unique_name): Likewise.
30877 (make_dispatcher_decl): Likewise.
30878 (is_function_default_version): Likewise.
30879 * config/i386/i386.c (attr_strcmp): Likewise.
30880 (sorted_attr_string): Likewise.
30881 (ix86_function_versions): Likewise.
30882 (make_name): Likewise.
30883 (make_dispatcher_decl): Likewise.
30884 (is_function_default_version): Likewise.
30885 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
30887 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30890 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
30891 Account for direct move costs for vec_construct of integer
30894 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
30897 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
30898 (UNSPEC_STX_ATOMIC): Ditto.
30899 (loaddi_via_sse): New insn.
30900 (storedi_via_sse): Ditto.
30901 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
30902 Update corresponding peephole2 patterns.
30903 (atomic_storedi_fpu): Ditto.
30905 2017-05-11 Julia Koval <julia.koval@intel.com>
30907 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
30908 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
30910 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
30911 (__builtin_ia32_rsqrt14ss_mask): New builtins.
30912 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
30914 2017-05-11 Nathan Sidwell <nathan@acm.org>
30916 * graphite-poly.c: Include dumpfile.h.
30918 * dumpfle.h (dump_function): Declare here ...
30919 * tree-dump.h (dump_function): ... not here.
30920 * dumpfile.c: #include tree-cfg.h.
30921 (dump_function): Move here from ...
30922 * tree-dump.c (dump_function): ... here.
30923 * gimplify.c: #include splay-tree.h, not tree-dump.h.
30924 * graphite-poly.c: Don't include tree-dump.h.
30925 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
30926 * print-tree.c: Likewise.
30927 * stor-layout.c: Likewise.
30928 * tree-nested.c: Likewise.
30930 * dumpfile.c (dump_start): Use TDF_FLAGS.
30931 (dump_enable_all): Fix TDF_KIND check thinko.
30933 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
30935 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30936 array entries to represent two legal parameterizations of the
30937 overloaded __builtin_cmpb function, as represented by the
30938 P6_OV_BUILTIN_CMPB constant.
30939 (altivec_resolve_overloaded_builtin): Add special case handling
30940 for the __builtin_cmpb function, as represented by the
30941 P6_OV_BUILTIN_CMPB constant.
30942 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
30943 (BU_P6_64BIT_2): New macro.
30944 (BU_P6_OVERLOAD_2): New macro
30945 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
30946 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
30947 (CMPB): Add overload support to represent both 32-bit and 64-bit
30948 compare-bytes function.
30949 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
30950 support for TARGET_CMPB.
30951 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
30952 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30953 documentation of the __builtin_cmpb overloaded built-in function.
30955 2017-05-11 Richard Biener <rguenther@suse.de>
30957 PR tree-optimization/80705
30958 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
30959 bases are not vectorizable.
30961 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30963 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
30964 when counting register pressure.
30966 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30968 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
30969 (iv_ca_more_deps): Renamed to ...
30970 (iv_ca_compare_deps): ... this.
30971 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
30973 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30975 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
30977 (determine_group_iv_costs): ... here.
30978 (find_inv_vars_cb): Record inv var if it's not recorded before.
30980 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30982 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
30983 (get_shiftadd_cost): Ditto.
30985 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30987 * tree-ssa-address.c: Include header file.
30988 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
30990 (add_to_parts): Refactor.
30991 (addr_to_parts): New parameter. Update use of move_hint_to_base.
30992 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
30995 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30997 PR tree-optimization/53090
30998 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
31000 (extract_cond_operands): Detect condition with IV on both sides
31001 and return COMP_IV_EXPR_2.
31002 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
31003 (rewrite_use_compare): Simplify by removing call to function
31004 extract_cond_operands.
31006 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31008 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
31009 (extract_cond_operands): Detect condition comparing against non-
31010 invariant bound and return appropriate enum value.
31011 (find_interesting_uses_cond): Update use of extract_cond_operands.
31012 Handle its return value accordingly.
31013 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
31015 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31017 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
31018 nonlinear iv_use computation in loop invariant sensitive way.
31020 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31022 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
31023 (find_iv_candidates): Call relate_compare_use_with_all_cands.
31025 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31027 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
31028 (dump_cand): Support iv_cand.inv_exprs.
31029 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
31031 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
31034 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31036 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
31038 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
31039 as local function. Include necessary header files.
31040 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
31042 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31044 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
31046 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31048 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
31049 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
31050 RSHIFT_EXPR and BIT_NOT_EXPR.
31052 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31054 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
31055 (adjust_setup_cost): New parameter supporting round up adjustment.
31056 (struct address_cost_data): Delete.
31057 (force_expr_to_var_cost): Don't bound cost with spill_cost.
31058 (split_address_cost, ptr_difference_cost): Delete.
31059 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
31060 (struct ainc_cost_data): New struct.
31061 (get_address_cost_ainc): New function.
31062 (get_address_cost, get_computation_cost): Reimplement.
31063 (determine_group_iv_cost_address): Record inv_expr for all uses of
31065 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
31066 (iv_ca_has_deps): Reimplemented to ...
31067 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
31069 (iv_ca_extend): Call iv_ca_more_deps.
31071 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31073 * tree-ssa-address.c (struct mem_address): Move to header file.
31074 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
31075 * tree-ssa-address.h (struct mem_address): Move from C file.
31076 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
31078 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31080 * tree-affine.h (aff_combination_type): New interface.
31081 (aff_combination_zero_p): Remove static.
31082 (aff_combination_const_p): New interface.
31083 (aff_combination_singleton_var_p): New interfaces.
31085 2017-05-11 Richard Biener <rguenther@suse.de>
31087 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31088 Skip unreachable blocks and destinations.
31089 (eliminate): Move stmt removal and fixup ...
31090 (fini_eliminate): ... here. Skip inserted exprs.
31091 (pass_pre::execute): Move fini_pre after fini_eliminate.
31092 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
31093 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
31094 PRE to get rid of dead code that has invalid SSA form and
31095 split critical edges again.
31097 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31099 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31101 2017-05-11 Richard Biener <rguenther@suse.de>
31103 * passes.c (execute_function_todo): Verify loops if they are
31104 said to be up-to-date.
31105 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
31106 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
31108 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
31111 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
31112 handle calling assemble_external ourself.
31115 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
31116 modes with zero size. Enhance comment.
31118 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31120 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
31121 built-ins for vec_xl and vec_xst with short and char pointer
31124 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31126 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
31127 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
31128 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
31129 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
31130 (_mm_maskz_min_round_ss): New intrinsics.
31131 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
31132 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
31133 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
31134 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
31135 (__builtin_ia32_minss_mask_round): New builtins.
31136 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31137 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31138 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
31140 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
31141 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
31143 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
31146 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31148 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
31149 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
31150 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
31151 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
31152 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
31153 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
31154 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
31155 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31156 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31157 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
31158 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
31159 (__builtin_ia32_mulss_mask_round): New builtins.
31160 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31161 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31162 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
31164 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
31165 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31167 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31170 2017-05-10 Julia Koval <julia.koval@intel.com>
31172 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
31173 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
31174 (_mm256_setr_m128i): New intrinsics.
31176 2017-05-10 Julia Koval <julia.koval@intel.com>
31178 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
31179 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
31180 (_mm_maskz_rcp14_ss): New intrinsics.
31181 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
31182 (__builtin_ia32_rcp14ss_mask): New builtins.
31183 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
31185 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
31187 PR tree-optimization/51513
31188 * tree-cfg.c (gimple_seq_unreachable_p): New function.
31189 (assert_unreachable_fallthru_edge_p): Use it.
31190 (group_case_labels_stmt): Likewise.
31191 * tree-cfg.h: Prototype it.
31192 * stmt.c: Include cfghooks.h and tree-cfg.h.
31193 (emit_case_dispatch_table) <gap_label>: New local variable.
31194 Use it to fill dispatch table gaps.
31195 Test for default_label before updating probabilities.
31196 (expand_case) <default_label>: Remove unneeded initialization.
31197 Test for unreachable default case statement and remove its edge.
31198 Set default_label accordingly.
31199 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
31201 2017-05-10 Carl Love <cel@us.ibm.com>
31203 * config/rs6000/rs6000-c: Add support for built-in functions
31204 vector signed char vec_neg (vector signed char)
31205 vector signed short int vec_neg (vector short int)
31206 vector signed int vec_neg (vector signed int)
31207 vector signed long long vec_neg (vector signed long long)
31208 vector float vec_neg (vector float)
31209 vector double vec_neg (vector double)
31210 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
31212 * config/rs6000/altivec.h: Add define for vec_neg
31213 * doc/extend.texi: Update the built-in documentation for the
31214 new built-in functions.
31216 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31218 PR tree-optimization/77644
31219 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
31221 2017-05-10 Nathan Sidwell <nathan@acm.org>
31223 * dumpfile.h (TDI_lang_all): New.
31224 (TDF_KIND): New. Renumber others
31225 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
31227 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
31229 (get_dump_file_name): Adjust suffix generation.
31230 (dump_enable_all): Use TDF_KIND.
31231 * doc/invoke.texi (-fdump-lang-all): Document.
31233 * dumpfile.h: Tabify.
31235 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
31238 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
31239 Move member access before delete.
31241 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
31243 * tree-inline.c (expand_call_inline): Split block at stmt
31246 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
31249 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
31250 are now unused after splitting mov{sf,sd}_hardfloat.
31251 (f32_lr2): Likewise.
31252 (f32_lm): Likewise.
31253 (f32_lm2): Likewise.
31254 (f32_li): Likewise.
31255 (f32_li2): Likewise.
31256 (f32_lv): Likewise.
31257 (f32_sr): Likewise.
31258 (f32_sr2): Likewise.
31259 (f32_sm): Likewise.
31260 (f32_sm2): Likewise.
31261 (f32_si): Likewise.
31262 (f32_si2): Likewise.
31263 (f32_sv): Likewise.
31264 (f32_dm): Likewise.
31265 (f32_vsx): Likewise.
31266 (f32_av): Likewise.
31267 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
31268 For movsf, order stores so the VSX stores occur before the GPR
31269 store which encourages the register allocator to use a traditional
31270 FPR instead of a GPR. For movsd, order the stores so that the GPR
31271 store comes before the VSX stores to allow the power6 to work.
31272 This is due to the power6 not having a 32-bit integer store
31273 instruction from a FPR.
31274 (movsf_hardfloat): Likewise.
31275 (movsd_hardfloat): Likewise.
31277 2017-05-09 Martin Sebor <msebor@redhat.com>
31279 PR translation/80280
31280 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
31283 PR translation/80280
31284 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
31285 data member added in r247778.
31286 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
31288 2017-05-09 Nathan Sidwell <nathan@acm.org>
31290 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
31292 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
31295 2017-05-09 Marek Polacek <polacek@redhat.com>
31297 * doc/invoke.texi: Fix typo.
31299 2017-05-09 Richard Biener <rguenther@suse.de>
31301 * tree-vrp.c (vrp_val_is_max): Adjust comment.
31302 (vrp_val_is_min): Likewise.
31303 (set_value_range_to_value): Likewise.
31304 (set_value_range_to_nonnegative): Likewise.
31305 (gimple_assign_nonzero_p): Likewise.
31306 (gimple_stmt_nonzero_p): Likewise.
31307 (vrp_int_const_binop): Likewise. Remove unreachable case.
31308 (adjust_range_with_scev): Adjust comments.
31309 (compare_range_with_value): Likewise.
31310 (extract_range_from_phi_node): Likewise.
31311 (test_for_singularity): Likewise.
31313 2017-05-09 Richard Biener <rguenther@suse.de>
31315 * tree-vrp.c (get_single_symbol): Add assert that we don't
31316 get overflowed constants as invariant part.
31317 (compare_values_warnv): Add comment before the TREE_NO_WARNING
31318 checks. Use wi::cmp instead of recursing for integer constants.
31319 (compare_values): Just ignore whether we assumed undefined
31320 overflow instead of failing the compare.
31321 (extract_range_for_var_from_comparison_expr): Add comment before the
31322 TREE_NO_WARNING sets.
31323 (test_for_singularity): Likewise.
31324 (extract_range_from_comparison): Do not disable optimization
31325 when we assumed undefined overflow.
31326 (extract_range_basic): Remove init of unused var.
31328 2017-05-09 Richard Biener <rguenther@suse.de>
31330 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
31331 (extract_range_from_multiplicative_op_1): Adjust.
31332 (extract_range_from_binary_expr_1): Use int_const_binop.
31334 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
31337 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
31338 rs6000_store_data_bypass_p in seven define_bypass directives and
31339 in several comments.
31340 * config/rs6000/rs6000-protos.h: Add prototype for
31341 rs6000_store_data_bypass_p function.
31342 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
31343 function implements slightly different (rs6000-specific) semantics
31344 than store_data_bypass_p, returning false rather than aborting
31345 with assertion error when arguments do not satisfy the
31346 requirements of store data bypass.
31347 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
31348 rs6000_store_data_bypass_p.
31350 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
31352 * config/xtensa/xtensa-protos.h
31353 (xtensa_initial_elimination_offset): New declaration.
31354 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
31355 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
31356 macro definition, add case for FRAME_POINTER_REGNUM when
31357 FRAME_GROWS_DOWNWARD.
31358 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
31359 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
31360 xtensa_initial_elimination_offset.
31362 2017-05-08 Nathan Sidwell <nathan@acm.org>
31364 * doc/invoke.texi: Alphabetize -fdump options.
31366 2017-05-08 Martin Sebor <msebor@redhat.com>
31368 PR translation/80280
31369 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
31371 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
31373 * target.def (compute_frame_layout): New optional target hook.
31374 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
31375 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
31376 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
31378 * reload1.c (verify_initial_elim_offsets): Likewise.
31379 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
31380 (use_simple_return_p): Call arm_compute_frame_layout if needed.
31381 (arm_get_frame_offsets): Split up into this ...
31382 (arm_compute_frame_layout): ... and this function.
31384 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
31386 * config/aarch64/constraints.md (Usa): New constraint.
31387 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
31389 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
31391 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
31392 with_multilib_list after it has been checked.
31394 2017-05-08 Richard Biener <rguenther@suse.de>
31396 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
31397 (bitmap_set_subtract_values): Likewise.
31399 2017-05-08 Richard Biener <rguenther@suse.de>
31401 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
31402 (gimple_assign_nonzero): ... this and remove strict_overflow_p
31404 (gimple_stmt_nonzero_warnv_p): Rename to ...
31405 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
31407 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
31408 (extract_range_basic): Adjust, do not disable propagation on
31409 strict overflow sensitive simplification.
31410 (vrp_visit_cond_stmt): Likewise.
31412 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
31414 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
31415 body size unconditionally.
31417 2017-05-07 Jeff Law <law@redhat.com>
31420 2017-05-06 Jeff Law <law@redhat.com>
31421 PR tree-optimization/78496
31422 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31425 PR tree-optimization/78496
31426 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31427 (simplify_stmt_using_ranges): Call it.
31428 (vrp_dom_walker::before_dom_children): Extract equivalences
31429 from an ASSERT_EXPR with an equality comparison against a
31432 2017-05-06 Jeff Law <law@redhat.com>
31434 PR tree-optimization/78496
31435 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31438 PR tree-optimization/78496
31439 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31440 (simplify_stmt_using_ranges): Call it.
31441 (vrp_dom_walker::before_dom_children): Extract equivalences
31442 from an ASSERT_EXPR with an equality comparison against a
31445 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31447 * lra-constraints.c (lra_copy_reg_equiv): New function.
31448 (split_reg): Use it to copy equivalence information from the
31449 original register to the spill register.
31451 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31453 PR rtl-optimization/75964
31454 * simplify-rtx.c (simplify_const_relational_operation): Remove
31455 invalid handling of comparisons of integer ABS.
31457 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
31459 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
31460 initialize to zero.
31461 (init_regs): Remove declaration.
31462 (function_arg_advance_32): Initialize error_p as boolean variable.
31464 2017-05-05 Nathan Sidwell <nathan@acm.org>
31466 * store-motion.c (remove_reachable_equiv_notes): Reformat long
31467 lines. Use for (;;).
31469 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31471 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
31472 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
31473 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
31474 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
31475 VF=2 that require versioning.
31477 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31479 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
31482 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31484 * diagnostic.h (diagnostic_override_option_index): Convert from
31485 macro to inline function.
31487 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31489 * diagnostic.c (last_module_changed_p): New function.
31490 (set_last_module): New function.
31491 (diagnostic_report_current_module): Convert macro usage to
31492 the above functions.
31493 * diagnostic.h (diagnostic_context::last_module): Strengthen
31494 from const line_map * to const line_map_ordinary *.
31495 (diagnostic_last_module_changed): Delete macro.
31496 (diagnostic_set_last_module): Delete macro.
31498 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31500 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
31501 with diagnostic_report_diagnostic.
31502 (diagnostic_n_impl_richloc): Likewise.
31503 * diagnostic.h (report_diagnostic): Delete macro.
31504 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
31505 with diagnostic_report_diagnostic.
31506 * substring-locations.c (format_warning_va): Likewise.
31508 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31510 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
31511 save/restor of format_spec. Move option-printing code to...
31512 (print_option_information): ...this new function, and
31513 reimplement by simply printing to the pretty_printer,
31514 rather than appending to the format string.
31516 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31518 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
31519 handling logic into...
31520 (update_effective_level_from_pragmas): ...this new function.
31522 2017-05-04 Andrew Waterman <andrew@sifive.com>
31524 * config/riscv/riscv.opt (mstrict-align): New option.
31525 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
31526 (SLOW_UNALIGNED_ACCESS): Define.
31527 (riscv_slow_unaligned_access): Declare.
31528 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
31530 (riscv_slow_unaligned_access): New variable.
31531 (rocket_tune_info): Set slow_unaligned_access to true.
31532 (optimize_size_tune_info): Set slow_unaligned_access to false.
31533 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
31534 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
31535 (riscv_option_override): Set riscv_slow_unaligned_access.
31536 * doc/invoke.texi: Add -mstrict-align to RISC-V.
31538 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
31540 * config/riscv/riscv.md: Unify indentation.
31542 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
31547 * config/rs6000/rs6000.md (u code attribute): Add FIX and
31549 (extendsi<mode>2): Add support for doing sign extension via
31550 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
31551 don't have ISA 3.0 instructions.
31552 (extendsi<mode>2 splitter): Likewise.
31553 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
31554 generate the normal insns since SImode can now go in vector
31555 registers. Disallow the special UNSPECs needed for previous
31556 machines to hide SImode being used. Add new insns
31557 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
31558 (fix_trunc<mode>si2_stfiwx): Likewise.
31559 (fix_trunc<mode>si2_internal): Likewise.
31560 (fixuns_trunc<mode>si2): Likewise.
31561 (fixuns_trunc<mode>si2_stfiwx): Likewise.
31562 (fctiw<u>z_<mode>_smallint): Likewise.
31563 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
31564 of floating point to 32-bit integer from doing a direct move to
31565 the GPR registers to do a store.
31566 (fctiwz_<mode>): Break long line.
31568 2017-05-05 Bin Cheng <bin.cheng@arm.com>
31570 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
31571 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
31572 (addr_list, addr_offset_valid_p): New.
31573 (split_address_groups): Check offset validity with above function.
31574 (gt-tree-ssa-loop-ivopts.h): Include header file.
31576 2017-05-05 Nathan Sidwell <nathan@acm.org>
31578 * config.gcc (arm*-*-*): Add missing 'fi'.
31580 2017-05-05 Steve Ellcey <sellcey@cavium.com>
31582 * doc/invoke.texi (-fopt-info): Explicitly say order of options
31583 included in -fopt-info does not matter.
31584 * doc/optinfo.texi (-fopt-info): Fix description of default
31585 behavour. Explicitly say order of options included in -fopt-info
31588 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
31590 * config.gcc: Allow combinations of aprofile and rmprofile values for
31591 --with-multilib-list.
31592 * config/arm/t-multilib: New file.
31593 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
31594 variables. Remove setting of ISA and floating-point ABI in
31595 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
31596 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
31597 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
31598 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
31600 * config/arm/t-rmprofile: Likewise except for the matches changes.
31601 * doc/install.texi (--with-multilib-list): Document the combination of
31602 aprofile and rmprofile values and warn about pitfalls in doing that.
31604 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31606 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
31607 (movdi_aarch64): Likewise.
31609 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31611 PR tree-optimization/80632
31612 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
31614 (build_arrays): Initialize it for virtual phis.
31615 (fix_phi_nodes): Use it for virtual phis.
31617 PR tree-optimization/80558
31618 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
31619 [x, y] op z into [x op, y op z] for op & or | if conditions
31622 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
31623 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31626 * config/arm/arm.md (use_literal_pool): Remove.
31627 (64-bit immediate split): No longer takes cost into consideration
31628 if arm_disable_literal_pool is enabled.
31629 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
31630 used when arm_disable_literal_pool is enabled.
31631 (arm_max_const_double_inline_cost): Remove use of
31632 arm_disable_literal_pool.
31633 (push_minipool_fix): Add assert.
31634 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
31635 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
31636 (no_literal_pool_sf_immediate): New.
31638 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31640 PR tree-optimization/80613
31641 * tree-ssa-dce.c (propagate_necessity): Remove cases for
31642 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
31644 2017-05-05 Richard Biener <rguenther@suse.de>
31646 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
31648 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
31650 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
31651 of this flag from insn conditions due to removal from r247495.
31653 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31655 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
31657 (arm_early_store_addr_dep_ptr): Likewise.
31658 * config/arm/aarch-common-protos.h
31659 (arm_early_load_addr_dep_ptr): Add prototype.
31660 (arm_early_store_addr_dep_ptr): Likewise.
31661 * config/arm/cortex-a53.md: Add new bypasses.
31663 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31665 * tree.c (next_type_uid): Change type to unsigned.
31666 (type_hash_canon): Decrement back next_type_uid if
31667 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
31668 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
31671 2017-05-04 Martin Sebor <msebor@redhat.com>
31673 * builtins.c: Fix a trivial typo in a comment.
31675 PR middle-end/79234
31676 * builtins.c (check_sizes): Adjust to handle reading past the end.
31677 Avoid printing excessive upper bound of ranges. Use %E to print
31678 tree nodes instead of converting them to %wu.
31679 (expand_builtin_memchr): New function.
31680 (compute_dest_size): Rename...
31681 (compute_objsize): ...to this.
31682 (expand_builtin_memcpy): Adjust.
31683 (expand_builtin_mempcpy): Adjust.
31684 (expand_builtin_strcat): Adjust.
31685 (expand_builtin_strcpy): Adjust.
31686 (check_strncat_sizes): Adjust.
31687 (expand_builtin_strncat): Adjust.
31688 (expand_builtin_strncpy): Adjust and simplify.
31689 (expand_builtin_memset): Adjust.
31690 (expand_builtin_bzero): Adjust.
31691 (expand_builtin_memcmp): Adjust.
31692 (expand_builtin): Handle memcmp.
31693 (maybe_emit_chk_warning): Check strncat just once.
31695 2017-05-04 Martin Sebor <msebor@redhat.com>
31697 PR preprocessor/79214
31698 PR middle-end/79222
31699 PR middle-end/79223
31700 * builtins.c (check_sizes): Add inlining context and issue
31701 warnings even when -Wno-system-headers is set.
31702 (check_strncat_sizes): Same.
31703 (expand_builtin_strncat): Same.
31704 (expand_builtin_memmove): New function.
31705 (expand_builtin_stpncpy): Same.
31706 (expand_builtin): Handle memmove and stpncpy.
31708 2017-05-04 Bin Cheng <bin.cheng@arm.com>
31710 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
31711 which is not used any more.
31713 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31715 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
31717 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31719 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
31720 (cortexa53_tunings): Likewise.
31721 (cortexa57_tunings): Likewise.
31722 (cortexa72_tunings): Likewise.
31723 (cortexa73_tunings): Likewise.
31725 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31727 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
31728 Set loop alignment to 8.
31730 2017-05-04 Martin Sebor <msebor@redhat.com>
31732 PR translation/80280
31733 * builtins.c (expand_builtin_object_size): Add missing quoting to
31734 %D and like directives.
31735 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
31736 (hsa_type_for_tree_type): Same.
31737 (verify_function_arguments): Same.
31738 * symtab.c (symbol_table::change_decl_assembler_name): Same.
31739 * varasm.c (get_section): Same.
31742 2017-05-04 Martin Sebor <msebor@redhat.com>
31744 PR translation/80280
31745 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
31747 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31749 * config/aarch64/aarch64.c (generic_addrcost_table):
31750 Change HI/TI mode setting.
31752 2017-05-04 Martin Jambor <mjambor@suse.cz>
31754 PR tree-optimization/80622
31755 * tree-sra.c (comes_initialized_p): New function.
31756 (build_accesses_from_assign): Only set write lazily when
31757 comes_initialized_p is false.
31758 (analyze_access_subtree): Use comes_initialized_p.
31759 (propagate_subaccesses_across_link): Assert !comes_initialized_p
31760 instead of testing for PARM_DECL.
31762 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31764 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
31765 constraint on operand 0 to allow more general addressing modes.
31766 Adjust output template.
31767 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
31769 * config/aarch64/aarch64-protos.h
31770 (aarch64_address_valid_for_prefetch_p): Declare prototype.
31771 * config/aarch64/constraints.md (Dp): New address constraint.
31772 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
31775 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
31777 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
31778 update use of estimate_ipcp_clone_size_and_time.
31779 (estimate_local_effects): Update use of
31780 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
31781 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
31782 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
31783 Return nonspecialized time.
31785 2017-05-04 Richard Biener <rguenther@suse.de>
31787 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
31788 for the last VUSE which def dominates the PHI. Directly call
31790 (get_continuation_for_phi_1): Remove.
31792 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
31794 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
31795 to explain the use of truncating division. Cap the number of
31796 iterations to the maximum given by nb_iterations_upper_bound,
31799 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31801 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
31802 * configure: Regenerate.
31803 * config.in: Regenerate.
31804 * config/i386/driver-mingw32.c: new file.
31805 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
31806 * config.host: Link driver-mingw32.o on MinGW host.
31807 * doc/install.texi: Document new --enable-mingw-wildcard configure
31810 2017-05-04 Marek Polacek <polacek@redhat.com>
31812 PR tree-optimization/80612
31813 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
31815 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31816 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
31818 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
31819 (movt splitter): Likewise.
31820 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
31821 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
31822 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
31823 block for Thumb-1 with MOVT.
31824 (thumb2_legitimate_address_p): Move code block ...
31825 (can_avoid_literal_pool_for_label_p): ... into this new function.
31826 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
31828 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
31829 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
31830 "M-profile targets with the MOVT instruction".
31832 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31834 * config/arm/arm-builtins.c (arm_init_builtins): Rename
31835 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
31836 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
31838 2017-05-04 Martin Liska <mliska@suse.cz>
31840 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
31841 variable cond_code.
31843 2017-05-04 Richard Biener <rguenther@suse.de>
31845 * tree.c (array_at_struct_end_p): Handle arrays at struct
31846 end with flexarrays more conservatively. Refactor and treat
31847 arrays of arrays or aggregates more strict. Fix
31848 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
31849 * tree.c (array_at_struct_end_p): Adjust prototype.
31850 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
31851 * gimple-fold.c (get_range_strlen): Likewise.
31852 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
31854 2017-05-04 Richard Biener <rguenther@suse.de>
31856 PR tree-optimization/31130
31857 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
31859 (supports_overflow_infinity): Likewise.
31860 (is_negative_overflow_infinity): Likewise.
31861 (is_positive_overflow_infinity): Likewise.
31862 (is_overflow_infinity): Likewise.
31863 (stmt_overflow_infinity): Likewise.
31864 (overflow_infinity_range_p): Likewise.
31865 (usable_range_p): Remove as always returning true.
31866 (make_overflow_infinity): Remove.
31867 (negative_overflow_infinity): Likewise.
31868 (positive_overflow_infinity): Likewise.
31869 (avoid_overflow_infinity): Likewise.
31870 (set_value_range): Adjust accordingly.
31871 (set_value_range_to_nonnegative): Likewise, remove now unused
31872 overflow_infinity arg.
31873 (vrp_operand_equal_p): Adjust.
31874 (update_value_range): Likewise.
31875 (range_int_cst_singleton_p): Likewise.
31876 (operand_less_p): Likewise.
31877 (compare_values_warnv): Likewise.
31878 (extract_range_for_var_from_comparison_expr): Likewise.
31879 (vrp_int_const_binop): Likewise.
31880 (zero_nonzero_bits_from_vr): Likewise.
31881 (extract_range_from_multiplicative_op_1): Likewise.
31882 (extract_range_from_binary_expr_1): Likewise.
31883 (extract_range_from_unary_expr): Likewise.
31884 (extract_range_from_comparison): Likewise.
31885 (extract_range_basic): Likewise.
31886 (adjust_range_with_scev): Likewise.
31887 (compare_ranges): Likewise.
31888 (compare_range_with_value): Likewise.
31889 (dump_value_range): Likewise.
31890 (test_for_singularity): Likewise, remove strict_overflow_p parameter
31892 (simplify_cond_using_ranges): Adjust.
31894 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31896 * brig-builtins.def: Added a builtin for class_f64.
31897 * builtin-types.def: Added a builtin type needed by class_f64.
31899 2017-05-03 Jason Merrill <jason@redhat.com>
31901 * timevar.def: Add TV_CONSTEXPR.
31903 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31905 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
31907 2017-05-03 Martin Jambor <mjambor@suse.cz>
31909 * ipa-prop.c (ipa_update_after_lto_read): Removed.
31910 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
31911 * ipa-cp.c (ipcp_propagate_stage): Do not call
31912 ipa_update_after_lto_read.
31913 * ipa-inline.c (ipa_inline): Likewise.
31915 2017-05-03 Martin Jambor <mjambor@suse.cz>
31917 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
31918 tag. Added a default constructor and a destructor.
31919 (ipa_edge_args_sum_t): New class;
31920 (ipa_edge_args_sum): Declare.
31921 (ipa_edge_args_vector): Remove declaration.
31922 (IPA_EDGE_REF): Use ipa_edge_args_sum.
31923 (ipa_free_edge_args_substructures): Remove declaration.
31924 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
31925 (ipa_edge_args_info_available_for_edge_p): Likewise.
31926 * ipa-prop.c (ipa_edge_args_vector): Removed.
31927 (edge_removal_hook_holder): Likewise.
31928 (edge_duplication_hook_holder): Likewise.
31929 (ipa_edge_args_sum): New variable.
31930 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
31931 ipa_edge_args_vector.
31932 (ipa_free_edge_args_substructures): Likewise.
31933 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
31934 ipa_edge_args_vector.
31935 (ipa_edge_removal_hook): Turned into method
31936 ipa_edge_args_sum_t::remove.
31937 (ipa_edge_duplication_hook): Turned into method
31938 ipa_edge_args_sum_t::duplicate.
31939 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
31940 registering edge hooks.
31941 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
31942 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
31943 ipa_edge_args_sum instead of ipa_edge_args_vector.
31944 * ipa-profile.c (ipa_profile): Likewise.
31946 2017-05-03 Martin Jambor <mjambor@suse.cz>
31948 * symbol-summary.h (function_summary): New method exists.
31949 (function_summary::symtab_removal): Deallocate through release.
31950 (call_summary): New class.
31951 (gt_ggc_mx): New overload.
31952 (gt_pch_nx): Likewise.
31953 (gt_pch_nx): Likewise.
31955 2017-05-03 Jeff Law <law@redhat.com>
31957 PR tree-optimization/78496
31958 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
31959 from simplify_cond_using_ranges. Split off code to walk
31960 backwards through casts into ...
31961 (simplify_cond_using_ranges_2): New function.
31962 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
31963 (execute_vrp): After identifying jump threads, call
31964 simplify_cond_using_ranges_2.
31966 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
31969 * ipa-inline.h (inline_summary): Add ctor.
31970 (create_ggc): Do not use ggc_cleared_alloc.
31972 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
31973 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31975 * gcc.c (handle_braces): Support escaping in switch matching
31977 * doc/invoke.texi (Spec Files): Document it.
31978 Remove superfluous @code markup in items.
31980 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31982 * diagnostic-show-locus.c (struct column_range): New struct.
31983 (get_affected_columns): New function.
31984 (get_printed_columns): New function.
31985 (struct correction): New struct.
31986 (correction::ensure_capacity): New function.
31987 (correction::ensure_terminated): New function.
31988 (struct line_corrections): New struct.
31989 (line_corrections::~line_corrections): New dtor.
31990 (line_corrections::add_hint): New function.
31991 (layout::print_trailing_fixits): Reimplement in terms of the new
31993 (selftest::test_overlapped_fixit_printing): New function.
31994 (selftest::diagnostic_show_locus_c_tests): Call it.
31996 2017-05-03 Nathan Sidwell <nathan@acm.org>
31998 Canonicalize canonical type hashing
31999 * tree.h (type_hash_canon_hash): Declare.
32000 * tree.c (type_hash_list, attribute_hash_list): Move into
32001 type_hash_canon_hash.
32002 (build_type_attribute_qual_variant): Break out hash code calc into
32003 type_hash_canon_hash.
32004 (type_hash_canon_hash): New. Generic type hash computation.
32005 (build_range_type_1, build_array_type_1, build_function_type,
32006 build_method_type_directly, build_offset_type, build_complex_type,
32007 make_vector_type): Call it.
32009 2017-05-03 Richard Biener <rguenther@suse.de>
32011 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
32012 When all DRs have unknown misaligned do not always peel
32013 when there is a store but apply the same costing model as if
32014 there were only loads.
32016 2017-05-03 Richard Biener <rguenther@suse.de>
32019 PR tree-optimization/80492
32020 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32021 compare_base_decls returning dont-know properly.
32023 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
32025 * config/arm/iterators.md (CCSI): New mode iterator.
32026 (arch): New mode attribute.
32027 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
32028 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
32029 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
32030 code iterator for success result mode.
32031 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
32032 the corresponding new insn generators.
32034 2017-05-03 Bin Cheng <bin.cheng@arm.com>
32037 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32038 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32040 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
32042 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
32043 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
32044 (DDR_A): Wrap DDR argument in brackets.
32045 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
32046 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
32047 (DDR_REVERSED_P): Likewise.
32049 2017-05-03 Jakub Jelinek <jakub@redhat.com>
32051 PR tree-optimization/79472
32052 * tree-switch-conversion.c (struct switch_conv_info): Add
32053 contiguous_range and default_case_nonstandard fields.
32054 (collect_switch_conv_info): Compute contiguous_range and
32055 default_case_nonstandard fields, don't clear final_bb if
32056 contiguous_range and only the default case doesn't have the required
32058 (check_all_empty_except_final): Set default_case_nonstandard instead
32059 of failing if contiguous_range and the default case doesn't have empty
32061 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
32062 and only the default case doesn't have the required constants. Skip
32064 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
32065 if default_case_nonstandard.
32066 (build_constructors): Build constant 1 just once. Assert that default
32067 values aren't inserted in between cases if contiguous_range. Skip
32069 (build_arrays): Skip virtual phis.
32070 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
32071 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
32072 Handle virtual phis.
32073 (gen_inbound_check): Handle default_case_nonstandard case.
32074 (process_switch): Adjust check_final_bb caller. Call
32075 gather_default_values with the first non-default case instead of
32076 default case if default_case_nonstandard.
32078 2017-05-02 Nathan Sidwell <nathan@acm.org>
32080 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
32081 check. Fix formatting.
32083 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
32085 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
32086 errors when comparing specialized and unspecialized times.
32088 2017-05-02 David Malcolm <dmalcolm@redhat.com>
32090 * diagnostic-show-locus.c
32091 (layout::should_print_annotation_line_p): Make private.
32092 (layout::print_annotation_line): Make private.
32093 (layout::annotation_line_showed_range_p): Make private.
32094 (layout::show_ruler): Make private.
32095 (layout::print_source_line): Make private. Pass in line and
32096 line_width, rather than calling location_get_source_line. Drop
32098 (layout::print_leading_fixits): New method.
32099 (layout::print_any_fixits): Rename to...
32100 (layout::print_trailing_fixits): ...this, and make private.
32101 Don't print newline fixits.
32102 (diagnostic_show_locus): Move logic for printing one row into...
