1 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
3 * combine.c (simplify_set): Do not transform subregs to zero_extends
4 if the destination is not a scalar int mode.
6 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
9 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
10 shared zero if the input has overflowed.
12 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
15 support for Cortex-A55 and Cortex-A75.
17 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19 * common/config/arm/arm-common.c (compare_opt_names): Add function
20 comment. Use strcmp instead of manual loop.
22 2017-12-21 Martin Liska <mliska@suse.cz>
25 * gcov-dump.c (dump_gcov_file): Do not read info about
26 support_unexecuted_blocks for gcda files.
28 2017-12-21 Jakub Jelinek <jakub@redhat.com>
30 PR rtl-optimization/82973
31 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
32 (valid_for_const_vector_p): ... this.
33 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
34 (valid_for_const_vector_p): ... this. Adjust function comment.
35 (gen_vec_duplicate): Adjust caller.
36 * optabs.c (expand_vector_broadcast): Likewise.
37 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
38 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
40 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
41 instead of GET_CODE == CONST_FIXED.
42 (simplify_subreg): Use CONST_FIXED_P macro instead of
43 GET_CODE == CONST_FIXED.
46 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
47 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
48 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
49 (ix86_option_override_internal): Adjust for
50 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
51 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
52 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
53 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
54 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
55 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
56 def_builtin2 for CET builtins.
57 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
58 up formatting in IX86_BUILTIN_RDPID code.
59 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
60 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
61 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
62 ix86_isa_flags2 to ix86_isa_flags.
63 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
65 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
66 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
67 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
69 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
70 Or in OPTION_MASK_ISA_AVX512F_SET.
71 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
72 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
73 (ix86_handle_option): Adjust for
74 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
75 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
77 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
78 Alan Hayward <alan.hayward@arm.com>
79 David Sherwood <david.sherwood@arm.com>
81 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
82 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
83 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
84 Track polynomial offsets.
86 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
87 Alan Hayward <alan.hayward@arm.com>
88 David Sherwood <david.sherwood@arm.com>
90 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
91 Treat drb->init as a poly_int. Fail if its misalignment wrt
92 vector_alignment isn't known.
94 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
95 Alan Hayward <alan.hayward@arm.com>
96 David Sherwood <david.sherwood@arm.com>
98 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
100 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
101 (vect_loop_versioning): Take the loop versioning threshold as a
103 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
104 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
105 versioning_threshold.
106 (vect_analyze_loop_2): Compute the loop versioning threshold
107 whenever loop versioning is needed, and store it in the new
108 field rather than combining it with the cost model threshold.
109 (vect_transform_loop): Update call to vect_loop_versioning.
110 Try to combine the loop versioning and cost thresholds here.
112 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
113 Alan Hayward <alan.hayward@arm.com>
114 David Sherwood <david.sherwood@arm.com>
116 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
117 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
118 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
119 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
120 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
121 (group_compare_offset): Update accordingly.
122 (split_small_address_groups_p): Likewise.
123 (record_use): Take addr_offset as a poly_uint64 rather than
124 an unsigned HOST_WIDE_INT.
125 (strip_offset): Return the offset as a poly_uint64 rather than
126 an unsigned HOST_WIDE_INT.
127 (record_group_use, split_address_groups): Track polynomial offsets.
128 (add_iv_candidate_for_use): Likewise.
129 (addr_offset_valid_p): Take the offset as a poly_int64 rather
130 than a HOST_WIDE_INT.
131 (strip_offset_1): Return the offset as a poly_int64 rather than
134 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
135 Alan Hayward <alan.hayward@arm.com>
136 David Sherwood <david.sherwood@arm.com>
138 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
139 rather than a HOST_WIDE_INT.
140 * tree.c (get_binfo_at_offset): Likewise.
142 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
143 Alan Hayward <alan.hayward@arm.com>
144 David Sherwood <david.sherwood@arm.com>
146 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
147 rather than a HOST_WIDE_INT.
148 * tree-sra.c (build_ref_for_offset): Likewise.
150 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
151 Alan Hayward <alan.hayward@arm.com>
152 David Sherwood <david.sherwood@arm.com>
154 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
156 * tree.c (mem_ref_offset): Likewise.
157 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
158 * builtins.c (get_object_alignment_2): Likewise.
159 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
160 * gimple-fold.c (get_base_constructor): Likewise.
161 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
162 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
164 * ipa-polymorphic-call.c
165 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
166 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
167 (get_ancestor_addr_info): Likewise.
168 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
169 * match.pd: Likewise.
170 * tree-data-ref.c (dr_analyze_innermost): Likewise.
171 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
172 * tree-eh.c (tree_could_trap_p): Likewise.
173 * tree-object-size.c (addr_object_size): Likewise.
174 * tree-ssa-address.c (copy_ref_info): Likewise.
175 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
176 (indirect_refs_may_alias_p): Likewise.
177 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
178 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
179 (non_rewritable_mem_ref_base): Likewise.
180 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
181 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
182 * varasm.c (decode_addr_const): Likewise.
184 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
185 Alan Hayward <alan.hayward@arm.com>
186 David Sherwood <david.sherwood@arm.com>
188 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
189 offsets for MEM_REFs.
191 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
192 Alan Hayward <alan.hayward@arm.com>
193 David Sherwood <david.sherwood@arm.com>
195 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
196 as a poly_uint64 rather than an unsigned int.
197 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
199 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
200 Alan Hayward <alan.hayward@arm.com>
201 David Sherwood <david.sherwood@arm.com>
203 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
205 (decode_addr_const): Update accordingly.
207 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
208 Alan Hayward <alan.hayward@arm.com>
209 David Sherwood <david.sherwood@arm.com>
211 * tree.h (bit_field_size, bit_field_offset): New functions.
212 * hsa-gen.c (gen_hsa_addr): Use them.
213 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
214 (simplify_vector_constructor): Likewise.
215 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
216 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
217 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
218 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
221 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
222 Alan Hayward <alan.hayward@arm.com>
223 David Sherwood <david.sherwood@arm.com>
225 * expr.h (emit_group_load, emit_group_load_into_temps)
226 (emit_group_store): Take the size as a poly_int64 rather than an int.
227 * expr.c (emit_group_load_1, emit_group_load): Likewise.
228 (emit_group_load_into_temp, emit_group_store): Likewise.
230 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
231 Alan Hayward <alan.hayward@arm.com>
232 David Sherwood <david.sherwood@arm.com>
234 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
235 an unsigned int to a poly_uint64.
236 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
237 sizes as poly_uint64s rather than unsigned ints.
238 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
241 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
242 Alan Hayward <alan.hayward@arm.com>
243 David Sherwood <david.sherwood@arm.com>
245 * lra-constraints.c (emit_inc): Change inc_amount from an int
248 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
249 Alan Hayward <alan.hayward@arm.com>
250 David Sherwood <david.sherwood@arm.com>
252 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
254 (add_stack_var, stack_var_cmp, partition_stack_vars)
255 (dump_stack_var_partition): Update accordingly.
256 (alloc_stack_frame_space): Take the size as a poly_int64 rather
257 than a HOST_WIDE_INT.
258 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
259 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
260 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
261 rather than a HOST_WIDE_INT.
263 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
264 Alan Hayward <alan.hayward@arm.com>
265 David Sherwood <david.sherwood@arm.com>
267 * target.def (return_pops_args): Treat both the input and output
268 sizes as poly_int64s rather than HOST_WIDE_INTS.
269 * targhooks.h (default_return_pops_args): Update accordingly.
270 * targhooks.c (default_return_pops_args): Likewise.
271 * doc/tm.texi: Regenerate.
272 * emit-rtl.h (incoming_args): Change pops_args, size and
273 outgoing_args_size from int to poly_int64_pod.
274 * function.h (expr_status): Change x_pending_stack_adjust and
275 x_stack_pointer_delta from int to poly_int64.
276 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
277 (ARGS_SIZE_RTX): Update accordingly.
278 * calls.c (highest_outgoing_arg_in_use): Change from int to
280 (stack_usage_watermark, stored_args_watermark): New variables.
281 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
282 (emit_call_1): Change the stack_size and rounded_stack_size
283 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
285 (save_fixed_argument_area): Check stack_usage_watermark.
286 (initialize_argument_information): Change old_pending_adj from
287 a HOST_WIDE_INT * to a poly_int64_pod *.
288 (compute_argument_block_size): Return the size as a poly_int64
290 (finalize_must_preallocate): Track polynomial argument sizes.
291 (compute_argument_addresses): Likewise.
292 (internal_arg_pointer_based_exp): Track polynomial offsets.
293 (mem_overlaps_already_clobbered_arg_p): Rename to...
294 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
295 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
296 Check stored_args_used_watermark.
297 (load_register_parameters): Update accordingly.
298 (check_sibcall_argument_overlap_1): Likewise.
299 (combine_pending_stack_adjustment_and_call): Take the unadjusted
300 args size as a poly_int64 rather than an int. Return a bool
301 indicating whether the optimization was possible and return
302 the new adjustment by reference.
303 (check_sibcall_argument_overlap): Track polynomail argument sizes.
304 Update stored_args_watermark.
305 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
306 (expand_call): Likewise. Maintain stack_usage_watermark and
307 stored_args_watermark. Update calls to
308 combine_pending_stack_adjustment_and_call.
309 (emit_library_call_value_1): Handle polynomial argument sizes.
310 Call stack_region_maybe_used_p and mark_stack_region_used.
311 Maintain stack_usage_watermark.
312 (store_one_arg): Likewise. Update call to
313 mem_overlaps_already_clobbered_arg_p.
314 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
316 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
317 * config/microblaze/microblaze.c (microblaze_function_prologue):
319 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
320 TARGET_RETURN_POPS_ARGS interface.
321 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
323 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
324 * config/i386/i386.c (ix86_return_pops_args): Update for new
325 TARGET_RETURN_POPS_ARGS interface.
326 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
327 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
328 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
329 TARGET_RETURN_POPS_ARGS interface.
330 * config/vax/vax.c (vax_return_pops_args): Likewise.
331 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
332 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
333 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
334 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
335 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
336 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
337 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
338 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
339 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
340 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
341 and x_stack_pointer_delta from int to poly_int64.
342 * dojump.c (do_pending_stack_adjust): Update accordingly.
343 * explow.c (allocate_dynamic_stack_space): Handle polynomial
344 stack_pointer_deltas.
345 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
346 (pad_to_arg_alignment): Track polynomial offsets.
347 (assign_parm_find_stack_rtl): Likewise.
348 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
349 * toplev.c (output_stack_usage): Update reference to
350 current_function_pushed_stack_size.
352 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
353 Alan Hayward <alan.hayward@arm.com>
354 David Sherwood <david.sherwood@arm.com>
356 * function.c (in_arg_offset, var_offset, dynamic_offset)
357 (out_arg_offset, cfa_offset): Change from int to poly_int64.
358 (instantiate_new_reg): Return the new offset as a poly_int64_pod
359 rather than a HOST_WIDE_INT.
360 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
361 (instantiate_virtual_regs_in_insn): Likewise.
363 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
364 Alan Hayward <alan.hayward@arm.com>
365 David Sherwood <david.sherwood@arm.com>
367 * rtl.h (get_args_size, add_args_size_note): New functions.
368 (find_args_size_adjust): Return a poly_int64 rather than a
370 (fixup_args_size_notes): Likewise. Make the same change to the
371 end_args_size parameter.
372 * rtlanal.c (get_args_size, add_args_size_note): New functions.
373 * builtins.c (expand_builtin_trap): Use add_args_size_note.
374 * calls.c (emit_call_1): Likewise.
375 * explow.c (adjust_stack_1): Likewise.
376 * cfgcleanup.c (old_insns_match_p): Update use of
377 find_args_size_adjust.
378 * combine.c (distribute_notes): Track polynomial arg sizes.
379 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
380 end_true_args_size, beg_delay_args_size and end_delay_args_size
381 from HOST_WIDE_INT to poly_int64.