32103 (layout::print_line): ...this new function. Move the
32104 location_get_source_line call and error-handling from
32105 print_source_line to here. Call print_leading_fixits, and rename
32106 print_any_fixits to print_trailing_fixits.
32107 (selftest::test_fixit_insert_containing_newline): Update now that
32108 newlines are partially supported.
32109 (selftest::test_fixit_insert_containing_newline_2): New test.
32110 (selftest::test_fixit_replace_containing_newline): Update comments.
32111 (selftest::diagnostic_show_locus_c_tests): Call the new test.
32112 * edit-context.c (class added_line): New class.
32113 (class edited_line): Describe newline handling in comment.
32114 (edited_line::actually_edited_p): New method.
32115 (edited_line::print_content): Delete redundant decl.
32116 (edited_line::m_predecessors): New field.
32117 (edited_file::print_content): Call edited_line::print_content.
32118 (edited_file::print_diff): Update to support newlines.
32119 (edited_file::print_diff_hunk): Likewise.
32120 (edited_file::print_run_of_changed_lines): New function.
32121 (edited_file::print_diff_line): Convert to...
32122 (print_diff_line): ...this.
32123 (edited_file::get_effective_line_count): New function.
32124 (edited_line::edited_line): Initialize new field m_predecessors.
32125 (edited_line::~edited_line): Clean up m_predecessors.
32126 (edited_line::apply_fixit): Handle newlines.
32127 (edited_line::get_effective_line_count): New function.
32128 (edited_line::print_content): New function.
32129 (edited_line::print_diff_lines): New function.
32130 (selftest::test_applying_fixits_insert_containing_newline): New
32132 (selftest::test_applying_fixits_replace_containing_newline): New
32134 (selftest::insert_line): New function.
32135 (selftest::test_applying_fixits_multiple_lines): Add example of
32137 (selftest::edit_context_c_tests): Call the new tests.
32139 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32141 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
32142 parameter cand. Update dump information.
32143 (get_computation_cost): Update uses.
32145 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32147 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
32148 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
32149 (get_computation_at, rewrite_use_address): Update use of
32150 get_computation_aff.
32152 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32154 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
32155 (get_computation): Delete.
32156 (get_computation_cost): Implement like get_computation_cost_at.
32157 Use get_computation_at.
32158 (get_computation_cost_at): Delete.
32159 (rewrite_use_nonlinear_expr): Use get_computation_at.
32160 (rewrite_use_compare, remove_unused_ivs): Ditto.
32162 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32164 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
32166 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32168 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
32169 (ivopts_global_cost_for_size): Rename parameter and update uses.
32170 (iv_ca_recount_cost): Update uses.
32171 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
32172 candidates seperately in n_invs and n_cands.
32173 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
32175 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32177 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
32178 (find_inv_vars_cb): New.
32179 (find_depends): Renamed to ...
32180 (find_inv_vars): ... this.
32181 (add_candidate_1, force_var_cost): Call find_inv_vars.
32182 (split_address_cost, determine_group_iv_cost_cond): Ditto.
32184 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32186 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
32187 inv_vars. Add inv_exprs.
32188 (struct iv_cand): Rename depends_on to inv_vars.
32189 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
32190 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
32191 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
32192 (dump_cand): Dump inv_vars.
32193 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
32194 (record_invariant, find_depends, add_candidate_1): Ditto.
32195 (set_group_iv_cost, force_var_cost): Ditto.
32196 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
32197 (get_computation_cost_at, get_computation_cost): Ditto.
32198 (determine_group_iv_cost_generic): Ditto.
32199 (determine_group_iv_cost_address): Ditto.
32200 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
32201 (determine_group_iv_costs): Ditto.
32202 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
32203 (iv_ca_set_remove_invariants): Renamed to ...
32204 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
32205 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
32206 (iv_ca_set_add_invariants): Renamed to ...
32207 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
32208 (iv_ca_set_cp): Use iv_ca_set_add_invs.
32209 (iv_ca_has_deps): Support inv_vars and inv_exprs.
32210 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
32211 (create_new_ivs): Remove useless dump.
32213 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32215 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
32217 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
32218 (iv_ca_set_no_cp, create_new_iv): Ditto.
32220 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32222 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32224 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32226 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
32227 function tree_check2.
32229 2017-05-02 Martin Liska <mliska@suse.cz>
32231 * doc/gcov.texi: Add missing preposition.
32232 * gcov.c (function_info::function_info): Properly fill up
32233 all member variables.
32235 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32237 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
32239 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32241 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
32243 2017-05-02 Martin Liska <mliska@suse.cz>
32246 * lto-streamer-in.c (lto_read_tree_1): Remove
32247 LTO_STREAMER_DEBUG.
32248 * lto-streamer.c (struct tree_hash_entry): Likewise.
32249 (struct tree_entry_hasher): Likewise.
32250 (tree_entry_hasher::hash): Likewise.
32251 (tree_entry_hasher::equal): Likewise.
32252 (lto_streamer_init): Likewise.
32253 (lto_orig_address_map): Likewise.
32254 (lto_orig_address_get): Likewise.
32255 (lto_orig_address_remove): Likewise.
32256 * lto-streamer.h: Likewise.
32257 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
32258 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
32260 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
32262 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
32263 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
32264 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
32265 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
32266 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
32267 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
32268 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
32269 (mm_maskz_sub_ss): New intrinsics.
32270 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32271 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32272 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
32273 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
32274 (__builtin_ia32_subss_mask_round): New builtins.
32275 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32276 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32277 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
32279 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
32280 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32282 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32285 2017-05-02 Martin Jambor <mjambor@suse.cz>
32287 PR tree-optimization/78687
32288 * tree-sra.c (access): New field parent.
32289 (process_subtree_disqualification): New function.
32290 (disqualify_candidate): Call it.
32291 (build_accesses_from_assign): Reset write flag if creating an
32293 (build_access_subtree): Fill in parent field and also prpagate
32294 down grp_write flag.
32295 (create_artificial_child_access): New parameter set_grp_write, set
32296 grp_write to its value.
32297 (propagate_subaccesses_across_link): Also propagate grp_write flag
32299 (propagate_all_subaccesses): Push the closest parent back to work
32300 queue if add_access_to_work_queue returned true.
32302 2017-05-02 Richard Biener <rguenther@suse.de>
32304 * common.opt (fstrict-overflow): Alias negative to fwrapv.
32305 * doc/invoke.texi (fstrict-overflow): Remove all traces of
32306 -fstrict-overflow documentation.
32307 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
32308 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
32309 flag_strict_overflow.
32310 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
32311 * lto-opts.c (lto_write_options): Do not stream it.
32312 * lto-wrapper.c (merge_and_complain): Do not handle it.
32313 * opts.c (default_options_table): Do not set -fstrict-overflow.
32314 (finish_options): Likewise do not clear it when sanitizing.
32315 * simplify-rtx.c (simplify_const_relational_operation): Do not
32316 test flag_strict_overflow.
32318 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32320 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
32321 using enabled attribute.
32322 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
32323 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
32324 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
32325 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
32326 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
32327 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
32328 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
32329 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
32330 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
32331 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
32333 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32335 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
32337 2017-05-02 Richard Biener <rguenther@suse.de>
32339 PR tree-optimization/80591
32341 2017-04-10 Richard Biener <rguenther@suse.de>
32343 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32346 2017-05-02 Richard Biener <rguenther@suse.de>
32348 PR tree-optimization/80549
32349 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
32350 (cleanup_tree_cfg_noloop): Create forwarders to known loop
32351 headers if they do not have a preheader.
32353 2017-05-02 Martin Liska <mliska@suse.cz>
32356 * common.opt: Fix typo.
32357 * doc/invoke.texi: Likewise.
32359 2017-05-01 Jan Beulich <jbeulich@suse.com>
32361 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
32362 swapping, add (x,x,m,x,n) alternative.
32364 2017-05-01 Nathan Sidwell <nathan@acm.org>
32366 * calls.c (combine_pending_stack_adjustment_and_call): Remove
32367 unnecessary unadjusted_alignment check.
32369 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32372 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
32374 * gimplify.c (gimplify_cilk_detach): New function.
32375 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
32376 * tree-core.h: Document EXPR_CILK_SPAWN.
32377 * tree.h (EXPR_CILK_SPAWN): Define.
32379 2017-05-01 David Malcolm <dmalcolm@redhat.com>
32381 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
32382 to use new fixit_hint representation, using the "replace" logic.
32383 (get_line_span_for_fixit_hint): Likewise.
32384 (layout::print_any_fixits): Likewise.
32385 (selftest::test_one_liner_many_fixits): Rename to...
32386 (selftest::test_one_liner_many_fixits_1): ...this, and update
32387 comment and expected output to reflect that the multiple fix-it
32388 hints are now consolidated into one insertion.
32389 (selftest::test_one_liner_many_fixits_2): New test.
32390 (selftest::test_diagnostic_show_locus_one_liner): Update for
32392 (selftest::test_fixit_consolidation): Update for fix-it API
32394 * diagnostic.c (print_parseable_fixits): Likewise.
32395 * edit-context.c (edited_line::m_line_events): Convert from
32396 auto_vec <line_event *> to auto_vec <line_event>.
32397 (class line_event): Convert from abstract base class to a concrete
32398 class, taking over the role of replace_event.
32399 (class insert_event): Delete.
32400 (class replace_event): Rename to class line_event. Convert to
32402 (edit_context::add_fixits): Reimplement.
32403 (edit_context::apply_insert): Delete.
32404 (edit_context::apply_replace): Rename to...
32405 (edit_context::apply_fixit): ...this. Convert to half-open range.
32406 (edited_file::apply_insert): Delete.
32407 (edited_file::apply_replace): Rename to...
32408 (edited_file::apply_fixit): ...this.
32409 (edited_line::~edited_line): Drop deletion of events.
32410 (edited_line::apply_insert): Delete.
32411 (edited_line::apply_replace): Rename to...
32412 (edited_line::apply_fixit): ...this. Convert to half-open range.
32413 Update for change to type of m_line_events.
32414 * edit-context.h (edit_context::apply_insert): Delete.
32415 (edit_context::apply_replace): Rename to...
32416 (edit_context::apply_fixit): ...this.
32418 2017-05-01 Martin Sebor <msebor@redhat.com>
32420 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
32423 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
32426 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
32427 __get_cpuid_max returns 0.
32428 (__get_cpuid_count): Ditto.
32430 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
32432 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
32433 replacement expression is another instance of one of its arguments.
32435 2017-05-01 Jakub Jelinek <jakub@redhat.com>
32438 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
32439 check for stack push/pop autoinc.
32440 * config/i386/i386.c (ix86_agi_dependent): Return false
32441 if the only reason why modified_in_p returned true is that
32442 addr is SP based and set_insn is a push or pop.
32444 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32446 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
32449 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32452 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
32453 (account_size_time): Use two predicates - exec_pred and
32455 (evaluate_conditions_for_known_args): Compute both clause and
32457 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
32458 (inline_summary_t::duplicate): Update.
32459 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
32461 (compute_inline_parameters): Likewise.
32462 (estimate_edge_size_and_time): Update caluclation of time.
32463 (estimate_node_size_and_time): Compute both time and nonspecialized
32465 (estimate_ipcp_clone_size_and_time): Update.
32466 (inline_merge_summary): Update.
32467 (do_estimate_edge_time): Update.
32468 (do_estimate_edge_size): Update.
32469 (do_estimate_edge_hints): Update.
32470 (inline_read_section, inline_write_summary): Stream both new predicates.
32471 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
32473 (compute_inlined_call_time): Cleanup.
32474 (big_speedup_p): Update.
32475 (edge_badness): Update.
32476 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
32477 (size_time_entry): Replace predicate by exec_predicate and
32478 nonconst_predicate.
32479 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
32480 (estimate_edge_time): Return also nonspec_time.
32481 (reset_edge_growth_cache): Update.
32483 2017-04-29 Jakub Jelinek <jakub@redhat.com>
32485 PR rtl-optimization/80491
32486 * ifcvt.c (noce_process_if_block): When looking for x setter
32487 with missing else_bb, don't check only the insn right before
32488 cond_earliest, but look for the last insn that x is modified in
32489 within the same bb.
32491 PR rtl-optimization/80491
32492 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
32494 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
32496 PR tree-optimization/80487
32497 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
32499 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32501 PR tree-optimization/79697
32502 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
32503 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
32504 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
32506 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
32507 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
32509 2017-04-28 Martin Sebor <msebor@redhat.com>
32511 PR tree-optimization/80523
32512 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
32513 (init_target_to_host_charmap, target_to_host, target_strtol10): New
32515 (maybe_warn, format_directive, parse_directive): Use new functions.
32516 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
32518 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
32520 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
32522 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
32524 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
32525 target_header_dir): Set correctly.
32526 * configure: Regenerated.
32527 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32528 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32529 instead of SYSTEM_HEADER_DIR.
32531 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
32533 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
32534 (estimate_local_effects): Likewise.
32535 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
32536 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
32537 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
32538 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
32539 do_estimate_edge_time, estimate_edge_time): Likewise.
32540 * ipa-inline-analysis.c (estimate_node_size_and_time,
32541 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
32542 (estimate_time_after_inlining): Remove.
32544 2017-04-28 Martin Liska <mliska@suse.cz>
32546 * doc/gcov.texi: Enhance documentation of gcov.
32548 2017-04-28 Martin Liska <mliska@suse.cz>
32550 * doc/gcov.texi: Sort options in alphabetic order.
32551 * doc/gcov-dump.texi: Likewise.
32552 * doc/gcov-tool.texi: Likewise.
32553 * gcov.c (print_usage): Likewise.
32554 * gcov-dump.c (print_usage): Likewise.
32555 * gcov-tool.c (print_merge_usage_message): Likewise.
32556 (print_rewrite_usage_message): Likewise.
32557 (print_overlap_usage_message): Likewise.
32559 2017-04-28 Martin Liska <mliska@suse.cz>
32561 PR gcov-profile/53915
32562 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
32564 2017-04-28 Martin Liska <mliska@suse.cz>
32566 PR gcov-profile/79891
32567 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
32568 is marked by compiler as living on a line.
32569 (get_cycles_count): Remove usage of the union.
32570 (output_intermediate_file): Likewise.
32571 (find_source): Fix GNU coding style.
32572 (accumulate_line_counts): Remove old non-all block mode.
32573 (output_lines): Remove usage of the union.
32574 * profile.c (output_location): Include all BBs, even if
32575 belonging to a same line (and file) as a previous BB.
32577 2017-04-28 Martin Liska <mliska@suse.cz>
32579 * gcov.c (process_args): Handle new argument 'w'.
32580 (read_graph_file): Assign ID to BBs.
32581 (output_branch_count): Display BB # if verbose flag is set.
32582 (output_lines): Likewise for arcs.
32583 (print_usage): Add '--verbose' option help.
32584 * doc/gcov.texi: Document --verbose (-w) option.
32586 2017-04-28 Martin Liska <mliska@suse.cz>
32588 * gcov.c (struct block_location_info): New struct.
32589 (process_file): Fill up the new structure.
32590 (read_graph_file): Replace usage of encoding by the newly added
32592 (add_line_counts): Likewise.
32593 (accumulate_line_counts): Remove usage of the union.
32594 (function_info::function_info): New function.
32595 (function_info::~function_info): Likewise.
32596 (process_file): Call delete instead of release_function.
32597 (release_function): Release the function.
32598 (release_structures): Call delete instead of release_function.
32599 (solve_flow_graph): Replace usage of num_blocks.
32600 (find_exception_blocks): Likewise.
32601 (output_lines): Fix GNU coding style.
32603 2017-04-28 Martin Liska <mliska@suse.cz>
32606 * coverage.c (coverage_remove_note_file): New function.
32607 * coverage.h: Declare the function.
32608 * toplev.c (finalize): Clean if an error has been seen.
32610 2017-04-28 Martin Liska <mliska@suse.cz>
32612 PR gcov-profile/80031
32613 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
32614 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
32615 * gcov.c (read_graph_file): Read just number of blocks.
32616 * profile.c (branch_prob): Do not stream 0 flags per a basic
32619 2017-04-28 Martin Liska <mliska@suse.cz>
32621 * gcov-dump.c (tag_*): Add new argument to declarations.
32622 (dump_gcov_file): Likewise.
32623 (tag_blocks): Add and use new argument depth.
32624 (tag_arcs): Likewise.
32625 (tag_lines): Likewise.
32626 (tag_counters): Likewise.
32627 (tag_summary): Likewise.
32628 (dump_working_sets): Use depth to do a proper indentation.
32630 2017-04-28 Jakub Jelinek <jakub@redhat.com>
32633 * cgraph.h (symtab_node::debug_symtab): No longer inline.
32634 * symtab.c (symtab_node::debug_symtab): Move definition here.
32636 2017-04-28 Richard Biener <rguenther@suse.de>
32638 * lto-streamer.h (LTO_major_version): Bump to 7.
32640 2017-04-28 Richard Biener <rguenther@suse.de>
32642 * tree-vrp.c (assert_info): New struct.
32643 (add_assert_info): New helper.
32644 (register_edge_assert_for_2): Refactor to add asserts to a vector
32646 (register_edge_assert_for_1): Likewise.
32647 (register_edge_assert_for): Likewise.
32648 (finish_register_edge_assert_for): New helper actually registering
32649 asserts where live on edge.
32650 (find_conditional_asserts): Adjust.
32651 (find_switch_asserts): Likewise.
32652 (evrp_dom_walker::try_find_new_range): Generalize.
32653 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
32655 2017-04-27 Marek Polacek <polacek@redhat.com>
32658 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
32659 arg10 and arg11 to itype.
32661 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
32663 * doc/extend.texi (Object Size Checking): Improve grammar.
32665 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
32668 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
32669 that the logic for permitting reciprocal estimates matches that
32672 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32675 * tree.c (type_cache_hasher::equal): Only compare
32676 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
32677 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
32678 non-aggregate element types.
32679 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
32680 about the flag on ARRAY_TYPEs in the comment, formatting fix.
32682 2017-04-27 Richard Biener <rguenther@suse.de>
32684 PR middle-end/80533
32685 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
32686 stripping ARRAY_REFs from MEM_EXPR make sure we're not
32687 keeping a reference to a trailing array.
32689 2017-04-27 Richard Biener <rguenther@suse.de>
32691 PR middle-end/80539
32692 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
32693 being in loop-closed SSA form conservatively.
32694 (chrec_fold_multiply_poly_poly): Likewise.
32696 2017-04-27 Tamar Christina <tamar.christina@arm.com>
32698 PR middle-end/79665
32699 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
32700 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
32702 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32705 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
32706 (aarch64_function_arg_alignment): Return unsigned int again, but still
32707 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
32708 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32709 Don't emit -Wpsabi note.
32710 (aarch64_function_arg_boundary): Likewise.
32711 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32714 2017-04-26 Nathan Sidwell <nathan@acm.org>
32716 * tree.h (crc32_unsigned_n): Declare.
32717 (crc32_unsigned, crc32_unsigned): Make inline.
32718 * tree.c (crc32_unsigned_bits): Replace with ...
32719 (crc32_unsigned_n): ... this.
32720 (crc32_unsigned, crc32_byte): Remove.
32721 (crc32_string): Remove unnecessary braces.
32723 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32725 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
32726 * ipa-inline-analysis.c (MAX_TIME): Remove.
32727 (account_size_time): Use sreal for time.
32728 (dump_inline_summary): Update.
32729 (estimate_function_body_sizes): Update.
32730 (estimate_edge_size_and_time): Update.
32731 (estimate_calls_size_and_time): Update.
32732 (estimate_node_size_and_time): Update.
32733 (inline_merge_summary): Update.
32734 (inline_update_overall_summary): Update.
32735 (estimate_time_after_inlining): Update.
32736 (inline_read_section): Update.
32737 (inline_write_summary): Update.
32738 * ipa-inline.c (compute_uninlined_call_time): Update.
32739 (compute_inlined_call_time): Update.
32740 (recursive_inlining): Update.
32741 (inline_small_functions): Update.
32742 (dump_overall_stats): Update.
32743 * ipa-inline.h: Include sreal.h.
32744 (size_time_entry): Turn time to sreal.
32745 (inline_summary): Turn self_time nad time to sreal.
32747 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32749 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
32751 (sreal::stream_out, sreal::stream_in): New.
32752 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
32754 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32756 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
32759 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
32762 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
32764 (dimode_scalar_chain::compute_convert_gain): Ditto.
32765 (dimode_scalar_chain::make_vector_copies): Ditto.
32766 (dimode_scalar_chain::convert_reg): Ditto.
32767 (dimode_scalar_chain::convert_insn): Ditto.
32768 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
32769 (VI248_AVX512BW_1): New mode iterator.
32770 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
32771 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
32774 2017-04-25 Martin Sebor <msebor@redhat.com>
32776 PR tree-optimization/80497
32777 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
32778 constants are representable in HOST_WIDE_INT.
32779 (parse_directive): Ditto.
32781 2017-04-25 Martin Sebor <msebor@redhat.com>
32784 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
32785 (new_zero_array): Adjust signature.
32786 (dom_info::dom_init): Used unsigned rather that size_t.
32787 (dom_info::dom_info): Same.
32789 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32790 Jakub Jelinek <jakub@redhat.com>
32793 * config/arm/arm.c: Include gimple.h.
32794 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32795 returns negative, increment ncrn only if it returned positive.
32796 (arm_needs_doubleword_align): Return int instead of bool,
32797 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
32798 members, but if there is any such non-FIELD_DECL
32799 > PARM_BOUNDARY aligned decl, return -1 instead of false.
32800 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32801 returns negative, increment nregs only if it returned positive.
32802 (arm_setup_incoming_varargs): Likewise.
32803 (arm_function_arg_boundary): Emit -Wpsabi note if
32804 arm_needs_doubleword_align returns negative, return
32805 DOUBLEWORD_ALIGNMENT only if it returned positive.
32807 2017-04-25 Marek Polacek <polacek@redhat.com>
32810 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32811 first argument to type.
32813 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
32816 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
32817 type checks to test for compatibility instead of equality.
32819 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32820 Jakub Jelinek <jakub@redhat.com>
32823 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
32825 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
32826 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
32827 the alignment computation, but return their maximum in warn_alignment.
32828 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32829 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
32831 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
32832 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32835 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32837 * config/arc/simdext.md (dmpyh): Fix typo.
32839 2017-04-25 Richard Biener <rguenther@suse.de>
32841 PR tree-optimization/80492
32842 * alias.c (compare_base_decls): Handle registers with asm
32843 specification conservatively.
32844 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32845 compare_base_decls returning dont-know properly.
32847 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32849 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
32850 (legitimate_offset_address_p): New function.
32851 (arc_legitimate_address_p): Use above function.
32853 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32855 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
32857 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32859 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
32860 ACCH registers whenever they are available.
32862 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32864 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
32865 double regs fix when not used.
32867 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32869 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
32871 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
32872 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
32874 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32876 * config/arc/arc.c (arc_output_addsi): Check for h-register class
32877 when emitting short ADD instructions.
32879 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32881 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
32883 (cmpsi_cc_c_insn): Likewise.
32884 (cbranchsi4_scratch): Compute proper instruction length using
32885 compact_hreg_operand.
32886 * config/arc/predicates.md (compact_hreg_operand): New predicate.
32888 2017-04-25 Richard Biener <rguenther@suse.de>
32890 PR middle-end/80509
32891 * passes.c (pass_manager::pass_manager): Initialize
32892 m_name_to_pass_map.
32894 2017-04-25 Richard Biener <rguenther@suse.de>
32896 PR tree-optimization/79201
32897 * tree-ssa-sink.c (statement_sink_location): Handle calls.
32899 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32902 * config/s390/vector.md: Split MEM->GPR vector moves for
32903 non-s_operand addresses.
32905 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32908 * config/s390/predicates.md (reload_const_wide_int_operand): New
32910 * config/s390/s390.md ("movti"): Remove d/P alternative.
32911 ("movti_bigconst"): New pattern definition.
32913 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32916 * s390-protos.h (s390_expand_cs_hqi): Removed.
32917 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
32918 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
32919 modes as well as CCZ1mode and CCZmode.
32920 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
32921 signature of s390_emit_compare_and_swap.
32922 (s390_expand_cs_hqi): Likewise, make static.
32923 (s390_expand_cs_tdsi): Generate an explicit compare before trying
32924 compare-and-swap, in some cases.
32925 (s390_expand_cs): Wrapper function.
32926 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
32928 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
32929 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
32930 patterns for small and large integers. Forbid symref memory operands.
32931 Move expander to s390.c. Require cc register.
32932 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
32933 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
32934 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
32935 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
32936 symref memory operands. Remove CC mode and call s390_match_ccmode
32938 ("atomic_exchange<mode>"): Allow and implement all integer modes.
32940 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32942 * config/s390/s390.md (define_peephole2): New peephole to help
32943 combining the load-and-test pattern with volatile memory.
32945 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32947 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
32948 with CCZmode for TARGET_Z196.
32950 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32952 PR rtl-optimization/80501
32953 * combine.c (make_compound_operation_int): Set subreg_code to SET
32954 even for AND with mask of the sign bit of mode.
32956 PR rtl-optimization/80500
32957 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
32958 sum's initial value.
32960 2017-04-25 Julian Brown <julian@codesourcery.com>
32961 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32963 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
32965 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
32967 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
32969 2017-04-25 Julian Brown <julian@codesourcery.com>
32970 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32972 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
32973 (thunderx2t99_sha): New Reservation.
32975 2017-04-25 Julian Brown <julian@codesourcery.com>
32976 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32978 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
32979 type for 1-element load.
32981 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
32983 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
32985 2017-04-24 Martin Jambor <mjambor@suse.cz>
32987 PR tree-optimization/80293
32988 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
32989 char arrays not totally scalarizable if it is false.
32990 (analyze_all_variable_accesses): Pass correct value in the new
32991 parameter. Add a statistics counter.
32993 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
32995 PR middle-end/79931
32996 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
32998 2017-04-24 Richard Biener <rguenther@suse.de>
33000 PR tree-optimization/80494
33001 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
33002 out for complex types.
33004 2017-04-24 Richard Biener <rguenther@suse.de>
33006 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
33007 * tree-ssa-sccvn.c (print_scc): Print SCC size.
33008 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
33009 (DFS): Adjust and never fail.
33010 (sccvn_dom_walker::fail): Remove.
33011 (sccvn_dom_walker::before_dom_children): Adjust.
33012 (run_scc_vn): Likewise and never fail.
33013 * tree-ssa-pre.c (pass_pre::execute): Adjust.
33014 (pass_fre::execute): Likewise.
33016 2017-04-24 Richard Biener <rguenther@suse.de>
33018 PR tree-optimization/79725
33019 * tree-ssa-sink.c (statement_sink_location): Return whether
33020 failure reason was zero uses. Move that check later.
33021 (sink_code_in_bb): Deal with zero uses by removing the stmt
33024 2017-04-24 Richard Biener <rguenther@suse.de>
33027 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
33028 pointer-based references.
33030 2017-04-24 Richard Biener <rguenther@suse.de>
33033 * pass_manager.h (pass_manager::operator new): Remove.
33034 (pass_manager::operator delete): Likewise.
33035 * passes.c (pass_manager::operator new): Remove.
33036 (pass_manager::operator delete): Likewise.
33037 (pass_manager::pass_manager): Zero individual pass members.
33039 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
33042 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
33043 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
33044 Check "XEXP (src, 1)" operand here.
33045 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
33046 Check "XEXP (src, 1)" operand here.
33047 (dimode_scalar_chain::make_vector_copies): Detect count register
33048 of a shift instruction. Zero extend count register from QImode
33049 to DImode to satisfy vector shift pattern count operand predicate.
33050 Substitute vector shift count operand with a DImode copy.
33051 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
33054 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
33056 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
33057 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33058 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
33059 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33060 (UNSPEC_NOREX_MEM): Remove definition.
33062 2017-04-21 Richard Biener <rguenther@suse.de>
33064 PR tree-optimization/79547
33065 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
33066 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
33067 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
33068 without any constraints.
33070 2017-04-21 Richard Biener <rguenther@suse.de>
33072 PR tree-optimization/78847
33073 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
33075 2017-04-21 Richard Biener <rguenther@suse.de>
33077 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
33078 (build_distinct_type_copy): Likewise.
33079 (build_variant_type_copy): Likewise.
33080 * tree.c (build_qualified_type): Pass down mem-stat info.
33081 (build_distinct_type_copy): Likewise.
33082 (build_variant_type_copy): Likewise.
33084 2017-04-21 Richard Biener <rguenther@suse.de>
33086 PR tree-optimization/80237
33087 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
33089 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
33090 for a simplified result.
33092 2016-04-21 Richard Biener <rguenther@suse.de>
33094 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
33095 sth as strict as a simple_iv but a chrec without symbols and an
33096 operand defined in the loop we are peeling (and not some subloop).
33097 (propagate_constants_for_unrolling): Propagate all constants.
33099 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33102 * config/i386/i386.c (print_reg): Remove assert for disalowed
33103 regno values, call output_operand_lossage instead.
33105 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33108 * config/i386/constraints.md (Yc): New register constraint.
33109 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
33110 Use Yc constraint for alternative 2 of operand 0. Remove
33111 preferred_for_speed attribute.
33113 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33115 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
33116 lastprivate clauses in SIMT case.
33118 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
33120 * doc/invoke.texi (-Wextra-semi): Document new warning option.
33122 2017-04-20 Richard Biener <rguenther@suse.de>
33124 PR tree-optimization/57796
33125 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
33126 as N scalar stores.
33127 (vect_model_load_cost): Cost gathers as N scalar loads.
33129 2017-04-20 Richard Biener <rguenther@suse.de>
33131 * ggc-page.c (ggc_allocated_p): Rename to ...
33132 (safe_lookup_page_table_entry): ... this and return the lookup
33134 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
33136 2017-04-20 Richard Biener <rguenther@suse.de>
33138 PR tree-optimization/80453
33139 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
33140 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
33141 from the conditions.
33142 (vn_phi_eq): Pass them down.
33143 (vn_phi_lookup): Record them.
33144 (vn_phi_insert): Likewise.
33146 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
33148 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
33149 uninitialized variable warning to avoid buffer overrun.
33151 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33154 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
33155 is suppressed for '{ 0 }' in C.
33157 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33159 * BASE-VER: Set to 8.0.0.
33161 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
33163 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
33164 priority .init_array and .fini_array section with SECTION_NOTYPE
33167 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33169 PR middle-end/80423
33170 * tree.h (build_array_type): Add typeless_storage default argument.
33171 * tree.c (type_cache_hasher::equal): Also compare
33172 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
33173 (build_array_type): Add typeless_storage argument, set
33174 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
33176 (build_nonshared_array_type): Adjust build_array_type_1 caller.
33177 (build_array_type): Likewise. Add typeless_storage argument.
33179 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33180 Jakub Jelinek <jakub@redhat.com>
33182 PR tree-optimization/80426
33183 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
33184 operation on symbolic operands, also compute the overflow for the
33185 invariant part when the operation degenerates into a negation.
33187 2017-04-19 Jakub Jelinek <jakub@redhat.com>
33190 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
33191 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
33194 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
33196 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
33199 * config/avr/avr.c (tree.h): Include it.
33200 (cgraph.h): Include it.
33201 (avr_encode_section_info): Don't warn for uninitialized progmem
33202 variable if it's just an alias.
33204 2017-04-19 Richard Biener <rguenther@suse.de>
33207 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
33208 when needed by AutoPGO.
33210 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
33213 * doc/lto.texi: Remove an extra 'that'.
33215 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
33217 PR rtl-optimization/80429
33218 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
33219 are only used in debug insns.
33221 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33222 Vladimir Makarov <vmakarov@redhat.com>
33224 * config/sparc/predicates.md (input_operand): Add comment. Return
33225 true for any memory operand when LRA is in progress.
33226 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
33228 2017-04-18 Jeff Law <law@redhat.com>
33231 * mips.md ({return,simple_return}_internal): Do not overwrite
33234 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33236 PR tree-optimization/80443
33237 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
33238 instead of adding 1, subtract -1 and similarly instead of subtracting
33241 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
33243 PR rtl-optimization/80357
33244 * haifa-sched.c (tmp_bitmap): New variable.
33245 (model_recompute): Handle duplicate use records.
33246 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
33247 (free_global_sched_pressure_data): Free it.
33249 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
33252 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33253 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33254 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33255 instead of SYSTEM_HEADER_DIR.
33257 2017-04-18 Jeff Law <law@redhat.com>
33259 PR middle-end/80422
33260 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
33261 predecessors after walking up the insn chain.
33263 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33266 * dwarf2out.c (modified_type_die): Try harder not to emit internal
33267 sizetype type into debug info.
33269 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33272 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
33273 unneeded test for TARGET_UPPER_REGS_SF.
33274 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
33276 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33279 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
33280 instead of gsi_after_labels.
33282 2017-04-18 Jeff Law <law@redhat.com>
33284 * regcprop.c (maybe_mode_change): Avoid creating copies of the
33288 2017-04-13 Jeff Law <law@redhat.com>
33289 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33290 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33292 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
33295 * config/avr/avr.c (intl.h): Include it.
33296 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
33298 2017-04-18 Martin Liska <mliska@suse.cz>
33300 PR gcov-profile/78783
33301 * gcov-tool.c (gcov_output_files): Validate that destination
33302 file is either removed by the tool or by a user.
33304 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33305 Guy Benyei <guybe@mellanox.com>
33307 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
33308 block, and do not negate it, the stored id is already negative.
33310 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33312 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
33314 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33317 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
33318 masks of options that should be turned off if the VSX vector
33319 options are turned off.
33320 (OTHER_P8_VECTOR_MASKS): Likewise.
33321 (OTHER_VSX_VECTOR_MASKS): Likewise.
33322 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
33323 rs6000_disable_incompatible_switches to validate no type switches
33325 (rs6000_incompatible_switch): New function to disallow turning on
33326 other vector options if -mno-vsx, -mno-power8-vector, or
33327 -mno-power9-vector are specified.
33329 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33331 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
33333 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33335 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
33336 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
33337 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
33338 (ARG_POINTER_CFA_OFFSET): Likewise.
33340 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33342 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
33343 conditions to take advantage of various optimizations.
33345 2017-04-13 Jeff Law <law@redhat.com>
33347 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33348 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33349 (zero_extendsidi2_dext): Likewise.
33351 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33354 * fold-const.c (fold_ternary_loc): Revert
33355 use op0 instead of fold_convert_loc (loc, type, arg0) part of
33358 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
33360 PR rtl-optimization/80343
33361 * lra-remat.c (update_scratch_ops): Assign original hard reg to
33362 new scratch pseudo.
33364 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
33367 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
33368 to ubsan_encode_value.
33370 2017-04-13 Jeff Law <law@redhat.com>
33372 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
33373 appearing in DEBUG_INSNs.
33375 2017-04-13 Martin Liska <mliska@suse.cz>
33377 PR gcov-profile/80413
33378 * gcov-io.c (gcov_write_string): Copy to buffer just when
33379 allocated size is greater than zero.
33381 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33384 * dwarf2out.c (decls_for_scope): Ignore declarations of
33385 current_function_decl in BLOCK_NONLOCALIZED_VARS.
33387 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
33390 * ipa-visibility.c (non_local_p): Fix typos.
33391 (localize_node): When localizing symbol in same comdat group,
33392 dissolve the group only when we know external symbols are going
33394 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
33396 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33398 PR tree-optimization/79390
33399 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
33400 order does not result in usable sequence, retry with reversed operand
33406 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
33407 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
33408 op0 instead of fold_convert_loc (loc, type, arg0).
33410 2017-04-12 Jeff Law <law@redhat.com>
33412 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
33413 has a delay slot in the generated code.
33415 * config/cris/cris.md (cris_preferred_reload_class): Return
33416 GENNONACR_REGS rather than GENERAL_REGS.
33418 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33421 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
33422 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
33423 signedness of the result type.
33425 2017-04-12 Richard Biener <rguenther@suse.de>
33426 Jeff Law <law@redhat.com>
33428 PR tree-optimization/80359
33429 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
33430 trim stores to TARGET_MEM_REFs.
33432 2017-04-12 Richard Biener <rguenther@suse.de>
33434 PR tree-optimization/79390
33435 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
33436 threading case even more.
33438 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
33441 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
33442 for quad_address_p for TImode, instead of just not indexed_address.
33444 2017-04-12 Richard Biener <rguenther@suse.de>
33445 Bernd Edlinger <bernd.edlinger@hotmail.de>
33447 PR middle-end/79671
33448 * alias.c (component_uses_parent_alias_set_from): Handle
33449 TYPE_TYPELESS_STORAGE.
33450 (get_alias_set): Likewise.
33451 * tree-core.h (tree_type_common): Add typeless_storage flag.
33452 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
33453 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
33454 for types containing members with TYPE_TYPELESS_STORAGE.
33455 (place_field): Likewise.
33456 (layout_type): Likewise for ARRAY_TYPE.
33457 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
33458 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
33459 TYPE_TYPELESS_STORAGE.
33460 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
33462 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33465 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33466 first argument to type.
33468 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33472 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
33473 CONST0_RTX (mode) rather than const0_rtx where appropriate.
33474 (rs6000_expand_binop_builtin): Likewise.
33475 (rs6000_expand_ternop_builtin): Likewise; also add missing
33476 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
33477 vshasigma built-ins.
33478 * doc/extend.texi: Document that vec_xxpermdi's third argument
33479 must be a constant.
33481 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
33483 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
33484 Use shift_const cost parameter when calculating gain of STV shifts.
33486 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
33488 PR rtl-optimization/70478
33489 * lra-constraints.c (process_alt_operands): Check memory for
33490 disfavoring memory insn operand.
33492 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33494 PR middle-end/80100
33495 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
33496 left shift in unsigned HOST_WIDE_INT type.
33498 PR rtl-optimization/80385
33499 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
33500 (not (neg X)) into (plus X -1) for complex or non-integral modes.
33503 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
33504 if they have any depend clauses.
33506 2017-04-11 Martin Liska <mliska@suse.cz>
33509 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
33510 * ipa-split.c (split_function): Create a local comdat symbol
33511 if caller is in a comdat group.
33513 2017-04-11 Martin Liska <mliska@suse.cz>
33516 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
33519 2017-04-11 Martin Sebor <msebor@redhat.com>
33521 PR middle-end/80364
33522 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
33523 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
33525 (directive::set_width, directive::set_precision, format_character):
33527 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
33530 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
33533 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
33534 conflict, set target->arch_name instead of target->cpu_name.
33536 2017-04-11 Richard Biener <rguenther@suse.de>
33538 PR tree-optimization/80374
33539 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
33540 build_zero_cst, remove fold_convertible_p check again.
33542 2017-04-11 Martin Liska <mliska@suse.cz>
33545 * ubsan.c (instrument_object_size): Do not instrument register
33548 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33551 * config/i386/i386-builtin-types.def
33552 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
33553 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
33554 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
33555 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
33556 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
33557 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
33558 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
33559 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
33560 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
33561 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
33562 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
33563 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
33564 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
33565 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
33566 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
33567 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
33568 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
33569 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
33570 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
33571 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
33572 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
33573 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
33574 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
33575 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
33576 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
33577 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
33578 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
33579 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
33580 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
33581 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
33582 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
33583 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
33584 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
33585 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
33586 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
33587 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
33588 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
33589 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
33590 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
33591 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
33592 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
33593 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
33594 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
33595 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
33596 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
33597 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
33599 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
33600 flag to second_arg_count, handle 4 argument function type _COUNT
33601 aliases, handle second_arg_count on second argument rather than last.
33603 2017-04-10 Jeff Law <law@redhat.com>
33605 PR tree-optimization/80374
33606 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
33607 record anything if we can not convert integer_zero_node to the
33610 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33613 * config/rs6000/rs6000.c (rs6000_option_override_internal):
33614 Enhance special handling given to the TARGET_P9_MINMAX option in
33615 relation to certain other options.
33617 2017-04-10 Bin Cheng <bin.cheng@arm.com>
33619 PR tree-optimization/80153
33620 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
33621 remove POINTER_PLUS_EXPR's base part directly, rather than through
33624 2017-04-10 Richard Biener <rguenther@suse.de>
33625 Bin Cheng <bin.cheng@arm.com>
33627 PR tree-optimization/80153
33628 * tree-affine.c (aff_combination_to_tree): Get base pointer from
33629 the first element of pointer type aff_tree. Build result expr in
33631 (add_elt_to_tree): Convert to type unconditionally. Remove other
33632 fold_convert calls.
33633 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
33634 (rewrite_use_nonlinear_expr): Check invariant using iv information.
33636 2017-04-10 Richard Biener <rguenther@suse.de>
33638 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33641 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
33643 PR rtl-optimization/70478
33644 * lra-constraints.c (curr_small_class_check): New.
33645 (update_and_check_small_class_inputs): New.
33646 (process_alt_operands): Update curr_small_class_check. Disfavor
33647 alternative insn memory operands. Check available regs for small
33650 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33653 * config/mips/mips.opt (-mvirt): Update description.
33654 * doc/invoke.texi (-mvirt): Likewise.
33656 2017-04-10 Richard Biener <rguenther@suse.de>
33658 PR middle-end/80362
33659 * fold-const.c (fold_binary_loc): Look at unstripped ops when
33660 looking for NEGATE_EXPR in -A / -B to A / B folding.
33662 2017-04-10 Martin Liska <mliska@suse.cz>
33664 PR gcov-profile/80224
33665 * gcov.c (print_usage): Fix usage string.
33666 (get_gcov_intermediate_filename): Remove.
33667 (output_gcov_file): Use both for normal and intermediate format.
33668 (generate_results): Do not initialize special file for
33669 intermediate format.
33671 2017-04-10 Richard Biener <rguenther@suse.de>
33673 PR tree-optimization/80304
33674 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
33677 2017-04-10 Nathan Sidwell <nathan@acm.org>
33680 * config/rs6000/rs6000.c (rs6000_vector_type): New.
33681 (rs6000_init_builtins): Use it.
33683 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33685 * config/arm/arm.md (<mrc>): Add mode to SET source.
33686 (<mrrc>): Likewise.
33688 2017-04-10 Richard Biener <rguenther@suse.de>
33690 PR middle-end/80344
33691 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
33693 2017-04-10 Jakub Jelinek <jakub@redhat.com>
33696 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
33697 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
33698 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
33699 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
33700 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
33701 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
33702 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
33703 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
33704 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
33705 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
33706 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
33707 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
33708 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
33709 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
33710 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
33711 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
33712 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
33713 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
33714 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
33715 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
33716 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
33717 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
33718 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
33720 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
33722 PR rtl-optimization/70478
33723 * lra-constraints.c: Reverse the last patch.
33725 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
33727 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
33728 Add comment for WCHAR_T.
33730 2017-04-08 Martin Liska <mliska@suse.cz>
33733 2017-04-07 Martin Liska <mliska@suse.cz>
33736 * ipa-split.c (split_function): Add function part to a same comdat
33739 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33742 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
33744 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
33746 * rs6000/rs6000.c (vec_load_pendulum): Rename...
33747 (vec_pairing): ...to this.
33748 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
33749 (rs6000_sched_init): Adjust for name change.
33750 (struct rs6000_sched_context): Likewise.
33751 (rs6000_init_sched_context): Likewise.
33752 (rs6000_set_sched_context): Likewise.
33754 2017-04-07 Jakub Jelinek <jakub@redhat.com>
33760 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
33762 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
33763 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
33764 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
33766 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
33768 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
33770 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33772 PR rtl-optimization/70703
33773 * ira-color.c (update_conflict_hard_regno_costs): Use
33774 int64_t instead of HOST_WIDE_INT.
33776 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33778 PR rtl-optimization/70478
33779 * lra-constraints.c (process_alt_operands): Disfavor alternative
33780 insn memory operands.
33782 2017-04-07 Jeff Law <law@redhat.com>
33784 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
33785 CALL and NOTE_INSN_CALL_ARG_LOCATION.
33787 2017-04-07 Martin Liska <mliska@suse.cz>
33790 * config/aarch64/aarch64.c (aarch64_process_target_attr):
33791 Show error message instead of an ICE.
33793 2017-04-07 Martin Liska <mliska@suse.cz>
33796 * ipa-split.c (split_function): Add function part to a same comdat
33799 2017-04-07 Richard Biener <rguenther@suse.de>
33801 PR middle-end/80341
33802 * tree.c (get_unwidened): Also handle ! for_type case for
33804 * convert.c (do_narrow): Split out from ...
33805 (convert_to_integer_1): ... here. Do not pass final truncation
33806 type to get_unwidened for TRUNC_DIV_EXPR.
33808 2017-04-07 Richard Biener <rguenther@suse.de>
33810 * tree-affine.c (wide_int_ext_for_comb): Take type rather
33812 (aff_combination_const): Adjust.
33813 (aff_combination_scale): Likewise.
33814 (aff_combination_add_elt): Likewise.
33815 (aff_combination_add_cst): Likewise.
33816 (aff_combination_convert): Likewise.
33817 (add_elt_to_tree): Likewise. Remove unused argument.
33818 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
33820 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
33822 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
33824 * config/arm/arm.c (arm_default_short_enums): Use
33825 ARM_DEFAULT_SHORT_ENUMS.
33826 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
33828 2017-04-06 Jakub Jelinek <jakub@redhat.com>
33831 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
33832 members with redundant out-of-class redeclaration.
33834 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33837 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
33838 * config/i386/i386.md (*zero_extendsidi2):
33839 Add (?*x,*x) and (?*v,*v) alternatives.
33841 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33844 * config/i386/i386.c (ix86_expand_builtin)
33845 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
33846 mode from insn data. Convert operands to insn operand mode.
33847 Copy operands that don't satisfy insn predicate to a register.
33849 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
33851 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
33854 2017-04-06 Richard Biener <rguenther@suse.de>
33856 PR tree-optimization/80334
33857 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
33858 preserve alignment of accesses.
33860 2017-04-06 Richard Biener <rguenther@suse.de>
33862 PR tree-optimization/80262
33863 * tree-sra.c (build_ref_for_offset): Preserve address-space
33865 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
33866 Drop useless address-space information on MEM_REF offsets.
33868 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
33870 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
33872 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33874 PR rtl-optimization/70703
33875 * ira-color.c (update_conflict_hard_regno_costs): Use
33876 HOST_WIDE_INT instead of long.
33878 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
33881 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
33882 not defined for x86_64 target. Add -mmmx target option when __SSE2__
33884 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
33885 for x86_64 target. Handle -m3dnowa option.
33887 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33889 PR rtl-optimization/70703
33890 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
33891 (update_conflict_hard_regno_costs): Use long instead of unsigned
33892 arithmetic for cost calculation.
33894 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33895 Bernd Edlinger <bernd.edlinger@hotmail.de>
33898 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
33901 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
33904 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
33905 ptr_mode with Pmode throughout.
33906 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
33907 into probe_stack_range and use DImode.
33909 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
33912 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
33913 call_eh_return is true.
33915 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33917 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
33918 Initialize last_match_fntype_index.
33920 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33923 * tree-nvr.c: Include internal-fn.h.
33924 (pass_return_slot::execute): Ignore internal calls without
33927 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33928 Richard Biener <rguenther@suse.de>
33931 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
33932 captures used multiple times, except for the last use.
33933 * generic-match-head.c: Include gimplify.h.
33935 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33937 PR tree-optimization/79390
33938 * target.h (struct noce_if_info): Declare.
33939 * targhooks.h (default_noce_conversion_profitable_p): Declare.
33940 * target.def (noce_conversion_profitable_p): New target hook.
33941 * ifcvt.h (struct noce_if_info): New type, moved from ...
33942 * ifcvt.c (struct noce_if_info): ... here.
33943 (noce_conversion_profitable_p): Renamed to ...
33944 (default_noce_conversion_profitable_p): ... this. No longer
33946 (noce_try_store_flag_constants, noce_try_addcc,
33947 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
33948 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
33949 instead of noce_conversion_profitable_p.
33950 * config/i386/i386.c: Include ifcvt.h.
33951 (ix86_option_override_internal): Don't override
33952 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
33953 (ix86_noce_conversion_profitable_p): New function.
33954 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
33955 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
33956 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
33957 * doc/tm.texi: Regenerated.
33959 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33961 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
33964 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
33967 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
33968 instructions for small multiply cores.
33970 2017-04-04 Jeff Law <law@redhat.com>
33972 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
33974 (mips_expand_vec_perm_const): Initialize elements in orig_perm
33975 that are not set by the loop over the elements.
33977 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33980 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
33981 int mode, convert_modes it to mode as unsigned, otherwise use
33982 lowpart_subreg to mode rather than SImode.
33983 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
33984 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
33985 Use DImode instead of SImode for the shift count operand.
33986 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
33989 2017-04-04 Richard Biener <rguenther@suse.de>
33991 PR middle-end/80281
33992 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
33993 arithmetic done for the negate or the plus. Simplify.
33994 (A - (-B) -> A + B): Likewise.
33995 * fold-const.c (split_tree): Make sure to not negate pointers.
33997 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
33999 PR rtl-optimization/60818
34000 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
34001 a compare of comparisons with the thing compared if this results
34002 in a different machine mode.
34004 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
34006 * alias.c (base_alias_check): Fix typo in comment.
34007 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
34008 * cgraphunit.c (symbol_table::compile): Likewise.
34009 * collect2.c (maybe_run_lto_and_relink): Likewise.
34010 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
34011 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
34012 * config/avr/avr.c (avr_map_op_t): Likewise.
34013 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
34014 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
34015 * config/epiphany/epiphany.md (movcc): Likewise.
34016 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
34017 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
34019 * config/mips/mips.c (mips_save_restore_reg): Likewise.
34020 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
34021 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
34022 * config/sh/sh.c (sh_rtx_costs): Likewise.
34023 * fold-const.c (fold_truth_andor): Likewise.
34024 * genautomata.c (collapse_flag): Likewise.
34025 * gengtype.h (struct type::u::s): Likewise.
34026 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
34027 * input.c (FORMAT_AMOUNT): Likewise.
34028 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
34029 (known_aggs_to_agg_replacement_list): Likewise.
34030 * ipa-inline-analysis.c: Likewise.
34031 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
34032 * ipa-polymorphic-call.c
34033 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
34034 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
34035 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
34037 * modulo-sched.c (apply_reg_moves): Likewise.
34038 * omp-expand.c (build_omp_regions_1): Likewise.
34039 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
34040 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
34041 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
34042 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
34043 * value-prof.c: Likewise.
34044 * var-tracking.c (val_reset): Likewise.
34046 2017-04-03 Richard Biener <rguenther@suse.de>
34048 PR tree-optimization/80275
34049 * fold-const.c (split_address_to_core_and_offset): Handle
34052 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
34054 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
34055 descriptors is at least equal to that of functions.
34057 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34059 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
34061 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34064 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
34065 (mov<IMOD4:mode>): New expander.
34066 (*mov<IMOD4:mode>_internal): New insn and split pattern.
34068 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
34070 PR rtl-optimization/79405
34071 * fwprop.c (propagations_left): New variable.
34072 (forward_propagate_into): Decrement it.
34073 (fwprop_init): Initialize it.
34074 (fw_prop): If the variable has reached zero, stop propagating.
34075 (fwprop_addr): Ditto.
34077 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34080 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
34081 a FUNCTION_DECL, pass it as decl instead of origin to
34084 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
34086 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
34089 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
34092 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
34093 TARGET_VSX_SMALL_INTEGER.
34095 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34097 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
34098 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
34100 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34102 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
34103 extraction from odd-numbered MSA register.
34105 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34107 PR middle-end/80173
34108 * expmed.c (store_bit_field_1): Don't attempt to create
34109 a word subreg out of hard registers wider than word if they
34110 have HARD_REGNO_NREGS of 1 for their mode.
34112 PR middle-end/80163
34113 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
34114 conversions to integer types wider than word and pointer.
34117 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
34118 (rtx_equal_for_cselib_p): Pass 0 to it.
34119 * cselib.c (cselib_hasher::equal): Likewise.
34120 (rtx_equal_for_cselib_1): Add depth argument. If depth
34121 is 128, don't look up VALUE locs and punt. Increment
34122 depth in recursive calls when walking VALUE locs.
34124 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
34126 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
34127 (make_gcov_file_name): Use the canonical path name for generating
34129 (read_line): Fix handling of files with ascii null bytes.
34131 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
34133 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
34134 to initialise a vector register instead
34135 of using a const_int.
34137 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34139 PR translation/80189
34140 * gimplify.c (omp_default_clause): Use %qs instead of %s in
34141 diagnostic messages.
34143 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
34146 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
34147 (dfp_diex_<mode>): Update mode of operand 1.
34148 * doc/extend.texi (dxex, dxexq): Document change to return type.
34149 (diex, diexq): Document change to argument type.
34151 2017-03-30 Martin Jambor <mjambor@suse.cz>
34154 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
34155 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
34156 it reflects the signature changes performed at the callee side.
34157 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
34158 to cgraph_build_function_type_skip_args.
34159 (build_function_decl_skip_args): Adjust call to the above function.
34161 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34164 * config/i386/sse.md
34165 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
34166 register as dest whenever it is a MEM not rtx_equal_p to the
34167 corresponding dup operand, and when forcing into reg move the
34168 reg into the memory afterwards.
34169 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
34170 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
34171 for the force_reg mode.
34172 (avx512vl_vextractf128<mode>): Use register as dest either
34173 always when a MEM, or when it is a MEM not rtx_equal_p to the
34174 corresponding dup operand, or even not when it is a CONST_VECTOR
34175 depending on the mode and lo vs. hi.
34176 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
34178 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
34179 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
34180 Likewise. Require that operands[2] is even.
34181 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
34182 Remove extraneous parens. Require that operands[2] is a multiple
34184 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
34185 operands[0] is a MEM if <mask_applied>, the predicates/constraints
34186 disallow memory then.
34188 2017-03-30 Richard Biener <rguenther@suse.de>
34190 PR tree-optimization/77498
34191 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
34192 to non-constants over backedges.
34194 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
34196 PR rtl-optimization/80233
34197 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
34198 as last_combined_insn. Do not test for BARRIER_P separately.
34200 2017-03-29 Andreas Schwab <schwab@suse.de>
34203 * calls.c (prepare_call_address): Convert funexp to Pmode before
34204 copying to temp reg.
34206 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34208 PR tree-optimization/80158
34209 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
34210 Handle possible future case of more than one alternate
34212 (replace_rhs_if_not_dup): Likewise.
34213 (replace_one_candidate): Likewise.
34215 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
34217 PR rtl-optimization/80193
34218 * ira.c (ira): Do not check allocation for LRA.
34220 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
34222 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
34223 (nvptx_output_simt_exit): Declare.
34224 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
34225 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
34226 (init_softstack_frame): Move initialization of crtl->is_leaf to...
34227 (nvptx_declare_function_name): ...here. Emit declaration of local
34228 memory space buffer for omp_simt_enter insn.
34229 (nvptx_output_unisimt_switch): New.
34230 (nvptx_output_softstack_switch): New.
34231 (nvptx_output_simt_enter): New.
34232 (nvptx_output_simt_exit): New.
34233 * config/nvptx/nvptx.h (struct machine_function): New fields
34234 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
34235 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
34236 (UNSPECV_SIMT_EXIT): Ditto.
34237 (omp_simt_enter_insn): New insn.
34238 (omp_simt_enter): New expansion.
34239 (omp_simt_exit): New insn.
34240 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
34242 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
34243 (expand_GOMP_SIMT_ENTER_ALLOC): New.
34244 (expand_GOMP_SIMT_EXIT): New.
34245 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
34246 (GOMP_SIMT_ENTER_ALLOC): Ditto.
34247 (GOMP_SIMT_EXIT): Ditto.
34248 * target-insns.def (omp_simt_enter): New insn.
34249 (omp_simt_exit): Ditto.
34250 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
34252 (lower_rec_simd_input_clauses): Implement SIMT privatization.
34253 (lower_rec_input_clauses): Likewise.
34254 (lower_lastprivate_clauses): Handle SIMT privatization.
34256 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
34257 (ompdevlow_adjust_simt_enter): New.
34258 (find_simtpriv_var_op): New.
34259 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
34260 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
34262 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
34263 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
34264 (copy_decl_for_dup_finish): Ditto.
34266 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
34268 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
34271 * config/i386/i386.c (ix86_option_override_internal): Always
34272 allow -mpreferred-stack-boundary=3 for 64-bit targets.
34274 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34276 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
34278 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34280 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
34281 mark new edge's irreducible flag accordign to it.
34282 (vect_do_peeling): Check loop preheader edge's irreducible flag
34283 and pass it to function slpeel_add_loop_guard.
34285 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
34287 PR tree-optimization/80218
34288 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34289 Update block frequencies and counts.
34291 2017-03-28 Richard Biener <rguenther@suse.de>
34293 PR tree-optimization/78644
34294 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
34295 of a simplification result we may not use it at all.
34297 2017-03-28 Richard Biener <rguenther@suse.de>
34300 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
34301 without arguments, generate default definition of a SSA name.
34303 2017-03-28 Richard Biener <rguenther@suse.de>
34305 PR middle-end/80222
34306 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
34307 TYPE_REF_CAN_ALIAS_ALL references.
34308 * fold-const.c (fold_indirect_ref_1): Likewise.
34310 2017-03-28 Martin Liska <mliska@suse.cz>
34313 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
34314 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
34316 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34317 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
34319 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
34320 (EXTRA_SPECS): Define.
34321 (SUBTARGET_EXTRA_SPECS): Likewise.
34322 (SUBTARGET_CPP_SPEC): Likewise.
34323 * config/arc/elf.h (EXTRA_SPECS): Renamed to
34324 SUBTARGET_EXTRA_SPECS.
34325 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
34327 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34329 * config/arc/simdext.md (vst64_insn): Update pattern.
34330 (vld32wh_insn): Likewise.
34331 (vld32wl_insn): Likewise.
34332 (vld64_insn): Likewise.
34333 (vld32_insn): Likewise.
34335 2017-03-28 Marek Polacek <polacek@redhat.com>
34338 * fold-const.c (fold_comparison): Use protected_set_expr_location
34339 instead of SET_EXPR_LOCATION.
34341 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
34343 * tree.c (add_expr): Avoid name lookup warning.
34345 2017-03-27 Jeff Law <law@redhat.com>
34347 PR tree-optimization/80216
34348 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
34349 function name. Limit recursion depth.
34350 (record_temporary_equivalences): Corresponding changes.
34352 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
34354 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
34357 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34360 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
34362 * cfgcleanup.c (reg_note_cfa_p): New array.
34363 (insns_have_identical_cfa_notes): New function.
34364 (old_insns_match_p): Don't cross-jump in between /f
34365 and non-/f instructions. If both i1 and i2 are frame related,
34366 verify all CFA notes, their order and content.
34368 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
34371 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
34372 HImode and SImode with zero extend to DImode to one insn.
34373 (bswap<mode>2_extenddi): Likewise.
34374 (bswapsi2_extenddi): Likewise.
34375 (bswaphi2_extendsi): Likewise.
34376 (bswaphi2): Combine bswap HImode and SImode into one insn.
34377 Separate memory insns from swapping register.
34378 (bswapsi2): Likewise.
34379 (bswap<mode>2): Likewise.
34380 (bswaphi2_internal): Delete, no longer used.
34381 (bswapsi2_internal): Likewise.
34382 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
34383 store, and gpr<-gpr swap insns.
34384 (bswap<mode>2_store): Likewise.
34385 (bswaphi2_reg): Register only splitter, combine with the splitter.
34386 (bswaphi2 splitter): Likewise.
34387 (bswapsi2_reg): Likewise.
34388 (bswapsi2 splitter): Likewise.
34389 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
34390 the insns into load, store, and register/register insns.
34391 (bswapdi2_ldbrx): Likewise.
34392 (bswapdi2_load): Likewise.
34393 (bswapdi2_store): Likewise.
34394 (bswapdi2_reg): Likewise.
34396 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
34398 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
34399 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
34401 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
34404 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
34406 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34407 special handling for target option conflicts between dform
34408 options (-mpower9-dform, -mpower9-dform-vector,
34409 -mpower9-dform-scalar) and -mno-direct-move.
34411 2017-03-27 Richard Biener <rguenther@suse.de>
34413 PR tree-optimization/80181
34414 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
34416 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34418 * config/arc/predicates.md (move_double_src_operand): Replace the
34419 call to move_double_src_operand with a call to address_operand.
34421 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34423 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
34424 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
34425 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
34427 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34429 * config/arc/predicates.md (long_immediate_loadstore_operand):
34430 Consider scaled addresses cases.
34432 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34434 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
34435 restored when in interrupt.
34436 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
34437 doesn't have delay slot.
34439 2017-03-27 Richard Biener <rguenther@suse.de>
34442 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
34443 inlined thunk clones.
34445 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34448 * asan.c (instrument_derefs): Copy over last operand from
34449 original COMPONENT_REF to the new COMPONENT_REF with
34450 DECL_BIT_FIELD_REPRESENTATIVE.
34451 * ubsan.c (instrument_object_size): Likewise.
34453 2017-03-27 Richard Biener <rguenther@suse.de>
34455 PR tree-optimization/80170
34456 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
34457 sure DR/SCEV didnt fold in constants we do not see when looking
34458 at the reference base alignment.
34460 2017-03-27 Richard Biener <rguenther@suse.de>
34462 PR middle-end/80171
34463 * gimple-fold.c (fold_ctor_reference): Properly guard against
34464 NULL return value from canonicalize_constructor_val.
34466 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
34469 * config/i386/i386.c (ix86_expand_builtin)
34470 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
34471 flags reg setting and flags reg using instructions.
34472 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
34473 clobbering instructions to zero extend op2.
34475 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
34477 * doc/install.texi (Configuration) <--with-aix-soname>:
34478 Update link to AIX ld.
34480 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
34482 PR rtl-optimization/80160
34483 PR rtl-optimization/80159
34484 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
34485 reg_alternate_class into account.
34487 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
34490 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
34491 to consider in curr_insn_transform.
34493 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34495 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
34496 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
34497 and emit_mode_inner.
34499 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34501 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
34502 argument to the overloaded builtin variants. Use the new flag to
34503 deprecate certain builtin variants.
34504 * config/s390/s390-builtin-types.def: Add new builtin types.
34505 * config/s390/s390-builtins.h: Support new flags field for
34506 overloaded builtins.
34507 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
34508 (s390_macro_to_expand): Enable vector float data type.
34509 (s390_cpu_cpp_builtins_internal): Indicate support of the new
34510 builtins by incrementing the __VEC__ version number.
34511 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
34513 (s390_resolve_overloaded_builtin): Emit error messages depending
34514 on the builtin flags.
34515 * config/s390/s390.c (s390_expand_builtin): Support additional
34516 flags argument. Change error message to match the messages
34517 emitted in s390-c.c.
34518 * config/s390/s390.md: New UNSPEC_* constants.
34519 (op_type): Add new instruction types.
34520 * config/s390/vecintrin.h: Add new builtins and test data class
34522 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
34523 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
34524 (VEC_INEXACT, VEC_NOINEXACT): New constants.
34525 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
34526 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
34527 ("vec_mergel<mode>"): V_HW -> VEC_HW.
34529 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
34530 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
34531 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
34532 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
34534 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
34535 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
34536 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
34537 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
34539 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
34540 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
34541 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
34542 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
34543 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
34544 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
34545 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
34547 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
34548 ("vec_scatter_element<V_HW_4:mode>_DI")
34549 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
34550 ("vec_fpint<mode>", "vflls")
34551 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
34552 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
34553 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
34554 ("*vec_cmphe<mode>_cc"): ... these.
34556 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
34557 mode constant instead of magic value.
34559 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34561 * config/s390/s390.c (s390_expand_vec_compare): Support other
34562 vector floating point modes than just V2DF.
34563 (s390_expand_vcond): Likewise.
34564 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
34565 (s390_cannot_change_mode_class): Prevent mode changes between TF
34566 and V1TF in vector registers.
34567 * config/s390/s390.md (DF, SF): New mode attributes.
34568 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
34569 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
34570 SFmode support for VRs.
34571 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
34573 (VFT, VF_HW): New mode iterators.
34574 (vw, sdx): New mode attributes.
34575 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
34576 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
34577 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
34578 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
34579 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
34580 also the new vector floating point modes. Renaming to ...
34582 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
34583 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
34584 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
34585 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
34586 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
34587 ("vec_unordered<mode>"): ... these.
34589 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
34590 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
34591 ("*vec_extendv2df"): New insn definitions.
34593 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34595 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
34596 ("mulditi3_2", "*muldi3_sign"): New patterns.
34597 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
34598 rename the pattern definition.
34600 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34602 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
34604 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
34606 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34608 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
34609 instruction if possible.
34610 * config/s390/vector.md (vec_halfnumelts): New mode
34612 ("*vec_vllezlf<mode>"): New pattern.
34614 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34616 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
34617 ("popcountv4si2", "popcountv2di2"): Rename to ...
34618 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
34619 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
34621 ("popcount<mode>2_vxe"): New pattern.
34623 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34625 * common/config/s390/s390-common.c (processor_flags_table): Add
34627 * config.gcc: Add arch12.
34628 * config/s390/driver-native.c (s390_host_detect_local_cpu):
34629 Default to arch12 for unknown CPU model numbers.
34630 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
34631 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
34632 PROCESSOR_max sanity check.
34633 * config/s390/s390-opts.h (enum processor_type): Add
34635 * config/s390/s390.c (processor_table): Add arch12.
34636 (s390_expand_builtin): Add check for B_VXE flag.
34637 (s390_issue_rate): Add PROCESSOR_ARCH12.
34638 (s390_get_sched_attrmask): Likewise.
34639 (s390_get_unit_mask): Likewise.
34640 (s390_sched_score): Enable z13 scheduling for arch12.
34641 (s390_sched_reorder): Likewise.
34642 (s390_sched_variable_issue): Likewise.
34643 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
34645 (s390_tune_attr): Use z13 scheduling also for arch12.
34646 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
34647 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
34648 (TARGET_VXE_P): New macros.
34649 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
34650 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
34651 * config/s390/s390.opt: Add arch12 as processor_type.
34653 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34655 * config/s390/s390.md
34656 ("fixuns_truncdddi2", "fixuns_trunctddi2")
34657 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
34658 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
34660 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
34661 Rename expanders to ...
34663 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
34664 ("fixuns_truncdddi2_emu"): ... these.
34666 ("fixuns_trunc<mode>si2_emu"): New expander.
34668 ("*fixuns_truncdfdi2_z13"): Rename to ...
34669 ("*fixuns_truncdfdi2_vx"): ... this.
34671 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34673 * config/s390/2964.md: Remove the single element vector compare
34674 instructions which are no longer used.
34675 * config/s390/s390.c (s390_select_ccmode): Remove handling of
34677 (s390_canonicalize_comparison): Remove handling of DFmode
34679 (s390_expand_vec_compare_scalar): Remove function.
34680 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
34681 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
34683 ("*cmp<mode>_ccs"): Add wfcdb instruction.
34685 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34687 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
34689 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
34690 will anyway by matched by mov<mode>_64dfp.
34692 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34694 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
34695 vlef/vstef. Add missing operand to vleif.
34697 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34699 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
34700 pair for all vector types with 64 bit elements.
34701 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
34702 * config/s390/vector.md (V_HW_64): ... here.
34703 (V_128_NOSINGLE): New mode iterator.
34704 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
34705 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
34706 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
34707 ("*vec_load_pairv2di"): Change to ...
34708 ("*vec_load_pair<mode>"): ... this one.
34710 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34712 * config/s390/constraints.md: Add comments.
34713 (jKK): Reject element sizes > 8 bytes.
34714 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
34716 * config/s390/s390.md: Add the s_operand checks formerly in
34717 s390_split_ok_p to various splitters where they are still
34719 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
34720 for 128 bit vectors. Plus two splitters.
34722 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34724 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
34727 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34730 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
34731 error if the boundary argument is not constant.
34733 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34735 PR rtl-optimization/80112
34736 * loop-doloop.c (doloop_condition_get): Don't check condition
34737 if cmp isn't SET with IF_THEN_ELSE src.
34739 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34741 PR tree-optimization/80158
34742 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
34743 replacing a candidate statement, also replace it for the
34744 candidate's alternate interpretation.
34745 (replace_rhs_if_not_dup): Likewise.
34746 (replace_one_candidate): Likewise.
34748 2017-03-24 Richard Biener <rguenther@suse.de>
34750 PR tree-optimization/80167
34751 * graphite-isl-ast-to-gimple.c
34752 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
34754 (translate_isl_ast_to_gimple::get_rename): Likewise.
34756 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34758 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
34759 handling of certain combinations of target options, including the
34760 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
34761 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
34763 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34766 * config/arm/arm.md (*load_multiple): Add reload_completed to
34767 matching condition.
34769 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34770 Richard Biener <rguenth@suse.de>
34772 PR tree-optimization/79908
34773 PR tree-optimization/80136
34774 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34775 been cast away, gimplify_and_add suffices.
34777 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
34779 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
34781 2017-03-23 Richard Biener <rguenther@suse.de>
34783 PR tree-optimization/80032
34784 * gimplify.c (gimple_push_cleanup): Forced unconditional
34785 cleanups still have to go to the conditional_cleanups
34788 2017-03-22 Jakub Jelinek <jakub@redhat.com>
34790 PR tree-optimization/80072
34791 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
34793 (next_operand_entry_id): Change type to unsigned int.
34794 (sort_by_operand_rank): Make sure to return the right return value
34795 even if unsigned fields are bigger than INT_MAX.
34796 (struct oecount): Change cnt and id type to unsigned int.
34797 (oecount_hasher::equal): Formatting fix.
34798 (oecount_cmp): Make sure to return the right return value
34799 even if unsigned fields are bigger than INT_MAX.
34800 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
34803 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
34804 TREE_READONLY on result if writing it more than once.
34807 * doc/invoke.texi (-fsanitize=thread): Document that with
34808 -fnon-call-exceptions atomics are not able to throw
34812 * tsan.c: Include tree-eh.h.
34813 (instrument_builtin_call): Call maybe_clean_eh_stmt or
34814 maybe_clean_or_replace_eh_stmt where needed.
34815 (instrument_memory_accesses): Add cfg_changed argument.
34816 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
34817 if it returned true.
34818 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
34820 PR rtl-optimization/63191
34821 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
34822 wrapper function, moved the whole old content into ...
34823 (ix86_delegitimize_address_1): ... this. New inline function.
34824 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
34825 true as last argument instead of ix86_delegitimize_address.
34827 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34829 * config/aarch64/aarch64.c (generic_branch_cost): Copy
34830 cortexa57_branch_cost.
34832 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34834 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
34836 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34839 * doc/md.texi (Constraints): Document wA constraint.
34840 * config/rs6000/constraints.md (wA): New.
34841 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
34842 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
34843 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
34844 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
34846 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
34849 * gimplify.c (is_oacc_declared): New function.
34850 (oacc_default_clause): Use it to set default flags for acc declared
34851 variables inside parallel regions.
34852 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
34853 declared variables.
34854 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
34855 declare attribute to any decl as necessary.
34857 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
34860 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
34861 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
34862 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
34863 (arm_arch_lpae): This.
34864 * config/arm/arm.c (arm_arch7ve): Rename into ...
34865 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
34866 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
34869 2017-03-22 Martin Liska <mliska@suse.cz>
34872 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
34873 error message instead of an ICE.
34875 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34877 * doc/extend.texi (6.11 Additional Floating Types): Revise.
34879 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
34881 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34883 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34886 2017-03-21 Martin Sebor <msebor@redhat.com>
34888 * doc/extend.texi: Use "cannot" instead of "can't."
34889 * doc/hostconfig.texi: Same.
34890 * doc/install.texi: Same.
34891 * doc/invoke.texi: Same.
34892 * doc/loop.texi: Same.
34893 * doc/md.texi: Same.
34894 * doc/objc.texi: Same.
34895 * doc/rtl.texi: Same.
34896 * doc/tm.texi: Same.
34897 * doc/tm.texi.in: Same.
34898 * doc/trouble.texi: Same.
34900 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
34903 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
34904 (collect_checksum_attributes): Set it.
34905 (die_checksum_ordered): Use it.