382 (add_cfi_args_size): Take the args_size as a poly_int64 rather
383 than a HOST_WIDE_INT.
384 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
385 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
386 polynomial arg sizes.
387 * emit-rtl.c (try_split): Use get_args_size.
388 * recog.c (peep2_attempt): Likewise.
389 * reload1.c (reload_as_needed): Likewise.
390 * expr.c (find_args_size_adjust): Return the adjustment as a
391 poly_int64 rather than a HOST_WIDE_INT.
392 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
393 to a poly_int64 and change the return type in the same way.
394 (emit_single_push_insn): Track polynomial arg sizes.
396 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
397 Alan Hayward <alan.hayward@arm.com>
398 David Sherwood <david.sherwood@arm.com>
400 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
401 from HOST_WIDE_INT to poly_int64.
402 * expr.c (push_block, emit_push_insn): Likewise.
404 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
405 Alan Hayward <alan.hayward@arm.com>
406 David Sherwood <david.sherwood@arm.com>
408 * function.h (frame_space): Change start and length from HOST_WIDE_INT
410 (get_frame_size): Return the size as a poly_int64 rather than a
412 (frame_offset_overflow): Take the offset as a poly_int64 rather
413 than a HOST_WIDE_INT.
414 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
415 (assign_stack_temp): Likewise for the size.
416 * function.c (get_frame_size): Return a poly_int64 rather than
418 (frame_offset_overflow): Take the offset as a poly_int64 rather
419 than a HOST_WIDE_INT.
420 (try_fit_stack_local): Take the start, length and size as poly_int64s
421 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
422 rather than a HOST_WIDE_INT.
423 (add_frame_space): Take the start and end as poly_int64s rather than
425 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
426 (assign_stack_temp): Likewise for the size.
427 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
429 (find_temp_slot_from_address): Handle polynomial offsets.
430 (combine_temp_slots): Likewise.
431 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
433 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
434 poly_int64 rather than a HOST_WIDE_INT.
435 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
436 than a HOST_WIDE_INT.
437 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
438 polynomial frame offsets.
439 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
440 as a poly_int64 rather than an int.
441 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
442 * config/v850/v850-protos.h (compute_frame_size): Likewise.
443 * config/v850/v850.c (compute_frame_size): Likewise.
444 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
445 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
446 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
447 * config/pa/pa.c (pa_compute_frame_size): Likewise.
448 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
449 rather than a HOST_WIDE_INT.
450 * explow.c (get_dynamic_stack_base): Likewise.
451 * final.c (final_start_function): Use the constant lower bound
452 of the frame size for -Wframe-larger-than.
453 * ira.c (do_reload): Adjust for new get_frame_size return type.
454 * lra.c (lra): Likewise.
455 * reload1.c (reload): Likewise.
456 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
457 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
458 * rtlanal.c (get_initial_register_offset): Return the offset as
459 a poly_int64 rather than a HOST_WIDE_INT.
461 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
462 Alan Hayward <alan.hayward@arm.com>
463 David Sherwood <david.sherwood@arm.com>
465 * reload1.c (elim_table): Change initial_offset, offset and
466 previous_offset from HOST_WIDE_INT to poly_int64_pod.
467 (offsets_at): Change the target array's element type from
468 HOST_WIDE_INT to poly_int64_pod.
469 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
470 (elimination_costs_in_insn, update_eliminable_offsets)
471 (verify_initial_elim_offsets, set_offsets_for_label)
472 (init_eliminable_invariants): Update after above changes.
474 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
475 Alan Hayward <alan.hayward@arm.com>
476 David Sherwood <david.sherwood@arm.com>
478 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
479 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
480 (decomposition): Change start and end from HOST_WIDE_INT
482 (decompose, immune_p): Update accordingly.
483 (find_inc_amount): Return a poly_int64 rather than an int.
484 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
487 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
488 Alan Hayward <alan.hayward@arm.com>
489 David Sherwood <david.sherwood@arm.com>
491 * tree.h (get_inner_reference): Return the bitsize and bitpos
492 as poly_int64_pods rather than HOST_WIDE_INT.
493 * fold-const.h (ptr_difference_const): Return the pointer difference
494 as a poly_int64_pod rather than a HOST_WIDE_INT.
495 * expr.c (get_inner_reference): Return the bitsize and bitpos
496 as poly_int64_pods rather than HOST_WIDE_INT.
497 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
499 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
500 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
501 (optimize_bit_field_compare): Update call to get_inner_reference.
502 (decode_field_reference): Likewise.
503 (fold_unary_loc): Track polynomial offsets and sizes.
504 (split_address_to_core_and_offset): Return the bitpos as a
505 poly_int64_pod rather than a HOST_WIDE_INT.
506 (ptr_difference_const): Likewise for the pointer difference.
507 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
508 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
509 * dbxout.c (dbxout_expand_expr): Likewise.
510 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
511 (loc_list_from_tree_1, fortran_common): Likewise.
512 * gimple-laddress.c (pass_laddress::execute): Likewise.
513 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
514 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
515 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
516 * tree-affine.c (tree_to_aff_combination): Likewise.
517 (get_inner_reference_aff): Likewise.
518 * tree-data-ref.c (split_constant_offset_1): Likewise.
519 (dr_analyze_innermost): Likewise.
520 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
521 * tree-sra.c (ipa_sra_check_caller): Likewise.
522 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
523 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
524 (instrument_bool_enum_load, instrument_object_size): Likewise.
525 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
526 to get_inner_reference.
527 * hsa-gen.c (gen_hsa_addr): Likewise.
528 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
529 * tsan.c (instrument_expr): Likewise.
530 * match.pd: Update call to ptr_difference_const.
532 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
533 Alan Hayward <alan.hayward@arm.com>
534 David Sherwood <david.sherwood@arm.com>
536 * fold-const.c (fold_comparison): Track sizes and offsets as
537 poly_int64s rather than HOST_WIDE_INTs when folding address
540 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
541 Alan Hayward <alan.hayward@arm.com>
542 David Sherwood <david.sherwood@arm.com>
544 * expr.h (get_bit_range): Return the bitstart and bitend as
545 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
546 as a poly_int64 rather than a HOST_WIDE_INT.
547 * expr.c (get_bit_range): Likewise.
548 (expand_assignment): Update call accordingly.
549 * fold-const.c (optimize_bit_field_compare): Likewise.
551 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
552 Alan Hayward <alan.hayward@arm.com>
553 David Sherwood <david.sherwood@arm.com>
555 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
556 (aarch64_legitimate_address_p): Use it instead of an rtx code,
557 as an optional final parameter.
558 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
559 (aarch64_legitimate_address_p): Likewise.
560 (aarch64_print_address_internal): Take an aarch64_addr_query_type
561 instead of an rtx code.
562 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
563 (aarch64_legitimate_address_hook_p): Likewise.
564 (aarch64_print_ldpstp_address): Likewise.
565 (aarch64_print_operand_address): Likewise.
566 (aarch64_address_cost): Likewise.
567 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
568 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
569 (aarch64_mem_pair_lanes_operand): Likewise.
571 2017-12-20 Richard Biener <rguenther@suse.de>
573 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
574 update_stmt_if_modified.
576 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
578 PR tree-optimization/83491
579 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
580 before walking uses. Improve coding style and comments.
582 2017-12-20 Tom de Vries <tom@codesourcery.com>
584 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
586 2017-12-20 Jakub Jelinek <jakub@redhat.com>
589 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
591 (pass_ipa_free_fn_summary): Add small_p private data member,
592 initialize to false in the ctor.
593 (pass_ipa_free_fn_summary::clone,
594 pass_ipa_free_fn_summary::set_pass_param,
595 pass_ipa_free_fn_summary::gate): New methods.
596 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
597 | TODO_dump_symtab if small_p.
598 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
599 entry and add another instance of the pass with false parm after
601 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
602 ipa_free_fn_summary here.
604 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
606 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
608 2017-12-20 Martin Sebor <msebor@redhat.com>
611 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
612 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
613 (expand_builtin_strncmp): Same.
615 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
618 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
619 allowed debug stmts before labels.
620 (expand_gimple_basic_block): Likewise.
621 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
622 * gimple-iterator.h (gsi_after_labels): Likewise.
623 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
624 rename reused variable, and simplify using gsi_move_before.
625 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
626 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
627 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
628 (gimple_verify_flow_info, gimple_block_label): Likewise.
629 (make_blocks): Move debug markers after adjacent labels.
630 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
631 allowed debug insns outside blocks.
632 * df-scan.c (df_insn_delete): Likewise.
633 * lra-constraints.c (update_ebb_live_info): Likewise.
634 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
635 (vt_initialize, delete_vta_debug_insns): Likewise.
636 (reemit_marker_as_note): Drop BB parm. Adjust callers.
638 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
639 Alan Hayward <alan.hayward@arm.com>
640 David Sherwood <david.sherwood@arm.com>
642 * poly-int-types.h (round_down_to_byte_boundary): New macro.
643 (round_up_to_byte_boundary): Likewise.
644 * expr.h (get_bit_range): Add temporary shim.
645 * gimple-ssa-store-merging.c (store_operand_info): Change the
646 bitsize, bitpos, bitregion_start and bitregion_end fields from
647 unsigned HOST_WIDE_INT to poly_uint64.
648 (merged_store_group): Likewise load_align_base.
649 (compatible_load_p, compatible_load_p): Update accordingly.
650 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
651 (split_group, imm_store_chain_info::output_merged_store): Likewise.
652 (mem_valid_for_store_merging): Return the bitsize, bitpos,
653 bitregion_start and bitregion_end as poly_uint64s rather than
654 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
655 (handled_load): Take the bitsize, bitpos,
656 bitregion_start and bitregion_end as poly_uint64s rather than
657 unsigned HOST_WIDE_INTs.
658 (pass_store_merging::process_store): Update call to
659 mem_valid_for_store_merging.
661 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
662 Alan Hayward <alan.hayward@arm.com>
663 David Sherwood <david.sherwood@arm.com>
665 * builtins.c (get_object_alignment_2): Track polynomial offsets
666 and sizes. Update the alignment handling.
668 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
669 Alan Hayward <alan.hayward@arm.com>
670 David Sherwood <david.sherwood@arm.com>
672 * tree.h (get_inner_reference): Add a version that returns the
673 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
674 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
675 the case in which bitpos is not associated with the first byte.
677 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
678 Alan Hayward <alan.hayward@arm.com>
679 David Sherwood <david.sherwood@arm.com>
681 * tree-affine.h (get_inner_reference_aff): Return the size as a
683 * tree-affine.c (get_inner_reference_aff): Likewise.
684 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
685 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
687 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
688 Alan Hayward <alan.hayward@arm.com>
689 David Sherwood <david.sherwood@arm.com>
691 * fold-const.c (pointer_may_wrap_p): Take the offset as a
692 HOST_WIDE_INT rather than a poly_int64.
694 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
695 Alan Hayward <alan.hayward@arm.com>
696 David Sherwood <david.sherwood@arm.com>
698 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
699 HOST_WIDE_INT to poly_int64_pod.
700 (perform_symbolic_merge): Update accordingly.
701 (bswap_replace): Likewise.
703 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
704 Alan Hayward <alan.hayward@arm.com>
705 David Sherwood <david.sherwood@arm.com>
707 * tree-affine.h (aff_tree::offset): Change from widest_int
709 (wide_int_ext_for_comb): Delete.
710 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
711 constants as poly_widest_int rather than widest_int.
712 (aff_combination_constant_multiple_p): Return the multiplier
713 as a poly_widest_int.
714 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
716 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
717 version static and add an overload for poly_widest_int.
718 (aff_combination_const, aff_combination_add_cst)
719 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
720 the constants as poly_widest_int rather than widest_int.
721 (tree_to_aff_combination): Generalize INTEGER_CST case to
723 (aff_combination_to_tree): Track offsets as poly_widest_ints.