34907 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34909 PR tree-optimization/79908
34910 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
34911 change: For a VA_ARG whose LHS has been cast away, use
34912 force_gimple_operand to construct the side effects.
34914 2017-03-21 David Malcolm <dmalcolm@redhat.com>
34916 PR translation/80001
34917 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
34918 more amenable to translation.
34919 (oacc_loop_auto_partitions): Likewise.
34921 2017-03-21 Marek Polacek <polacek@redhat.com>
34922 Martin Sebor <msebor@redhat.com>
34924 PR tree-optimization/80109
34925 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
34926 on INTEGRAL_TYPE_P.
34928 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34929 Segher Boessenkool <segher@kernel.crashing.org>
34932 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
34933 check reg_used_between_p between insn and one of succ or succ2
34934 depending on if succ is artificial insn not inserted into insn
34937 2017-03-21 Martin Liska <mliska@suse.cz>
34939 PR gcov-profile/80081
34940 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
34941 * doc/gcc.texi: Include gcov-dump stuff.
34942 * doc/gcov-dump.texi: New file.
34944 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
34946 PR rtl-optimization/79150
34947 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
34948 conditional jump, if the jump is the last insn of the loop.
34950 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34951 Richard Biener <rguenth@suse.de>
34953 PR tree-optimization/79908
34954 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34955 been cast away, use force_gimple_operand to construct the side
34958 2017-03-21 Martin Liska <mliska@suse.cz>
34960 PR libfortran/79956
34961 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
34964 2017-03-21 Brad Spengler <spender@grsecurity.net>
34967 * plugin.c (htab_hash_plugin): New function.
34968 (add_new_plugin): Use it and adjust.
34969 (parse_plugin_arg_opt): Adjust.
34970 (init_one_plugin): Likewise.
34972 2017-03-21 Richard Biener <rguenther@suse.de>
34974 PR tree-optimization/80032
34975 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
34976 if set force the cleanup to happen unconditionally.
34977 (gimplify_target_expr): Push inserted clobbers with force_uncond
34978 to avoid them being removed by control-dependent DCE.
34980 2017-03-21 Richard Biener <rguenther@suse.de>
34982 PR tree-optimization/80122
34983 * tree-inline.c (copy_bb): Do not expans va-arg packs or
34984 va_arg_pack_len when the inlined call stmt requires pack
34986 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
34988 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34991 * tsan.c (instrument_builtin_call): If the memory model argument
34992 is not a constant, assume it is valid.
34995 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
34998 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35000 PR rtl-optimization/79910
35001 * combine.c (can_combine_p): Do not allow combining an I0 or I1
35002 if its dest is used by an insn before I2 (other than the combined
35003 insns themselves, which are properly handled already).
35005 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35008 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35010 * combine.c (record_used_regs): New static function.
35011 (try_combine): Handle situations where there is an additional
35012 instruction between I2 and I3 which needs to have a LOG_LINK
35016 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35018 * combine.c (try_combine): Delete redundant i1 test. Call
35019 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35021 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35024 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
35025 alternatives 13/14.
35027 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35029 PR tree-optimization/80054
35030 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
35031 the optimization if a PHI or any of its arguments is not dominated
35032 by the candidate's basis. Use gphi* rather than gimple* as
35034 (replace_profitable_candidates): Clean up a gimple* variable that
35035 should be a gphi* variable.
35037 2017-03-20 Martin Sebor <msebor@redhat.com>
35040 * doc/extend.texi (attribute constructor): Document present limitation.
35042 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
35045 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
35046 __POWER9_VECTOR__ #ifdef control, change template definition to
35047 use Power9-specific built-in function.
35048 (vec_any_eq): Likewise.
35049 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
35050 to control outcomes from this test.
35051 (vector_ae_<mode>p): For VEC_F modes, likewise.
35053 2017-03-20 Ian Lance Taylor <iant@google.com>
35055 * config/i386/i386.c (ix86_function_regparm): Save an extra
35056 register for -fsplit-stack with DECL_STATIC_CHAIN.
35058 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35061 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
35062 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
35064 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35066 * config/riscv/riscv.c (riscv_print_operand): Use "fence
35068 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
35071 2017-03-20 Marek Polacek <polacek@redhat.com>
35074 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
35076 2017-03-20 Richard Biener <rguenther@suse.de>
35078 PR tree-optimization/80113
35079 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
35080 allocate extra SSA name for PHI def.
35081 (add_close_phis_to_outer_loops): Likewise.
35082 (add_close_phis_to_merge_points): Likewise.
35083 (copy_loop_close_phi_args): Likewise.
35084 (copy_cond_phi_nodes): Likewise.
35086 2017-03-20 Martin Liska <mliska@suse.cz>
35088 PR middle-end/79753
35089 * tree-chkp.c (chkp_build_returned_bound): Do not build
35090 returned bounds for a LHS that's not a BOUNDED_P type.
35092 2017-03-20 Martin Liska <mliska@suse.cz>
35096 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
35097 COMPLEX_CST and VECTOR_CST.
35099 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35102 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
35103 target operand. A new splitter adds the clobber statement in case
35104 the target operand is dead anyway.
35106 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
35108 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
35109 to age-old versions of binutils and glibc.
35111 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
35113 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
35115 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35117 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
35119 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35121 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
35122 requirement for binutils 2.13.
35124 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35126 * combine.c (try_combine): Delete redundant i1 test. Call
35127 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35129 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
35131 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
35132 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
35134 <riscv64-*-elf>: Re-arrange section
35135 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
35136 <riscv32-*-linux>: Likewise.
35137 <riscv64-*-elf>: Likewise
35138 <riscv64-*-linux>: Likewise.
35140 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
35143 * aarch64.opt(verbose-cost-dump): Fix typo.
35145 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
35148 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
35149 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
35151 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35153 * reload.c (find_reloads): When reloading a nonoffsettable address,
35154 use RELOAD_OTHER for it and its address reloads.
35156 PR rtl-optimization/79910
35157 * combine.c (record_used_regs): New static function.
35158 (try_combine): Handle situations where there is an additional
35159 instruction between I2 and I3 which needs to have a LOG_LINK
35162 2017-03-17 Jeff Law <law@redhat.com>
35164 PR tree-optimization/71437
35165 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
35166 conditional in the hash table first.
35167 (vrp_dom_walker::before_dom_children): Extract condition from
35168 ASSERT_EXPR. Record condition, its inverion and any implied
35169 conditions as well.
35171 2017-03-17 Marek Polacek <polacek@redhat.com>
35172 Markus Trippelsdorf <markus@trippelsdorf.de>
35174 PR tree-optimization/80079
35175 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
35178 2017-03-17 Richard Biener <rguenther@suse.de>
35180 PR middle-end/80075
35181 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
35182 Properly verify the LHS before the RHS possibly claims to be
35184 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
35187 2017-03-17 Martin Jambor <mjambor@suse.cz>
35189 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
35190 (List of -O2 options): Likewise.
35191 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
35194 2017-03-17 Tom de Vries <tom@codesourcery.com>
35196 * gcov-dump.c (print_usage): Print bug_report_url.
35198 2017-03-17 Richard Biener <rguenther@suse.de>
35200 PR middle-end/80050
35201 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
35202 (parser::peek): Likewise.
35204 2017-03-17 Richard Biener <rguenther@suse.de>
35206 PR tree-optimization/80048
35207 * sese.c (free_sese_info): Properly release rename_map and
35208 copied_bb_map elements.
35210 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
35212 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
35213 Add linked-list forward and backlinks. Insert on
35214 construction, remove on destruction.
35215 (class pass_store_merging): Add m_stores_head field.
35216 (pass_store_merging::terminate_and_process_all_chains):
35217 Iterate over m_stores_head list.
35218 (pass_store_merging::terminate_all_aliasing_chains):
35220 (pass_store_merging::execute): Check for debug stmts first.
35221 Push new chains onto the m_stores_head stack.
35223 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
35226 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
35227 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
35228 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
35230 2017-03-16 Jeff Law <law@redhat.com>
35232 PR tree-optimization/71437
35233 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
35234 member function. Implementation moved into after_dom_children
35235 member function and into the threader's thread_outgoing_edges
35237 (dom_opt_dom_walker::after_dom_children): Simplify by moving
35238 some code into new thread_outgoing_edges.
35239 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
35240 definition. Simplify marker handling (do it here). Assume we always
35241 have the available expression and the const/copies tables.
35242 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
35244 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
35245 * tree-vrp.c (equiv_stack): No longer file scoped.
35246 (vrp_dom_walker): New class.
35247 (vrp_dom_walker::before_dom_children): New member function.
35248 (vrp_dom_walker::after_dom_children): Likewise.
35249 (identify_jump_threads): Setup domwalker. Use it rather than
35250 walking edges in a random order by hand. Simplify setup/finalization.
35251 (finalize_jump_threads): Remove.
35252 (vrp_finalize): Do not call identify_jump_threads here.
35253 (execute_vrp): Do it here instead and call thread_through_all_blocks
35256 PR tree-optimization/71437
35257 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
35259 (simplify_stmt_for_jump_threading): Add basic_block argument. All
35261 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
35262 (dom_opt_dom_walker::thread_across_edge): Remove
35263 handle_dominating_asserts argument. All callers changed.
35264 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
35265 changes. Remove calls to lhs_of_dominating_assert. Other
35266 uses of handle_dominating_asserts turn into unconditional code
35267 (simplify_control_stmt_condition_1): Likewise.
35268 (simplify_control_stmt_condition): Likewise.
35269 (thread_through_normal_block, thread_across_edge): Likewise.
35270 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
35271 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
35272 object if it is not an SSA_NAME.
35273 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
35274 before calling into the VRP specific simplifiers.
35275 (identify_jump_threads): Remove handle_dominating_asserts
35278 2017-03-16 Jakub Jelinek <jakub@redhat.com>
35281 * tree-diagnostic.c (default_tree_printer): No longer static.
35282 * tree-diagnostic.h (default_tree_printer): New prototype.
35284 2017-03-16 Tamar Christina <tamar.christina@arm.com>
35286 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
35287 Change ins into fmov.
35289 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35291 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
35292 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
35293 Use h_con constraint for operand 1.
35294 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
35295 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
35297 2017-03-15 Jeff Law <law@redhat.com>
35299 PR tree-optimization/71437
35300 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
35301 (record_temporary_equivalences): Use it.
35303 PR tree-optimization/71437
35304 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
35305 tree-ssa-scopedtables.
35306 (lookup_avail_expr, build_and_record_new_cond): Likewise.
35307 (record_conditions, record_cond, vuse_eq): Likewise.
35308 (record_edge_info): Adjust to API tweak of record_conditions.
35309 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
35310 (record_temporary_equivalences, optimize_stmt): Likewise.
35311 (eliminate_redundant_computations): Likewise.
35312 (record_equivalences_from_stmt): Likewise.
35313 * tree-ssa-scopedtables.c: Include options.h and params.h.
35314 (vuse_eq): New function, moved from tree-ssa-dom.c
35315 (build_and_record_new_cond): Likewise.
35316 (record_conditions): Likewise. Accept vector of conditions rather
35317 than edge_equivalence structure for first argument.
35318 for the first argument.
35319 (avail_exprs_stack::lookup_avail_expr): New member function, moved
35320 from tree-ssa-dom.c.
35321 (avail_exprs_stack::record_cond): Likewise.
35322 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
35323 from tree-ssa-dom.c.
35324 (avail_exprs_stack): Add new member functions lookup_avail_expr
35326 (record_conditions): Declare.
35328 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
35331 * lra-constraints.c (process_alt_operands): Increase reject for
35332 reloading an input/output operand.
35334 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
35337 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
35338 insns to convert from signed/unsigned char/short to IEEE 128-bit
35340 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
35342 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
35345 * config/i386/i386.c (ix86_vector_duplicate_value): Create
35346 subreg of inner mode for values already in registers.
35348 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
35350 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
35351 iteration reg is used after the loop.
35353 2017-03-14 Martin Sebor <msebor@redhat.com>
35355 PR tree-optimization/79800
35356 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
35357 precision in negative-positive range.
35358 (format_floating): Call non-const overload with adjusted precision.
35360 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
35363 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
35366 2017-03-14 Martin Sebor <msebor@redhat.com>
35368 PR middle-end/80020
35369 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
35370 * builtins.def (aligned_alloc): Use it.
35373 * Makefile.in (GTFILES): Add calls.c.
35374 * calls.c: Include "gt-calls.h".
35376 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
35378 PR rtl-optimization/79728
35379 * regs.h (struct target_regs): New field
35380 x_contains_allocatable_regs_of_mode.
35381 (contains_allocatable_regs_of_mode): New macro.
35382 * reginfo.c (init_reg_sets_1): Initialize it, and change
35383 contains_reg_of_mode so it includes global regs as well.
35384 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
35385 rather than contains_regs_of_mode.
35387 2017-03-14 Martin Liska <mliska@suse.cz>
35389 * doc/invoke.texi: Document options that can't be combined with
35390 -fcheck-pointer-bounds.
35392 2017-03-14 Martin Liska <mliska@suse.cz>
35394 PR middle-end/79831
35395 * doc/invoke.texi (-Wchkp): Document the option.
35397 2017-03-14 Martin Liska <mliska@suse.cz>
35399 * Makefile.in: Install gcov-dump.
35401 2017-03-14 Martin Liska <mliska@suse.cz>
35403 * multiple_target.c (expand_target_clones): Bail out for
35404 an invalid attribute.
35406 2017-03-14 Richard Biener <rguenther@suse.de>
35408 * alias.c (struct alias_set_entry): Pack properly.
35409 * cfgloop.h (struct loop): Likewise.
35410 * cse.c (struct set): Likewise.
35411 * ipa-utils.c (struct searchc_env): Likewise.
35412 * loop-invariant.c (struct invariant): Likewise.
35413 * lra-remat.c (struct cand): Likewise.
35414 * recog.c (struct change_t): Likewise.
35415 * rtl.h (struct address_info): Likewise.
35416 * symbol-summary.h (function_summary): Likewise.
35417 * tree-loop-distribution.c (struct partition): Likewise.
35418 * tree-object-size.c (struct object_size_info): Likewise.
35419 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
35420 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
35421 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
35422 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
35423 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
35424 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
35425 (struct _stmt_vec_info): Likewise.
35427 2017-03-14 Martin Liska <mliska@suse.cz>
35430 * multiple_target.c (create_dispatcher_calls): Check that
35431 a target can create a function dispatcher.
35433 2017-03-14 Martin Liska <mliska@suse.cz>
35436 * multiple_target.c (expand_target_clones): Drop local.local
35437 flag for default implementation.
35439 2017-03-14 Richard Biener <rguenther@suse.de>
35441 PR tree-optimization/80030
35442 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
35444 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
35446 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
35447 gcc_fallthrough() instead of __attribute__((fallthrough));
35449 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35451 * doc/gcc.texi: Remove "up" link to (DIR).
35452 * doc/gccint.texi: Ditto.
35454 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35456 * doc/install.texi (Specific) <avr>: Remove reference to
35459 2017-03-13 Jeff Law <law@redhat.com>
35461 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
35462 attribute rather than comments.
35464 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
35465 match_scratch operand is highest.
35467 2017-03-13 Martin Liska <mliska@suse.cz>
35469 PR middle-end/78339
35470 * ipa-pure-const.c (warn_function_noreturn): If the declarations
35471 is a CHKP clone, use original declaration.
35473 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35475 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
35476 (arc_conditional_register_usage): Use a different allocation order
35477 when optimizing for size.
35478 * common/config/arc/arc-common.c (arc_option_optimization_table):
35479 Section anchors default on when optimizing for size.
35481 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35483 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
35485 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35487 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
35488 * config/arc/arc.md (cpu_facility): Add cd variant.
35489 (*movqi_insn): Add code density variant.
35490 (*movhi_insn): Likewise.
35491 (*movqi_insn): Likewise.
35492 (*addsi3_mixed): Likewise.
35493 (subsi3_insn): Likewise.
35495 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35497 * config/arc/arc.md (movsi_cond_exec): Update constraint.
35499 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35501 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
35502 expressions with MINUS and UNARY ops.
35504 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35507 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
35509 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
35510 between vec_select and vector argument.
35511 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
35512 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
35513 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
35514 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
35515 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
35516 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
35518 2017-03-13 Richard Biener <rguenther@suse.de>
35521 * params.def (vect-max-peeling-for-alignment): Fix typo.
35523 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35525 * doc/install.texi (Specific) <mips-*-*>: Remove description of
35526 issue that only occurred with binutils below 2.18.
35528 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35530 * doc/install.texi (Specific) <cris-axis-elf>: No longer
35531 refer to binutils 2.11/2.12 minimum.
35533 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35535 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
35536 ftp.kernel.org and simplify binutils requirement.
35538 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
35540 * doc/invoke.texi (Warning Options): Fix spelling of link-time
35542 (Optimize Options): Ditto. Also remove redundancy.
35544 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35546 PR translation/79848
35547 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
35549 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
35550 to G_ to avoid double translation.
35552 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35554 PR translation/79923
35555 * auto-profile.c (get_combined_location): Convert leading
35556 character of diagnostics to lower case and remove trailing period.
35557 (read_profile): Likewise for various diagnostics.
35558 * config/arm/arm.c (arm_option_override): Remove trailing period
35559 from various diagnostics.
35560 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
35561 (msp430_expand_delay_cycles): Likewise.
35563 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35566 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
35567 full command-line argument, rather than just "str".
35568 (aarch64_validate_march): Likewise.
35569 (aarch64_validate_mtune): Likewise.
35571 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
35573 PR rtl-optimization/78911
35574 * lra-assigns.c (must_not_spill_p): New function.
35575 (spill_for): Use it.
35577 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35579 PR tree-optimization/79981
35580 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
35581 ATOMIC_COMPARE_EXCHANGE ifn result.
35582 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
35583 IFN_ATOMIC_COMPARE_EXCHANGE.
35585 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35588 * opts.c (parse_sanitizer_options): Add missing question mark to
35589 "did you mean" message.
35591 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35593 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
35595 (VMULEUH_UNS): Likewise.
35596 (VMULOUB_UNS): Likewise.
35597 (VMULOUH_UNS): Likewise.
35598 * config/rs6000/rs6000.c (builtin_function_type): Remove
35599 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
35601 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35604 * read-rtl-function.c (function_reader::read_rtx_operand): Update
35605 x with result of extra_parsing_for_operand_code_0.
35606 (function_reader::extra_parsing_for_operand_code_0): Convert
35607 return type from void to rtx, returning x. When reading
35608 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
35609 larger size containing struct block_symbol.
35611 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
35613 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
35614 -mfloat128-hardware without -m64.
35616 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
35619 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
35620 entries to the case statement that marks unsigned arguments to
35621 overloaded functions.
35623 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
35625 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
35626 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
35628 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
35631 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
35632 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
35634 2017-03-10 Martin Liska <mliska@suse.cz>
35638 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
35639 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
35642 2017-03-10 Olivier Hainque <hainque@adacore.com>
35644 * tree-switch-conversion (array_value_type): Start by resetting
35645 candidate type to it's main variant.
35647 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35649 PR rtl-optimization/79909
35650 * combine.c (try_combine): Use simplify_replace_rtx on individual
35651 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
35652 of the whole CALL_INSN_FUNCTION_USAGE.
35654 PR tree-optimization/79972
35655 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
35656 get_range_info on SSA_NAMEs. Formatting fixes.
35658 2017-03-10 Richard Biener <rguenther@suse.de>
35659 Jakub Jelinek <jakub@redhat.com>
35661 PR tree-optimization/77975
35662 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
35663 edge to be constant.
35664 (get_val_for): For constant x return it. Formatting fix.
35665 (loop_niter_by_eval): Avoid pointless looping if the next iteration
35666 would use the same bases as the current one.
35668 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35670 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
35671 instead of vec_select for V1TImode.
35672 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
35674 (VSX_LE_128): Add V1TI to this mode iterator.
35675 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
35676 (*vsx_le_perm_store_<mode>): Likewise.
35677 (pre-reload splitter for VSX stores): Likewise.
35678 (post-reload splitter for VSX stores): Likewise.
35679 (*vsx_xxpermdi2_le_<mode>): Likewise.
35680 (*vsx_lxvd2x2_le_<mode>): Likewise.
35681 (*vsx_stxvd2x2_le_<mode>): Likewise.
35683 2017-03-09 Michael Eager <eager@eagercon.com>
35685 Correct failures with --enable-checking=yes,rtl.
35687 * config/microblaze/microblaze.c (microblaze_expand_shift):
35688 Replace GET_CODE test with CONST_INT_P and INTVAL test with
35689 test for const0_rtx.
35690 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
35691 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
35693 2017-03-09 Richard Biener <rguenther@suse.de>
35695 PR tree-optimization/79977
35696 * graphite-scop-detection.c (scop_detection::merge_sese):
35697 Handle the case of extra exits to blocks dominating the entry.
35699 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
35701 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
35704 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
35706 PR rtl-optimization/79949
35707 * lra-constraints.c (process_alt_operands): Check memory when
35708 trying to predict a cycle. Print about the overall increase.
35710 2017-03-09 Richard Biener <rguenther@suse.de>
35712 PR middle-end/79971
35713 * gimple-expr.c (useless_type_conversion_p): Preserve
35714 TYPE_SATURATING for fixed-point types.
35716 2017-03-09 Richard Biener <rguenther@suse.de>
35719 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
35720 alignment of BLKmode params.
35722 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35725 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
35726 (VALL_NO_V2Q): Likewise.
35728 * config/aarch64/aarch64-simd.md
35729 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
35731 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
35732 VALL_NO_V2Q mode iterator.
35733 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
35735 2017-03-09 Martin Liska <mliska@suse.cz>
35737 PR tree-optimization/79631
35738 * tree-chkp-opt.c (chkp_is_constant_addr): Call
35739 tree_int_cst_sign_bit just for INTEGER constants.
35741 2017-03-09 Martin Liska <mliska@suse.cz>
35745 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
35748 2017-03-09 Marek Polacek <polacek@redhat.com>
35751 * tree.c (inchash::add_expr): Handle TREE_VEC.
35753 2017-03-09 Martin Liska <mliska@suse.cz>
35756 (chkp_narrow_size_and_offset): New function.
35757 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
35758 (void chkp_parse_bit_field_ref): New function.
35759 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
35760 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
35762 2017-03-09 Martin Liska <mliska@suse.cz>
35765 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
35766 (chkp_find_bounds_1): Remove gcc_unreachable.
35768 2017-03-09 Jakub Jelinek <jakub@redhat.com>
35771 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
35772 BUILT_IN_SYNC*, determine the access type from the size suffix and
35773 always build a MEM_REF with that type. Handle forgotten
35774 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
35777 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
35778 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
35779 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
35780 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
35781 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
35782 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
35783 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
35784 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
35785 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
35786 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
35787 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
35788 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
35789 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
35790 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
35791 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
35792 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
35793 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
35794 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
35795 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
35796 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
35797 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
35798 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
35799 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
35800 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
35801 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
35802 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
35803 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
35804 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
35805 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
35806 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
35807 definitions outside of __OPTIMIZE__ guarded section.
35810 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
35811 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
35812 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
35813 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
35816 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35818 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
35819 ("vfenez<mode>"): Add missing constraints.
35821 2017-03-08 Martin Sebor <msebor@redhat.com>
35824 * config/nds32/nds32.c (nds32_option_override):
35825 Fix misspelled diagnostic.
35827 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35830 * gimplify.c (gimplify_omp_for): Replace index var in outer
35831 taskloop statement with an artificial variable and add
35832 OMP_CLAUSE_PRIVATE clause for it.
35834 2017-03-08 Richard Biener <rguenther@suse.de>
35836 PR tree-optimization/79955
35837 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
35838 for accesses that are completely outside of the variable.
35840 2017-03-08 Andrew Haley <aph@redhat.com>
35842 PR tree-optimization/79943
35843 * tree-ssa-loop-split.c (compute_new_first_bound): When
35844 calculating the new upper bound, (END-BEG) should be added, not
35847 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35849 * config/avr/avr.md (setmemhi): Make sure match_dup
35850 operand number comes before match_scratch.
35852 2017-03-08 Richard Biener <rguenther@suse.de>
35854 PR tree-optimization/79920
35855 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
35856 with ncopies == 1 to ...
35857 (vect_transform_slp_perm_load): ... here. Properly compute
35858 all element loads by iterating VF times over the group. Do
35859 not handle ncopies (computed in a broken way) in
35860 vect_create_mask_and_perm.
35862 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35865 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
35866 is a uniform vector, use uniform_vector_p return value instead of
35867 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
35869 2017-03-07 Marek Polacek <polacek@redhat.com>
35871 PR middle-end/79809
35872 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
35873 (alloca_call_type): Likewise.
35875 2017-03-07 Martin Liska <mliska@suse.cz>
35877 * gcov.c (process_args): Put comment to correct location.
35879 2017-03-07 Martin Liska <mliska@suse.cz>
35881 PR middle-end/68270
35882 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
35883 Use array_at_struct_end_p instead of DECL_CHAIN (field).
35884 (chkp_narrow_bounds_for_field): Likewise.
35885 (chkp_parse_array_and_component_ref): Pass one more argument to
35888 2017-03-07 Richard Biener <rguenther@suse.de>
35890 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
35893 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
35895 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
35896 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
35898 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35901 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
35902 to end of description.
35903 (PARAM_MAX_STORES_TO_MERGE): Likewise.
35905 2017-03-07 Jakub Jelinek <jakub@redhat.com>
35907 PR rtl-optimization/79901
35908 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
35910 (*avx512f_<code><mode>3<mask_name>): ... this.
35911 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
35912 iterator instead of VI8_AVX2_AVX512BW.
35914 PR rtl-optimization/79901
35915 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
35916 min/max expander, expand it using expand_vec_cond_expr.
35919 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
35922 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35925 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
35926 to void * for PCH reasons.
35927 * dwarf2out.c (output_loc_operands, output_die): Cast
35928 v.val_vec.array to unsigned char *.
35930 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
35933 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
35936 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
35938 PR rtl-optimization/79571
35939 * lra-constraints.c (process_alt_operands): Calculate static
35940 reject and subtract it from overall when only addresses will be
35943 2017-03-06 Julia Koval <julia.koval@intel.com>
35946 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
35947 incoming stack boundary to 128 for 64-bit targets.
35949 2017-03-06 Richard Biener <rguenther@suse.de>
35951 PR tree-optimization/79894
35952 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
35953 to NULL after folding it.
35955 2017-03-06 Richard Biener <rguenther@suse.de>
35957 PR tree-optimization/79824
35958 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
35959 check disabling peeling for gaps.
35961 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
35963 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
35964 attributes): Document gettimeofday.
35966 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35968 * config/s390/s390.c (s390_option_override_internal): Set
35969 PARAM_MIN_VECT_LOOP_BOUND
35971 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35973 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
35974 * config/s390/s390.md: Likewise.
35976 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35979 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
35980 (<avx2_avx512>_perm<mode>): Rename to ...
35981 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
35983 (<avx512>_perm<mode>_mask): Rename to ...
35984 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
35986 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
35987 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
35988 instead of VI8F_256_512.
35989 (avx512f_perm<mode>): New define_expand.
35990 (avx512f_perm<mode>_mask): Likewise.
35991 (avx512f_perm<mode>_1<mask_name>): New define_insn.
35992 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
35994 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35996 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
35997 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
35999 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
36001 2017-03-06 Martin Liska <mliska@suse.cz>
36004 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
36005 when having a SSA NAME w/o VAR_DECL assigned to it.
36007 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36009 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
36010 msa_dpsub_<su>_d): Fix MODE for vec_select.
36012 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36014 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
36016 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
36018 2017-03-06 Richard Biener <rguenther@suse.de>
36020 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
36021 * plugin.c (register_plugin_info): Likewise.
36022 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
36024 2017-03-05 Jakub Jelinek <jakub@redhat.com>
36026 * config/i386/sse.md (sse_storehps, sse_storelps,
36027 avx_<castmode><avxsizesuffix>_<castmode>,
36028 avx512f_<castmode><avxsizesuffix>_<castmode>,
36029 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
36030 in condition that at least one operand is not a MEM.
36032 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36034 PR middle-end/79805
36035 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
36036 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
36038 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
36039 gimple_call_nothrow_p flag based on whether original builtin can throw.
36040 If it can, emit following stmts on the fallthrough edge.
36041 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
36042 don't create new bb if inserting just debug stmts on the edge, try to
36043 insert them on the fallthru bb or just reset debug stmts.
36045 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
36048 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
36049 restore recog_data (including the operand rtxes inside it) around
36050 the call to get_insn_template.
36052 2017-03-03 Martin Sebor <msebor@redhat.com>
36054 PR tree-optimization/79699
36055 * context.c (context::~context): Free MPFR caches to avoid
36056 a memory leak on program exit.
36058 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36060 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
36061 Use wide_int::ulow () instead of .elt (0).
36063 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36065 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
36066 (*pushxf): Limit oF constraint to 32bit targets and add oC
36067 constraint for 64bit targets.
36068 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
36069 (*pushdf): Change rmF constraint to rmC.
36071 2017-03-03 Martin Liska <mliska@suse.cz>
36073 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
36074 Remove unused variable.
36076 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36079 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
36080 is a memory operand, increase num_memory.
36081 (ix86_expand_args_builtin): Likewise.
36083 2017-03-03 Jan Hubicka <jh@suse.cz>
36086 * ipa-devirt.c (maybe_record_node): Properly handle
36087 __cxa_pure_virtual visibility.
36089 2017-03-03 Martin Liska <mliska@suse.cz>
36091 PR tree-optimization/79803
36092 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
36094 (pass_loop_prefetch::execute): Disabled optimization if an
36095 assumption about L1 cache size is not met.
36097 2017-03-03 Martin Liska <mliska@suse.cz>
36099 PR rtl-optimization/79574
36100 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
36101 (hash_scan_set): Likewise.
36102 (dump_hash_table): Likewise.
36103 (hoist_code): Likewise.
36105 2017-03-03 Richard Biener <rguenther@suse.de>
36107 * fixed-value.c (fixed_from_string): Restore use of elt (1)
36108 in place of uhigh ().
36109 (fixed_convert_from_real): Likewise.
36111 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36114 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
36116 2017-03-03 Richard Biener <rguenther@suse.de>
36118 PR middle-end/79818
36119 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
36120 TYPE_OVERFLOW_UNDEFINED check.
36122 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36124 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
36126 (vector_ae_<mode>_p): Likewise.
36127 (vector_nez_<mode>_p): Likewise.
36128 (vector_ne_v2di_p): Likewise.
36129 (vector_ae_v2di_p): Likewise.
36130 (vector_ne_<mode>_p): Likewise.
36131 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
36133 (vsx_tsqrt<mode>2_fe): Likewise.
36135 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
36138 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
36140 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36142 PR rtl-optimization/79780
36143 * cprop.c (one_cprop_pass): When second and further conditional trap
36144 in a single basic block is turned into an unconditional trap, turn it
36145 into a deleted note to avoid RTL verification failures.
36147 2017-03-02 Richard Biener <rguenther@suse.de>
36149 * fold-const.c (const_binop): Use ulow () instead of elt (0).
36151 2017-03-02 Richard Biener <rguenther@suse.de>
36153 PR tree-optimization/79345
36155 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
36156 param and abort the walk, returning -1 if it is hit.
36157 (walk_aliased_vdefs): Take a limit param and pass it on.
36158 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
36159 defaulting to 0 and return a signed int.
36160 * tree-ssa-uninit.c (struct check_defs_data): New struct.
36161 (check_defs): New helper.
36162 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
36163 about uninitialized memory.
36164 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
36165 bogus uninitialized warning.
36166 (fixed_convert_from_real): Likewise.
36168 2017-03-02 Bin Cheng <bin.cheng@arm.com>
36170 PR tree-optimization/66768
36171 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
36172 iv_use if base object can't be determined.
36174 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36176 PR tree-optimization/79345
36177 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
36178 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
36179 (get_pattern_stats): Initialize it.
36180 * genemit.c (gen_expand): Verify match_scratch numbers come after
36181 match_operand/match_dup numbers.
36182 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
36183 match_scratch numbers.
36184 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
36186 * config/s390/s390.md (trunctdsd2): Likewise.
36188 2017-03-02 Richard Biener <rguenther@suse.de>
36190 * wide-int.h (wide_int_storage::operator=): Implement in terms
36193 2017-03-02 Richard Biener <rguenther@suse.de>
36195 PR tree-optimization/79777
36196 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
36197 the to insert expression to sth existing.
36199 2017-03-01 Martin Sebor <msebor@redhat.com>
36201 PR middle-end/79692
36202 * gimple-ssa-sprintf.c
36203 (directive::known_width_and_precision): New function.
36204 (format_integer): Use it.
36205 (get_mpfr_format_length): Consider the full range of precision
36206 when computing %g output with the # flag. Set the likely byte
36207 count to 3 rather than 1 when precision is indeterminate.
36208 (format_floating): Correct the lower bound of precision.
36210 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36212 * doc/invoke.texi: Document default code model for 64-bit Linux.
36214 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36217 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
36218 udiv rather than div since input pattern is unsigned.
36220 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
36222 * config/i386/i386.c (print_reg): Warn for values of
36223 unsupported size in integer register.
36225 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
36228 * config/rs6000/predicates.md (current_file_function_operand): Do
36229 not allow self calls to be local if the function is replaceable.
36231 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
36234 * config/rs6000/altivec.h (vec_ctz and others): Change the
36235 preprocessor macro that controls conditional compilation from
36236 _ARCH_PWR9 to __POWER9_VECTOR__.
36237 (vec_all_ne): Change parameterization of __altivec_scalar_pred
36238 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
36239 control (instead of _ARCH_PWR9 control) so that template
36240 definition uses power9-specific function.
36241 (vec_any_eq): Likewise.
36242 (vec_all_ne): Change macro definition to use a power9-specific
36243 expansion under #ifdef __POWER9_VECTOR__ control (instead of
36244 _ARCH_PWR9 control).
36245 (vec_any_eq) Likewise.
36246 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
36247 expansion for CMPNEF to remove support for xvcmpnesp instruction.
36248 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
36249 support for xvcmpnedp instruction.
36250 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
36251 macro expansion so that Power9 implementation of vec_all_ne does
36252 not use the AltiVec predicate framework.
36253 (VCMPNEH_P): Likewise.
36254 (VCMPNEW_P): Likewise.
36255 (VCMPNED_P): Likewise.
36256 (VCMPNEFP_P): Likewise.
36257 (VCMPNEDP_P): Likewise.
36258 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
36259 implementation of vec_any_eq to not use AltiVec predicate
36261 (VCMPAEH_P): Likewise.
36262 (VCMPAEW_P): Likewise.
36263 (VCMPAED_P): Likewise.
36264 (VCMPAEFP_P): Likewise.
36265 (VCMPAEDP_P): Likewise.
36266 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
36267 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
36268 not use the AltiVec predicate framework.
36269 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
36270 of vec_any_eq to not use AltiVec predicate framework.
36271 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36272 support for predefined __POWER9_VECTOR__ macro to indicate that
36273 Power9 instruction selection is enabled.
36274 (altivec_overloaded_builtins): Remove extraneous
36275 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
36276 function argument types RS6000_BTI_bool_V16QI and
36277 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
36278 entry for overloaded function argument types RS6000_BTI_bool_V4SI
36279 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
36280 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
36281 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
36282 Power9 for implementations of vec_cmpne. Change the signature for
36283 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
36284 (representing vec_all_ne) to remove the previously described first
36285 argument of type RS6000_BTI_INTSI, as this was an artifact of
36286 reliance on the AltiVec predicate framework, which is no longer
36287 used in the implementation of these functions. Add
36288 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
36289 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
36290 since, unlike the AltiVec predicate framework implementation, we
36291 do not share function descriptors between vec_alle and vec_anyeq.
36292 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
36293 set of modes that receive special treatment even when
36294 TARGET_P9_VECTOR is true. The special treatment emits code that
36295 does not depend on Power9 instructions.
36296 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
36297 define_expand to not rely on AltiVec predicate framework.
36298 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36300 (vector_ne_v2di_p): Change this define_expand to not rely on
36301 AltiVec predicate framework.
36302 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
36304 (vector_ne_<mode>_p): Change this define_expand to not rely on
36305 AltiVec predicate framework.
36306 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36308 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
36309 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
36310 define_insn pattern.
36311 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
36312 define_insn pattern because the xvcmpne<VSs>. instruction is not
36314 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
36315 instruction is not supported.
36317 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36319 * config/nvptx/nvptx.c: Include intl.h.
36321 2017-03-01 Martin Jambor <mjambor@suse.cz>
36324 * ipa-prop.h (ipa_bits): Removed field known.
36325 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
36326 to pointers. Adjusted their comments to warn about their sharing.
36327 (ipcp_transformation_summary): Change bits to a vector of pointers.
36328 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
36329 (ipa_get_ipa_bits_for_value): Declare.
36330 * tree-vrp.h (value_range): Mark as GTY((for_user)).
36331 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
36332 (ipa_bits_hash_table): Likewise.
36333 (ipa_vr_ggc_hash_traits): Likewise.
36334 (ipa_vr_hash_table): Likewise.
36335 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
36336 being pointers and vr_known being removed.
36337 (ipa_set_jf_unknown): Likewise.
36338 (ipa_get_ipa_bits_for_value): New function.
36339 (ipa_set_jfunc_bits): Likewise.
36340 (ipa_get_value_range): New overloaded functions.
36341 (ipa_set_jfunc_vr): Likewise.
36342 (ipa_compute_jump_functions_for_edge): Use the above functions to
36343 construct bits and vr parts of jump functions.
36344 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
36345 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36347 (ipcp_grow_transformations_if_necessary): Also allocate
36348 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36350 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
36351 them. Fix too long lines.
36352 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
36353 vr_known being removed.
36354 (ipa_read_jump_function): Use new setter functions to construct bits
36355 and vr parts of jump functions or set them to NULL.
36356 (write_ipcp_transformation_info): Adjust for bits being pointers.
36357 (read_ipcp_transformation_info): Likewise.
36358 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
36360 Include gt-ipa-prop.h.
36361 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
36363 (ipcp_store_bits_results): Likewise.
36364 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
36365 Do not write to existing jump functions but use a temporary instead.
36367 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36370 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
36371 attempt to use its first operand as BIT_FIELD_REF base.
36373 2017-03-01 Richard Biener <rguenther@suse.de>
36375 PR middle-end/79721
36376 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
36377 interpolating formula in wrapping arithmetic.
36378 (chrec_apply): Convert chrec_evaluate return value to wanted type.
36380 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36382 PR tree-optimization/79734
36383 * tree-vect-generic.c (expand_vector_condition): Optimize
36384 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
36385 Handle VEC_COND_EXPR where comparison has different inner width from
36386 type's inner width.
36388 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
36390 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
36391 markup, and similar issues. Remove @opindex entries for things
36392 that aren't options. Add missing -mmpy-option entries.
36394 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36396 PR tree-optimization/79737
36397 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
36398 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
36399 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
36400 instead of byte_size. Formatting fix.
36401 (shift_bytes_in_array_right): Formatting fix.
36403 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
36406 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
36407 condition on optimize for the leaf function test.
36409 2017-02-28 Martin Liska <mliska@suse.cz>
36412 * read-rtl-function.c (function_reader::handle_unknown_directive):
36413 Bail out when one uses -flto.
36415 2017-02-28 Martin Liska <mliska@suse.cz>
36417 * common.opt: Replace space with tabular for options of <number>
36419 * config/i386/i386.opt: Show <number> value for
36420 -mlarge-data-threshold.
36421 * opts.c (print_filtered_help): Do not display number in hexadecimal
36424 2017-02-28 Martin Liska <mliska@suse.cz>
36426 * common.opt: Fix --help=option -Q for options which are of
36429 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
36431 * config/i386/i386.c (print_reg): Error out for values
36432 of 8-bit size in invalid integer register.
36434 2017-02-28 Martin Sebor <msebor@redhat.com>
36436 PR tree-optimization/79691
36437 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
36439 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36442 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
36443 gcc_unreachable with output_operand_lossage.
36445 2017-02-28 Richard Biener <rguenther@suse.de>
36447 PR tree-optimization/79740
36448 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
36450 (visit_nary_op): Insert the nary into the hashtable if we
36451 pattern-matched sth.
36452 * tree-ssa-pre.c (eliminate_insert): Robustify.
36454 2017-02-28 Richard Biener <rguenther@suse.de>
36456 PR middle-end/79731
36457 * fold-const.c (decode_field_reference): Reject out-of-bound
36460 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36462 * config/i386/i386.c: Include intl.h.
36463 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
36464 instead of just cond ? "..." : "...".
36465 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
36466 * coverage.c (read_counts_file): Likewise.
36467 * omp-offload.c: Include intl.h.
36468 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
36469 of just cond ? "..." : "...".
36470 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
36471 of just cond ? "..." : "...".
36473 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
36476 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
36478 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
36479 'tune for' CPU name.
36480 * config/arm/arm-cpu-data.h: Regenerated.
36482 2017-02-28 Richard Biener <rguenther@suse.de>
36484 PR tree-optimization/79732
36485 * tree-inline.c (expand_call_inline): Do not shadow var.
36487 2017-02-28 Richard Biener <rguenther@suse.de>
36489 PR tree-optimization/79723
36490 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
36491 address-space properly.
36493 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
36495 * doc/optinfo.texi (Optimization groups): Fix option used for
36497 * doc/invoke.texi (-fopt-info): Document "omp".
36498 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
36499 (OPTGROUP_ALL): Add OPTGROUP_OMP.
36500 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
36501 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
36502 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
36504 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
36506 * dumpfile.c (optgroup_options): Instead of "openmp", associate
36507 OPTGROUP_OMP with "omp".
36509 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
36512 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
36513 for arithmetic shift of unsigned V2DI.
36515 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
36517 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
36518 arc/linux.h headers.
36519 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
36520 (LINK_SPEC): Likewise.
36521 (ARC_TLS_EXTRA_START_SPEC): Likewise.
36522 (EXTRA_SPECS): Likewise.
36523 (STARTFILE_SPEC): Likewise.
36524 (ENDFILE_SPEC): Likewise.
36525 (LIB_SPEC): Likewise.
36526 (TARGET_SDATA_DEFAULT): Likewise.
36527 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
36528 (MULTILIB_DEFAULTS): Likewise.
36529 (DWARF2_UNWIND_INFO): Likewise.
36530 * config/arc/big.h: New file.
36531 * config/arc/elf.h: Likewise.
36532 * config/arc/linux.h: Likewise.
36533 * config/arc/t-uClibc: Remove.
36535 2017-02-27 Bin Cheng <bin.cheng@arm.com>
36537 PR tree-optimization/77536
36538 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
36539 (tree_transform_and_unroll_loop): Use above function to compute the
36540 estimated niter of unrolled loop and use it when scaling profile.
36541 Also use count info rather than frequency if it's non-zero.
36542 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
36543 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
36544 (vect_transform_loop): Call above function.
36546 2017-02-27 Richard Biener <rguenther@suse.de>
36548 PR tree-optimization/45397
36549 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
36550 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
36551 (visit_nary_op): Add pattern matching for CSEing sign-changed
36552 or truncated operations with wider ones.
36554 2017-02-27 Richard Biener <rguenther@suse.de>
36556 PR tree-optimization/79690
36557 * tree-vect-stmts.c (vectorizable_store): Use vector type
36558 built from the DR with address-space.
36560 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
36562 * doc/invoke.texi (Optimize Options): Refine the description
36563 of asan-use-after-return.
36565 2017-02-25 Alan Modra <amodra@gmail.com>
36567 PR rtl-optimization/79584
36568 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
36569 base, not ad->base_term, the reg within base. Remove assertion
36570 that ad->base == ad->base_term. Replace gen_int_mode using
36571 bogus mode with const0_rtx.
36573 2017-02-25 Jakub Jelinek <jakub@redhat.com>
36575 PR middle-end/79396
36576 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
36577 FMA_EXPR like tcc_binary or tcc_unary.
36579 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
36582 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
36584 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
36585 (output_loc_operands): Handle DW_OP_call_ref and
36586 DW_OP_GNU_variable_value.
36587 (struct variable_value_struct): New type.
36588 (struct variable_value_hasher): Likewise.
36589 (variable_value_hash): New variable.
36590 (string_types): Remove.
36591 (copy_loc_descr): New function.
36592 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
36593 (prepend_loc_descr_to_each): New function.
36594 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
36595 instead of add_loc_descr_to_each if the first argument is single
36596 location list and the second has multiple.
36597 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
36598 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
36599 when looking for variable value which doesn't have other location info.
36600 (loc_list_from_tree): Formatting fix.
36601 (gen_array_type_die): Simplify DW_AT_string_length handling.
36602 (adjust_string_types): Remove.
36603 (gen_subprogram_die): Don't call adjust_string_types nor test/set
36604 string_types. Call resolve_variable_values.
36605 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
36606 (resolve_addr_in_expr): Likewise. Add A argument.
36607 (copy_deref_exprloc): Remove deref argument. Adjust for the
36608 original expression being DW_OP_GNU_variable_value with optionally
36609 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
36610 optionally after it.
36611 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
36612 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
36613 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
36614 (variable_value_hasher::hash, variable_value_hasher::equal): New
36616 (resolve_variable_value_in_expr, resolve_variable_value,
36617 resolve_variable_values, note_variable_value_in_expr,
36618 note_variable_value): New functions.
36619 (dwarf2out_early_finish): Call note_variable_value on all toplevel
36622 2017-02-24 Jakub Jelinek <jakub@redhat.com>
36625 * opts.h (handle_generated_option): Add GENERATED_P argument.
36626 * opts-common.c (handle_option): Adjust function comment.
36627 (handle_generated_option): Add GENERATED_P argument, pass it to
36629 (control_warning_option): Pass false to handle_generated_option
36631 * opts.c (maybe_default_option): Pass true to handle_generated_option
36633 * optc-gen.awk: Likewise.
36635 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36637 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
36638 a REG, look at the REG it is a SUBREG of.
36639 (splitter for cmpeqsi_t): Ditto.
36641 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36643 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
36644 the special USEs with the pattern of the insn, not the insn itself.
36646 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
36649 * doc/invoke.texi: Document -mload-store-pairs.
36651 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36652 Sandra Loosemore <sandra@codesourcery.com>
36654 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
36655 argument isn't a CONST_INT.
36656 (nios2_alternate_compare_const): Assert op is a CONST_INT.
36657 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
36658 (nios2_validate_compare): Bypass alternate compare logic if *op2
36659 is not a CONST_INT.
36660 (ldstwm_operation_p): Return false if first_base is not a REG or
36661 if first_offset is not a CONST_INT.
36663 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36665 * config/cris/cris.md: Use correct operand in a define_peephole2.
36667 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36669 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
36671 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36673 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
36674 this_insn if it is an INSN or JUMP_INSN.
36675 (force_offsettable): Look at base, not at addr.
36676 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
36677 on things that aren't necessarily CONST_INTs.
36679 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
36681 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
36682 -mfpmath=sse is the default also for x86-32 targets with SSE2
36683 instruction set when @option{-ffast-math} is enabled
36685 2017-02-24 Jeff Law <law@redhat.com>
36687 PR rtl-optimizatoin/79286
36688 * ira.c (update_equiv_regs): Drop may_trap_p exception to
36691 2017-02-24 Richard Biener <rguenther@suse.de>
36693 PR tree-optimization/79389
36694 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
36697 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
36699 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
36700 function comment to reflect reality.
36701 (loop_exits_before_overflow): Fix typo in function description.
36703 2017-02-24 Richard Biener <rguenther@suse.de>
36705 PR tree-optimization/79389
36706 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
36707 properly that a threading opportunity exists. Detect conditional
36708 copy/constant propagation opportunities.
36710 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
36712 * config/visium/visium.md (type): Add trap.
36713 (b): New mode attribute.
36714 (*btst): Rename into...
36715 (*btst<mode>): ...this and adjust.
36716 (*cbranchsi4_btst_insn): Rename into...
36717 (*cbranch<mode>4_btst_insn): ...this and adjust.
36718 (trap): New define_insn.
36720 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36722 PR tree-optimization/79389
36723 * ifcvt.c (struct noce_if_info): Add rev_cond field.
36724 (noce_reversed_cond_code): New function.
36725 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
36726 reversed_comparison_code. Formatting fix.
36727 (noce_try_store_flag): Test rev_cond != NULL in addition to
36728 reversed_comparison_code.
36729 (noce_try_store_flag_constants): Likewise.
36730 (noce_try_store_flag_mask): Likewise.
36731 (noce_try_addcc): Use rev_cond if non-NULL instead of
36732 reversed_comparison_code.
36733 (noce_try_cmove_arith): Likewise. Formatting fixes.
36734 (noce_try_minmax, noce_try_abs): Clear rev_cond.
36735 (noce_find_if_block): Initialize rev_cond.
36736 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
36737 instead of false as last argument never attempt to reverse it
36740 2017-02-23 Bin Cheng <bin.cheng@arm.com>
36742 PR tree-optimization/79663
36743 * tree-predcom.c (combine_chains): Process refs in reverse order
36744 only for ZERO length chains, and add explaining comment.
36746 2017-02-23 Jeff Law <law@redhat.com>
36748 PR tree-optimization/79578
36749 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
36750 in call to operand_equal_p.
36752 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
36755 * config/i386/cpuid.h: Fix another undefined behavior.
36757 2017-02-23 Richard Biener <rguenther@suse.de>
36759 PR tree-optimization/79683
36760 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
36761 vector types for data-refs.
36763 2017-02-23 Martin Liska <mliska@suse.cz>
36765 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
36767 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36769 PR middle-end/79665
36770 * internal-fn.c (get_range_pos_neg): Moved to ...
36771 * tree.c (get_range_pos_neg): ... here. No longer static.
36772 * tree.h (get_range_pos_neg): New prototype.
36773 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
36774 are known to be in between 0 and signed maximum inclusive, try to
36775 expand both unsigned and signed divmod and use the cheaper one from
36778 2017-02-22 Jeff Law <law@redhat.com>
36780 PR tree-optimization/79578
36781 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
36782 to compare base operands.
36784 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
36787 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
36788 gpc_reg_operand instead of fpr_reg_operand.
36790 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
36792 * config/mips/mips.c (mips_return_in_memory): Force FP
36793 vector types to be returned in memory for o32 ABI.
36795 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36797 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
36798 instead of DW_TAG_member for static data member declarations and don't
36799 set no_linkage_name for static inline data members.
36800 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
36803 2017-02-22 Martin Liska <mliska@suse.cz>
36805 * doc/invoke.texi: Replace inequality signs with square brackets
36808 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
36811 * lra-constraints.c (simplify_operand_subreg): Handle
36812 WORD_REGISTER_OPERATIONS targets.
36814 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36817 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
36818 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
36819 elimination by swapping fld*.
36821 2017-02-22 Richard Biener <rguenther@suse.de>
36823 PR tree-optimization/79673
36824 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
36825 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
36826 irrelevant address-space qualifiers and avoiding a
36827 ADDR_SPACE_CONVERT_EXPR from fold_convert.
36829 2017-02-22 Richard Biener <rguenther@suse.de>
36831 PR tree-optimization/79666
36832 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
36833 to not symbolically negate if that may introduce undefined
36836 2017-02-22 Martin Liska <mliska@suse.cz>
36839 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
36840 * data-streamer-out.c (streamer_write_gcov_count_stream):
36842 * value-prof.c (stream_out_histogram_value): Make assert more
36843 precise based on type of counter.
36845 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
36848 * config/i386/i386.md (standard_x87sse_constant_load splitter):
36849 Use nonimmediate_operand instead of memory_operand for operand 1.
36850 (float-extend standard_x87sse_constant_load splitter): Ditto.
36852 2017-02-21 Jeff Law <law@redhat.com>
36854 PR tree-optimization/79621
36855 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
36856 blocks with edges to themselves.
36858 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36861 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
36862 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
36863 Use gimple_call_builtin_p.
36866 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
36867 on temporarily removed DEBUG_INSNs.
36869 PR tree-optimization/79649
36870 * tree-loop-distribution.c (classify_partition): Give up on
36871 non-generic address space loads/stores.
36873 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
36875 * doc/loop.texi (Loop manipulation): Remove nonexistent
36876 tree_ssa_loop_version from the documentation.
36877 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
36879 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36882 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
36883 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
36884 * config/rs6000/rs6000.c: Include except.h.
36885 (rs6000_expand_split_stack_prologue): Call
36886 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
36888 2017-02-21 Martin Jambor <mjambor@suse.cz>
36891 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
36892 have been analyzed.
36894 2017-02-21 Martin Jambor <mjambor@suse.cz>
36896 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
36897 for backward compatibility only.
36898 * doc/invoke.texi (Option Summary): Remove all references to
36899 -fipa-cp-alignment.
36901 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
36905 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36907 * lra-constraints.c (curr_insn_transform): Handle
36908 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36910 2017-02-21 Martin Liska <mliska@suse.cz>
36912 * config/i386/i386.opt: Replace -masm-dialect with -masm.
36914 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
36916 PR translation/79638
36917 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
36919 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
36922 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
36923 (arm_function_ok_for_sibcall): Return false for an indirect call by
36924 descriptor if all the argument registers are used.
36925 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
36926 alignment of the function.
36928 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36930 PR tree-optimization/61441
36931 * simplify-rtx.c (simplify_const_unary_operation): For
36932 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
36933 the sNaN unmodified.
36935 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
36937 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
36938 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
36939 instead of SYSTEM_HEADER_DIR.
36941 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
36942 Martin Liška <mliska@suse.cz>
36944 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
36945 Fix typos and grammar, use active voice, and clarify.
36947 2017-02-20 Marek Polacek <polacek@redhat.com>
36949 PR middle-end/79537
36950 * gimplify.c (gimplify_expr): Handle unused *&&L;.
36953 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
36955 2017-02-20 Jakub Jelinek <jakub@redhat.com>
36958 * config/i386/i386.c (ix86_expand_builtin): Handle
36959 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
36960 ix86_builtins_isa[fcode].isa as a requirement of those
36961 flags and any other flag in the bitmask.
36962 (ix86_init_mmx_sse_builtins): Use 0 instead of
36963 ~OPTION_MASK_ISA_64BIT as mask.
36964 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
36965 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
36966 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
36967 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
36969 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36972 * lra-constraints.c (split_reg): Check requested split mode
36973 is supported by the register.
36975 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36977 * lra-constraints.c (simplify_operand_subreg): Remove early
36980 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36983 * lra-constraints.c (curr_insn_transform): Tighten condition
36984 for converting SUBREG reloads from OP_OUT to OP_INOUT.
36986 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36989 * lra-constraints.c (curr_insn_transform): Handle
36990 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36992 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
36995 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
36997 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
36999 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
37002 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
37005 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37007 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
37010 2017-02-18 Jakub Jelinek <jakub@redhat.com>
37012 * final.c (last_columnnum, override_columnnum): New variables.
37013 (final_start_function): Set last_columnnum, pass it to begin_prologue
37014 hook and pass 0 to dwarf2out_begin_prologue.
37015 (final_scan_insn): Update override_columnnum. Pass last_columnnum
37016 to source_line debug hook.
37017 (notice_source_line): Compute last_columnnum and for debug_column_info
37018 return true on column changes.
37019 * debug.h (struct gcc_debug_hooks): Add column argument to
37020 source_line and begin_prologue hooks.
37021 (debug_nothing_int_charstar_int_bool): Remove prototype.
37022 (debug_nothing_int_int_charstar,
37023 debug_nothing_int_int_charstar_int_bool): New prototypes.
37024 (dwarf2out_begin_prologue): Add column argument.
37025 * debug.c (do_nothing_debug_hooks): Adjust source_line and
37026 begin_prologue hooks.
37027 (debug_nothing_int_charstar_int_bool): Remove.
37028 (debug_nothing_int_int_charstar,
37029 debug_nothing_int_int_charstar_int_bool): New functions.
37030 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
37031 through to dwarf2out_source_line.
37032 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
37033 (dwarf2out_source_line): Add column argument, emit it if requested.
37034 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
37036 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37037 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37038 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
37039 through to dwarf2out_begin_prologue.
37040 (vmsdbgout_source_line): Add column argument, pass it through to
37041 dwarf2out_source_line.
37042 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
37043 dbxout_source_line caller.
37044 (dbxout_source_line): Add column argument.
37046 * common.opt (gno-column-info, gcolumn-info): New options.
37047 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
37048 (check_die): Also test for multiple DW_AT_decl_column attributes.
37049 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
37050 DW_AT_decl_column if requested.
37051 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
37053 (gen_variable_die): Likewise.
37054 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
37055 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
37058 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
37059 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
37060 (ix86_handle_option): Handle OPT_m3dnowa.
37061 * doc/invoke.texi (-m3dnowa): Document.
37062 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
37063 -m3dnowa instead of -m3dnow -march=athlon.
37066 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
37067 instead of gcc_assert for K, r and R code checks. Formatting fixes.
37069 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37072 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
37073 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
37074 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
37075 generator for vsx_xxpermdi_<mode>_be.
37076 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
37077 force big-endian semantics.
37078 (vsx_xxpermdi_<mode>_be): New define_expand with same
37079 implementation as previous version of vsx_xxpermdi_<mode>.
37081 2017-02-17 Jakub Jelinek <jakub@redhat.com>
37083 PR tree-optimization/79327
37084 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
37085 variable, its initialization and use.
37087 2017-02-17 Julia Koval <julia.koval@intel.com>
37089 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
37090 (OPTION_MASK_ISA_PKU_UNSET): New.
37091 (ix86_handle_option): Handle -mrdpid.
37092 * config/i386/cpuid.h (bit_RDPID): New.
37093 * config/i386/driver-i386.c (host_detect_local_cpu):
37094 Detect RDPID feature.
37095 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
37096 * config/i386/i386-c.c (ix86_target_macros_internal):
37098 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
37099 (ix86_valid_target_attribute_inner_p): Add "rdpid".
37100 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
37101 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
37102 * config/i386/i386.md (define_insn "rdpid"): New.
37103 * config/i386/i386.opt Add -mrdpid.
37104 * config/i386/immintrin.h (_rdpid_u32): New.
37106 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
37108 PR rtl-optimization/79541
37109 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
37110 instead of transforming it into USE.
37112 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
37114 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
37115 If HONOR_SNANS (SFmode) force the input to a register.
37116 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
37117 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
37118 an frsp or similar insn.
37120 2017-02-17 Martin Liska <mliska@suse.cz>
37122 PR rtl-optimization/79577
37123 * params.def (selsched-max-sched-times): Increase minimum to 1.
37125 2017-02-17 Martin Liska <mliska@suse.cz>
37127 PR rtl-optimization/79574
37128 * gcse.c (want_to_gcse_p): Prevent integer overflow.
37130 2017-02-17 Martin Liska <mliska@suse.cz>
37132 PR tree-optimization/79529
37133 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
37134 ssa_defined_default_def_p to handle cases which are implicitly
37136 * tree-ssa.c (ssa_defined_default_def_p): New function.
37137 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
37138 which are implicitly defined.
37139 * tree-ssa.h (ssa_defined_default_def_p): Declare.
37141 2017-02-17 Richard Biener <rguenther@suse.de>
37143 PR middle-end/79576
37144 * params.def (max-ssa-name-query-depth): Limit to 10.
37146 2017-02-17 Richard Biener <rguenther@suse.de>
37148 PR tree-optimization/79552
37149 * tree-ssa-structalias.c (visit_loadstore): Properly verify
37152 2017-02-17 Richard Biener <rguenther@suse.de>
37155 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
37157 2017-02-17 Marek Polacek <polacek@redhat.com>
37159 PR middle-end/79536
37160 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
37161 (fold_negate_expr): New wrapper.
37163 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
37165 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
37166 Correct terminology and de-emphasize pre-standard behavior.
37168 2017-02-16 Alan Modra <amodra@gmail.com>
37170 PR rtl-optimization/79286
37171 * ira.c (def_dominates_uses): New function.
37172 (update_equiv_regs): Don't create an equivalence for insns that
37173 may trap where the register def does not dominate the use.
37175 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
37177 PR rtl-optimization/78127
37178 * lra.c (lra): Call lra_eliminate before finish the loop after
37181 2017-02-16 Richard Biener <rguenther@suse.de>
37183 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
37185 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
37186 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
37187 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
37188 (isl_val_int_from_wi): New function.
37189 (extract_affine_gmp): Rename to ...
37190 (extract_affine_wi): ... this, take a widest_int.
37191 (extract_affine_int): Just wrap extract_affine_wi.
37192 (add_param_constraints): Use isl_val_int_from_wi.
37193 (add_loop_constraints): Likewise, and extract_affine_wi.
37195 2017-02-15 Jeff Law <law@redhat.com>
37197 PR middle-end/79521
37198 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
37199 ira_init_register_move_cost_if_necessary.
37201 2017-02-15 Martin Sebor <msebor@redhat.com>
37203 PR middle-end/32003
37204 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
37205 removed in a prior commit.
37207 2017-02-15 Bin Cheng <bin.cheng@arm.com>
37209 PR tree-optimization/79347
37210 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
37211 counters during peeling.
37213 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
37215 * Makefile.in (site.exp): Remove "set ISLVER".
37217 2017-02-15 Jakub Jelinek <jakub@redhat.com>
37220 * real.c (real_from_integer): Call real_convert even for decimal.
37222 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
37225 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
37227 2017-02-14 Andrew Pinski <apinski@cavium.com>
37229 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
37230 cores and change the partno/implementer to be correct.
37231 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
37232 the 'B" as the implementer.
37233 * config/aarch64/aarch64-tune.md: Regenerate.
37235 2017-02-14 Carl Love <cel@us.ibm.com>
37237 * config/rs6000/rs6000.c: Add case statement entry to make the
37238 xvcvuxdsp built-in argument unsigned.
37239 * config/rs6000/vsx.md: Fix the source and return operand types so they
37240 match the instruction definitions from the ISA document. Fix typo
37241 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
37244 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
37247 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
37248 member early_clobber_alts.
37249 * lra-lives.c (reg_early_clobber_p): New.
37250 (process_bb_lives): Use it.
37251 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
37252 (debug_operand_data): Initialize early_clobber_alts.
37253 (setup_operand_alternative): Set up early_clobber_alts.
37254 (collect_non_operand_hard_regs): Ditto. Pass early clobber
37255 alternatives to new_insn_reg.
37256 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
37258 (lra_update_insn_regno_info): Pass the new arg.
37260 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37262 PR middle-end/79505
37263 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
37264 (new_oacc_loop_raw): Don't clear already cleared fields.
37267 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
37268 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
37269 _mm512_prefetch_i64gather_ps): New inline functions and macros.
37271 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
37274 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
37276 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
37279 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
37280 the extra instruction to the right place to store 128-bit constant
37283 2017-02-14 Martin Sebor <msebor@redhat.com>
37285 PR middle-end/79448
37286 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
37287 warning for strings of unknown length.
37289 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
37291 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
37293 2017-02-14 Jeff Law <law@redhat.com>
37296 * ira-costs.c (scan_one_insn): Initialize register move costs
37297 for pseudos seen in USE/CLOBBER insns.
37299 PR tree-optimization/79095
37300 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
37301 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
37302 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
37303 if the operands are known to be not equal, then the resulting range
37305 (intersect_ranges): If the new range is ~[0,0] and the old range is
37306 wide, then prefer ~[0,0].
37307 * tree-vrp.c (overflow_comparison_p_1): New function.
37308 (overflow_comparison_p): New function.
37309 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
37310 if NAME is used in an overflow test.
37311 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
37312 overflow check that can be expressed as an equality test, then adjust
37313 ops to be that equality test.
37315 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37317 * config/s390/s390-builtin-types.def: Remove flags argument.
37318 * config/s390/s390.c (s390_init_builtins): Likewise.
37320 2017-02-14 Martin Liska <mliska@suse.cz>
37322 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
37323 vector. Fix trailing white spaces.
37325 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
37327 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
37330 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37332 PR rtl-optimization/68664
37333 * config/arm/arm.c (arm_sched_can_speculate_insn):
37334 New function. Declare prototype.
37335 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37337 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37339 PR rtl-optimization/68664
37340 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
37342 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37344 2017-02-14 Amit Pawar <amit.pawar@amd.com>
37346 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
37347 max skip bytes for function, loop and jump.
37349 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37351 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
37352 ABS_EXPR for gimple dump.
37354 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37357 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
37359 PR tree-optimization/79408
37360 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
37361 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
37362 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
37363 also if rhs1 is INTEGER_CST.
37365 2017-02-14 Richard Biener <rguenther@suse.de>
37367 PR middle-end/79432
37368 * tree-into-ssa.c (insert_phi_nodes): When the function can
37369 have abnormal edges rewrite SSA names with broken use-def
37370 dominance out of SSA and register them for PHI insertion.
37372 2017-02-13 Martin Sebor <msebor@redhat.com>
37374 PR middle-end/79496
37375 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
37376 clearing info.nowrite flag when snprintf size argument is a range.
37378 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37380 * cprop.c (cprop_jump): Add missing space in string literal.
37381 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
37382 (get_constraint_for_component_ref): Likewise.
37383 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
37384 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
37385 * lra-constraints.c (process_alt_operands): Likewise.
37386 * ipa-inline.c (inline_small_functions): Likewise.
37387 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
37388 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
37389 * trans-mem.c (diagnose_tm_1_op): Likewise.
37390 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
37391 (grid_parallel_clauses_gridifiable): Likewise.
37393 * config/nvptx/mkoffload.c (process): Add space in between
37396 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
37397 "MOD4_SSE_REGS" and "ALL_REGS".
37399 * spellcheck.c (test_data): Add , in between "foo" and "food".
37401 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37404 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
37405 boundary crossing check and subsequent code generation agree.
37407 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37409 * config/aarch64/aarch64.c (has_memory_op): Delete.
37410 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
37413 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37415 PR rtl-optimization/79388
37416 PR rtl-optimization/79450
37417 * combine.c (distribute_notes): When removing TEM_INSN for which
37418 corresponding dest has last value recorded, invalidate that last
37421 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37423 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
37424 of explicit '@'. Add missing assembly comment marker on branch costs
37427 2017-02-13 Nathan Sidwell <nathan@acm.org>
37429 * gengtype-lex.l (<in_struct>): Add '/'.
37431 2017-02-13 Martin Liska <mliska@suse.cz>
37434 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
37436 2017-02-13 Richard Biener <rguenther@suse.de>
37438 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
37440 * configure: Re-generate.
37441 * config.in: Likewise.
37442 * graphite-dependences.c: Simplify as if
37443 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
37444 * graphite-isl-ast-to-gimple.c: Likewise.
37445 * graphite-optimize-isl.c: Likewise.
37446 * graphite-poly.c: Likewise.
37447 * graphite-sese-to-poly.c: Likewise.
37448 * graphite.h: Likewise.
37449 * toplev.c: Include isl/version.h and use isl_version () for
37450 printing the ISL version.
37451 * doc/install.texi: Update ISL requirement.
37453 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37455 * doc/standards.texi (Standards): Update reference to
37458 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37460 * doc/extend.texi (Named Address Spaces): sourceware.org now
37462 * doc/install.texi (Binaries): Ditto.
37465 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37467 * doc/cpp.texi: Replace "stringify"/"stringification" with C
37468 standard terminology "stringize"/"stringizing" throughout.
37469 * doc/cppinternals.texi: Likewise.
37471 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37473 * doc/extend.texi: Fix some spelling mistakes and typos.
37474 * doc/invoke.texi: Likewise.
37476 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37479 * params.def (inline-min-speedup) Change from 10 to 8.
37481 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37483 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
37486 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37489 * ipa-inline-analysis.c (get_minimal_bb): New function.
37490 (record_modified): Use it.
37491 (remap_edge_change_prob): Handle also ancestor functions.
37493 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
37495 * doc/contrib.texi (Contributors): Remove broken link into
37496 the Mauve CVS repository.
37498 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37500 PR middle-end/79454
37501 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
37502 result computation whenever lhs doesn't have vector mode, not
37503 just when it has BLKmode.
37505 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37507 * doc/makefile.texi (profiledbootstrap): Refer to the
37508 installation instructions only in textual form.
37510 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37513 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
37515 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37517 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
37518 (Specific): Update mingw-w64 reference.
37520 (Specific): Remove broken link to Renesas RX processor.
37522 2017-02-10 Richard Biener <rguenther@suse.de>
37524 * toplev.c (process_options): Do not mention obsolete graphite
37525 options when printing sorry message about missing graphite support.
37526 Mention -floop-nest-optimize.
37528 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
37530 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
37531 (vtst_p16): Likewise.
37532 (vtstq_p8): Likewise.
37533 (vtstq_p16): Likewise.
37535 (vtstq_p64): Likewise.
37536 * config/arm/arm_neon.h (vgetq_lane_p64): New.
37537 (vset_lane_p64): New.
37538 (vsetq_lane_p64): New.
37540 2017-02-10 Jakub Jelinek <jakub@redhat.com>
37542 PR tree-optimization/79411
37543 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
37544 stmt operands are SSA_NAMEs used in abnormal phis.
37545 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
37548 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37551 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
37554 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37556 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
37558 2017-02-09 Jakub Jelinek <jakub@redhat.com>
37560 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
37564 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
37565 not arbitrary TREE_CONSTANT.
37568 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
37569 "omp declare target link" attribute unless is_global_var.
37570 * omp-offload.c (find_link_var_op): Likewise.
37572 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
37573 Chung-Lin Tang <cltang@codesourcery.com>
37575 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
37577 (gimplify_adjust_omp_clauses): Don't delete TILE.
37578 (gimplify_omp_for): Deal with TILE.
37579 * internal-fn.c (expand_GOACC_TILE): New function.
37580 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
37582 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
37583 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
37585 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
37586 avoid DIV for outermost collapse var.
37587 (expand_oacc_for): Insert tile element loop as needed. Adjust.
37588 Remove out of date comments, fix whitespace.
37589 * omp-general.c (omp_extract_for_data): Deal with tiling.
37590 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
37591 adjust OLF_DIM_BASE value.
37592 (struct omp_for_data): Add tiling field.
37593 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
37594 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
37595 for auto loops. Remove default auto determining, moved to
37596 oacc_loop_fixed_partitions.
37597 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
37598 stmts, add e_mask field.
37599 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
37600 (oacc_thread_numbers): Use oacc_dim_call.
37601 (oacc_xform_tile): New.
37602 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
37603 (finish_oacc_loop): Adjust for ifns vector.
37604 (oacc_loop_discover_walk): Append loop abstraction sites to list,
37605 add case for GOACC_TILE fns.
37606 (oacc_loop_xform_loop): Delete.
37607 (oacc_loop_process): Iterate over call list directly, and add
37608 handling for GOACC_TILE fns.
37609 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
37611 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
37612 vector partitioning to outer loops. Assign 2 partitions to loops
37613 when available. Add TILE handling.
37614 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
37615 (execite_oacc_device_lower): Process GOACC_TILE fns,
37616 ignore unknown specs.
37617 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
37618 * tree.c (omp_clause_num_ops): Adjust TILE ops.
37619 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
37621 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
37623 * configure.ac (ACX_BUGURL): Update.
37624 * configure: Regenerate.
37626 2017-02-09 Richard Biener <rguenther@suse.de>
37628 PR tree-optimization/69823
37629 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
37630 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
37632 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37634 * config/arc/arc-c.def: Add __NPS400__ definition.
37635 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
37636 (TARGET_NPS400): Define.
37638 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37640 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
37642 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
37643 pointer, arch_info.
37644 (arc_cpu_types): Fill the arch_info field with a pointer into the
37645 arc_arch_types table.
37646 (arc_selected_cpu): Declare.
37647 * config/arc/arc.c (arc_selected_cpu): Make global.
37648 (arc_selected_arch): Delete.
37649 (arc_base_cpu): Delete.
37650 (arc_override_options): Remove references to deleted variables,
37651 update access to arch information.
37652 (ARC_OPT): Update access to arch information.
37653 (ARC_OPTX): Likewise.
37654 * config/arc/arc.h (arc_base_cpu): Remove declaration.
37655 (TARGET_ARC600): Update access to arch information.
37656 (TARGET_ARC601): Likewise.
37657 (TARGET_ARC700): Likewise.
37658 (TARGET_EM): Likewise.
37659 (TARGET_HS): Likewise.