724 (aff_combination_add_product, aff_combination_mult): Handle
726 (aff_combination_constant_multiple_p): Return the multiplier
727 as a poly_widest_int.
728 * tree-predcom.c (determine_offset): Return the offset as a
730 (split_data_refs_to_components, suitable_component_p): Update
732 (valid_initializer_p): Update call to
733 aff_combination_constant_multiple_p.
734 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
735 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
736 as a poly_int64 rather than a HOST_WIDE_INT.
737 (get_address_cost): Handle polynomial offsets.
738 (iv_elimination_compare_lt): Likewise.
739 (rewrite_use_nonlinear_expr): Likewise.
741 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
742 Alan Hayward <alan.hayward@arm.com>
743 David Sherwood <david.sherwood@arm.com>
745 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
746 as a poly_int64_pod rather than a HOST_WIDE_INT.
747 (get_addr_base_and_unit_offset): Likewise.
748 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
749 (get_addr_base_and_unit_offset): Likewise.
750 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
751 to poly_int64 in example.
752 * fold-const.c (fold_binary_loc): Update call to
753 get_addr_base_and_unit_offset.
754 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
755 (maybe_canonicalize_mem_ref_addr): Likewise.
756 (gimple_fold_stmt_to_constant_1): Likewise.
757 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
759 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
760 * match.pd: Likewise.
761 * omp-low.c (lower_omp_target): Likewise.
762 * tree-sra.c (build_ref_for_offset): Likewise.
763 (build_debug_ref_for_model): Likewise.
764 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
765 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
766 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
767 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
768 (constant_pointer_difference): Likewise.
769 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
770 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
771 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
772 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
773 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
774 (set_ssa_val_to): Likewise.
775 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
776 (maybe_diag_stxncpy_trunc): Likewise.
777 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
778 * tree.c (build_simple_mem_ref_loc): Likewise.
779 (array_at_struct_end_p): Likewise.
781 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
782 Alan Hayward <alan.hayward@arm.com>
783 David Sherwood <david.sherwood@arm.com>
785 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
786 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
787 (get_ref_base_and_extent_hwi): Declare.
788 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
789 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
790 (get_ref_base_and_extent_hwi): New function.
791 * cfgexpand.c (expand_debug_expr): Update call to
792 get_ref_base_and_extent.
793 * dwarf2out.c (add_var_loc_to_decl): Likewise.
794 * gimple-fold.c (get_base_constructor): Return the offset as a
795 poly_int64_pod rather than a HOST_WIDE_INT.
796 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
797 * ipa-polymorphic-call.c
798 (ipa_polymorphic_call_context::set_by_invariant)
799 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
800 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
801 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
802 rather than get_ref_base_and_extent.
803 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
804 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
805 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
807 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
808 call to get_ref_base_and_extent.
809 * tree-sra.c (create_access, get_access_for_expr): Likewise.
810 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
811 (stmt_kills_ref_p): Likewise.
812 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
813 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
815 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
816 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
817 when calling native_encode_expr.
818 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
819 call to get_ref_base_and_extent.
820 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
821 get_ref_base_and_extent.
822 * var-tracking.c (track_expr_p): Likewise.
824 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
825 Alan Hayward <alan.hayward@arm.com>
826 David Sherwood <david.sherwood@arm.com>
828 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
829 HOST_WIDE_INT to poly_int64_pod.
830 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
831 polynomail parameter offsets.
833 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
834 Alan Hayward <alan.hayward@arm.com>
835 David Sherwood <david.sherwood@arm.com>
837 * gengtype.c (main): Handle poly_int64_pod.
838 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
839 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
840 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
841 from HOST_WIDE_INT to poly_int64_pod.
842 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
843 (copy_cfa): New function.
844 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
845 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
846 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
847 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
848 (initial_return_save): Treat offsets as poly_ints.
849 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
850 offset is nonconstant.
851 (reg_save): Take the offset as a poly_int64. Fall back to
852 DW_CFA_expression for nonconstant offsets.
853 (queue_reg_save): Take the offset as a poly_int64.
854 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
856 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
857 Alan Hayward <alan.hayward@arm.com>
858 David Sherwood <david.sherwood@arm.com>
860 * rtl.h (operand_subword, operand_subword_force): Take the offset
861 as a poly_uint64 an unsigned int.
862 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
864 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
865 Alan Hayward <alan.hayward@arm.com>
866 David Sherwood <david.sherwood@arm.com>
868 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
869 'p' format code. Use INT_LIST rather than SUBREG as the example of
870 a code with an XINT and an XEXP. Remove the implication that
871 accessing an rtx field using XINT is expected to work.
872 * rtl.def (SUBREG): Change format from "ei" to "ep".
873 * rtl.h (rtunion::rt_subreg): New field.
874 (XCSUBREG): New macro.
875 (SUBREG_BYTE): Use it.
876 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
877 Update constructor accordingly.
878 (subreg_shape::operator ==): Update accordingly.
879 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
880 than an unsigned int.
881 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
882 a poly_uint64 rather than an unsigned int.
883 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
884 than an unsigned int.
885 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
886 (subreg_size_highpart_offset): Return a poly_uint64 rather than
887 an unsigned int. Take the sizes as poly_uint64s.
888 (subreg_offset_from_lsb): Return a poly_uint64 rather than
889 an unsigned int. Take the shift as a poly_uint64 rather than
891 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
892 as a poly_uint64 rather than an unsigned int.
893 (simplify_subreg_regno): Likewise.
894 (byte_lowpart_offset): Return the memory offset as a poly_int64
896 (subreg_memory_offset): Likewise. Take the subreg offset as a
897 poly_uint64 rather than an unsigned int.
898 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
899 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
900 poly_uint64 rather than an unsigned int.
901 * rtl.c (rtx_format): Describe 'p' in comment.
902 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
903 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
904 offset as a poly_uint64 rather than an unsigned int.
905 (byte_lowpart_offset): Return the memory offset as a poly_int64
907 (subreg_memory_offset): Likewise. Take the subreg offset as a
908 poly_uint64 rather than an unsigned int.
909 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
910 mode sizes as poly_uint64s rather than unsigned ints. Return a
911 poly_uint64 rather than an unsigned int.
912 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
913 (copy_insn_1): Handle 'p'.
914 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
915 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
916 an unsigned int. Return the shift in the same way.
917 (subreg_lsb): Return the shift as a poly_uint64 rather than an
919 (subreg_size_offset_from_lsb): Take the sizes and shift as
920 poly_uint64s rather than unsigned ints. Return the offset as
922 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
923 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
925 * rtlhash.c (add_rtx): Handle 'p'.
926 * genemit.c (gen_exp): Likewise.
927 * gengenrtl.c (type_from_format, gendef): Likewise.
928 * gensupport.c (subst_pattern_match, get_alternatives_number)
929 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
930 (subst_dup): Likewise.
931 * gengtype.c (adjust_field_rtx_def): Likewise.
932 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
933 (match_pattern_2): Likewise.
934 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
935 (rtx_test::subreg_field): New function.
936 (operator ==, safe_to_hoist_p, transition_parameter_type)
937 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
938 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
939 * genpeep.c (match_rtx): Likewise.
940 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
941 (rtx_writer::print_rtx_operand): Handle 'p'.
942 (print_value): Handle SUBREG.
943 * read-rtl.c (apply_int_iterator): Likewise.
944 (rtx_reader::read_rtx_operand): Handle 'p'.
945 * alias.c (rtx_equal_for_memref_p): Likewise.
946 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
947 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
949 * calls.c (expand_call): Likewise.
950 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
951 (make_extraction, gen_lowpart_for_combine): Likewise.
952 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
954 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
955 rather than an unsigned int. Treat subreg offsets as poly_ints.
956 (exp_equiv_p): Handle 'p'.
957 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
958 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
959 * dse.c (find_shift_sequence): Likewise.
960 * dwarf2out.c (rtl_for_decl_location): Likewise.
961 * expmed.c (extract_low_bits): Likewise.
962 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
963 (expand_expr_real_2): Likewise.
964 * final.c (alter_subreg): Likewise.
965 (leaf_renumber_regs_insn): Handle 'p'.
966 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
967 Treat subreg offsets as poly_ints.
968 * fwprop.c (forward_propagate_and_simplify): Likewise.
969 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
970 * ira.c (get_subreg_tracking_sizes): Likewise.
971 * ira-conflicts.c (go_through_subreg): Likewise.
972 * ira-lives.c (process_single_reg_class_operands): Likewise.
973 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
974 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
975 as a poly_uint64 rather than an unsigned int.
976 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
977 subreg offsets as poly_ints.
978 * lra-constraints.c (operands_match_p): Handle 'p'.
979 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
980 * lra-spills.c (assign_mem_slot): Likewise.
981 * postreload.c (move2add_valid_value_p): Likewise.
982 * recog.c (general_operand, indirect_operand): Likewise.
983 * regcprop.c (copy_value, maybe_mode_change): Likewise.
984 (copyprop_hardreg_forward_1): Likewise.
985 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
986 (record_subregs_of_mode): Likewise.
987 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
988 * reload.c (operands_match_p): Handle 'p'.
989 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
990 * reload1.c (alter_reg, choose_reload_regs): Likewise.
991 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
992 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
993 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
994 (simplify_const_poly_int_tests<N>::run): Likewise.
995 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
996 a poly_uint64 rather than an unsigned int.
997 * valtrack.c (debug_lowpart_subreg): Likewise.
998 * var-tracking.c (var_lowpart): Likewise.
999 (loc_cmp): Handle 'p'.
1001 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1002 Alan Hayward <alan.hayward@arm.com>
1003 David Sherwood <david.sherwood@arm.com>
1005 * ira.c (get_subreg_tracking_sizes): New function.
1006 (init_live_subregs): Take an integer size rather than a register.
1007 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
1008 to init_live_subregs.
1010 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1011 Alan Hayward <alan.hayward@arm.com>
1012 David Sherwood <david.sherwood@arm.com>
1014 * expr.c (store_constructor_field): Change bitsize from a
1015 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
1016 HOST_WIDE_INT to a poly_int64.
1017 (store_constructor): Change size from a HOST_WIDE_INT to
1019 (store_field): Likewise bitsize and bitpos.
1021 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1022 Alan Hayward <alan.hayward@arm.com>
1023 David Sherwood <david.sherwood@arm.com>
1025 * expmed.h (store_bit_field): Change bitregion_start and
1026 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
1027 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
1028 (store_bit_field_1, store_integral_bit_field, store_bit_field)
1029 (store_fixed_bit_field, store_split_bit_field): Likewise.
1030 * expr.c (store_constructor_field, store_field): Likewise.
1031 (optimize_bitfield_assignment_op): Likewise. Make the same change
1032 to bitsize and bitpos.
1033 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
1034 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
1035 same change in the constructor arguments.
1036 (get_best_mode): Change bitregion_start and bitregion_end from
1037 unsigned HOST_WIDE_INT to poly_uint64.
1038 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
1039 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
1041 (bit_field_mode_iterator::next_mode): Update for new types
1042 of m_bitregion_start and m_bitregion_end.
1043 (get_best_mode): Change bitregion_start and bitregion_end from
1044 unsigned HOST_WIDE_INT to poly_uint64.
1046 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1047 Alan Hayward <alan.hayward@arm.com>
1048 David Sherwood <david.sherwood@arm.com>
1050 * rtl.h (simplify_gen_subreg): Add a temporary overload that
1051 accepts poly_uint64 offsets.
1052 * expmed.h (extract_bit_field): Take bitsize and bitnum as
1053 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1054 * expmed.c (lowpart_bit_field_p): Likewise.
1055 (extract_bit_field_as_subreg): New function, split out from...
1056 (extract_bit_field_1): ...here. Take bitsize and bitnum as
1057 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
1058 extractions, check that BITSIZE matches the size of the extracted
1059 value and that BITNUM is an exact multiple of that size.
1060 If all else fails, try forcing the value into memory if
1061 BITNUM is variable, and adjusting the address so that the
1062 offset is constant. Split the part that can only handle constant
1063 bitsize and bitnum out into...