37660 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
37663 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
37666 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
37667 condition/operands for integer GE/LE/GEU/LEU operations.
37669 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
37671 PR translation/79397
37672 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
37675 2017-02-08 Martin Jambor <mjambor@suse.cz>
37678 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
37679 whether allocation happened.
37680 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
37681 nothing was allocated.
37683 2017-02-08 Jakub Jelinek <jakub@redhat.com>
37685 PR tree-optimization/79408
37686 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
37687 constant, but SSA_NAME with a known integer range, use the minimum
37688 of that range instead of op1 to determine if modulo can be replaced
37689 with its first operand.
37691 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37693 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
37695 2017-02-08 Richard Biener <rguenther@suse.de>
37697 PR tree-optimization/71824
37698 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37699 Check all loops contained in the merged region.
37701 2017-02-07 Andrew Pinski <apinski@cavium.com>
37703 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
37705 2017-02-07 Andrew Pinski <apinski@cavium.com>
37707 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
37708 (thunderxt88): Likewise.
37709 (thunderxt81): Disable LSE and change v8.1 to v8.
37710 (thunderxt83): Likewise.
37712 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37713 Richard Biener <rguenther@suse.de>
37715 PR middle-end/79399
37716 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
37717 type from int to size_t.
37718 * ira-costs.c (struct_costs_size): Change type from int to size_t.
37720 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37722 PR rtl-optimization/79386
37723 * cprop.c (bypass_conditional_jumps): Initialize
37724 bypass_last_basic_block already before splitting bbs after
37725 unconditional traps...
37726 (bypass_conditional_jumps): ... rather than here.
37729 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
37730 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
37731 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
37732 fix -masm=intel patterns.
37734 2017-02-07 Richard Biener <rguenther@suse.de>
37736 PR tree-optimization/79256
37737 PR middle-end/79278
37738 * builtins.c (get_object_alignment_2): Use min_align_of_type
37739 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
37740 and ADJUST_FIELD_ALIGN.
37742 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
37744 * doc/tm.texi: Regenerate.
37745 * stor-layout.c (layout_decl): Adjust.
37746 (update_alignment_for_field): Likewise.
37747 (place_field): Likewise.
37748 (min_align_of_type): Likewise.
37749 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
37750 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
37751 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
37752 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
37753 * config/frv/frv.c (frv_adjust_field_align): Likewise.
37754 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
37755 * config/i386/i386.c (x86_field_alignment): Likewise.
37756 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
37757 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
37758 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
37759 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
37760 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
37761 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
37765 2017-01-30 Richard Biener <rguenther@suse.de>
37767 PR tree-optimization/79256
37768 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37769 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37772 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
37774 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
37775 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
37776 builtins to SImode and emit a zero-extend, if necessary.
37778 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37780 * docs/invoke.texi (RISC-V Options): Alphabetize.
37782 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37784 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
37787 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37789 * config/riscv/riscv.c: New file.
37790 * common/config/riscv/riscv-common.c: Likewise.
37791 * config.gcc: Likewise.
37792 * config/riscv/constraints.md: Likewise.
37793 * config/riscv/elf.h: Likewise.
37794 * config/riscv/generic.md: Likewise.
37795 * config/riscv/linux.h: Likewise.
37796 * config/riscv/multilib-generator: Likewise.
37797 * config/riscv/peephole.md: Likewise.
37798 * config/riscv/pic.md: Likewise.
37799 * config/riscv/predicates.md: Likewise.
37800 * config/riscv/riscv-builtins.c: Likewise.
37801 * config/riscv/riscv-c.c: Likewise.
37802 * config/riscv/riscv-ftypes.def: Likewise.
37803 * config/riscv/riscv-modes.def: Likewise.
37804 * config/riscv/riscv-opts.h: Likewise.
37805 * config/riscv/riscv-protos.h: Likewise.
37806 * config/riscv/riscv.h: Likewise.
37807 * config/riscv/riscv.md: Likewise.
37808 * config/riscv/riscv.opt: Likewise.
37809 * config/riscv/sync.md: Likewise.
37810 * config/riscv/t-elf-multilib: Likewise.
37811 * config/riscv/t-linux: Likewise.
37812 * config/riscv/t-linux-multilib: Likewise.
37813 * config/riscv/t-riscv: Likewise.
37814 * configure.ac: Likewise.
37815 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
37816 Waterman as RISC-V maintainers.
37817 * doc/install.texi: Add RISC-V entries.
37818 * doc/invoke.texi: Add RISC-V options section.
37819 * doc/md.texi: Add RISC-V constraints section.
37820 * configure: Regenerated.
37822 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
37825 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
37826 false values to be constant vectors with all 0 or all 1 bits set.
37827 (vcondu<mode><mode>): Likewise.
37828 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
37830 (fpmask_comparison_operator): Update comment.
37831 (vecint_comparison_operator): New predicate.
37832 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
37833 vector conditionals when the true and false values are constant
37834 vectors with all 0 bits or all 1 bits set.
37836 2017-02-06 Martin Sebor <msebor@redhat.com>
37838 PR tree-optimization/79376
37839 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
37841 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
37843 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
37844 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
37845 to simplify split condition.
37847 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37849 * omp-expand.c (oxpand_omp_atomic_fetch_op,
37850 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
37853 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
37855 PR rtl-optimization/68664
37856 * target.def (can_speculate_insn): New hook.
37857 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
37858 * doc/tm.texi: Regenerate.
37859 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
37860 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
37861 (rs6000_sched_can_speculate_insn): New function.
37863 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37865 PR tree-optimization/79284
37866 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
37867 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
37868 vectorizable_mask_load_store, vectorizable_operation,
37869 vect_is_simple_cond, get_same_sized_vectype): Use it instead
37870 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
37871 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
37872 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
37873 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
37874 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
37875 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
37876 is_gimple_assign (stmt). Replace another such test with
37877 is_gimple_assign (stmt).
37879 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37882 * config/avr/avr.c (rtl-iter.h): Include it.
37883 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
37884 (avr_legitimate_combined_insn): ...and implementation.
37886 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
37888 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
37889 * config/s390/s390.c (s390_const_operand_ok)
37890 (s390_canonicalize_comparison, s390_extract_part)
37891 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
37892 (s390_contiguous_bitmask_p, s390_rtx_costs)
37893 (legitimize_pic_address): Likewise.
37894 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
37895 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
37896 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
37897 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
37898 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
37900 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37902 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
37903 REGNO($0) == REGNO($1).
37905 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37907 * config/s390/linux.h(SIZE_TYPE): Add comment.
37909 2017-02-06 Julian Brown <julian@codesourcery.com>
37910 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37911 Virendra Pathak <virendra.pathak@broadcom.com>
37913 * config/aarch64/aarch64-cores.def: Change the scheduler
37915 * config/aarch64/aarch64.md: Include thunderx2t99.md.
37916 * config/aarch64/thunderx2t99.md: New file.
37918 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37920 * doc/standards.texi (Go Language): Update link to language
37923 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37925 * tree-eh.c (lower_resx): Sanitize profile.
37926 (cleanup_empty_eh_move_lp): Likewise.
37928 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37931 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
37933 * cfgloopmanip.h (loop_version): Update prototype.
37934 * modulo-sched.c (sms_schedule): Update call of loop_version.
37935 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
37936 * tree-parloops.c (gen_parallel_loop): Likewise.
37937 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
37938 * tree-ssa-loop-split.c (split_loop): Likewise.
37939 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
37940 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
37942 2017-02-05 Martin Liska <mliska@suse.cz>
37945 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
37947 (print_operand_address): Initialize a struct to zero.
37949 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37951 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
37952 garbage collector only in textual form.
37954 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37956 * doc/extend.texi (x86 specific memory model extensions for
37957 transactional memory): Simplify a phrase.
37959 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
37962 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
37963 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
37964 (atomic_storedi_1): Likewise.
37966 2017-02-04 Jakub Jelinek <jakub@redhat.com>
37968 PR tree-optimization/79338
37969 * tree-parloops.c (gather_scalar_reductions): Don't call
37970 vect_analyze_loop_form for loop->inner before destroying loop's
37973 2017-02-03 Martin Sebor <msebor@redhat.com>
37975 PR tree-optimization/79327
37976 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
37977 when precision has resulted in leading zeros.
37978 (format_integer): Adjust the likely counter to assume an unknown
37979 argument that may be zero is non-zero.
37981 2017-02-03 Jason Merrill <jason@redhat.com>
37984 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
37985 avoid copying non-taken branch.
37987 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37989 PR tree-optimization/79340
37990 * tree-vect-loop.c (vectorizable_reduction): Release
37991 vec_defs elements after safe_splicing them into other vectors.
37994 PR tree-optimization/79327
37995 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
37996 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
37998 (format_integer): Use wide_int_to_tree instead of build_int_cst
37999 + to_?hwi. If argmin is NULL, just set argmin and argmax to
38000 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
38001 of shortest and longest sequence.
38003 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
38005 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
38006 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
38008 2017-02-03 Walter Lee <walt@tilera.com>
38011 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
38012 after initial stackframe link reg save.
38013 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
38015 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38018 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
38019 wu for stxssp alternative.
38021 2017-02-03 Martin Sebor <msebor@redhat.com>
38023 PR tree-optimization/79352
38024 * gimple-fold.c (get_range_strlen): Add argument.
38025 (get_range_strlen): Change return type to bool.
38026 (get_maxval_strlen): Pass in a dummy argument.
38027 * gimple-fold.h (get_range_strlen): Change return type to bool.
38028 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
38029 * tree.h (array_at_struct_end_p): Add argument.
38030 * tree.c (array_at_struct_end_p): Handle it.
38032 2017-02-03 Martin Liska <mliska@suse.cz>
38035 * multiple_target.c (create_dispatcher_calls): Redirect edge
38036 from a caller of a dispatcher.
38037 (expand_target_clones): Make the clones local.
38038 (ipa_target_clone): Do both target clones and resolvers.
38039 (ipa_dispatcher_calls): Remove the pass.
38040 (pass_dispatcher_calls::gate): Likewise.
38041 (make_pass_dispatcher_calls): Likewise.
38042 * passes.def (pass_target_clone): Put as very first IPA early
38045 2017-02-03 Martin Liska <mliska@suse.cz>
38047 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
38048 in case of a function with ifunc attribute.
38050 2017-02-03 Martin Liska <mliska@suse.cz>
38052 * cgraph.c (cgraph_node::dump): Dump function version info.
38053 * symtab.c (symtab_node::dump_base): Add missing new line.
38055 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38057 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
38058 (ifcombine_ifandif): Use it.
38060 2017-02-03 Martin Liska <mliska@suse.cz>
38062 * doc/invoke.texi: Document default value for
38063 use-after-scope-direct-emission-threshold.
38065 2017-02-03 Martin Liska <mliska@suse.cz>
38067 PR tree-optimization/79339
38068 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
38069 (format_floating): Likewise.
38071 2017-02-03 Martin Liska <mliska@suse.cz>
38074 * ipa-prop.c (ipa_node_params_t::insert): Remove current
38076 (ipa_node_params_t::remove): Likewise.
38077 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
38078 initialization from removed ipa_node_params_t::insert.
38079 (ipa_node_params::~ipa_node_params): Move from removed
38080 ipa_node_params_t::release.
38081 * symbol-summary.h (symbol_summary::m_released): New member.
38082 Do not release a summary twice. Do not allow to call finalizer
38083 for types of a summary that live in GGC memory.
38085 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38087 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
38090 2017-02-02 Martin Sebor <msebor@redhat.com>
38092 PR middle-end/79275
38093 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
38094 (format_string): Tighten up the range of output for non-constant
38095 strings and correct the expected range for wide non-constant strings.
38097 2017-02-02 Martin Sebor <msebor@redhat.com>
38099 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
38101 PR middle-end/32003
38102 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
38104 (-fdump-tree-@var): Add to index and document how to come up
38105 with pass-specific option and dump file names.
38106 (-fdump-passes): Clarify where to look for output.
38108 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38110 PR middle-end/77445
38111 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
38112 statistics of the analyzed path; allow threading for speed when
38113 any of BBs along the path are optimized for speed.
38115 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
38117 PR middle-end/78468
38118 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
38119 settings of the virtual registers.
38122 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
38124 * explow.c (get_dynamic_stack_size): Take known alignment of stack
38125 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
38128 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38130 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
38131 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
38133 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38135 * config/s390/s390.md: Add missing comments with the expanded
38137 * config/s390/vector.md: Likewise.
38138 * config/s390/vx-builtins.md: Likewise.
38140 2017-02-02 Jakub Jelinek <jakub@redhat.com>
38143 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
38144 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
38145 conditions on a single line.
38147 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38149 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38150 __S390_VX__ to __VX__.
38152 2017-02-01 Andrew Pinski <apinski@cavium.com>
38154 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
38155 stmt_info to record_stmt_cost.
38156 (vect_get_known_peeling_cost): Pass stmt_info if known to
38158 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
38159 cpu_vector_cost field into
38160 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
38161 field into vec_int_stmt_cost and vec_fp_stmt_cost.
38162 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
38163 splitting of scalar_stmt_cost and vec_stmt_cost.
38164 (thunderx_vector_cost): Likewise.
38165 (cortexa57_vector_cost): LIkewise.
38166 (exynosm1_vector_cost): Likewise.
38167 (xgene1_vector_cost): Likewise.
38168 (thunderx2t99_vector_cost): Improve after the splitting of the two
38170 (aarch64_builtin_vectorization_cost): Update for the splitting of
38171 scalar_stmt_cost and vec_stmt_cost.
38173 2017-02-01 Torvald Riegel <triegel@redhat.com>
38174 Richard Henderson <rth@redhat.com>
38176 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
38177 conditional on existance of a fast atomic load.
38178 * optabs-query.c (can_atomic_load_p): New function.
38179 * optabs-query.h (can_atomic_load_p): Declare it.
38180 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
38181 no fast atomic load is available for the particular size of access.
38182 (expand_atomic_compare_and_swap): Likewise.
38183 (expand_atomic_load): Likewise.
38184 (expand_atomic_store): Likewise.
38185 (expand_atomic_fetch_op): Likewise.
38186 * testsuite/lib/target-supports.exp
38187 (check_effective_target_sync_int_128): Remove x86 because it provides
38188 no fast atomic load.
38189 (check_effective_target_sync_int_128_runtime): Likewise.
38191 2017-02-01 Richard Biener <rguenther@suse.de>
38193 * graphite.c: Include tree-vectorizer.h for find_loop_location.
38194 (graphite_transform_loops): Provide opt-info for optimized nests.
38195 * tree-parloop.c (parallelize_loops): Provide opt-info for
38196 parallelized loops.
38198 2017-02-01 Richard Biener <rguenther@suse.de>
38200 PR middle-end/79315
38201 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
38202 was not set before.
38204 2017-02-01 Richard Biener <rguenther@suse.de>
38206 PR tree-optimization/71824
38207 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38208 Verify the loops are valid in the merged SESE region.
38209 (scop_detection::can_represent_loop_1): Check analyzing the
38210 evolution of the number of iterations in the region succeeds.
38212 2017-01-31 Ian Lance Taylor <iant@golang.org>
38214 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
38215 REG_ARGS_SIZE note to 32-bit push insns and call insn.
38217 2017-01-31 David Malcolm <dmalcolm@redhat.com>
38219 PR preprocessor/79210
38220 * input.c (get_substring_ranges_for_loc): Replace line_width
38221 assertion with error-handling.
38223 2017-01-31 Richard Biener <rguenther@suse.de>
38225 PR tree-optimization/77318
38226 * graphite-sese-to-poly.c (extract_affine): Fix assert.
38227 (create_pw_aff_from_tree): Take loop parameter.
38228 (add_condition_to_pbb): Pass loop of the condition to
38229 create_pw_aff_from_tree.
38231 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38233 * config/s390/s390.c (s390_asan_shadow_offset): New function.
38234 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
38236 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
38240 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
38242 (convert_int_to_float128): Likewise.
38243 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
38244 (convert_int_to_float128): Likewise.
38245 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
38246 (UNSPEC_IEEE128_CONVERT): Likewise.
38247 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
38248 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
38249 Use local variables for IBM extended format.
38250 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
38251 (fix_trunc<mode>si2_fprs): Likewise.
38252 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
38253 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
38254 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
38255 to know that we can now have integers of all sizes in vector
38257 (fix<uns>_<mode>di2_hw): Likewise.
38258 (float<uns>_<mode>si2_hw): Likewise.
38259 (fix_<mode>si2_hw): Likewise.
38260 (fixuns_<mode>si2_hw): Likewise.
38261 (float<uns>_<mode>di2_hw): Likewise.
38262 (float_<mode>di2_hw): Likewise.
38263 (float_<mode>si2_hw): Likewise.
38264 (floatuns_<mode>di2_hw): Likewise.
38265 (floatuns_<mode>si2_hw): Likewise.
38266 (xscvqp<su>wz_<mode>): Delete, no longer used.
38267 (xscvqp<su>dz_<mode>): Likewise.
38268 (xscv<su>dqp_<mode>): Likewise.
38269 (ieee128_mfvsrd_64bit): Likewise.
38270 (ieee128_mfvsrd_32bit): Likewise.
38271 (ieee128_mfvsrwz): Likewise.
38272 (ieee128_mtvsrw): Likewise.
38273 (ieee128_mtvsrd_64bit): Likewise.
38274 (ieee128_mtvsrd_32bit): Likewise.
38276 2017-01-31 Martin Liska <mliska@suse.cz>
38279 * ipa-prop.c (ipa_free_all_node_params): Call release method
38280 instead of ~sumbol_summary to not to trigger double times
38283 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
38285 PR tree-optimization/71691
38286 * bitmap.h (class auto_bitmap): New.
38287 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
38288 is_maybe_undefined instead of ssa_undefined_value_p.
38290 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38292 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38293 __S390_ARCH_LEVEL__ to __ARCH__.
38295 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38297 PR tree-optimization/79267
38298 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
38299 if should_remove_lhs_p is true.
38301 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
38304 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
38305 (add_alignment_attribute): New.
38306 (base_type_die): Add alignment attribute.
38307 (subrange_type_die): Likewise.
38308 (modified_type_die): Likewise.
38309 (gen_array_type_die): Likewise.
38310 (gen_descr_array_type_die: Likewise.
38311 (gen_enumeration_type_die): Likewise.
38312 (gen_subprogram_die): Likewise.
38313 (gen_variable_die): Likewise.
38314 (gen_field_die): Likewise.
38315 (gen_ptr_to_mbr_type_die): Likewise.
38316 (gen_struct_or_union_type_die): Likewise.
38317 (gen_subroutine_type_die): Likewise.
38318 (gen_typedef_die): Likewise.
38319 (base_type_cmp): Compare alignment attribute.
38321 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38324 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
38325 (setb_unsigned) New pattern for setb with CCUNS.
38326 * config/rs6000/rs6000.c (expand_block_compare): Use a different
38327 subfc./subfe sequence to avoid overflow problems. Generate a
38328 shorter sequence with cmpld/setb for power9.
38329 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
38330 for generating subfc. instruction.
38331 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
38332 now uses this instruction.
38334 2017-01-30 Ian Lance Taylor <iant@google.com>
38337 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
38338 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
38340 2017-01-30 Martin Sebor <msebor@redhat.com>
38342 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
38343 Move constant to the right of a relational operator.
38344 (get_mpfr_format_length, format_character, format_string): Ditto.
38345 (should_warn_p, maybe_warn): Same.
38347 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
38349 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
38352 * asan.c (get_translation_unit_decl): Remove function.
38353 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
38355 2017-01-30 Martin Liska <mliska@suse.cz>
38357 PR gcov-profile/79259
38358 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
38359 -fprofile-generate.
38361 2017-01-30 Martin Liska <mliska@suse.cz>
38364 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
38365 Initialize variables with NULL value.
38367 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
38370 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
38372 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
38374 2017-01-30 Richard Biener <rguenther@suse.de>
38376 PR tree-optimization/79276
38377 * tree-vrp.c (process_assert_insertions): Properly adjust common
38378 when removing a duplicate.
38380 2017-01-30 Richard Biener <rguenther@suse.de>
38382 PR tree-optimization/79256
38383 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38384 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38386 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
38388 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
38391 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
38392 ("*r<noxa>sbg_<mode>_sll_bitmask")
38393 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
38394 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
38395 Use contiguous_bitmask_nowrap_operand.
38397 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38400 * config/rs6000/altivec.h (vec_xl): Revise #define.
38401 (vec_xst): Likewise.
38403 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
38405 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
38407 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
38409 PR rtl-optimization/79194
38410 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
38411 traps before call to bypass_conditional_jumps.
38413 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38415 PR tree-optimization/71374
38416 * lra-constraints.c (check_conflict_input_operands): New.
38417 (match_reload): Use it.
38419 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38422 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
38423 account to calculate conflict_set.
38425 2017-01-27 Bin Cheng <bin.cheng@arm.com>
38427 PR rtl-optimization/78559
38428 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
38429 other_insn in combine.
38431 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
38433 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
38434 uint16_type_node for BT_UINT16.
38436 2017-01-27 David Malcolm <dmalcolm@redhat.com>
38438 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
38439 "RTL Tests" to menu.
38440 (GIMPLE Tests): New node.
38441 (RTL Tests): New node.
38443 2017-01-27 Richard Biener <rguenther@suse.de>
38445 PR tree-optimization/79245
38446 * tree-loop-distribution.c (distribute_loop): Apply cost
38447 modeling also to detected patterns.
38449 2017-01-27 Richard Biener <rguenther@suse.de>
38451 PR tree-optimization/71433
38452 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
38453 (compare_assert_loc): New function.
38454 (process_assert_insertions): Sort and optimize assert locations
38455 to remove duplicates and push down identical assertions on
38456 edges to their destination block.
38458 2017-01-27 Richard Biener <rguenther@suse.de>
38460 PR tree-optimization/79244
38461 * tree-vrp.c (remove_range_assertions): Forcefully propagate
38462 out SSA names even if abnormal.
38464 2017-01-27 Jakub Jelinek <jakub@redhat.com>
38466 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
38467 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
38468 instead of MPFR_RNDN.
38470 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
38473 * arm.c (arm_option_override): Don't call build_target_option_node
38474 until after doing all option overrides.
38475 (arm_valid_target_attribute_tree): Likewise.
38477 2017-01-27 Martin Liska <mliska@suse.cz>
38479 * doc/invoke.texi (-fprofile-arcs): Document profiling support
38480 for {cd}tors and C++ {cd}tors.
38482 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
38484 * config/s390/s390.md ("*setmem_long_and")
38485 ("*setmem_long_and_31z"): Use zero_extend instead of and.
38487 2017-01-26 Martin Sebor <msebor@redhat.com>
38489 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
38492 2017-01-26 Martin Sebor <msebor@redhat.com>
38494 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
38495 HAVE_DFmode before using XFmode or DFmode.
38496 (parse_directive): Avoid using the z length modifier to avoid
38497 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
38499 PR middle-end/78703
38500 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
38501 to accept adjustment as an array.
38502 (get_int_range): New function.
38503 (struct directive): Make width and prec arrays.
38504 (directive::set_width, directive::set_precision): Call get_int_range.
38505 (format_integer, format_floating): Handle width and precision ranges.
38506 (format_string, parse_directive): Same.
38508 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38511 * dwarf2out.c (generate_skeleton_bottom_up): For children with
38512 comdat_type_p set, just clone them, but keep the children in the
38516 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
38517 which have direct callers with -fvar-tracking-assignments enabled
38519 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
38520 inside of type units.
38522 2017-01-26 Martin Sebor <msebor@redhat.com>
38524 PR middle-end/78703
38525 * gimple-ssa-sprintf.c (struct result_range): Add likely and
38527 (struct format_result): Replace number_chars, number_chars_min,
38528 and number_chars_max with a single member of struct result_range.
38530 (format_result::operator+=): Adjust.
38531 (struct fmtresult): Remove bounded. Handle likely and unlikely
38533 (fmtresult::adjust_for_width_or_precision): New function.
38534 (fmtresult:type_max_digits): New function.
38535 (bytes_remaining): Handle likely and unlikely counters.
38536 (min_bytes_remaining): Remove.
38537 (format_percent): Simplify.
38538 (format_integer, format_floating): Set likely and unlikely counters.
38539 (get_string_length, format_character, format_string): Same.
38540 (format_plain, should_warn_p): New function.
38541 (maybe_warn): Call should_warn_p. Update diagnostic messages
38542 and handle those for all directives, including plain strings.
38543 (format_directive): Handle likely and unlikely counters.
38544 Remove unnecessary quoting from diagnostics. Add an informational
38546 (add_bytes): Remove.
38547 (pass_sprintf_length::compute_format_length): Simplify.
38548 (try_substitute_return_value): Handle likely and unlikely counters.
38550 2017-01-26 Carl Love <cel@us.ibm.com>
38552 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
38553 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
38555 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
38558 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
38559 endianess for subregs into account.
38560 * lra-constraints.c (lra_constraints): Do risky transformations
38561 always on the first iteration.
38562 * lra-lives.c (check_pseudos_live_through_calls): Add arg
38563 last_call_used_reg_set.
38564 (process_bb_lives): Define and use last_call_used_reg_set.
38565 * lra.c (lra): Always continue after lra_constraints on the first
38568 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
38570 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
38572 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
38574 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38576 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
38577 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
38578 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
38579 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
38580 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
38581 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
38582 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
38583 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
38584 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
38586 2017-01-26 Marek Polacek <polacek@redhat.com>
38589 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
38590 for the third operand.
38592 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38594 PR middle-end/79236
38595 * omp-low.c (struct omp_context): Add simt_stmt field.
38596 (scan_omp_for): Return omp_context *.
38597 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
38598 context to the _simt_ SIMD stmt.
38599 (lower_omp_for): For combined SIMD with sibling _simt_
38600 SIMD, make sure to use the same decls in _looptemp_
38601 clauses as in the sibling.
38603 2017-01-26 David Sherwood <david.sherwood@arm.com>
38605 PR middle-end/79212
38606 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
38609 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38612 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
38613 emit fld b; fld a; if possible.
38615 * brig-builtins.def: Update copyright years.
38616 * config/arm/arm_acle_builtins.def: Update copyright years.
38618 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
38621 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
38622 constraint instead of o for the stxsd instruction.
38624 2017-01-25 Carl Love <cel@us.ibm.com>
38626 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
38627 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
38629 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
38631 * doc/invoke.texi (C++ Dialect Options): Fix typo.
38633 2017-01-25 Richard Biener <rguenther@suse.de>
38635 PR tree-optimization/69264
38636 * target.def (vector_alignment_reachable): Improve documentation.
38637 * doc/tm.texi: Regenerate.
38638 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
38640 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
38641 earlier changes with respect to TYPE_USER_ALIGN.
38642 (vector_alignment_reachable_p): Likewise. Improve dumping.
38644 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38647 * config/arm/arm.md (xordi3): Force constant operand into a register
38650 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38652 * doc/invoke.texi (-fstore-merging): Correct default optimization
38653 levels at which it is enabled.
38654 (-O): Move -fstore-merging from list to...
38657 2017-01-25 Richard Biener <rguenther@suse.de>
38660 * omp-expand.c: Include debug.h.
38661 (expand_omp_taskreg): Make sure to generate early debug before
38662 outlining anything from a function.
38663 (expand_omp_target): Likewise.
38664 (grid_expand_target_grid_body): Likewise.
38666 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
38669 * asan.c (get_translation_unit_decl): New function.
38670 (asan_add_global): Extract modules file name from globals
38671 TRANSLATION_UNIT_DECL name.
38673 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
38676 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
38677 for long calls with APCS frame and VFP.
38679 2017-01-24 David Malcolm <dmalcolm@redhat.com>
38681 * cfg.c (original_copy_tables_initialized_p): New function.
38682 * cfg.h (original_copy_tables_initialized_p): New decl.
38683 * cfgrtl.c (relink_block_chain): Guard the call to
38684 free_original_copy_tables with a call to
38685 original_copy_tables_initialized_p.
38686 * cgraph.h (symtab_node::native_rtl_p): New decl.
38687 * cgraphunit.c (symtab_node::native_rtl_p): New function.
38688 (symtab_node::needed_p): Don't assert for early assembly output
38689 for __RTL functions.
38690 (cgraph_node::finalize_function): Set "force_output" for __RTL
38692 (cgraph_node::analyze): Bail out early for __RTL functions.
38693 (analyze_functions): Update assertion to support __RTL functions.
38694 (cgraph_node::expand): Bail out early for __RTL functions.
38695 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
38697 * function.h (struct function): Update comment for field
38699 * gimple-expr.c: Include "tree-pass.h".
38700 (gimple_has_body_p): Return false for __RTL functions.
38701 * Makefile.in (OBJS): Add run-rtl-passes.o.
38702 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
38704 (gcc::pass_manager::get_clean_slate): New accessor.
38705 * passes.c: Include "insn-addr.h".
38706 (should_skip_pass_p): Add logging. Update logic for running
38707 "expand" to be compatible with both __GIMPLE and __RTL. Guard
38708 property-provider override so it is only done for gimple passes.
38710 (skip_pass): New function.
38711 (execute_one_pass): Call skip_pass when skipping passes.
38712 * read-md.c (md_reader::read_char): Support filtering
38713 the input to a subset of line numbers.
38714 (md_reader::md_reader): Initialize fields
38715 m_first_line and m_last_line.
38716 (md_reader::read_file_fragment): New function.
38717 * read-md.h (md_reader::read_file_fragment): New decl.
38718 (md_reader::m_first_line): New field.
38719 (md_reader::m_last_line): New field.
38720 * read-rtl-function.c (function_reader::create_function): Only
38721 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
38722 curr_properties. Set DECL_INITIAL to a dummy block.
38723 (read_rtl_function_body_from_file_range): New function.
38724 * read-rtl-function.h (read_rtl_function_body_from_file_range):
38726 * run-rtl-passes.c: New file.
38727 * run-rtl-passes.h: New file.
38729 2017-01-24 Jeff Law <law@redhat.com>
38731 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
38734 2017-01-24 Bin Cheng <bin.cheng@arm.com>
38736 PR tree-optimization/79159
38737 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
38738 (record_nonwrapping_iv): Improve boundary using above function if no
38739 value range information.
38741 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
38742 Martin Jambor <mjambor@suse.cz>
38744 * brig-builtins.def: New file.
38745 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
38746 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
38747 (DEF_HSAIL_SAT_BUILTIN): Likewise.
38748 (DEF_HSAIL_INTR_BUILTIN): Likewise.
38749 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
38750 * builtin-types.def (BT_INT8): New.
38751 (BT_INT16): Likewise.
38752 (BT_UINT8): Likewise.
38753 (BT_UINT16): Likewise.
38754 (BT_FN_ULONG): Likewise.
38755 (BT_FN_UINT_INT): Likewise.
38756 (BT_FN_UINT_ULONG): Likewise.
38757 (BT_FN_UINT_LONG): Likewise.
38758 (BT_FN_UINT_PTR): Likewise.
38759 (BT_FN_ULONG_PTR): Likewise.
38760 (BT_FN_INT8_FLOAT): Likewise.
38761 (BT_FN_INT16_FLOAT): Likewise.
38762 (BT_FN_UINT32_FLOAT): Likewise.
38763 (BT_FN_UINT16_FLOAT): Likewise.
38764 (BT_FN_UINT8_FLOAT): Likewise.
38765 (BT_FN_UINT64_FLOAT): Likewise.
38766 (BT_FN_UINT16_UINT32): Likewise.
38767 (BT_FN_UINT32_UINT16): Likewise.
38768 (BT_FN_UINT16_UINT16_UINT16): Likewise.
38769 (BT_FN_INT_PTR_INT): Likewise.
38770 (BT_FN_UINT_PTR_UINT): Likewise.
38771 (BT_FN_LONG_PTR_LONG): Likewise.
38772 (BT_FN_ULONG_PTR_ULONG): Likewise.
38773 (BT_FN_VOID_UINT64_UINT64): Likewise.
38774 (BT_FN_UINT8_UINT8_UINT8): Likewise.
38775 (BT_FN_INT8_INT8_INT8): Likewise.
38776 (BT_FN_INT16_INT16_INT16): Likewise.
38777 (BT_FN_INT_INT_INT): Likewise.
38778 (BT_FN_UINT_FLOAT_UINT): Likewise.
38779 (BT_FN_FLOAT_UINT_UINT): Likewise.
38780 (BT_FN_ULONG_UINT_UINT): Likewise.
38781 (BT_FN_ULONG_UINT_PTR): Likewise.
38782 (BT_FN_ULONG_ULONG_ULONG): Likewise.
38783 (BT_FN_UINT_UINT_UINT): Likewise.
38784 (BT_FN_VOID_UINT_PTR): Likewise.
38785 (BT_FN_UINT_UINT_PTR: Likewise.
38786 (BT_FN_UINT32_UINT64_PTR): Likewise.
38787 (BT_FN_INT_INT_UINT_UINT): Likewise.
38788 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
38789 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
38790 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
38791 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
38792 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
38793 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
38794 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
38795 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
38796 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
38797 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
38798 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
38799 * doc/frontends.texi: List BRIG FE.
38800 * doc/install.texi (Testing): Add BRIG tesring requirements.
38801 * doc/invoke.texi (Overall Options): Mention BRIG.
38802 * doc/standards.texi (Standards): Doucment BRIG HSA version.
38804 2017-01-24 Richard Biener <rguenther@suse.de>
38806 PR translation/79208
38807 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
38809 2017-01-24 Martin Jambor <mjambor@suse.cz>
38812 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
38813 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
38814 and known_contexts.
38816 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
38818 PR middle-end/79123
38819 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
38820 casts from signed to unsigned really don't have a range.
38822 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
38824 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
38825 GMP_RNDx for compatiblity.
38827 2017-01-24 Martin Liska <mliska@suse.cz>
38830 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
38831 that would prevent us to call alloca with -1 as argument.
38833 2017-01-24 Jakub Jelinek <jakub@redhat.com>
38835 * dwarf2out.c (output_compilation_unit_header, output_file_names):
38836 Avoid -Wformat-security warning.
38838 2017-01-23 Andrew Pinski <apinski@cavium.com>
38840 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
38843 2017-01-23 Martin Sebor <msebor@redhat.com>
38845 PR middle-end/78703
38846 * gimple-ssa-sprintf.c (warn_level): New global.
38847 (format_integer): Use it here and throughout the rest of the file.
38848 Use the same switch to compute sign as base.
38849 (maybe_warn): New function.
38850 (format_directive): Factor out warnings into maybe_warn.
38851 Add debugging output. Use warn_level.
38852 (add_bytes): Use warn_level.
38853 (pass_sprintf_length::compute_format_length): Add debugging output.
38854 (try_substitute_return_value): Same.
38855 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
38857 PR middle-end/78703
38858 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
38859 (struct fmtresult, format_integer, format_floating): Adjust.
38860 (fmtresult::fmtresult): Set max correctly in two argument ctor.
38861 (get_string_length, format_string,format_directive): Same.
38862 (pass_sprintf_length::compute_format_length): Same.
38863 (try_substitute_return_value): Simplify slightly.
38865 PR middle-end/78703
38866 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
38867 (fmtresult::operator+=): Outlined.
38868 (struct fmtresult): Add ctors.
38869 (struct conversion_spec): Rename...
38870 (struct directive): ...to this. Add and remove data members.
38871 (directive::set_width, directive::set_precision): New functions.
38872 (format_percent): Use fmtresult ctor.
38873 (get_width_and_precision): Remove.
38874 (format_integer): Make naming changes. Avoid computing width and
38876 (format_floating): Same. Adjust indentation.
38877 (format_character, format_none): New functions.
38878 (format_string): Moved character handling to format_character.
38879 (format_directive): Remove arguments, change return type.
38880 (parse_directive): New function.
38881 (pass_sprintf_length::compute_format_length): Move directive
38882 parsing to parse_directive.
38884 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38886 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
38887 (assign_assembler_name_if_needed): ... this.
38888 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
38889 (assign_assembler_name_if_needed): ... this.
38890 (free_lang_data_in_cgraph): Adjust callers.
38891 * cgraphunit.c (cgraph_node::analyze): Likewise.