1064 (extract_integral_bit_field): ...this new function.
1065 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
1066 rather than unsigned HOST_WIDE_INTs.
1068 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1069 Alan Hayward <alan.hayward@arm.com>
1070 David Sherwood <david.sherwood@arm.com>
1072 * expmed.h (store_bit_field): Take bitsize and bitnum as
1073 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1074 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
1075 that returns the byte size.
1076 (store_bit_field_1): Take bitsize and bitnum as
1077 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
1078 to simple_mem_bitfield_p. Split the part that can only handle
1079 constant bitsize and bitnum out into...
1080 (store_integral_bit_field): ...this new function.
1081 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
1082 than unsigned HOST_WIDE_INTs.
1083 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
1085 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1086 Alan Hayward <alan.hayward@arm.com>
1087 David Sherwood <david.sherwood@arm.com>
1089 * lra-int.h (lra_reg): Change offset from int to poly_int64.
1090 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
1092 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1093 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
1094 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
1095 offset as a poly_int64 rather than an int.
1096 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
1097 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1098 * lra-constraints.c (equiv_address_substitution): Track offsets
1100 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
1101 (curr_insn_transform): Handle the new form of sp_offset.
1102 * lra-eliminations.c (lra_elim_table): Change previous_offset
1103 and offset from HOST_WIDE_INT to poly_int64.
1104 (print_elim_table, update_reg_eliminate): Update accordingly.
1105 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
1106 (get_elimination): Update accordingly.
1107 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
1108 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1109 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
1110 poly_int64 offsets generally.
1111 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
1112 (mark_not_eliminable, init_elimination): Update accordingly.
1113 (remove_reg_equal_offset_note): Return a bool and pass the new
1114 offset back by pointer as a poly_int64.
1115 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
1116 rather than a HOST_WIDE_INT.
1117 (do_remat): Track offsets poly_int64s.
1118 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
1120 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1121 Alan Hayward <alan.hayward@arm.com>
1122 David Sherwood <david.sherwood@arm.com>
1124 * rtl.h (mem_attrs): Add a default constructor. Change size and
1125 offset from HOST_WIDE_INT to poly_int64.
1126 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
1127 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
1128 (widen_memory_access): Take the sizes and offsets as poly_int64s
1129 rather than HOST_WIDE_INTs.
1130 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
1131 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
1133 (adjust_offset_for_component_ref): Change the offset from a
1134 HOST_WIDE_INT to a poly_int64.
1135 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
1136 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
1137 * dce.c (find_call_stack_args): Likewise.
1138 * dse.c (record_store): Likewise.
1139 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
1140 * print-rtl.c (rtx_writer::print_rtx): Likewise.
1141 * read-rtl-function.c (test_loading_mem): Likewise.
1142 * rtlanal.c (may_trap_p_1): Likewise.
1143 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1144 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
1145 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
1146 (mem_attrs::mem_attrs): New function.
1147 (set_mem_attributes_minus_bitpos): Change bitpos from a
1148 HOST_WIDE_INT to poly_int64.
1149 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
1150 (clear_mem_offset, clear_mem_size, change_address)
1151 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
1152 initialize mem_attrs.
1153 (set_mem_offset, set_mem_size, adjust_address_1)
1154 (adjust_automodify_address_1, offset_address, widen_memory_access):
1155 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
1157 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1158 Alan Hayward <alan.hayward@arm.com>
1159 David Sherwood <david.sherwood@arm.com>
1161 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
1162 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
1163 rather than 0 to represent an unknown size. Assert that the size
1164 is known when the mode isn't BLKmode.
1165 (may_trap_p_1): Use -1 for unknown sizes.
1166 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
1168 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1169 Alan Hayward <alan.hayward@arm.com>
1170 David Sherwood <david.sherwood@arm.com>
1172 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
1173 to poly_int64. Update commentary for positions_needed.large.
1174 (read_info_type): Change offset and width from HOST_WIDE_INT
1176 (set_usage_bits): Likewise.
1177 (canon_address): Return the offset as a poly_int64 rather than
1178 a HOST_WIDE_INT. Use strip_offset_and_add.
1179 (set_all_positions_unneeded, any_positions_needed_p): Use
1180 positions_needed.large to track stores with non-constant widths.
1181 (all_positions_needed_p): Likewise. Take the offset and width
1182 as poly_int64s rather than ints. Assert that rhs is nonnull.
1183 (record_store): Cope with non-constant offsets and widths.
1184 Nullify the rhs of an earlier store if we can't tell which bytes
1186 (find_shift_sequence): Take the access_size and shift as poly_int64s
1188 (get_stored_val): Take the read_offset and read_width as poly_int64s
1189 rather than HOST_WIDE_INTs.
1190 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
1191 non-constant offsets and widths.
1193 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1194 Alan Hayward <alan.hayward@arm.com>
1195 David Sherwood <david.sherwood@arm.com>
1197 * inchash.h (inchash::hash::add_poly_int): New function.
1198 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
1199 Use poly_int64 rather than HOST_WIDE_INT.
1200 (ao_ref::max_size_known_p): New function.
1201 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
1202 rather than HOST_WIDE_INT.
1203 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
1204 to temporaries until its interface is adjusted to match.
1205 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
1206 (aliasing_component_refs_p, decl_refs_may_alias_p)
1207 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
1208 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
1209 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
1211 * alias.c (ao_ref_from_mem): Likewise.
1212 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1213 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
1214 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
1215 (maybe_trim_complex_store, maybe_trim_constructor_store)
1216 (live_bytes_read, dse_classify_store): Likewise.
1217 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
1218 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
1219 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
1220 (vn_reference_lookup_3): Likewise.
1221 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
1223 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1224 Alan Hayward <alan.hayward@arm.com>
1225 David Sherwood <david.sherwood@arm.com>
1227 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
1228 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
1229 instead of ranges_overlap_p.
1231 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1232 Alan Hayward <alan.hayward@arm.com>
1233 David Sherwood <david.sherwood@arm.com>
1235 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
1236 sizes as poly_int64s rather than HOST_WIDE_INTs.
1238 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1239 Alan Hayward <alan.hayward@arm.com>
1240 David Sherwood <david.sherwood@arm.com>
1242 * gimple-fold.h (fold_ctor_reference): Take the offset and size
1243 as poly_uint64 rather than unsigned HOST_WIDE_INT.
1244 * gimple-fold.c (fold_ctor_reference): Likewise.
1246 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1247 Alan Hayward <alan.hayward@arm.com>
1248 David Sherwood <david.sherwood@arm.com>
1250 * target.def (dwarf_poly_indeterminate_value): New hook.
1251 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
1252 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
1253 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
1254 * doc/tm.texi: Regenerate.
1255 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
1256 offset as a poly_int64.
1257 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
1258 offset as a poly_int64.
1259 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
1260 Take the offset as a poly_int64.
1261 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
1262 (frame_pointer_fb_offset): Change to a poly_int64.
1263 (int_loc_descriptor): Take the offset as a poly_int64. Use
1264 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
1265 (based_loc_descr): Take the offset as a poly_int64.
1266 Use strip_offset_and_add to handle (plus X (const)).
1267 Use new_reg_loc_descr instead of an open-coded version of the
1268 previous implementation.
1269 (mem_loc_descriptor): Handle CONST_POLY_INT.
1270 (compute_frame_pointer_to_fb_displacement): Take the offset as a
1271 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
1273 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1274 Alan Hayward <alan.hayward@arm.com>
1275 David Sherwood <david.sherwood@arm.com>
1277 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
1278 (gen_rtx_REG_offset): Take the offset as a poly_int64.
1279 * inchash.h (inchash::hash::add_poly_hwi): New function.
1280 * gengtype.c (main): Register poly_int64.
1281 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
1282 offset as a poly_int.
1283 (reg_attr_hasher::equal): Use must_eq to compare offsets.
1284 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
1285 offset as a poly_int64.
1286 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
1287 * print-rtl.c (print_poly_int): New function.
1288 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
1290 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
1292 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
1293 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
1294 (vt_get_decl_and_offset): Return the offset as a poly_int64.
1295 Enforce track_offset_p for parts of a PARALLEL.
1296 (vt_add_function_parameter): Use const_offset for the final
1297 offset to track. Use get_tracked_reg_offset for the parts
1300 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1301 Alan Hayward <alan.hayward@arm.com>
1302 David Sherwood <david.sherwood@arm.com>
1304 * target.def (truly_noop_truncation): Take poly_uint64s instead of
1305 unsigned ints. Change default to hook_bool_puint64_puint64_true.
1306 * doc/tm.texi: Regenerate.
1307 * hooks.h (hook_bool_uint_uint_true): Delete.
1308 (hook_bool_puint64_puint64_true): Declare.
1309 * hooks.c (hook_bool_uint_uint_true): Delete.
1310 (hook_bool_puint64_puint64_true): New function.
1311 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
1312 instead of unsigned ints.
1313 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
1314 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
1316 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1317 Alan Hayward <alan.hayward@arm.com>
1318 David Sherwood <david.sherwood@arm.com>
1320 * optabs.h (expand_operand): Add an int_value field.
1321 (create_expand_operand): Add an int_value parameter and use it
1322 to initialize the new expand_operand field.
1323 (create_integer_operand): Replace with a declaration of a function
1324 that accepts poly_int64s. Move the implementation to...
1325 * optabs.c (create_integer_operand): ...here.
1326 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
1327 the mode preserves the value of int_value, instead of calling
1328 const_int_operand on the rtx. Use gen_int_mode to generate
1331 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1332 Alan Hayward <alan.hayward@arm.com>
1333 David Sherwood <david.sherwood@arm.com>
1335 * dumpfile.h (dump_dec): Declare.
1336 * dumpfile.c (dump_dec): New function.
1337 * pretty-print.h (pp_wide_integer): Turn into a function and
1338 declare a poly_int version.
1339 * pretty-print.c (pp_wide_integer): New function for poly_ints.
1341 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1342 Alan Hayward <alan.hayward@arm.com>
1343 David Sherwood <david.sherwood@arm.com>
1345 * doc/generic.texi (POLY_INT_CST): Document.
1346 * tree.def (POLY_INT_CST): New tree code.
1347 * treestruct.def (TS_POLY_INT_CST): New tree layout.
1348 * tree-core.h (tree_poly_int_cst): New struct.
1349 (tree_node): Add a poly_int_cst field.
1350 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
1351 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
1352 instead of a wide_int_ref.
1353 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
1355 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1356 instead of an unsigned HOST_WIDE_INT.
1357 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
1358 (ptrdiff_tree_p): Declare.
1359 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
1360 extern inline implementations if the target doesn't use POLY_INT_CST.
1361 (poly_int_tree_p): New function.
1362 (wi::unextended_tree): New class.
1363 (wi::int_traits <unextended_tree>): New override.
1364 (wi::extended_tree): Add a default constructor.
1365 (wi::extended_tree::get_tree): New function.
1366 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
1367 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
1368 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
1369 (wi::tree_to_poly_wide_ref): New typedefs.
1370 (wi::ints_for): Provide overloads for extended_tree and
1372 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
1373 (wi::to_wide): New functions.
1374 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
1375 * tree.c (poly_int_cst_hasher): New struct.
1376 (poly_int_cst_hash_table): New variable.
1377 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
1378 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
1380 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
1381 (init_ttree): Initialize poly_int_cst_hash_table.
1382 (build_int_cst, build_int_cst_type, build_invariant_address): Take
1383 a poly_int64 instead of a HOST_WIDE_INT.
1384 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1385 instead of an unsigned HOST_WIDE_INT.
1386 (wide_int_to_tree): Rename to...
1387 (wide_int_to_tree_1): ...this.
1388 (build_new_poly_int_cst, build_poly_int_cst): New functions.
1389 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
1390 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
1391 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
1392 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
1393 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
1395 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
1396 (streamer_read_tree_body): Likewise.
1397 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
1398 (streamer_write_tree_body): Likewise.
1399 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
1400 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
1401 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
1402 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
1403 * gimple-expr.h (is_gimple_constant): Likewise.
1404 * gimplify.c (maybe_with_size_expr): Likewise.
1405 * print-tree.c (print_node): Likewise.
1406 * tree-data-ref.c (data_ref_compare_tree): Likewise.
1407 * tree-pretty-print.c (dump_generic_node): Likewise.
1408 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1409 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
1410 * tree-vrp.c (compare_values_warnv): Likewise.
1411 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
1412 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
1413 (force_expr_to_var_cost): Likewise.
1414 * tree-ssa-loop.c (for_each_index): Likewise.
1415 * fold-const.h (build_invariant_address, size_int_kind): Take a
1416 poly_int64 instead of a HOST_WIDE_INT.
1417 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
1418 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
1420 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
1421 (int_const_binop_2): New function, split out from...
1422 (int_const_binop_1): ...here. Handle POLY_INT_CST.
1423 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
1424 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
1425 * gimple-ssa-strength-reduction.c (slsr_process_add)
1426 (slsr_process_mul): Check for INTEGER_CSTs before using them
1428 * stor-layout.c (bits_from_bytes): New function.
1429 (bit_from_pos): Use it.
1430 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
1431 by BITS_PER_UNIT to get the TYPE_SIZE.
1432 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
1433 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
1435 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1436 Alan Hayward <alan.hayward@arm.com>
1437 David Sherwood <david.sherwood@arm.com>
1439 * doc/rtl.texi (const_poly_int): Document. Also document the
1440 rtl sharing behavior.
1441 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
1442 * rtl.h (const_poly_int_def): New struct.
1443 (rtx_def::u): Add a cpi field.
1444 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
1445 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
1446 (wi::rtx_to_poly_wide_ref): New typedef
1447 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
1448 (poly_int_rtx_p): New functions.
1449 (trunc_int_for_mode): Declare a poly_int64 version.
1450 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1451 (immed_wide_int_const): Take a poly_wide_int_ref rather than
1453 (strip_offset): Declare.
1454 (strip_offset_and_add): New function.
1455 * rtl.def (CONST_POLY_INT): New rtx code.
1456 * rtl.c (rtx_size): Handle CONST_POLY_INT.
1457 (shared_const_p): Use poly_int_rtx_p.
1458 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
1460 (gen_int_shift_amount): Likewise.
1461 * emit-rtl.c (const_poly_int_hasher): New class.
1462 (const_poly_int_htab): New variable.
1463 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
1464 (const_poly_int_hasher::hash): New function.
1465 (const_poly_int_hasher::equal): Likewise.
1466 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
1467 (immed_wide_int_const): Rename to...
1468 (immed_wide_int_const_1): ...this and make static.
1469 (immed_wide_int_const): New function, taking a poly_wide_int_ref
1470 instead of a wide_int_ref.
1471 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
1472 (gen_lowpart_common): Handle CONST_POLY_INT.
1473 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
1474 * cselib.c (cselib_hash_rtx): Likewise.
1475 * dwarf2out.c (const_ok_for_output_1): Likewise.
1476 * expr.c (convert_modes): Likewise.
1477 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
1478 * rtlhash.c (add_rtx): Likewise.
1479 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
1480 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1481 Handle existing CONST_POLY_INT rtxes.
1482 * expmed.h (expand_shift): Take a poly_int64 instead of a
1484 * expmed.c (expand_shift): Likewise.
1485 * rtlanal.c (strip_offset): New function.
1486 (commutative_operand_precedence): Give CONST_POLY_INT the same
1487 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
1489 * rtl-tests.c (const_poly_int_tests): New struct.
1490 (rtl_tests_c_tests): Use it.
1491 * simplify-rtx.c (simplify_const_unary_operation): Handle
1493 (simplify_const_binary_operation): Likewise.
1494 (simplify_binary_operation_1): Fold additions of symbolic constants
1495 and CONST_POLY_INTs.
1496 (simplify_subreg): Handle extensions and truncations of
1498 (simplify_const_poly_int_tests): New struct.
1499 (simplify_rtx_c_tests): Use it.
1500 * wide-int.h (storage_ref): Add default constructor.
1501 (wide_int_ref_storage): Likewise.
1502 (trailing_wide_ints): Use GTY((user)).
1503 (trailing_wide_ints::operator[]): Add a const version.
1504 (trailing_wide_ints::get_precision): New function.
1505 (trailing_wide_ints::extra_size): Likewise.
1507 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1508 Alan Hayward <alan.hayward@arm.com>
1509 David Sherwood <david.sherwood@arm.com>
1511 * emit-rtl.h (gen_int_shift_amount): Declare.
1512 * emit-rtl.c (gen_int_shift_amount): New function.
1513 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
1515 * calls.c (shift_return_value): Likewise.
1516 * cse.c (fold_rtx): Likewise.
1517 * dse.c (find_shift_sequence): Likewise.
1518 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
1519 (expand_shift, expand_smod_pow2): Likewise.
1520 * lower-subreg.c (shift_cost): Likewise.
1521 * optabs.c (expand_superword_shift, expand_doubleword_mult)
1522 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
1523 (expand_vec_perm_var): Likewise.
1524 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1525 (simplify_binary_operation_1): Likewise.
1526 * combine.c (try_combine, find_split_point, force_int_to_mode)
1527 (simplify_shift_const_1, simplify_shift_const): Likewise.
1528 (change_zero_ext): Likewise. Use simplify_gen_binary.
1530 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1532 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
1534 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1536 * doc/invoke.texi (ARM Options): Document accepted extension options
1537 for -march=armv8.3-a.
1539 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
1542 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
1543 or --with-float=softfp, set the default CPU to arm10e.
1545 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
1547 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
1548 * config/visium/predicates.md (const_shift_operand): Likewise.
1549 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
1550 (visium_legitimize_reload_address): Likewise.
1552 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1555 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
1557 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1560 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
1562 2017-12-20 Julia Koval <julia.koval@intel.com>
1564 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
1565 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
1566 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
1567 * config.gcc: Include vpclmulqdqintrin.h.
1568 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
1569 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
1571 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
1572 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
1573 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
1574 * config/i386/i386.c (isa2_opts): Add -mcx16.
1575 (isa_opts): Add -mpclmulqdq, remove -mcx16.
1576 (ix86_option_override_internal): Move mcx16 to flags2.
1577 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
1578 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
1579 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
1580 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
1581 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
1582 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
1583 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
1584 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
1585 * doc/invoke.texi: Add -mvpclmulqdq.
1587 2017-12-20 Tom de Vries <tom@codesourcery.com>
1590 * config/i386/i386.c (ix86_static_chain): Move DECL_STATIC_CHAIN test ...
1591 * calls.c (rtx_for_static_chain): ... here. New function.
1592 * calls.h (rtx_for_static_chain): Declare.
1593 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
1594 instead of targetm.calls.static_chain.
1595 * df-scan.c (df_get_entry_block_def_set): Same.
1597 2017-12-19 Tom de Vries <tom@codesourcery.com>
1599 PR tree-optimization/83493
1600 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
1603 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1605 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
1606 inform with hardcoded english plural handling.
1608 2017-12-18 Jeff Law <law@redhat.com>
1610 PR tree-optimization/83477
1611 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
1612 a non-virtual PHI, always push a new range.
1614 2017-12-19 Martin Sebor <msebor@redhat.com>
1617 * builtins.c (compute_objsize): Handle non-constant offsets.
1619 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1621 PR tree-optimization/83444
1622 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
1623 character load case, if get_stridx on MEM_REF's operand doesn't
1624 look usable, retry with get_addr_stridx.
1626 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
1629 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
1630 (variable_tracking_main_1): Keep markers even when VTA fails.
1633 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
1634 even if there are markers before it.
1635 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
1637 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1639 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
1640 typical order conditions.
1641 * sel-sched.c (extract_new_fences_from): Likewise.
1642 * config/visium/constraints.md (J, K, L): Likewise.
1643 * config/visium/predicates.md (const_shift_operand): Likewise.
1644 * config/visium/visium.c (visium_legitimize_address,
1645 visium_legitimize_reload_address): Likewise.
1646 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
1647 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
1648 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
1649 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
1650 SET_DUMP_DETAIL): Likewise.
1651 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
1652 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
1653 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
1654 avr_set_core_architecture, avr_set_current_function,
1655 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
1656 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
1657 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
1658 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
1659 avr_map_decompose, avr_fold_builtin): Likewise.
1660 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
1661 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
1662 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
1663 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
1664 * config/m32c/m32c.c (m32c_conditional_register_usage,
1665 m32c_address_cost): Likewise.
1666 * config/m32c/predicates.md (shiftcount_operand,
1667 longshiftcount_operand): Likewise.
1668 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
1669 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
1670 can_use_cdx_ldstw): Likewise.
1671 * config/nios2/nios2.h (CDX_REG_P): Likewise.
1672 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
1674 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
1675 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
1676 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
1677 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
1678 * config/vax/vax.c (adjacent_operands_p): Likewise.
1679 * config/ft32/constraints.md (L, b, KA): Likewise.
1680 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
1682 * cfgexpand.c (expand_stack_alignment): Likewise.
1683 * gcse.c (insert_expr_in_table): Likewise.
1684 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
1685 * cgraphunit.c (cgraph_node::expand): Likewise.
1686 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
1687 * emit-rtl.c (add_insn): Likewise.
1688 * input.c (dump_location_info): Likewise.
1689 * passes.c (NEXT_PASS): Likewise.
1690 * read-rtl-function.c (parse_note_insn_name,
1691 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
1693 * sched-rgn.c (sched_rgn_init): Likewise.
1694 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
1695 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
1696 if_then_else_cond, simplify_shift_const_1, simplify_comparison): Likewise.
1697 * explow.c (eliminate_constant_term): Likewise.
1698 * final.c (leaf_renumber_regs_insn): Likewise.
1699 * cfgrtl.c (print_rtl_with_bb): Likewise.
1700 * genhooks.c (emit_init_macros): Likewise.
1701 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
1702 * tree-data-ref.c (conflict_fn): Likewise.
1703 * selftest.c (assert_streq): Likewise.
1704 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
1705 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
1706 fold_binary_loc, multiple_of_p): Likewise.
1707 * reload.c (push_reload, find_equiv_reg): Likewise.
1708 * et-forest.c (et_nca, et_below): Likewise.
1709 * dbxout.c (dbxout_symbol_location): Likewise.
1710 * reorg.c (relax_delay_slots): Likewise.
1711 * dojump.c (do_compare_rtx_and_jump): Likewise.
1712 * gengtype-parse.c (type): Likewise.
1713 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
1714 simplify_const_relational_operation): Likewise.
1715 * reload1.c (do_output_reload): Likewise.
1716 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
1717 * gengtype.c (type_for_name): Likewise.
1718 * gimple-ssa-sprintf.c (format_directive): Likewise.
1720 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1723 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
1724 accessing it. Adjust comment.
1726 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1729 * predict.c (zero_one_minusone): New function.
1730 (apply_return_prediction): Avoid return prediction for functions
1731 returning only -1, 0 and 1 values, unless they only return -1 and 0
1734 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1736 * config/arc/arc.c (legitimate_scaled_address_p): Clean
1737 fall-through warning.
1738 (arc_compute_frame_size): Remove unused variables.
1739 (arc_print_operand): Fix fprintif format.
1740 (arc_can_follow_jump): Clean fall-through warning.
1742 2017-12-19 Marek Polacek <polacek@redhat.com>
1745 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
1748 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1750 * config/arc/arc.c (overriderregs): New variable.
1751 (arc_override_options): Track fixed/call saved/call options.
1752 (arc_conditional_register_usage): Check against overrideregs
1753 variable whenever we change register properties.
1755 2017-12-19 Nathan Sidwell <nathan@acm.org>
1757 * opts.c (finish_options): Don't prefix dump_base_name if it
1758 already contains directories.