38892 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
38895 2017-01-23 Richard Biener <rguenther@suse.de>
38897 PR tree-optimization/79088
38898 PR tree-optimization/79188
38899 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
38900 resetting loop bounds after last path deletion. Reset loop
38901 bounds of the target loop, make code match the comments.
38902 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
38903 Make sure loops need no fixups.
38905 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
38907 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
38908 exponent support with double type for first argument.
38909 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
38910 type returned by __builtin_vec_extract_sig,
38911 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
38912 functions from "vector int" to "vector unsigned int" or from
38913 "vector long long int" to "vector unsigned long long int".
38914 Changed type returned by __builtin_vec_extract_exp,
38915 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
38916 functions from "vector int" to "vector unsigned int" or from
38917 "vector long long int" to "vector unsigned long long int".
38918 Changed return type of __builtin_vec_test_data_class,
38919 __builtin_vec_test_data_class_sp, and
38920 __builtin_vec_test_data_class_dp from "vector int" to
38921 "vector bool int" or from "vector long long int" to "vector bool
38922 long long int" and changed second argument type from "unsigned
38923 int" to "int". Added new overloaded function forms "vector float
38924 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
38925 "vector float __builtin_vec_insert_exp_sp (vector float, vector
38926 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
38927 double, vector unsigned long long int)" and "vector double
38928 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
38929 long int)". Changed return type of
38930 __builtin_scalar_test_data_class and
38931 __builtin_scalar_test_data_class_sp and
38932 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
38933 int" and changed second argument from "unsigned int" to "int".
38934 Changed type returned by __builtin_scalar_test_neg,
38935 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
38936 from "int" to "bool int". Added new overloaded function form
38937 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
38938 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
38939 exponent double-precision with floating point first argument.
38940 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
38941 documentation of scalar_test_data_class, scalar_test_neg,
38942 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
38943 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
38944 vec_test_data_class built-in functions to reflect refinements in
38945 their type signatures.
38947 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
38949 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
38951 (aarch64_elf_asm_destructor): Likewise.
38953 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
38955 PR rtl-optimization/78634
38956 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
38957 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
38958 * ifcvt.c (noce_try_cmove): Add missing cost check.
38960 PR rtl-optimization/71724
38961 * combine.c (if_then_else_cond): Look for situations where it is
38962 beneficial to undo the work of one of the recursive calls.
38964 2017-01-23 Bin Cheng <bin.cheng@arm.com>
38966 PR tree-optimization/70754
38967 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
38968 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
38969 combined stmt before it if not NULL.
38970 (combine_chains): Process refs reversely and compute dominance point
38973 2017-01-23 Martin Liska <mliska@suse.cz>
38975 PR tree-optimization/79196
38976 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
38977 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
38979 (strlen_optimize_stmt): Call the renamed function.
38981 2017-01-23 Michael Matz <matz@suse.de>
38983 PR tree-optimization/78384
38984 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
38986 2017-01-23 Richard Biener <rguenther@suse.de>
38988 PR tree-optimization/79186
38989 * tree-vrp.c (register_new_assert_for): Make sure we've seen
38990 both incoming edges before moving an assert.
38992 2017-01-23 Martin Jambor <mjambor@suse.cz>
38994 * ipa-prop.c (load_from_param_1): Removed.
38995 (load_from_unmodified_param): Bits from load_from_param_1 put back
38997 (load_from_param): Removed.
38998 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
38999 with stmt. Reverted back to use of load_from_unmodified_param.
39001 2017-01-23 Martin Jambor <mjambor@suse.cz>
39004 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
39005 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
39006 field a pointer to garbage collected vector, mark lattices and
39007 ipcp_orig_node with GTY((skip)).
39008 (ipa_get_param_count): Adjust to descriptors being a pointer.
39009 (ipa_get_param): Likewise.
39010 (ipa_get_type): Likewise.
39011 (ipa_get_param_move_cost): Likewise.
39012 (ipa_set_param_used): Likewise.
39013 (ipa_get_controlled_uses): Likewise.
39014 (ipa_set_controlled_uses): Likewise.
39015 (ipa_is_param_used): Likewise.
39016 (ipa_node_params_t): Move into garbage collector. New methods insert
39018 (ipa_node_params_sum): Annotate wth GTY(()).
39019 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
39021 (ipa_load_from_parm_agg): Adjust declaration.
39022 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
39023 * ipa-profile.c (ipa_profile): Likewise.
39024 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
39025 (ipa_populate_param_decls): Make descriptors parameter garbage
39027 (ipa_dump_param): Adjust to descriptors being a pointer.
39028 (ipa_alloc_node_params): Likewise.
39029 (ipa_initialize_node_params): Likewise.
39030 (load_from_param_1): Make descriptors parameter garbage collected.
39031 (load_from_unmodified_param): Likewise.
39032 (load_from_param): Likewise.
39033 (ipa_load_from_parm_agg): Likewise.
39034 (ipa_node_params::~ipa_node_params): Removed.
39035 (ipa_free_all_node_params): Remove call to delete operator.
39036 (ipa_node_params_t::insert): New.
39037 (ipa_node_params_t::remove): Likewise.
39038 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
39039 copy known_csts and known_contexts vectors.
39040 (ipa_read_node_info): Adjust to descriptors being a pointer.
39041 (ipcp_modif_dom_walker): Make m_descriptors field garbage
39043 (ipcp_transform_function): Make descriptors variable garbage
39046 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
39048 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
39049 * config/i386/avx512dqintrin.h: Ditto.
39050 * config/i386/avx512fintrin.h: Ditto.
39051 * config/i386/i386.c: Handle new builtins.
39052 * config/i386/i386-builtin.def: Add new builtins.
39053 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
39054 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
39056 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39057 Martin Liska <mliska@suse.cz>
39059 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
39060 * asan.c (asan_expand_poison_ifn): Support stores and use
39061 appropriate ASAN report function.
39062 * internal-fn.c (expand_ASAN_POISON_USE): New function.
39063 * internal-fn.def (ASAN_POISON_USE): Declare.
39064 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
39065 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
39066 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
39067 ASAN_POISON calls w/o LHS.
39068 * tree-ssa.c (execute_update_addresses_taken): Create clobber
39069 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
39070 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
39071 * gimplify.c (asan_poison_variables): Add attribute
39072 use_after_scope_memory to variables that really needs to live
39074 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
39075 having the attribute.
39077 2017-01-23 Martin Liska <mliska@suse.cz>
39079 * asan.c (create_asan_shadow_var): New function.
39080 (asan_expand_poison_ifn): Likewise.
39081 * asan.h (asan_expand_poison_ifn): New declaration.
39082 * internal-fn.c (expand_ASAN_POISON): Likewise.
39083 * internal-fn.def (ASAN_POISON): New builtin.
39084 * sanopt.c (pass_sanopt::execute): Expand
39085 asan_expand_poison_ifn.
39086 * tree-inline.c (copy_decl_for_dup_finish): Make function
39088 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
39089 * tree-ssa.c (is_asan_mark_p): New function.
39090 (execute_update_addresses_taken): Rewrite local variables
39091 (identified just by use-after-scope as addressable) into SSA.
39093 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39095 * doc/install.texi (Specific): opensource.apple.com uses https
39096 now. Remove trailing slash.
39098 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39100 * README.Portability: Remove note on an Irix compatibility issue.
39102 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
39104 * gcov.c (INCLUDE_ALGORITHM): Define.
39105 (INCLUDE_VECTOR): Define.
39106 No longer include <vector> and <algorithm> directly.
39108 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
39110 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
39112 * doc/invoke.texi (Code Gen Options): Ditto.
39114 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
39117 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
39119 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
39121 rtl-optimization/79125
39122 * cprop.c (local_cprop_pass): Handle cases where we make an
39123 unconditional trap.
39125 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
39129 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
39130 read from, for big endian.
39132 2017-01-20 Jiong Wang <jiong.wang@arm.com>
39134 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
39135 register pauth builtins for LP64 only.
39137 2017-01-20 Marek Polacek <polacek@redhat.com>
39140 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
39143 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39145 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
39147 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
39148 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
39149 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
39151 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39154 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
39155 in big-endian mode when they are not a single duplicated value.
39157 2017-01-20 Richard Biener <rguenther@suse.de>
39159 * BASE-VER: Bump to 7.0.1.
39161 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39163 * omp-low.c (omplow_simd_context): New struct. Use it...
39164 (lower_rec_simd_input_clauses): ...here and...
39165 (lower_rec_input_clauses): ...here to hold common data. Adjust all
39166 references to idx, lane, max_vf, is_simt.
39168 2017-01-20 Graham Markall <graham.markall@embecosm.com>
39170 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
39173 2017-01-20 Martin Jambor <mjambor@suse.cz>
39175 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
39176 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
39178 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
39179 (GTFILES): Rename hsa.c to hsa-common.c.
39180 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
39181 * hsa-dump.c: Likewise.
39182 * hsa-gen.c: Likewise.
39183 * hsa-regalloc.c: Likewise.
39184 * ipa-hsa.c: Likewise.
39185 * omp-expand.c: Likewise.
39186 * omp-low.c: Likewise.
39187 * toplev.c: Likewise.
39189 2017-01-20 Marek Polacek <polacek@redhat.com>
39192 * doc/invoke.texi: Document -Wduplicated-branches.
39193 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
39194 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
39195 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
39196 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
39197 return 0 only when not OEP_LEXICOGRAPHIC.
39198 (fold_build_cleanup_point_expr): Use the expression
39199 location when building CLEANUP_POINT_EXPR.
39200 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
39201 * tree.c (add_expr): Handle error_mark_node.
39203 2017-01-20 Martin Liska <mliska@suse.cz>
39206 * tree-profile.c (init_ic_make_global_vars): Do not call
39208 (gimple_init_gcov_profiler): Likewise.
39210 2017-01-20 Martin Liska <mliska@suse.cz>
39213 * cgraph.h (maybe_create_reference): Remove argument and
39215 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
39217 * ipa-cp.c (create_specialized_node): Likewise.
39218 * symtab.c (symtab_node::maybe_create_reference): Handle
39219 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
39221 2017-01-20 Martin Liska <mliska@suse.cz>
39223 * read-rtl-function.c (function_reader::create_function): Use
39224 build_decl instread of build_decl_stat.
39226 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
39228 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
39229 * config/i386/avx512dqintrin.h: Ditto.
39230 * config/i386/avx512fintrin.h: Ditto.
39231 * config/i386/i386-builtin-types.def: Add new types.
39232 * config/i386/i386.c: Handle new types.
39233 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
39234 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
39235 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
39236 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
39237 (__builtin_ia32_kshiftridi): New.
39238 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
39240 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
39244 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
39245 define to rs6000_init_stack_protect_guard.
39246 (rs6000_init_stack_protect_guard): New function.
39248 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39249 Yunqiang Su <yunqiang.su@imgtec.com>
39251 * config.gcc (supported_defaults): Add madd4.
39252 (with_madd4): Add validation.
39253 (all_defaults): Add madd4.
39254 * config/mips/mips.opt (mmadd4): New option.
39255 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39257 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39259 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
39260 (ISA_HAS_FUSED_MADD4): Likewise.
39261 * doc/invoke.texi (-mmadd4): Document the new option.
39262 * doc/install.texi (--with-madd4): Document the new option.
39264 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39266 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
39267 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
39268 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
39269 (aarch64_init_pauth_hint_builtins): New.
39270 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
39271 (aarch64_expand_builtin): Expand new builtins.
39273 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39275 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
39276 * combine-stack-adj.c (no_unhandled_cfa): Handle
39277 REG_CFA_TOGGLE_RA_MANGLE.
39278 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
39279 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
39280 info for return address signing.
39281 (aarch64_expand_epilogue): Likewise.
39283 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39285 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
39286 * config/aarch64/aarch64-protos.h
39287 (aarch64_return_address_signing_enabled): New declaration.
39288 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
39290 (aarch64_expand_prologue): Sign return address before it's pushed onto
39292 (aarch64_expand_epilogue): Authenticate return address fetched from
39294 (aarch64_override_options): Sanity check for ILP32 and ISA level.
39295 (aarch64_attributes): New function attributes for "sign-return-address".
39296 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
39297 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
39298 ("*do_return"): Generate combined instructions according to key index.
39299 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
39300 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
39302 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
39303 * config/aarch64/aarch64.opt (msign-return-address=): New.
39304 * doc/extend.texi (AArch64 Function Attributes): Documents
39305 "sign-return-address=".
39306 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
39308 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39310 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
39311 overall option summary.
39313 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39315 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
39316 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
39317 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
39318 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
39320 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
39322 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
39323 -mpower9-minmax by default for -mcpu=power9.
39324 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
39325 128-bit floating point.
39327 2017-01-20 Alan Modra <amodra@gmail.com>
39329 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
39330 optimizing for size.
39332 2017-01-20 Alan Modra <amodra@gmail.com>
39335 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
39336 for strcmp and strncmp from corresponding builtin decl.
39338 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39340 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
39341 instead of i386/rtems-64.h.
39342 * config/i386/rtems-64.h: Remove.
39344 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39348 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
39350 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
39352 2017-01-19 Tamar Christina <tamar.christina@arm.com>
39354 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
39355 Change int to HOST_WIDE_INT.
39356 * config/aarch64/aarch64-protos.h
39357 (aarch64_simd_gen_const_vector_dup): Likewise.
39358 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
39360 2017-01-19 David Malcolm <dmalcolm@redhat.com>
39362 * langhooks-def.h (lhd_type_for_size): New decl.
39363 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
39364 * langhooks.c (lhd_type_for_size): New function, taken from
39367 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
39369 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
39370 define_bypass for CR latency.
39371 (power9-cracked-alu): Update bypass latency and remove power9-branch.
39372 (power9-alu2): Add define_bypass for CR latency.
39374 (power9-mul): Update insn latency.
39375 (power9-mul-compare): Update insn latency, bypass latency and remove
39378 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39380 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
39382 * config/aarch64/aarch64.md
39383 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
39384 aarch64_nopcrelative_literal_loads.
39385 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
39387 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
39389 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
39390 TARGET_LOONGSON_3A.
39391 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
39393 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
39396 * config.gcc (supported_defaults): Add lxc1-sxc1.
39397 (with_lxc1_sxc1): Add validation.
39398 (all_defaults): Add lxc1-sxc1.
39399 * config/mips/mips.opt (mlxc1-sxc1): New option.
39400 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39402 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39403 __mips_no_lxc1_sxc1.
39404 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
39405 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
39406 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
39408 2017-01-19 Richard Biener <rguenther@suse.de>
39410 PR tree-optimization/72488
39411 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
39412 sure to restore SSA info.
39413 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
39415 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
39417 PR rtl-optimization/79121
39418 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
39419 of the inner type when shifting an extended value.
39421 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39424 * symtab.c (symtab_node::equal_address_to): Fix comparing of
39425 interposable aliases.
39427 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
39430 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
39431 Use the evmergelohi instruction.
39432 (mov_si<mode>_e500_subreg4_2_le): Likewise.
39433 (mov_sitf_e500_subreg8_2_be): Likewise.
39434 (mov_sitf_e500_subreg12_2_le): Likewise.
39435 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
39436 (mov_si<mode>_e500_subreg4_2_be): Likewise.
39437 (mov_sitf_e500_subreg8_2_le): Likewise.
39438 (mov_sitf_e500_subreg12_2_be): Likewise.
39440 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39442 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
39443 attribute from vecsimple to vecperm.
39444 (altivec_vbpermq2): Likewise.
39446 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39449 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
39451 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39452 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
39453 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
39454 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
39455 case where N arg is SIZE_MAX.
39456 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
39457 (cmpstrsi): Add pattern.
39459 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
39461 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
39462 __builtin_vec_revb builtins.
39463 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
39464 built-in functions to support generation of the ISA 3.0 XXBR<x>
39465 vector byte reverse instructions.
39466 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
39467 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
39468 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
39469 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
39470 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
39471 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
39472 (P9V_BUILTIN_VEC_REVB): Likewise.
39473 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
39474 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
39475 (p9_xxbrq_v16qi): Likewise.
39476 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
39477 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
39478 (p9_xxbrh_v8hi): Likewise.
39479 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
39480 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
39481 vec_revb built-in functions.
39483 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
39485 PR rtl-optimization/78952
39486 * config/i386/i386.md (any_extract): New code iterator.
39487 (*insvqi_2): Use any_extract for source operand.
39488 (*insvqi_3): Use any_shiftrt for source operand.
39490 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
39492 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
39494 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
39496 2017-01-18 Matthias Klose <doko@ubuntu.com>
39498 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
39500 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39502 * config/rs6000/altivec.h (vec_bperm): Change #define.
39503 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
39504 (altivec_vbpermq2): New define_insn.
39505 (altivec_vbpermd): Likewise.
39506 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
39507 function interface.
39508 (VBPERMD): Likewise.
39509 (VBPERM): New polymorphic function interface.
39510 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
39511 Add entries for P9V_BUILTIN_VEC_VBPERM.
39512 * doc/extend.texi: Add interfaces for vec_bperm.
39514 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39516 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
39517 first letter of error messages.
39518 (s390_resolve_overloaded_builtin): Likewise.
39519 * config/s390/s390.c (s390_expand_builtin): Likewise.
39520 (s390_invalid_arg_for_unprototyped_fn): Likewise.
39521 (s390_valid_target_attribute_inner_p): Likewise.
39522 * config/s390/s390.md ("tabort"): Likewise.
39524 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
39526 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
39527 (ISA_AVOID_DIV_HILO): New macro.
39528 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
39529 (ISA_HAS_DDIV): Likewise.
39531 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39533 * doc/invoke.texi (fabi-version): Correct number of occurrences.
39535 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39537 * doc/invoke.texi (fabi-version): Spelling fix.
39539 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39542 * doc/invoke.texi (fabi-version): Mention mangling fix for
39545 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39548 * doc/invoke.texi (fabi-version): Document discriminator mangling.
39550 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
39553 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
39554 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
39556 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
39558 (stack_protect_test): Ditto.
39559 * config/rs6000/rs6000.opt (mstack-protector-guard=,
39560 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
39562 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
39563 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
39564 -mstack-protector-guard-offset=.
39565 (RS/6000 and PowerPC Options): Ditto.
39567 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39569 * config/i386/i386.h (MASK_CLASS_P): New define.
39570 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
39571 there are no registers from different register sets also when
39572 mask registers are used. Update function comment.
39573 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
39574 to (*k/*r) and (*k/*km) alternatives.
39576 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
39578 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
39579 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
39580 (EH_RETURN_HANDLER_RTX): New define.
39581 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
39582 Force frame pointer in EH return functions.
39583 (aarch64_expand_epilogue): Add barrier for eh_return.
39584 (aarch64_final_eh_return_addr): Remove.
39585 (aarch64_eh_return_handler_rtx): New function.
39586 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
39588 (aarch64_eh_return_handler_rtx): New prototype.
39590 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39592 * config/rs6000/altivec.h (vec_rlmi): New #define.
39593 (vec_vrlnm): Likewise.
39594 (vec_rlnm): Likewise.
39595 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
39596 (UNSPEC_VRLNM): Likewise.
39597 (VIlong): New mode iterator.
39598 (altivec_vrl<VI_char>mi): New define_insn.
39599 (altivec_vrl<VI_char>nm): Likewise.
39600 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
39602 (VRLDNM): Likewise.
39603 (RLNM): New polymorphic function entry.
39604 (VRLWMI): New monomorphic function entry.
39605 (VRLDMI): Likewise.
39606 (RLMI): New polymorphic function entry.
39607 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
39608 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
39609 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
39612 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39615 * dwarf2out.c (field_byte_offset): Restore the
39616 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
39617 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
39618 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
39621 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39624 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
39626 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39629 * dwarf2out.c (add_data_member_location_attribute): For constant
39630 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
39631 instead of DW_AT_data_member_location, DW_AT_bit_offset and
39632 DW_AT_byte_size attributes.
39634 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39636 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
39637 after forcing to constant memory when the code model is medium.
39639 2017-01-17 Julia Koval <julia.koval@intel.com>
39642 * config/i386/avx512fintrin.h
39643 (_mm512_i32gather_ps): Change __addr type to void const*.
39644 (_mm512_mask_i32gather_ps): Ditto.
39645 (_mm512_i32gather_pd): Ditto.
39646 (_mm512_mask_i32gather_pd): Ditto.
39647 (_mm512_i64gather_ps): Ditto.
39648 (_mm512_mask_i64gather_ps): Ditto.
39649 (_mm512_i64gather_pd): Ditto.
39650 (_mm512_mask_i64gather_pd): Ditto.
39651 (_mm512_i32gather_epi32): Ditto.
39652 (_mm512_mask_i32gather_epi32): Ditto.
39653 (_mm512_i32gather_epi64): Ditto.
39654 (_mm512_mask_i32gather_epi64): Ditto.
39655 (_mm512_i64gather_epi32): Ditto.
39656 (_mm512_mask_i64gather_epi32): Ditto.
39657 (_mm512_i64gather_epi64): Ditto.
39658 (_mm512_mask_i64gather_epi64): Ditto.
39659 (_mm512_i32scatter_ps): Change __addr type to void*.
39660 (_mm512_mask_i32scatter_ps): Ditto.
39661 (_mm512_i32scatter_pd): Ditto.
39662 (_mm512_mask_i32scatter_pd): Ditto.
39663 (_mm512_i64scatter_ps): Ditto.
39664 (_mm512_mask_i64scatter_ps): Ditto.
39665 (_mm512_i64scatter_pd): Ditto.
39666 (_mm512_mask_i64scatter_pd): Ditto.
39667 (_mm512_i32scatter_epi32): Ditto.
39668 (_mm512_mask_i32scatter_epi32): Ditto.
39669 (_mm512_i32scatter_epi64): Ditto.
39670 (_mm512_mask_i32scatter_epi64): Ditto.
39671 (_mm512_i64scatter_epi32): Ditto.
39672 (_mm512_mask_i64scatter_epi32): Ditto.
39673 (_mm512_i64scatter_epi64): Ditto.
39674 (_mm512_mask_i64scatter_epi64): Ditto.
39675 * config/i386/avx512pfintrin.h
39676 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
39677 (_mm512_mask_prefetch_i32gather_ps): Ditto.
39678 (_mm512_mask_prefetch_i64gather_pd): Ditto.
39679 (_mm512_mask_prefetch_i64gather_ps): Ditto.
39680 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
39681 (_mm512_prefetch_i32scatter_ps): Ditto.
39682 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
39683 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
39684 (_mm512_prefetch_i64scatter_pd): Ditto.
39685 (_mm512_prefetch_i64scatter_ps): Ditto.
39686 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
39687 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
39688 * config/i386/avx512vlintrin.h
39689 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
39690 (_mm_mmask_i32gather_ps): Ditto.
39691 (_mm256_mmask_i32gather_pd): Ditto.
39692 (_mm_mmask_i32gather_pd): Ditto.
39693 (_mm256_mmask_i64gather_ps): Ditto.
39694 (_mm_mmask_i64gather_ps): Ditto.
39695 (_mm256_mmask_i64gather_pd): Ditto.
39696 (_mm_mmask_i64gather_pd): Ditto.
39697 (_mm256_mmask_i32gather_epi32): Ditto.
39698 (_mm_mmask_i32gather_epi32): Ditto.
39699 (_mm256_mmask_i32gather_epi64): Ditto.
39700 (_mm_mmask_i32gather_epi64): Ditto.
39701 (_mm256_mmask_i64gather_epi32): Ditto.
39702 (_mm_mmask_i64gather_epi32): Ditto.
39703 (_mm256_mmask_i64gather_epi64): Ditto.
39704 (_mm_mmask_i64gather_epi64): Ditto.
39705 (_mm256_i32scatter_ps): Change __addr type to void*.
39706 (_mm256_mask_i32scatter_ps): Ditto.
39707 (_mm_i32scatter_ps): Ditto.
39708 (_mm_mask_i32scatter_ps): Ditto.
39709 (_mm256_i32scatter_pd): Ditto.
39710 (_mm256_mask_i32scatter_pd): Ditto.
39711 (_mm_i32scatter_pd): Ditto.
39712 (_mm_mask_i32scatter_pd): Ditto.
39713 (_mm256_i64scatter_ps): Ditto.
39714 (_mm256_mask_i64scatter_ps): Ditto.
39715 (_mm_i64scatter_ps): Ditto.
39716 (_mm_mask_i64scatter_ps): Ditto.
39717 (_mm256_i64scatter_pd): Ditto.
39718 (_mm256_mask_i64scatter_pd): Ditto.
39719 (_mm_i64scatter_pd): Ditto.
39720 (_mm_mask_i64scatter_pd): Ditto.
39721 (_mm256_i32scatter_epi32): Ditto.
39722 (_mm256_mask_i32scatter_epi32): Ditto.
39723 (_mm_i32scatter_epi32): Ditto.
39724 (_mm_mask_i32scatter_epi32): Ditto.
39725 (_mm256_i32scatter_epi64): Ditto.
39726 (_mm256_mask_i32scatter_epi64): Ditto.
39727 (_mm_i32scatter_epi64): Ditto.
39728 (_mm_mask_i32scatter_epi64): Ditto.
39729 (_mm256_i64scatter_epi32): Ditto.
39730 (_mm256_mask_i64scatter_epi32): Ditto.
39731 (_mm_i64scatter_epi32): Ditto.
39732 (_mm_mask_i64scatter_epi32): Ditto.
39733 (_mm256_i64scatter_epi64): Ditto.
39734 (_mm256_mask_i64scatter_epi64): Ditto.
39735 (_mm_i64scatter_epi64): Ditto.
39736 (_mm_mask_i64scatter_epi64): Ditto.
39737 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
39738 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
39739 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
39740 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
39741 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
39742 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
39743 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
39744 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
39745 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
39746 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
39747 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
39748 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
39749 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
39750 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
39751 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
39752 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
39753 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
39754 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
39755 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
39756 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
39757 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
39758 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
39759 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
39760 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
39761 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
39762 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
39763 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
39764 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
39765 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
39766 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
39767 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
39768 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
39769 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
39770 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
39771 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
39772 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
39773 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
39774 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
39775 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
39776 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
39777 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
39778 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
39779 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
39780 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
39781 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
39782 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
39783 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
39784 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
39785 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
39786 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
39787 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
39788 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
39789 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
39790 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
39791 definitions accordingly.
39793 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
39794 Kuan-Lin Chen <kuanlinchentw@gmail.com>
39797 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
39800 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
39803 * ira-conflicts.c (ira_build_conflicts): Update total conflict
39804 hard regs for inner regno.
39806 2017-01-17 Martin Liska <mliska@suse.cz>
39809 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
39810 assumption and add comment.
39812 2017-01-17 Nathan Sidwell <nathan@acm.org>
39814 * ipa-visibility.c (localize_node): New function, broken out of ...
39815 (function_and_variable_visibility): ... here. Call it.
39817 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39819 PR middle-end/77445
39820 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
39821 correctly set frequency of oudgoing edge.
39822 (duplicate_thread_path): Fix profile updating.
39824 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39827 * configure.ac: Add GCC_BASE_VER.
39828 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
39829 version from BASE-VER file.
39830 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
39831 (gcc.o): Depend on $(BASEVER).
39832 * common.opt (dumpfullversion): New option.
39833 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
39834 * doc/invoke.texi: Document -dumpfullversion.
39835 * doc/install.texi: Document --with-gcc-major-version-only.
39836 * configure: Regenerated.
39838 2017-01-17 Richard Biener <rguenther@suse.de>
39840 PR tree-optimization/71433
39841 * tree-vrp.c (register_new_assert_for): Merge same asserts
39842 on all incoming edges.
39843 (process_assert_insertions_for): Handle insertions at the
39846 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
39848 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
39849 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
39851 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
39854 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
39857 2017-01-17 Alan Modra <amodra@gmail.com>
39860 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
39861 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
39862 symbolic stack limit when pic.
39864 2017-01-16 Martin Sebor <msebor@redhat.com>
39866 PR tree-optimization/78608
39867 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
39869 2017-01-16 Jeff Law <law@redhat.com>
39872 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
39873 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
39874 for several include directories that may be relative to sysroot.
39875 * config/i386/x-mingw32 (gplus_includedir): Define.
39876 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
39877 (native_system_includedir): Likewise.
39878 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
39879 override if TARGET_SYSTEM_ROOT is defined.
39880 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
39882 PR tree-optimization/79090
39883 PR tree-optimization/33562
39884 PR tree-optimization/61912
39885 PR tree-optimization/77485
39886 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
39887 and computed trims into the dump file.
39889 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39891 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
39893 2017-01-16 Jakub Jelinek <jakub@redhat.com>
39896 * gimplify.c (gimplify_init_constructor): If want_value and
39897 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
39901 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
39902 sequence. Formatting fixes.
39903 (doloop_optimize): Formatting fixes.
39906 * gcc.c (debug_level_greater_than_spec_func): New function.
39907 (static_spec_functions): Add debug-level-gt spec function.
39908 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
39910 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39911 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39912 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
39913 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
39914 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
39915 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
39917 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
39919 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
39920 QImode fixups to general and mask registers only.
39922 2017-01-16 Carl Love <cel@us.ibm.com>
39924 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
39925 for built-in functions
39926 vector signed char vec_nabs (vector signed char)
39927 vector signed short vec_nabs (vector signed short)
39928 vector signed int vec_nabs (vector signed int)
39929 vector signed long long vec_nabs (vector signed long long)
39930 vector float vec_nabs (vector float)
39931 vector double vec_nabs (vector double)
39932 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
39934 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
39935 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
39936 * doc/extend.texi: Update the documentation file for the new built-in
39939 2017-01-16 Martin Sebor <msebor@redhat.com>
39941 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
39944 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39946 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
39947 UNSPEC_VSX__XXSPLTD to require special splat handling.
39949 2017-01-16 David Malcolm <dmalcolm@redhat.com>
39952 * system.h: Poison strndup.
39954 2017-01-16 Alan Modra <amodra@gmail.com>
39957 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
39960 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
39962 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
39964 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
39966 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
39967 call recog here. Assert that INSN_CODE (insn) is non-negative.
39969 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39972 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
39974 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
39975 in the currently scheduled RTL fragment.
39977 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39979 PR rtl-optimization/78751
39980 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
39983 2017-01-14 Jeff Law <law@redhat.com>
39985 PR tree-optimization/79090
39986 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
39987 variable length stores.
39988 (compute_trims): Delete dead assignment to *trim_tail.
39989 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
39992 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
39994 PR rtl-optimization/78626
39995 PR rtl-optimization/78727
39996 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
39997 of a block, and split such blocks after everything else is finished.
39999 2017-01-14 Alan Modra <amodra@gmail.com>
40002 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
40003 target legitimate_combined_insn.
40004 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
40005 (rs6000_legitimate_combined_insn): New function.
40006 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
40008 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
40009 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
40010 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
40012 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
40014 * doc/frontends.texi (G++ and GCC): Remove references to Java.
40016 2017-01-13 Jeff Law <law@redhat.com>
40018 PR tree-optimization/33562
40019 PR tree-optimization/61912
40020 PR tree-optimization/77485
40021 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
40023 (delete_dead_assignment): Likewise.
40024 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
40025 statement to delete_dead_call and delete_dead_assignment.
40027 2017-01-13 David Malcolm <dmalcolm@redhat.com>
40030 * substring-locations.c (format_warning_va): Strengthen case 1 so
40031 that both endpoints of the substring must be within the format
40032 range for just the substring to be printed.
40034 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
40036 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
40037 * config/i386/i386.c (ix86_target_string): Add missing options
40038 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
40039 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
40040 flags_other and ix86_target_other to flags2_other. Display unknown
40042 (ix86_valid_target_attribute_inner_p): Add missing options and
40043 reorder options by implied ISAs, as in ix86_target_string.
40045 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40047 * hash-table.h (hash_table::too_empty_p): New function.
40048 (hash_table::expand): Use it.
40049 (hash_table::traverse): Likewise.
40050 (hash_table::empty_slot): Use sizeof (value_type) instead of
40051 sizeof (PTR) to convert bytes to elements. Shrink the table
40052 if the current size is excessive for the current number of
40055 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40057 * ira-costs.c (record_reg_classes): Break from the inner loop
40058 early once alt_fail is known to be true. Update outer loop
40059 handling accordingly.
40061 2017-01-13 Jeff Law <law@redhat.com>
40063 * tree-ssa-dse.c (decrement_count): New function.
40064 (increment_start_addr, maybe_trim_memstar_call): Likewise.
40065 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
40066 when we know the partially dead statement is a mem* function.
40068 PR tree-optimization/61912
40069 PR tree-optimization/77485
40070 * tree-ssa-dse.c: Include expr.h.
40071 (maybe_trim_constructor_store): New function.
40072 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
40074 PR tree-optimization/33562
40075 PR tree-optimization/61912
40076 PR tree-optimization/77485
40077 * doc/invoke.texi: Document new dse-max-object-size param.
40078 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
40079 * tree-ssa-dse.c: Include params.h.
40080 (dse_store_status): New enum.
40081 (initialize_ao_ref_for_dse): New, partially extracted from
40083 (valid_ao_ref_for_dse, normalize_ref): New.
40084 (setup_live_bytes_from_ref, compute_trims): Likewise.
40085 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
40086 (maybe_trim_partially_dead_store): Likewise.
40087 (maybe_trim_complex_store): Likewise.
40088 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
40089 Track what bytes live from the original store. Return tri-state
40090 for dead, partially dead or live.
40091 (dse_dom_walker): Add constructor, destructor and new private members.
40092 (delete_dead_call, delete_dead_assignment): New extracted from
40094 (dse_optimize_stmt): Make a member of dse_dom_walker.
40095 Use initialize_ao_ref_for_dse.
40097 PR tree-optimization/33562
40098 PR tree-optimization/61912
40099 PR tree-optimization/77485
40100 * sbitmap.h (bitmap_count_bits): Prototype.
40101 (bitmap_clear_range, bitmap_set_range): Likewise.
40102 * sbitmap.c (bitmap_clear_range): New function.
40103 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
40105 2017-01-13 Martin Liska <mliska@suse.cz>
40108 * function.c (set_cfun): Add new argument force.
40109 * function.h (set_cfun): Likewise.
40110 * ipa-inline-transform.c (inline_call): Use the function when
40111 strict alising from is dropped for function we inline to.
40113 2017-01-13 Richard Biener <rguenther@suse.de>
40115 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
40116 for dumping GIMPLE INTEGER_CSTs.
40118 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
40120 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
40121 to 201112L since C++17.
40123 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
40126 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
40127 if -fsanitize=kernel-address is present.
40129 2017-01-13 Richard Biener <rguenther@suse.de>
40131 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
40132 as _Literal ( type ) number in case usual suffixes do not
40133 preserve all information.
40135 2017-01-13 Richard Biener <rguenther@suse.de>
40137 PR tree-optimization/77283
40138 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
40139 and ssa-iterators.h.
40140 (is_feasible_trace): Implement a cost model based on joiner
40143 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
40146 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
40147 char or short to __float128/_Float128 directly.
40149 2017-01-12 Martin Sebor <msebor@redhat.com>
40151 to -Wformat-overflow.
40152 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
40153 (min_bytes_remaining): Same.
40154 (get_string_length): Same.
40155 (format_string): Same.
40156 (format_directive): Same.
40158 (pass_sprintf_length::handle_gimple_call): Same.
40160 2017-01-12 Jakub Jelinek <jakub@redhat.com>
40162 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
40163 info.nowrite calls with no lhs that can't throw. Return bool
40164 whether gsi_remove has been called or not.
40165 (pass_sprintf_length::handle_gimple_call): Return bool whether
40166 try_substitute_return_value called gsi_remove. Formatting fix.
40167 (pass_sprintf_length::execute): Don't use gsi_remove if
40168 handle_gimple_call returned true.
40171 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
40172 be removed due to side-effects, don't remove following barrier nor
40173 turn the successor edge into fallthru edge.
40175 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40178 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
40179 element-reversing loads and stores as not swappable.
40181 2017-01-12 Nathan Sidwell <nathan@acm.org>
40182 Nicolai Stange <nicstange@gmail.com>
40184 * combine.c (try_combine): Don't ignore result of overlap checking
40185 loop. Combine overlap & asm check into single loop.