1760 2017-12-19 Martin Liska <mliska@suse.cz>
1762 PR rtl-optimization/82675
1763 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
1764 more element in sbitmap.
1766 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
1768 * gimplify.c (gimplify_expr): Use error_operand_p.
1770 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
1773 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
1776 2017-12-19 Daniel Cederman <cederman@gaisler.com>
1778 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
1779 the jump is to a label.
1781 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1783 PR tree-optimization/83444
1784 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
1789 * ipa-inline.c (flatten_remove_node_hook): New function.
1790 (ipa_inline): Keep only nodes with flatten attribute at the end of
1791 the array in the order from ipa_reverse_postorder, only walk that
1792 portion of array for flattening, if there is more than one such
1793 node, temporarily register a removal hook and ignore removed nodes.
1795 PR tree-optimization/80631
1796 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
1797 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
1798 IFN_REDUC_MAX or IFN_REDUC_MIN.
1800 2017-12-18 Jeff Law <law@redhat.com>
1802 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
1803 record symbolic equivalences from backedges in the CFG.
1806 2017-11-19 Jeff Law <law@redhat.com>
1808 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1809 of degenerates resulting from ignoring an edge.
1811 2017-12-18 Martin Sebor <msebor@redhat.com>
1814 PR tree-optimization/78450
1815 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
1816 (handle_builtin_strlen): Call it.
1818 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
1820 PR rtl-optimization/83424
1821 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
1823 2017-12-18 Marek Polacek <polacek@redhat.com>
1826 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1827 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
1830 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1832 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
1833 (umaddsidi4, umaddsidi_split): Likewise.
1835 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1837 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
1840 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1843 * config/s390/s390.c (s390_option_override): Avoid strncpy.
1845 2017-12-18 Richard Biener <rguenther@suse.de>
1847 PR tree-optimization/81877
1848 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
1849 (outermost_indep_loop): Adjust.
1850 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
1851 (can_sm_ref_p): Adjust.
1853 2017-12-18 Richard Biener <rguenther@suse.de>
1856 * tree.c (array_at_struct_end_p): Return true if the underlying
1857 object has space for at least one element in excess of what
1858 the array domain specifies.
1860 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1862 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
1863 example to avoid overfull hbox.
1864 * doc/invoke.texi (Option Summary): Add missing @gol.
1865 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
1868 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1869 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1871 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
1872 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
1873 (AArch64 Options): Document -mverbose-cost-dump.
1874 (ARM Options): Likewise, plus -mflip-thumb.
1876 2017-12-17 Martin Sebor <msebor@redhat.com>
1879 * gimple-ssa-warn-restrict.c
1880 (builtin_memref::offset_out_of_bounds): Correct the handling of
1883 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1885 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
1887 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
1889 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
1891 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
1893 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
1896 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
1898 * doc/invoke.texi: Fix some typos.
1900 2017-12-16 Martin Sebor <msebor@redhat.com>
1902 PR tree-optimization/78918
1903 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
1904 * builtins.c (check_sizes): Rename...
1905 (check_access): ...to this. Rename function arguments for clarity.
1906 (check_memop_sizes): Adjust names.
1907 (expand_builtin_memchr, expand_builtin_memcpy): Same.
1908 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
1909 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
1910 (check_strncat_sizes, expand_builtin_strncat): Same.
1911 (expand_builtin_strncpy, expand_builtin_memset): Same.
1912 (expand_builtin_bzero, expand_builtin_memcmp): Same.
1913 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
1914 (maybe_emit_sprintf_chk_warning): Same.
1915 (expand_builtin_strcpy): Adjust.
1916 (expand_builtin_stpcpy): Same.
1917 (expand_builtin_with_bounds): Detect out-of-bounds accesses
1918 in pointer-checking forms of memcpy, memmove, and mempcpy.
1919 (gcall_to_tree_minimal, max_object_size): Define new functions.
1920 * builtins.h (max_object_size): Declare.
1921 * calls.c (alloc_max_size): Call max_object_size instead of
1922 hardcoding ssizetype limit.
1923 (get_size_range): Handle new argument.
1924 * calls.h (get_size_range): Add a new argument.
1925 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
1926 * doc/invoke.texi (-Wrestrict): Adjust, add example.
1927 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
1929 (gimple_fold_builtin_memory_chk): Same.
1930 (gimple_fold_builtin_stxcpy_chk): New function.
1931 * gimple-ssa-warn-restrict.c: New source.
1932 * gimple-ssa-warn-restrict.h: New header.
1933 * gimple.c (gimple_build_call_from_tree): Propagate location.
1934 * passes.def (pass_warn_restrict): Add new pass.
1935 * tree-pass.h (make_pass_warn_restrict): Declare.
1936 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
1938 (handle_builtin_strcat): Same.
1939 (strlen_optimize_stmt): Rename...
1940 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
1941 stpncpy, strncpy, and their checking forms.
1943 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
1945 PR rtl-optimization/82849
1946 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
1947 and get_max_loop_iterations_int.
1949 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1950 Alan Hayward <alan.hayward@arm.com>
1951 David Sherwood <david.sherwood@arm.com>
1953 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
1954 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
1955 size as a poly_uint64.
1956 (mode_for_vector, mode_for_int_vector): Take the number of vector
1957 elements as a poly_uint64.
1958 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
1959 size as a poly_uint64.
1960 (mode_for_vector, mode_for_int_vector): Take the number of vector
1961 elements as a poly_uint64.
1963 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1964 Alan Hayward <alan.hayward@arm.com>
1965 David Sherwood <david.sherwood@arm.com>
1967 * machmode.h (MACRO_MODE): New macro.
1968 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
1969 * targhooks.c (default_libcall_value, default_secondary_reload)
1970 (default_memory_move_cost, default_register_move_cost)
1971 (default_class_max_nregs): Likewise.
1973 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1974 Alan Hayward <alan.hayward@arm.com>
1975 David Sherwood <david.sherwood@arm.com>
1977 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
1978 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
1980 * genautomata.c (main): Likewise.
1981 * genconditions.c (write_header): Likewise.
1982 * genemit.c (main): Likewise.
1983 * genextract.c (print_header): Likewise.
1984 * genopinit.c (main): Likewise.
1985 * genoutput.c (output_prologue): Likewise.
1986 * genpeep.c (main): Likewise.
1987 * genpreds.c (write_insn_preds_c): Likewise.
1988 * genrecog.c (writer_header): Likewise.
1989 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
1990 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
1991 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
1992 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
1993 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
1994 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
1995 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
1996 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
1997 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
1998 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
1999 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
2000 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
2001 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
2002 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
2003 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
2004 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
2005 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
2006 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
2007 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
2008 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
2009 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
2010 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
2011 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
2012 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
2013 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
2014 * config/darwin.c (IN_TARGET_CODE): Likewise.
2015 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
2016 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
2017 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
2018 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
2019 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
2020 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
2021 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
2022 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
2023 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
2024 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
2025 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
2026 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
2027 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
2028 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
2029 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
2030 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
2031 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
2032 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
2033 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
2034 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
2035 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
2036 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
2037 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
2038 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
2039 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
2040 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
2041 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
2042 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
2043 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
2044 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
2045 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
2046 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
2047 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
2048 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
2049 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
2050 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
2051 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
2052 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
2053 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
2054 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
2055 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
2056 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
2057 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
2058 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
2059 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
2060 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
2061 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
2062 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
2063 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
2064 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
2065 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
2066 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
2067 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
2068 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
2069 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
2070 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
2071 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
2072 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
2073 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
2074 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
2075 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2076 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
2077 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
2078 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
2079 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
2080 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
2081 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
2082 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
2083 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
2084 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
2085 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
2086 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2087 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
2088 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
2089 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
2090 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
2091 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
2092 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
2093 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
2094 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
2095 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
2096 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
2097 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
2098 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
2099 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
2100 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
2101 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
2102 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
2103 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
2104 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
2105 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
2106 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
2107 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
2108 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
2109 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
2110 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
2111 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
2112 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
2113 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
2114 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
2115 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
2116 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
2117 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
2118 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
2119 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
2120 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
2122 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2123 Alan Hayward <alan.hayward@arm.com>
2124 David Sherwood <david.sherwood@arm.com>
2126 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
2129 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2130 Alan Hayward <alan.hayward@arm.com>
2131 David Sherwood <david.sherwood@arm.com>
2133 * doc/generic.texi (VEC_SERIES_EXPR): Document.
2134 * doc/md.texi (vec_series@var{m}): Document.
2135 * tree.def (VEC_SERIES_EXPR): New tree code.
2136 * tree.h (build_vec_series): Declare.
2137 * tree.c (build_vec_series): New function.
2138 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
2139 * tree-pretty-print.c (dump_generic_node): Likewise.
2140 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
2141 * tree-inline.c (estimate_operator_cost): Likewise.
2142 * expr.c (expand_expr_real_2): Likewise.
2143 * optabs-tree.c (optab_for_tree_code): Likewise.
2144 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
2145 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
2146 * expmed.c (make_tree): Handle VEC_SERIES.
2147 * optabs.def (vec_series_optab): New optab.
2148 * optabs.h (expand_vec_series_expr): Declare.
2149 * optabs.c (expand_vec_series_expr): New function.
2150 * tree-vect-generic.c (expand_vector_operations_1): Check that
2151 the operands also have vector type.
2153 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2154 Alan Hayward <alan.hawyard@arm.com>
2155 David Sherwood <david.sherwood@arm.com>
2157 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
2158 (VEC_COND_EXPR): Add missing @tindex.
2159 * doc/md.texi (vec_duplicate@var{m}): Document.
2160 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
2161 * tree.c (build_vector_from_val): Add stubbed-out handling of
2162 variable-length vectors, using VEC_DUPLICATE_EXPR.
2163 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
2164 * cfgexpand.c (expand_debug_expr): Likewise.
2165 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2166 * tree-inline.c (estimate_operator_cost): Likewise.
2167 * tree-pretty-print.c (dump_generic_node): Likewise.
2168 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
2169 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
2170 (test_vec_duplicate_folding): New function.
2171 (fold_const_c_tests): Call it.
2172 * optabs.def (vec_duplicate_optab): New optab.
2173 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
2174 * optabs.h (expand_vector_broadcast): Declare.
2175 * optabs.c (expand_vector_broadcast): Make non-static. Try using
2176 vec_duplicate_optab.
2177 * expr.c (store_constructor): Try using vec_duplicate_optab for
2179 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
2181 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
2184 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
2185 div/mod latencies a bit.
2187 2017-12-15 Jeff Law <law@redhat.com>
2189 PR tree-optimization/36550
2190 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
2191 (mark_threaded_blocks): Rewrite code to avoid block copying when
2192 optimizing for size. Don't pessimize blocks which will be
2193 copied, but all the statements will be dead.
2195 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
2197 PR tree-optimization/81165
2198 * tree-ssa-threadupdate.c (uses_in_bb): New.
2199 (estimate_threading_killed_stmts): New.
2200 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
2201 * tree-ssa-threadedge.c
2202 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
2205 2017-12-15 Jeff Law <law@redhat.com>
2207 PR tree-optimization/83410
2208 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
2209 threads when parallelizing loops.
2211 2017-12-15 Jakub Jelinek <jakub@redhat.com>
2213 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
2215 * config/alpha/alpha.c (vms_attribute_table): Swap
2216 affects_type_identity and handler fields, adjust comments.
2217 * config/mips/mips.c (mips_attribute_table): Likewise.
2218 * config/visium/visium.c (visium_attribute_table): Likewise.
2219 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
2220 * config/microblaze/microblaze.c (microblaze_attribute_table):
2222 * config/spu/spu.c (spu_attribute_table): Likewise.
2223 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
2224 * config/arc/arc.c (arc_attribute_table): Likewise.
2225 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
2226 * config/v850/v850.c (v850_handle_interrupt_attribute,
2227 v850_handle_data_area_attribute): Formatting fixes.
2228 (v850_attribute_table): Swap affects_type_identity and handler
2229 fields, adjust comments.
2230 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
2231 * config/arm/arm.c (arm_attribute_table): Likewise.
2232 * config/avr/avr.c (avr_attribute_table): Likewise.
2233 * config/s390/s390.c (s390_attribute_table): Likewise.
2234 * config/sh/sh.c (sh_attribute_table): Likewise.
2235 * config/i386/i386.c (ix86_handle_cconv_attribute,
2236 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
2237 (ix86_attribute_table): Swap affects_type_identity and handler
2238 fields, adjust comments.
2239 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2240 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
2241 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
2242 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
2243 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
2244 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2245 * config/rx/rx.c (rx_attribute_table): Likewise.
2246 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
2247 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
2248 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
2249 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
2250 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2251 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
2252 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
2253 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
2254 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2255 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2256 * doc/plugins.texi (user_attr): Likewise. Add NULL for
2258 * attribs.c (empty_attribute_table): Swap affects_type_identity and
2260 (register_scoped_attributes, decl_attributes): Formatting fixes.
2262 PR tree-optimization/83269
2263 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
2264 subtraction in arg0's type if type is signed and arg0 is unsigned.
2268 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
2269 convert? on @0 instead of convert. Check type of @1, not @0.
2270 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
2271 convert? on @0 instead of convert. Check type of @1, not @0.
2272 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
2273 only optimize if either both @1 and @2 types are narrower
2274 precision, or both are wider or equal precision, and in the former
2275 case only if both have undefined overflow.
2277 2017-12-15 Richard Biener <rguenther@suse.de>
2280 * internal-fn.def (IFN_NOP): Add.
2281 * internal-fn.c (expand_NOP): Do nothing.
2282 * lto-streamer-in.c (input_function): Instead of removing
2283 sanitizer calls replace them with IFN_NOP calls.
2285 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
2286 Alan Hayward <alan.hayward@arm.com>
2287 David Sherwood <david.sherwood@arm.com>
2289 * dse.c (store_info, read_info_type): Replace begin and end with
2291 (print_range): New function.
2292 (set_all_positions_unneeded, any_positions_needed_p)
2293 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
2295 (record_store): Likewise. Optimize the case in which all positions
2297 (get_stored_val): Replace read_begin and read_end with read_offset
2299 (replace_read): Update call accordingly.
2301 2017-12-15 Bin Cheng <bin.cheng@arm.com>
2303 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
2304 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
2305 (loop_cand::loop_cand): Initialize above members.
2306 (loop_cand::supported_operations): Delete.
2307 (loop_cand::can_interchange_p): Inline above function.
2308 (loop_cand::classify_simple_reduction): Record number of constant
2309 initialized simple reductions.
2310 (should_interchange_loops): New parameters. Check stmt cost of loops
2312 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
2313 Update call to should_interchange_loops.
2314 (should_interchange_loop_nest): Update call to
2315 should_interchange_loops.
2317 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
2320 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
2321 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
2322 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
2324 2017-12-15 Julia Koval <julia.koval@intel.com>
2326 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
2327 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
2328 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
2329 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
2330 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
2332 2017-12-15 Julia Koval <julia.koval@intel.com>
2334 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
2335 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
2336 * config/i386/sse.md (vaesenc_<mode>): New pattern.
2337 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
2338 _mm_aesenc_epi128): New intrinsics.
2340 2017-12-15 Julia Koval <julia.koval@intel.com>
2342 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
2343 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
2344 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
2345 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
2346 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
2348 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2350 * gimple-ssa-strength-reduction.c (analyze_increments):
2351 Distinguish replacement costs for constant strides from those for
2354 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2356 * var-tracking.c (variable_tracking_main_1): Formatting fix.
2358 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2360 * doc/invoke.texi: Document -Wcast-function-type.
2361 * recog.h (stored_funcptr): Change signature.
2362 * tree-dump.c (dump_node): Avoid warning.
2363 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
2365 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
2368 * gimple-fold.c (get_range_strlen): Add the handling of non-member
2371 2017-12-14 David Malcolm <dmalcolm@redhat.com>
2373 PR tree-optimization/83312
2374 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
2375 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
2376 "val" param, and to cope with arbitrary basic blocks.
2377 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
2378 handle NULL_TREE for "val", dropping "bb" param.
2379 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
2380 drop "bb" param. Handle NULL_TREE for "val".
2381 (find_case_label_for_value): Make "switch_stmt" param const.
2382 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
2384 (vrp_prop::check_all_array_refs): Reimplement as...
2385 (check_array_bounds_dom_walker::before_dom_children): ...this new
2386 vfunc. Replace linear search through BB block list, excluding
2387 those with non-executable in-edges via dominator walk.
2389 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2391 * config/arm/arm.opt (mverbose-cost-dump): New option.
2392 * config/arm/arm.c (arm_rtx_costs): Use it.
2394 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
2397 * reload1.c (emit_input_reload_insns): Skip debug markers.
2399 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
2401 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
2405 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
2406 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
2407 * config/sh/sh.c: Skip debug insns besides notes.
2408 * config/sh/sh.md: Likewise.
2409 * config/sh/sh_treg_combine.cc: Likewise.
2410 * config/sh/sync.md: Likewise.
2412 2017-12-14 Tom de Vries <tom@codesourcery.com>
2414 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2415 Add item for weak_undefined.
2417 2017-12-14 Richard Biener <rguenther@suse.de>
2419 PR tree-optimization/67842
2420 * sese.h (bb_in_region): Remove #if 0'ed code.
2422 2017-12-14 Richard Biener <rguenther@suse.de>
2424 PR tree-optimization/83326
2425 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2426 may_be_zero parameter and handle it by not marking the first
2427 peeled copy as not exiting the loop.
2428 (try_peel_loop): Likewise.
2429 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
2430 to handle the case of constant or zero iterations and perform
2431 loop header copying on-the-fly.
2433 2017-12-14 Richard Biener <rguenther@suse.de>
2435 PR tree-optimization/83418
2436 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
2437 Instead of asserting we don't get unfolded comparisons deal with
2440 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2443 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
2444 basic blocks. Assert debug bind insns don't appear outside of bbs,
2445 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
2448 PR tree-optimization/83198
2449 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
2450 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
2451 value if arg is a REAL_CST with incompatible type.
2453 2017-12-14 Sudakshina Das <sudi.das@arm.com>
2454 Bin Cheng <bin.cheng@arm.com>
2457 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
2459 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
2462 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2464 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
2465 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
2467 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
2468 Alan Hayward <alan.hayward@arm.com>
2469 David Sherwood <david.sherwood@arm.com>
2471 * poly-int.h: New file.
2472 * poly-int-types.h: Likewise.
2473 * coretypes.h: Include them.
2474 (POLY_INT_CONVERSION): Define.
2475 * target.def (estimated_poly_value): New hook.
2476 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
2477 * doc/tm.texi: Regenerate.
2478 * doc/poly-int.texi: New file.
2479 * doc/gccint.texi: Include it.
2480 * doc/rtl.texi: Describe restrictions on subreg modes.
2481 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
2482 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
2483 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
2484 * targhooks.h (default_estimated_poly_value): Declare.
2485 * targhooks.c (default_estimated_poly_value): New function.
2486 * target.h (estimated_poly_value): Likewise.
2487 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
2488 (wi::unary_traits): Delete.
2489 (wi::binary_traits::signed_shift_result_type): Define for
2490 offset_int << HOST_WIDE_INT, etc.
2491 (generic_wide_int::operator <<=): Define for all types and use
2492 wi::lshift instead of <<.
2493 (wi::hwi_with_prec): Add a default constructor.
2494 (wi::ints_for): New class.
2495 (operator <<): Define for all wide-int types.
2496 (operator /): New function.
2497 (operator %): Likewise.
2498 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
2499 (ASSERT_MAYBE_NE_AT): New macros.
2501 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
2502 Dominik Vogt <vogt@linux.vnet.ibm.com>
2505 * emit-rtl.c (init_emit): Remove ??? comment.
2506 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2507 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
2508 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
2509 alignment of 3 virtual registers to BITS_PER_WORD.
2511 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
2513 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
2515 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
2516 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
2518 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2521 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
2523 2017-12-13 Jakub Jelinek <jakub@redhat.com>
2525 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
2526 with the exception of debug begin stmt markers appear before
2530 * final.c (rest_of_handle_final): Call variable_tracking_main only
2531 if !flag_var_tracking.
2533 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2534 Jakub Jelinek <jakub@redhat.com>
2538 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
2539 labels debug stmts that can only appear after labels.
2541 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
2543 PR rtl-optimization/82398
2544 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
2545 EXPR_USEFULNESS in priority comparison.
2547 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
2549 PR rtl-optimization/83393
2550 * combine.c (move_deaths): If reg_stat points to a too new insn in
2551 last_death, do not use it: find the proper insn instead.
2553 2017-12-12 Jeff Law <law@redhat.com>
2555 PR tree-optimization/83298
2556 PR tree-optimization/83362
2557 PR tree-optimization/83383
2558 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
2559 push_value_range a public interface. Add new argument to
2560 record_ranges_from_stmt.
2561 * gimple-ssa-evrp-analyze.c
2562 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
2563 Update comments. Handle recording temporary equivalences.
2564 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
2565 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
2566 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
2567 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
2568 gimple-ssa-evrp-analyze.h.
2569 (record_temporary_equivalences_from_phis): Add new argument. When
2570 the PHI arg is an SSA_NAME, set the result's range to the range
2572 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
2573 from statements too.
2574 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
2575 Pass it down to children as needed.
2576 (thread_outgoing_edges): Likewise.
2577 (thread_across_edge): Likewise. Push/pop range state as needed.
2578 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
2580 2017-12-12 Julia Koval <julia.koval@intel.com>
2582 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
2583 (PTA_CANNONLAKE): Remove PTA_CLWB.
2585 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2588 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
2589 (vcondv2div2df): Likewise.
2590 (vconduv2dfv2di): Likewise.
2592 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2594 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
2595 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
2596 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
2597 ATTR_MATHFN_FPROUNDING_ERRNO.
2599 2017-12-12 Richard Biener <rguenther@suse.de>
2601 PR tree-optimization/83385
2602 * graphite-scop-detection.c (get_order, order): Remove.
2603 (bb_to_rpo): New global.
2605 (build_scops): Sort pbbs in RPO order.
2607 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
2609 * combine.c (simplify_set): Do not transform subregs to zero_extends
2610 if the destination mode is a vector mode.
2612 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2614 PR tree-optimization/83359
2615 * tree-cfg.h (fold_loop_internal_call): Declare.
2616 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
2617 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
2618 (find_loop_dist_alias): New function.
2619 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
2620 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
2623 PR tree-optimization/80631
2624 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
2625 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
2626 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
2627 hardcoding zero as the value if COND_EXPR is never true. For
2628 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
2629 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
2630 hardcoding MAX_EXPR as the reduction operation.
2631 (is_nonwrapping_integer_induction): Allow negative step.
2632 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
2633 vect_create_epilog_for_reduction, if no value is suitable, don't
2634 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
2636 2017-12-12 Richard Biener <rguenther@suse.de>
2638 PR tree-optimization/81889
2639 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
2640 range info from the non-wrapping IV instead of just the range
2643 2017-12-12 Julia Koval <julia.koval@intel.com>
2645 * config.gcc: Add vaesintrin.h.
2646 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2648 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2649 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2651 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2652 * config/i386/immintrin.h: Include vaesintrin.h.
2653 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2654 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2655 _mm_aesdec_epi128): New intrinsics.
2657 2017-12-12 Julia Koval <julia.koval@intel.com>
2659 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
2660 OPTION_MASK_ISA_VAES_UNSET): New.
2661 (ix86_handle_option): Handle -mvaes.
2662 * config/i386/cpuid.h: Define bit_VAES.
2663 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
2664 * config/i386/i386-c.c (__VAES__): New.