40187 2017-01-12 Richard Biener <rguenther@suse.de>
40189 * tree-pretty-print.c (dump_generic_node): Provide -gimple
40190 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
40192 2017-01-12 Richard Biener <rguenther@suse.de>
40194 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
40195 and TS_TARGET_OPTION directly derive from TS_BASE.
40196 * tree-core.h (tree_optimization_option): Derive from tree_base.
40197 (tree_target_option): Likewise.
40199 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
40201 * config/i386/i386.c (memory_address_length): Increase len
40202 only when rip_relative_addr_p returns false.
40204 2017-01-11 Julia Koval <julia.koval@intel.com>
40206 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
40207 (OPTION_MASK_ISA_SGX_SET): New.
40208 (ix86_handle_option): Handle OPT_msgx.
40209 * config.gcc: Added sgxintrin.h.
40210 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
40211 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
40212 * config/i386/i386.c (ix86_target_string): Add -msgx.
40214 (ix86_option_override_internal): Handle new options.
40215 (ix86_valid_target_attribute_inner_p): Add sgx.
40216 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
40217 * config/i386/i386.opt: Add msgx.
40218 * config/i386/sgxintrin.h: New file.
40219 * config/i386/x86intrin.h: Add sgxintrin.h.
40221 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40224 * fold-const.c (maybe_nonzero_address): Return 1 for function
40226 (tree_single_nonzero_warnv_p): Don't handle function local objects
40230 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
40233 2017-01-11 David Malcolm <dmalcolm@redhat.com>
40236 * opts.c: Include "spellcheck.h"
40237 (struct string_fragment): New struct.
40238 (struct edit_distance_traits<const string_fragment &>): New
40240 (get_closest_sanitizer_option): New function.
40241 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
40243 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40245 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
40247 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
40248 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
40249 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
40250 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
40251 for initial die_offset if dwarf_split_debug_info.
40252 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
40253 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
40255 (output_skeleton_debug_sections): Formatting fix. Use
40256 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
40257 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
40259 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
40261 * config/arm/cortex-a53.md: Add bypasses for
40262 cortex_a53_r2f_cvt.
40263 (cortex_a53_r2f): Only use for transfers.
40264 (cortex_a53_f2r): Likewise.
40265 (cortex_a53_r2f_cvt): Add reservation for conversions.
40266 (cortex_a53_f2r_cvt): Likewise.
40268 2017-01-11 Tamar Christina <tamar.christina@arm.com>
40270 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
40271 to all inlined functions, change static to extern.
40273 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
40276 * config/arm/arm.c (legitimize_pic_address): Handle reference to
40278 (arm_assemble_integer): Likewise.
40280 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
40282 * config.gcc: Use new awk script to check CPU, FPU and architecture
40283 parameters for --with-... options.
40284 * config/arm/parsecpu.awk: New file
40285 * config/arm/arm-cpus.in: New file.
40286 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
40288 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
40290 * config/arm/t-arm: Update dependency rules.
40291 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
40292 of processing .def files.
40293 * config/arm/genopt.sh: Deleted.
40294 * config/arm/gentune.sh: Deleted.
40295 * config/arm/arm-cores.def: Deleted.
40296 * config/arm/arm-arches.def: Deleted.
40297 * config/arm/arm-fpus.def: Deleted.
40298 * config/arm/arm-tune.md: Regenerated.
40299 * config/arm/arm-tables.opt: Regenerated.
40300 * config/arm/arm-cpu.h: New generated file.
40301 * config/arm/arm-cpu-data.h: New generated file.
40302 * config/arm/arm-cpu-cdata.h: New generated file.
40304 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
40307 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
40309 (input_varpool_node): Unpack dynamically_initialized bit.
40311 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
40313 PR rtl-optimization/79032
40314 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
40315 the alignment of the adjusted memory reference against that of MODE,
40316 instead of the alignment of the original memory reference.
40318 2017-01-11 Martin Jambor <mjambor@suse.cz>
40320 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
40322 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
40323 decorated functions.
40325 2017-01-11 Richard Biener <rguenther@suse.de>
40327 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
40328 set range/nonnull info for PHI results. Do not set it on
40329 stmts marked for removal.
40331 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
40333 * expr.c (store_field): In the bitfield case, fetch the return value
40334 from the registers before applying a single big-endian adjustment.
40335 Always do a final load for a BLKmode value not larger than a word.
40337 2017-01-10 David Malcolm <dmalcolm@redhat.com>
40340 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40341 that we correctly handle column numbers greater than
40342 LINE_MAP_MAX_COLUMN_NUMBER.
40344 2017-01-10 Martin Sebor <msebor@redhat.com>
40346 PR middle-end/78245
40347 * gimple-ssa-sprintf.c (get_destination_size): Call
40348 {init,fini}object_sizes.
40349 * tree-object-size.c (addr_object_size): Adjust.
40350 (pass_through_call): Adjust.
40351 (pass_object_sizes::execute): Adjust.
40352 * tree-object-size.h (fini_object_sizes): Declare.
40354 2017-01-10 Martin Sebor <msebor@redhat.com>
40356 PR tree-optimization/78775
40357 * builtins.c (get_size_range): Move...
40358 * calls.c: ...to here.
40359 (alloc_max_size): Accept zero argument.
40360 (operand_signed_p): Remove.
40361 (maybe_warn_alloc_args_overflow): Call get_size_range.
40362 * calls.h (get_size_range): Declare.
40364 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
40366 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
40367 from TI's devices.csv file as of September 2016.
40368 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
40370 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
40372 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
40373 * doc/invoke.texi: Likewise.
40374 * doc/md.texi: Likewise.
40375 * doc/objc.texi: Likewise.
40377 2017-01-10 Joshua Conner <joshconner@google.com>
40379 * config/arm/fuchsia-elf.h: New file.
40380 * config/fuchsia.h: New file.
40381 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
40382 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
40384 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
40386 2016-01-10 Richard Biener <rguenther@suse.de>
40388 PR tree-optimization/79034
40389 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
40390 Propagate out degenerate PHIs in the joiner.
40392 2017-01-10 Martin Liska <mliska@suse.cz>
40394 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
40395 (sort_congruence_classes_by_decl_uid): Likewise.
40396 (sort_congruence_class_groups_by_decl_uid): Likewise.
40397 (sem_item_optimizer::merge_classes): Sort class, groups in these
40398 classes and members in the groups by DECL_UID of declarations.
40399 This would make merge operations stable.
40401 2017-01-10 Martin Liska <mliska@suse.cz>
40403 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
40404 usage of m_classes_vec.
40405 (sem_item_optimizer::~sem_item_optimizer): Likewise.
40406 (sem_item_optimizer::get_group_by_hash): Likewise.
40407 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
40408 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
40409 (sem_item_optimizer::verify_classes): Likewise.
40410 (sem_item_optimizer::process_cong_reduction): Likewise.
40411 (sem_item_optimizer::dump_cong_classes): Likewise.
40412 (sem_item_optimizer::merge_classes): Likewise.
40413 * ipa-icf.h (congruence_class_hash): Rename from
40414 congruence_class_group_hash. Remove declaration of m_classes_vec.
40416 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
40418 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
40419 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
40420 * config.gcc: Add avx512vpopcntdqintrin.h.
40421 * config/i386/avx512vpopcntdqintrin.h: New.
40422 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
40423 * config/i386/i386-builtin-types.def: Add new types.
40424 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
40425 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
40426 __builtin_ia32_vpopcountq_v8di_mask): New.
40427 * config/i386/i386-c.c (ix86_target_macros_internal): Define
40428 __AVX512VPOPCNTDQ__.
40429 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
40430 (PTA_AVX512VPOPCNTDQ): Define.
40431 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
40432 TARGET_AVX512VPOPCNTDQ_P): Define.
40433 * config/i386/i386.opt: Add mavx512vpopcntdq.
40434 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
40435 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
40437 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40439 PR middle-end/77484
40440 * predict.def (PRED_CALL): Set to 67.
40442 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
40444 * expr.c (store_field): In the bitfield case, if the value comes from
40445 a function call and is of an aggregate type returned in registers, do
40446 not modify the field mode; extract the value in all cases if the mode
40447 is BLKmode and the size is not larger than a word.
40449 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
40452 * config/i386/cpuid.h: Fix undefined behavior.
40454 2017-01-04 Jeff Law <law@redhat.com>
40456 PR tree-optimization/79007
40457 PR tree-optimization/67955
40458 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
40459 conservative for pt.null when flag_non_call_exceptions is on.
40461 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40463 PR translation/79019
40464 PR translation/79020
40465 * params.def (PARAM_INLINE_MIN_SPEEDUP,
40466 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
40467 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
40469 * config/avr/avr.opt (maccumulate-args): Likewise.
40470 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
40471 * common.opt (freport-bug): Likewise.
40472 * cif-code.def (CIF_FINAL_ERROR): Likewise.
40473 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
40474 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
40475 translatable string.
40476 * config/i386/i386.c (function_value_32): Likewise.
40477 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
40478 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
40480 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
40481 * common/config/msp430/msp430-common.c (msp430_handle_option):
40483 * symtab.c (symtab_node::verify_base): Likewise.
40484 * opts.c (set_debug_level): Likewise.
40485 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
40486 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
40487 missing whitespace to translatable strings.
40488 * config/avr/avr.md (bswapsi2): Fix typo in comment.
40489 * config/sh/superh.h: Likewise.
40490 * config/i386/xopintrin.h: Likewise.
40491 * config/i386/znver1.md: Likewise.
40492 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
40493 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
40494 * double-int.h (struct double_int): Likewise.
40495 * double-int.c (div_and_round_double): Likewise.
40496 * wide-int.cc: Likewise.
40497 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
40498 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
40499 * cfgcleanup.c (crossjumps_occured): Renamed to ...
40500 (crossjumps_occurred): ... this.
40501 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
40504 PR tree-optimization/78899
40505 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
40506 returning bool return struct loop *, NULL for failure and the new
40508 (versionable_outer_loop_p): Don't version outer loop if it has
40509 dont_vectorized bit set.
40510 (tree_if_conversion): When versioning outer loop, ensure
40511 tree_if_conversion is performed also on the inner loop of the
40512 non-vectorizable outer loop copy.
40513 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
40514 LOOP_VECTORIZED in inner loop of the scalar outer loop and
40515 prevent vectorization of it.
40516 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
40517 the outer loop vectorization of the non-scalar version is attempted
40518 before vectorization of the inner loop in scalar version. If
40519 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
40520 vectorization of its inner loop.
40521 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
40522 has 2 inner loops, rename also on edges from bb whose single pred
40523 is outer_loop->header. Fix typo in function comment.
40525 2017-01-09 Martin Sebor <msebor@redhat.com>
40528 * asan.c (asan_emit_stack_protection): Increase local buffer size
40529 to avoid snprintf truncation warning.
40531 2017-01-09 Andrew Pinski <apinski@cavium.com>
40533 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
40534 to reference thunderx2t99 for the tuning structure
40535 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
40537 (thunderx2t99_extra_costs): This.
40538 * config/aarch64/aarch64-tune.md: Regenerate.
40539 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
40540 (vulcan_addrcost_table): This.
40541 (vulcan_regmove_cost): Rename to ...
40542 (thunderx2t99_regmove_cost): This.
40543 (vulcan_vector_cost): Rename to ...
40544 (thunderx2t99_vector_cost): this.
40545 (vulcan_branch_cost): Rename to ...
40546 (thunderx2t99_branch_cost): This.
40547 (vulcan_tunings): Rename to ...
40548 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
40549 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
40551 2017-01-09 Martin Jambor <mjambor@suse.cz>
40555 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
40556 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
40557 (propagate_vr_accross_jump_function): Use the above function for all
40558 value range computations for pass-through jump functions and type
40559 converasion from explicit value range values.
40560 (ipcp_propagate_stage): Do not attempt to deduce types of formal
40561 parameters from TYPE_ARG_TYPES.
40562 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
40563 (ipa_write_node_info): Stream type of the actual argument.
40564 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
40566 2017-01-09 Martin Liska <mliska@suse.cz>
40569 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
40570 (lookup_compiler): Do not show error message with have_E.
40572 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40574 PR tree-optimization/78938
40575 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
40576 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
40577 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
40578 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
40581 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40583 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
40586 2017-01-09 Richard Biener <rguenther@suse.de>
40588 PR tree-optimization/78997
40589 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
40590 name condition properly.
40592 2017-01-09 Richard Biener <rguenther@suse.de>
40595 * dwarf2out.c (is_cxx): New overload with context.
40596 (is_naming_typedef_decl): Use it.
40598 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40600 * invoke.texi (Option Summary): Correct spacing in option lists
40601 and add line breaks to fix over-long lines.
40603 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40605 PR middle-end/17660
40607 * extend.texi (Common Variable Attributes): Add xref to GCC
40608 Internals manual to explain mode attribute keywords.
40610 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40613 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
40614 and Preprocessor Options.
40615 (Options for Linking): Document -pthread here....
40616 (RS/6000 and PowerPC Options): ...not here.
40617 (Solaris 2 Options): ...or here.
40618 * doc/cppopts.texi: Document -pthread.
40620 2017-01-08 Martin Sebor <msebor@redhat.com>
40622 PR middle-end/77708
40623 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
40624 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
40625 New member functions.
40626 (format_directive): Used them.
40628 (pass_sprintf_length::handle_gimple_call): Same.
40629 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
40630 to avoid truncation for any argument.
40631 (extract_affine_mul): Same.
40632 * tree.c (get_file_function_name): Same.
40634 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40636 PR middle-end/77484
40637 * predict.def (PRED_INDIR_CALL): Set to 86.
40639 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40641 PR preprocessor/54124
40642 * doc/cppopts.texi: Reformat -d subtable to list the full name
40643 of the options. Add cross-reference to the docs for the general
40644 compiler -d options.
40645 * doc/invoke.texi (Developer Options): Add cross-reference to the
40646 preprocessor-specific -d option documentation.
40648 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40650 PR preprocessor/13498
40651 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
40652 redudant material, and reflect new command-line options.
40653 (System Headers): Likewise.
40655 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40657 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
40658 -isystem, and -idirafter. Copy-edit.
40659 * doc/cppopts.texi: Copy-edit. Remove contradiction about
40660 default for -ftrack-macro-expansion. Delete obsolete and
40661 badly-formatted implementation details about -fdebug-cpp output.
40662 * doc/cppwarnopts.texi: Copy-edit.
40664 2017-01-07 David Malcolm <dmalcolm@redhat.com>
40667 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40668 that the transition from a max line width >= 1<<10 to narrower
40669 lines works correctly.
40671 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
40673 * doc/options.texi (PerFunction): New.
40674 * opt-functions.awk (switch_flags): Map both Optimization and
40675 PerFunction to CL_OPTIMIZATION.
40676 * opth-gen.awk: Test for PerFunction flag along with
40678 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
40679 it only when the latter is present. Skip those that don't in
40680 the hash function generator.
40681 * common.opt (fvar-tracking): Mark as PerFunction instead of
40683 (fvar-tracking-assignments): Likewise.
40684 (fvar-tracking-assignments-toggle): Likewise.
40685 (fvar-tracking-uninit): Likewise.
40687 2017-01-07 Jakub Jelinek <jakub@redhat.com>
40689 PR translation/79018
40690 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
40693 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
40696 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
40697 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
40698 TARGET_LONG_JUMP_TABLE_OFFSETS.
40699 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
40700 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
40701 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
40702 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
40703 * config/m68k/m68k.md (tablejump expander): Likewise.
40704 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
40705 TARGET_LONG_JUMP_TABLE_OFFSETS.
40706 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
40707 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
40709 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
40710 David Holsgrove <david.holsgrove@xilinx.com>
40712 * common/config/microblaze/microblaze-common.c
40713 (TARGET_EXCEPT_UNWIND_INFO): Remove.
40714 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
40716 * config/microblaze/microblaze.c (microblaze_must_save_register)
40717 (microblaze_expand_epilogue, microblaze_return_addr): Handle
40719 (microblaze_eh_return): New function.
40720 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
40721 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
40722 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
40723 * config/microblaze/microblaze.md (eh_return): New pattern.
40725 2017-01-06 Jakub Jelinek <jakub@redhat.com>
40727 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
40728 GCC_DIAGNOSTIC_STRINGIFY): Define.
40730 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
40732 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40734 * config/arm/arm.md (<mcrr>): New.
40736 * config/arm/arm.c (arm_arch5te): New.
40737 (arm_option_override): Set arm_arch5te.
40738 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
40740 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
40741 (arm_mcrr_qualifiers): ... this. New.
40742 (MRRC_QUALIFIERS): Define to...
40743 (arm_mrrc_qualifiers): ... this. New.
40744 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
40746 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
40747 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
40748 (MRRCI, mrrc, MRRC): New.
40749 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
40750 VUNSPEC_MRRC2): New.
40752 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40754 * config/arm/arm.md (<mcr>): New.
40756 * config/arm/arm.c (arm_coproc_builtin_available): Add
40757 support for mcr, mrc, mcr2 and mrc2.
40758 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
40759 (arm_mcr_qualifiers): ... this. New.
40760 (MRC_QUALIFIERS): Define to ...
40761 (arm_mrc_qualifiers): ... this. New.
40762 (MCR_QUALIFIERS): Define to ...
40763 (arm_mcr_qualifiers): ... this. New.
40764 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
40766 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
40767 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
40768 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
40769 VUNSPEC_MRC2): New.
40771 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40773 * config/arm/arm.md (*ldc): New.
40777 * config/arm/arm.c (arm_coproc_builtin_available): Add
40778 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
40779 (arm_coproc_ldc_stc_legitimate_address): New.
40780 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
40781 'qualifier_const_pointer'.
40782 (LDC_QUALIFIERS): Define to...
40783 (arm_ldc_qualifiers): ... this. New.
40784 (STC_QUALIFIERS): Define to...
40785 (arm_stc_qualifiers): ... this. New.
40786 * config/arm/arm-protos.h
40787 (arm_coproc_ldc_stc_legitimate_address): New.
40788 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
40789 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
40790 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
40791 stc2, stcl, stc2l): New.
40792 * config/arm/constraints.md (Uz): New.
40793 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
40794 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
40795 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
40796 VUNSPEC_STC2L): New.
40798 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40800 * config/arm/arm.md (<cdp>): New.
40801 * config/arm/arm.c (neon_const_bounds): Rename this ...
40802 (arm_const_bounds): ... this.
40803 (arm_coproc_builtin_available): New.
40804 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
40805 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
40806 (CDP_QUALIFIERS): Define to...
40807 (arm_cdp_qualifiers): ... this. New.
40809 (arm_expand_builtin_args): Add case for 6 arguments.
40810 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
40811 (arm_const_bounds): ... this.
40812 (arm_coproc_builtin_available): New.
40813 * config/arm/arm_acle.h (__arm_cdp): New.
40815 * config/arm/arm_acle_builtins.def (cdp): New.
40817 * config/arm/iterators.md (CDPI,CDP,cdp): New.
40818 * config/arm/neon.md: Rename all 'neon_const_bounds' to
40819 'arm_const_bounds'.
40820 * config/arm/types.md (coproc): New.
40821 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
40822 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
40823 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
40824 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
40826 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40828 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
40829 (UBINOP_QUALIFIERS): New.
40831 (acle_builtin_data): New. Change comment.
40832 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
40833 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
40834 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
40835 arm_acle_builtins.def.
40836 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
40837 (arm_init_acle_builtins): New.
40838 (CRC32_BUILTIN): Remove.
40839 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
40840 crc32cb, crc32ch and crc32cw.
40841 (arm_init_crc32_builtins): Remove.
40842 (arm_init_builtins): Use arm_init_acle_builtins rather
40843 than arm_init_crc32_builtins.
40844 (arm_expand_acle_builtin): New.
40845 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
40846 * config/arm/arm_acle_builtins.def: New.
40848 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40850 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
40851 (arm_builtin_datum): ... this.
40852 (arm_init_neon_builtin): Rename to ...
40853 (arm_init_builtin): ... this. Add a new parameters PREFIX
40854 and USE_SIG_IN_NAME.
40855 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
40856 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
40857 'arm_builtin_datum'.
40858 (arm_init_vfp_builtins): Likewise.
40859 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
40860 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
40861 (arm_expand_neon_args): Rename to ...
40862 (arm_expand_builtin_args): ... this. Rename builtin_arg
40863 enum values and differentiate between ARG_BUILTIN_MEMORY
40864 and ARG_BUILTIN_NEON_MEMORY.
40865 (arm_expand_neon_builtin_1): Rename to ...
40866 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
40867 values, arm_expand_builtin_args and add bool parameter NEON.
40868 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
40869 (arm_expand_vfp_builtin): Likewise.
40870 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
40872 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40874 PR middle-end/77484
40875 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
40876 * predict.c (tree_estimate_probability_bb): Reverse direction of
40877 polymorphic call predictor.
40879 2017-01-06 David Malcolm <dmalcolm@redhat.com>
40881 * passes.c (execute_one_pass): Split out pass-skipping logic into...
40882 (determine_pass_name_match): ...this new function and...
40883 (should_skip_pass_p): ...this new function.
40885 2017-01-06 Nathan Sidwell <nathan@acm.org>
40887 * ipa-visibility.c (function_and_variable_visibility): Reformat
40888 comments and long lines. Remove extrneous if.
40889 * symtab.c (symtab_node::make_decl_local): Fix code format.
40890 (symtab_node::set_section_for_node): Fix comment typo.
40892 2017-01-06 Martin Liska <mliska@suse.cz>
40895 * lra-constraints.c: Rename invariant to lra_invariant.
40896 * predict.c (set_even_probabilities): Initialize e to NULL.
40898 2017-01-05 Martin Sebor <msebor@redhat.com>
40900 PR tree-optimization/78910
40901 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
40902 (format_integer): Correct off-by-one error in the handling
40903 of precision with negative numbers in signed conversions..
40905 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
40907 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
40909 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40911 PR tree-optimization/71016
40912 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
40913 factor_out_conditional_conversion. Formatting fix.
40914 (factor_out_conditional_conversion): Add cond_stmt argument.
40915 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
40916 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
40919 2017-01-05 David Malcolm <dmalcolm@redhat.com>
40921 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
40922 read-rtl-function.o, and selftest-rtl.o.
40923 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
40924 (selftest::aarch64_test_loading_full_dump): New function.
40925 (selftest::aarch64_run_selftests): New function.
40926 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
40927 selftest::aarch64_run_selftests.
40928 * config/i386/i386.c
40929 (selftest::ix86_test_loading_dump_fragment_1): New function.
40930 (selftest::ix86_test_loading_call_insn): New function.
40931 (selftest::ix86_test_loading_full_dump): New function.
40932 (selftest::ix86_test_loading_unspec): New function.
40933 (selftest::ix86_run_selftests): Call the new functions.
40934 * emit-rtl.c (maybe_set_max_label_num): New function.
40935 * emit-rtl.h (maybe_set_max_label_num): New decl.
40936 * function.c (instantiate_decls): Guard call to
40937 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
40938 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
40940 * gensupport.c (gen_reader::gen_reader): Pass "false"
40941 for new "compact" param of rtx_reader.
40942 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
40943 rather than an empty string for NULL strings.
40944 * read-md.c: Potentially include config.h rather than bconfig.h.
40945 Wrap include of errors.h with #ifdef GENERATOR_FILE.
40946 (have_error): New global, copied from errors.c.
40947 (md_reader::read_name): Rename to...
40948 (md_reader::read_name_1): ...this, adding "out_loc" param,
40949 and converting "missing name or number" to returning false, rather
40951 (md_reader::read_name): Reimplement in terms of read_name_1.
40952 (md_reader::read_name_or_nil): New function.
40953 (md_reader::read_string): Handle "(nil)" by returning NULL.
40954 (md_reader::md_reader): Add new param "compact".
40955 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
40956 (md_reader::read_file): New method.
40957 * read-md.h (md_reader::md_reader): Add new param "compact".
40958 (md_reader::read_file): New method.
40959 (md_reader::is_compact): New accessor.
40960 (md_reader::read_name): Convert return type from void to file_location.
40961 (md_reader::read_name_or_nil): New decl.
40962 (md_reader::read_name_1): New decl.
40963 (md_reader::m_compact): New field.
40964 (noop_reader::noop_reader): Pass "false" for new "compact" param
40966 (rtx_reader::rtx_reader): Add new "compact" param.
40967 (rtx_reader::read_rtx_operand): Make virtual and convert return
40968 type from void to rtx.
40969 (rtx_reader::read_until): New decl.
40970 (rtx_reader::handle_any_trailing_information): New virtual function.
40971 (rtx_reader::postprocess): New virtual function.
40972 (rtx_reader::finalize_string): New virtual function.
40973 (rtx_reader::m_in_call_function_usage): New field.
40974 (rtx_reader::m_reuse_rtx_by_id): New field.
40975 * read-rtl-function.c: New file.
40976 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
40977 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
40978 (selftest::verify_three_block_rtl_cfg): New decl.
40979 * read-rtl-function.h: New file.
40980 * read-rtl.c: Potentially include config.h rather than bconfig.h.
40981 For host, include function.h, memmodel.h, and emit-rtl.h.
40982 (one_time_initialization): New function.
40983 (struct compact_insn_name): New struct.
40984 (compact_insn_names): New array.
40985 (find_code): Handle insn codes in compact dumps.
40986 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
40987 (bind_subst_iter_and_attr): Likewise.
40988 (add_condition_to_string): Likewise.
40989 (add_condition_to_rtx): Likewise.
40990 (apply_attribute_uses): Likewise.
40991 (add_current_iterators): Likewise.
40992 (apply_iterators): Likewise.
40993 (initialize_iterators): Guard usage of apply_subst_iterator with
40994 #ifdef GENERATOR_FILE.
40995 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
40996 (md_reader::read_mapping): Likewise.
40997 (add_define_attr_for_define_subst): Likewise.
40998 (add_define_subst_attr): Likewise.
40999 (read_subst_mapping): Likewise.
41000 (check_code_iterator): Likewise.
41001 (rtx_reader::read_rtx): Likewise. Move one-time initialization
41003 (one_time_initialization): New function.
41004 (rtx_reader::read_until): New method.
41005 (read_flags): New function.
41006 (parse_reg_note_name): New function.
41007 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
41008 Handle reuse_rtx ids.
41009 Wrap iterator lookup within #ifdef GENERATOR_FILE.
41010 Add parsing support for RTL dumps, mirroring the special-cases in
41011 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
41012 values, and calling handle_any_trailing_information.
41013 (rtx_reader::read_rtx_operand): Convert return type from void
41014 to rtx, returning return_rtx. Handle case 'e'. Call
41015 finalize_string on XSTR and XTMPL fields.
41016 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
41017 "(nil)" values were omitted. Call the postprocess vfunc on the
41019 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
41020 class ctor. Initialize m_in_call_function_usage. Call
41021 one_time_initialization.
41022 * rtl-tests.c (selftest::test_uncond_jump): Call
41023 set_new_first_and_last_insn.
41024 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
41025 * selftest-rtl.c: New file.
41026 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
41027 (selftest::get_insn_by_uid): New decl.
41028 * selftest-run-tests.c (selftest::run_tests): Call
41029 read_rtl_function_c_tests.
41030 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
41031 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
41034 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
41036 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
41037 operands in a special way. Assert that pos+len <= mode precision.
41039 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41041 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
41042 3 argument Alias with unlimited for the negative form.
41043 (fno-vect-cost-model): Removed.
41045 2017-01-05 Martin Liska <mliska@suse.cz>
41047 * hsa-gen.c (gen_hsa_divmod): New function.
41048 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
41050 2017-01-05 Martin Liska <mliska@suse.cz>
41053 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
41056 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41058 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
41059 small constant length operands.
41061 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41063 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
41064 between loop iterations.
41066 2017-01-05 Martin Liska <mliska@suse.cz>
41069 * gimplify.c (gimplify_decl_expr): Compare to
41070 asan_poisoned_variables instread of checking flags.
41071 (gimplify_target_expr): Likewise.
41072 (gimplify_expr): Likewise.
41073 (gimplify_function_tree): Conditionally initialize
41074 asan_poisoned_variables.
41076 2017-01-04 Jeff Law <law@redhat.com>
41078 PR tree-optimizatin/78812
41079 * rtl.h (contains_mem_rtx_p): Prototype.
41080 * ifcvt.c (containts_mem_rtx_p): Move from here to...
41081 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
41082 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
41083 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
41084 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
41086 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41088 * input.c (assert_char_at_range): Default-initialize actual_range.
41090 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41092 * df-scan.c (df_ref_create_structure): Make regno unsigned,
41093 to match the caller.
41095 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41097 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
41098 insns after final jump in test to emit dummy move.
41100 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41102 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
41103 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
41105 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41107 * multiple_target.c (create_dispatcher_calls): Init e_next.
41108 * tree-ssa-loop-split.c (split_loop): Init border.
41109 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
41112 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
41117 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
41118 (altivec_register_operand): Do not return true if the operand
41119 contains a SUBREG mixing SImode and SFmode.
41120 (vsx_register_operand): Likewise.
41121 (vsx_reg_sfsubreg_ok): New predicate.
41122 (vfloat_operand): Do not return true if the operand contains a
41123 SUBREG mixing SImode and SFmode.
41124 (vint_operand): Likewise.
41125 (vlogical_operand): Likewise.
41126 (gpc_reg_operand): Likewise.
41127 (int_reg_operand): Likewise.
41128 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
41129 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
41130 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
41132 (rs6000_emit_move_si_sf_subreg): New helper function.
41133 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
41134 fixup SUBREGs involving SImode and SFmode.
41135 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
41136 numbers for the new peephole2 optimization.
41137 (peephole2 for SFmode unions): New peephole2 to optimize cases in
41138 the GLIBC math library that do AND/IOR/XOR operations on single
41139 precision floating point.
41140 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
41141 target macros to say whether we need to avoid SUBREGs mixing
41143 (TARGET_ALLOW_SF_SUBREG): Likewise.
41144 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
41145 (UNSPEC_SI_FROM_SF): Likewise.
41146 (iorxor): Change spacing.
41147 (and_ior_xor): New iterator for AND, IOR, and XOR.
41148 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
41149 (movdi_from_sf_zero_ext): Likewise.
41150 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
41151 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
41152 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
41153 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
41154 (fms<mode>4): Likewise.
41155 (fnma<mode>4): Likewise.
41156 (fnms<mode>4): Likewise.
41157 (nfma<mode>4): Likewise.
41158 (nfms<mode>4): Likewise.
41160 2017-01-04 Marek Polacek <polacek@redhat.com>
41163 * doc/invoke.texi: Document -Wpointer-compare.
41165 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41167 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
41170 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
41171 descriptions for -gdwarf-5 and emit them as uleb128 instead of
41174 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
41177 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
41178 documentation of the powerpc_popcntb_ok attribute.
41179 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
41180 code to issue warning messages if a requested CPU configuration is
41181 not supported by the binary (assembler and loader) toolchain.
41182 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
41183 made to define a built-in function that has been disabled.
41184 (paired_init_builtins): Add assertion to prevent ICE if attempt is
41185 made to define a built-in function that has been disabled.
41186 (altivec_init_builtins): Add comment explaining why definition
41187 of the DST built-in functions is not preceded by an assertion
41188 check. Add assertions to prevent ICE if attempts are made to
41189 define an altivec predicate or an abs* built-in function that has
41191 (htm_init_builtins): Add comment explaining why definition of the
41192 htm built-in functions is not preceded by an assertion check.
41194 2017-01-04 Jeff Law <law@redhat.com>
41196 PR tree-optimizatin/67955
41197 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
41198 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
41199 the points-to solution does not include pt_null. Use DECL_PT_UID
41202 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
41204 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
41205 Use gen_int_mode instead of gen_lopwart for const_int operands.
41207 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41209 PR tree-optimization/71563
41210 * match.pd: Simplify X << Y into X if Y is known to be 0 or
41211 out of range value - has low bits known to be zero.
41213 2017-01-04 Alan Modra <amodra@gmail.com>
41215 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
41216 * configure: Regenerate.
41217 * config.in: Regenerate.
41219 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41222 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
41223 a substring of the message, but strcmp with the whole message. Ifdef
41224 ENABLE_NLS, translate the message first using dgettext.
41226 2017-01-03 Jeff Law <law@redhat.com>
41228 PR tree-optimizatin/78856
41229 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
41230 (mark_threaded_blocks): Remove code to truncate thread paths that
41231 cross multiple loop headers. Instead invalidate the cached loop
41232 iteration information and handle case of a thread path walking
41233 into an irreducible region.
41235 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
41238 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
41239 assertions. Add support for doing the signbit if the IEEE 128-bit
41240 floating point value is in a GPR.
41241 * config/rs6000/rs6000.md (Fsignbit): Delete.
41242 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
41243 Update the length attribute if the value is in a GPR.
41244 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
41245 the sign or zero extension instruction, since the value is always 0/1.
41246 (signbit<mode>2_dm2): Delete using <Fsignbit>.
41249 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
41250 extracting SImode to a GPR register so that we can generate a
41251 store, limit the vector to be in a traditional Altivec register
41252 for the vextuwrx instruction.
41254 2017-01-03 Ian Lance Taylor <iant@google.com>
41256 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
41258 2017-01-03 Martin Sebor <msebor@redhat.com>
41260 PR tree-optimization/78696
41261 * gimple-ssa-sprintf.c (format_floating): Correct handling of
41262 precision. Use MPFR for %f for greater fidelity. Correct handling
41264 (pass_sprintf_length::compute_format_length): Set width and precision
41265 specified by asrerisk to void_node for vararg functions.
41266 (try_substitute_return_value): Adjust dump output.
41268 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
41270 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
41272 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
41274 * doc/invoke.texi (SPARC options): Document -mlra as the default.
41275 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
41276 -mlra/-mno-lra was passed to the compiler.
41278 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
41280 PR rtl-optimization/65618
41281 * emit-rtl.c (try_split): Move initialization of "before" and
41282 "after" to just before the call to emit_insn_after_setloc.
41284 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
41286 * doc/md.texi (Standard Names): Remove reference to Java frontend.
41288 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
41290 * dwarf2out.c (gen_enumeration_type_die): When
41291 -gno-strict-dwarf, add a DW_AT_encoding attribute.
41293 2017-01-03 Jakub Jelinek <jakub@redhat.com>
41295 PR tree-optimization/78965
41296 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
41297 Change first argument from const call_info & to call_info &. For %n
41298 set info.nowrite to false.
41300 PR middle-end/78901
41301 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
41302 possibly throwing calls.
41304 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
41305 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
41306 and fns handling, rather than in a separate case SSA_NAME.
41308 2017-01-02 Jeff Law <law@redhat.com>
41310 * config/darwin-driver.c (darwin_driver_init): Const-correctness
41311 fixes for first_period and second_period variables.
41313 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
41316 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
41317 (*insvqi_1): New insn pattern.
41318 (*insvqi_1_mem_rex64): Ditto.
41319 (*insvqi_2): Ditto.
41320 (*insvqi_3): Rename from *insvqi.
41322 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
41324 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
41326 * doc/cfg.texi (Edges): Remove reference to Java.
41327 (Maintaining the CFG): Ditto.
41329 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41331 PR middle-end/77674
41332 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
41333 transparent aliases.
41335 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41337 PR middle-end/77484
41338 * predict.def (PRED_CALL): Update hitrate.
41339 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
41340 * predict.c (tree_estimate_probability_bb): Split CALL predictor
41341 into direct/indirect/polymorphic variants.
41343 2017-01-01 Jakub Jelinek <jakub@redhat.com>
41345 Update copyright years.
41347 * gcc.c (process_command): Update copyright notice dates.
41348 * gcov-dump.c (print_version): Ditto.
41349 * gcov.c (print_version): Ditto.
41350 * gcov-tool.c (print_version): Ditto.
41351 * gengtype.c (create_file): Ditto.
41352 * doc/cpp.texi: Bump @copying's copyright year.
41353 * doc/cppinternals.texi: Ditto.
41354 * doc/gcc.texi: Ditto.
41355 * doc/gccint.texi: Ditto.
41356 * doc/gcov.texi: Ditto.
41357 * doc/install.texi: Ditto.
41358 * doc/invoke.texi: Ditto.
41360 Copyright (C) 2017 Free Software Foundation, Inc.
41362 Copying and distribution of this file, with or without modification,
41363 are permitted in any medium without royalty provided the copyright
41364 notice and this notice are preserved.