2665 * config/i386/i386.c (ix86_target_string): Add -mvaes.
2666 (ix86_valid_target_attribute_inner_p): Ditto.
2667 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
2668 * config/i386/i386.opt: Add -mvaes.
2669 * doc/invoke.texi: Ditto.
2671 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
2673 * debug.h (gcc_debug_hooks): Add inline_entry.
2674 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
2675 * debug.c (do_nothing_debug_hooks): Likewise.
2676 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2677 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
2678 (dwarf2_lineno_debug_hooks): Likewise.
2680 * common.opt (gstatement-frontiers): New, setting
2681 debug_nonbind_markers_p.
2682 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
2683 * toplev.c (process_options): Autodetect value for debug statement
2685 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
2686 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
2688 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
2689 markers. Integrate source bind into debug stmt expand loop.
2690 (pass_expand::execute): Check debug marker limit. Avoid deep
2691 TER and expand debug locations for debug bind insns only.
2692 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
2694 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
2695 * final.c (reemit_insn_block_notes): Take current block from
2696 nonbind markers. Declare note where it's first set.
2697 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
2698 begin stmt markers if enabled.
2699 (notice_source_line): Handle nonbind markers. Fail if their
2700 location is unknown or that of builtins.
2701 (rest_of_handle_final): Convert begin stmt markers to notes if
2702 var-tracking didn't run.
2703 (rest_of_clean_state): Skip begin stmt markers.
2704 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
2706 * function.c (allocate_struct_function): Set begin_stmt_markers.
2707 * function.h (struct function): Add debug_marker_count counter
2708 and debug_nonbind_markers flag.
2709 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
2710 * gimple-low.c (lower_function_body): Adjust
2711 debug_nonbind_markers.
2712 (lower_stmt): Drop or skip gimple debug stmts.
2713 (lower_try_catch): Skip debug stmts.
2714 * gimple.c (gimple_build_debug_begin_stmt): New.
2715 (gimple_copy): Increment debug_marker_count if copying one.
2716 * gimple.h (gimple_build_debug_begin_stmt): Declare.
2717 * gimplify.c (rexpr_location): New.
2718 (rexpr_has_location): New.
2719 (warn_switch_unreachable_r): Handle gimple debug stmts.
2720 (shortcut_cond_r): Call expr_location.
2722 (find_goto_label): New.
2723 (shortcut_cond_expr): Call expr_has_location, expr_location, and
2725 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
2727 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
2728 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
2729 (LANG_HOOKS_INITIALIZER): ... this.
2730 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
2731 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
2733 (update_ebb_live_info): Skip debug insn markers.
2734 * lra.c (debug_insn_static_data): Rename to...
2735 (debug_bind_static_data): ... this.
2736 (debug_marker_static_data): New.
2737 (lra_set_insn_recog_data): Select one of the above depending
2739 (lra_update_isn_regno_info): Don't assume debug insns have
2741 (push_insns): Skip debug insns.
2742 * lto-streamer-in.c (input_function): Drop debug stmts
2743 depending on active options. Adjust debug_nonbind_markers.
2744 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
2745 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
2746 begin stmt marker notes.
2747 (print_insn): Likewise.
2748 * recog.c (extract_insn): Recognize rtl for debug markers.
2749 * rtl.def (DEBUG_MARKER): New.
2750 * tree-inline.c: Include params.h.
2751 (remap_gimple_stmt): Handle nonbind markers.
2752 (maybe_move_debug_stmts_to_successors): Likewise.
2753 (copy_debug_stmt): Likewise.
2754 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
2755 markers regardless of no side effects.
2756 (tsi_link_before): Don't update container's side effects when adding
2757 a begin stmt marker.
2758 (tsi_link_after): Likewise.
2759 (expr_first): Skip begin stmt markers.
2760 (expr_last): Likewise.
2761 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
2762 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2763 Disregard nonbind markers.
2764 * tree.c (make_node_stat): Don't set side effects for begin stmt
2766 (build1_stat): Likewise.
2767 * tree.def (DEBUG_BEGIN_STMT): New.
2768 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
2769 * var-tracking.c (delete_debug_insns): Renamed to...
2770 (delete_vta_debug_insns): ... this.
2771 (reemit_marker_as_note): New.
2772 (vt_initialize): Reemit markers.
2773 (delete_vta_debug_insns): Likewise.
2774 (vt_debug_insns_local): Reemit or delete markers.
2775 (variable_tracking_main_1): Likewise.
2776 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
2777 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
2778 (gimple_debug_nonbind_marker_p): New.
2779 (gimple_build_debug_bind): Adjust.
2780 (gimple_build_debug_begin_stmt): New.
2781 * doc/invoke.texi (max-debug-marker-count): New param.
2782 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
2783 (debug_parameter_ref, debug_marker): New.
2784 (NOTE_INSN_BEGIN_STMT): New.
2785 (DEBUG_INSN): Describe begin stmt markers.
2787 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
2788 without debug insns, we wouldn't, but clean up debug insns
2789 after a control flow insn nevertheless.
2791 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
2792 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
2793 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
2794 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
2795 (rtl_tidy_fallthru_edge): Likewise.
2796 (rtl_verify_fallthru): Likewise.
2797 (rtl_verify_bb_layout): Likewise.
2798 (skip_insns_after_block): Likewise.
2799 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
2800 * dwarf2out.c: Include print-rtl.h.
2801 (dwarf2out_next_real_insn): New.
2802 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
2803 Dump debug binds in asm comments.
2804 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
2805 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
2806 callers to use gsi_start_nondebug_bb instead.
2807 (gsi_after_labels): Skip gimple debug stmts.
2808 (gsi_start_nondebug): New.
2809 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
2810 (proper_loop_form_for_interchange): Adjust.
2811 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
2812 * gimple.h (gimple_seq_last_nondebug_stmt): New.
2813 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
2814 (collect_fallthrough_labels): Likewise.
2815 (should_warn_for_implicit_fallthrough): Likewise.
2816 (warn_implicit_fallthrough_r): Likewise.
2817 (expand_FALLTHROUGH_r): Likewise.
2818 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
2819 (graphite_copy_stmts_from_block): Skip nonbind markers.
2820 * haifa-sched.c (sched_extend_bb): Skip debug insns.
2821 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
2822 * jump.c (clean_barriers): Skip debug insns.
2823 * omp-expand.c (expand_parallel_call): Skip debug insns.
2824 (expand_task_call): Likewise.
2825 (remove_exit_barrier): Likewise.
2826 (expand_omp_taskreg): Likewise.
2827 (expand_omp_for_init_counts): Likewise.
2828 (expand_omp_for_generic): Likewise.
2829 (expand_omp_for_static_nochunk): Likewise.
2830 (expand_omp_for_static_chunk): Likewise.
2831 (expand_omp_simd): Likewise.
2832 (expand_omp_taskloop_for_outer): Likewise.
2833 (expand_omp_taskloop_for_inner): Likewise.
2834 (expand_oacc_for): Likewise.
2835 (expand_omp_sections): Likewise.
2836 (expand_omp_single): Likewise.
2837 (expand_omp_synch): Likewise.
2838 (expand_omp_atomic_load): Likewise.
2839 (expand_omp_atomic_store): Likewise.
2840 (expand_omp_atomic_fetch_op): Likewise.
2841 (expand_omp_atomic_pipeline): Likewise.
2842 (expand_omp_atomic_mutex): Likewise.
2843 (expand_omp_target): Likewise.
2844 (grid_expand_omp_for_loop): Likewise.
2845 (grid_expand_target_grid_body): Likewise.
2846 (build_omp_regions_1): Likewise.
2847 * omp-low.c (check_combined_parallel): Skip debug stmts.
2848 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
2849 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
2850 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
2852 * tree-cfg.c (make_blobs_1): Skip debug stmts.
2853 (make_edges): Likewise.
2854 (cleanup_dead_labels): Likewise.
2855 (gimple_can_merge_blocks_p): Likewise.
2856 (stmt_starts_bb_p): Likewise.
2857 (gimple_block_label): Likewise.
2858 (gimple_redirect_edge_and_branch): Likewise.
2859 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
2861 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
2863 * tree-pretty-print (print_declaration): Omit initializer in slim
2865 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
2867 (eliminate_unnecessary_stmts): Stabilize block removal order.
2868 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
2869 * var-tracking.c (get_first_insn): New.
2870 (vt_emit_notes): Call it.
2871 (vt_initialize): Walk any insns before the first BB.
2872 (delete_debug_insns): Likewise.
2874 * gimple.h (enum gimple_debug_subcode): Add
2875 GIMPLE_DEBUG_BEGIN_STMT.
2876 (gimple_debug_begin_stmt_p): New.
2877 (gimple_debug_nonbind_marker_p): New.
2878 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
2879 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
2880 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
2881 * insn-notes.def (BEGIN_STMT): New.
2882 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
2883 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
2884 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
2885 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
2886 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
2887 (INSN_DEBUG_MARKER_KIND): New.
2888 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
2889 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
2890 (INSN_VAR_LOCATION_PTR): New.
2891 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
2893 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
2894 deps only if debug bind insns are enabled.
2895 (pass_expand::execute): Avoid deep TER and expand
2896 debug locations for debug bind insns only.
2897 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
2898 debug stmts special handling down to debug bind stmts.
2899 * combine.c (try_combine): Narrow debug insns special handling
2900 down to debug bind insns.
2901 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
2902 Narrow debug insns preexisting special handling down to debug
2904 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
2905 handling down to debug bind insns.
2906 * function.c (instantiate_virtual_regs): Skip debug markers,
2907 adjust handling of debug binds.
2908 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
2909 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
2910 * haifa-sched.c (schedule_insn): Narrow special handling of debug
2911 insns to debug bind insns.
2912 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
2913 special handling of debug stmts to debug bind stmts.
2914 * ipa-split.c (split_function): Likewise.
2915 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
2916 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
2918 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
2919 * regrename.c (build_def_use): Likewise.
2920 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
2921 (pass_cprop_hardreg): Narrow special casing of debug insns to
2923 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
2924 * reload1.c (reload): Likewise.
2925 * sese.c (sese_insert_phis_for_liveouts): Narrow special
2926 casing of debug stmts to debug bind stmts.
2927 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
2928 * ssa-iterators.h (num_imm_uses): Likewise.
2929 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
2930 debug stmts to debug bind stmts.
2931 * tree-inline.c (tree_function_versioning): Narrow special casing
2932 of debug stmts to debug bind stmts.
2933 * tree-loop-distribution.c (generate_loops_for_partition):
2934 Narrow special casing of debug stmts to debug bind stmts.
2935 * tree-sra.c (analyze_access_subtree): Narrow special casing
2936 of debug stmts to debug bind stmts.
2937 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
2938 stmts to debug bind stmts.
2939 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
2940 casing of debug stmts to debug bind stmts.
2941 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
2942 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
2943 casing of debug stmts to debug bind stmts.
2944 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2946 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
2947 debug stmts to debug bind stmts.
2948 (gimple_replace_ssa_lhs): Likewise.
2949 (insert_debug_temp_for_var_def): Likewise.
2950 (insert_debug_temps_for_defs): Likewise.
2951 (reset_debug_uses): Likewise.
2952 * tree-ssanames.c (release_ssa_name_fn): Likewise.
2953 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
2954 (adjust_debug_stmts): Likewise.
2955 (adjust_phi_and_debug_stmts): Likewise.
2956 (vect_do_peeling): Likewise.
2957 * tree-vect-loop.c (vect_transform_loop): Likewise.
2958 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
2959 * var-tracking.c (adjust_mems): Narrow special casing of debug
2960 insns to debug bind insns.
2961 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
2962 (compute_bb_dataflow, vt_find_locations): Likewise.
2963 (vt_expand_loc, emit_notes_for_changes): Likewise.
2964 (vt_init_cfa_base): Likewise.
2965 (vt_emit_notes): Likewise.
2966 (vt_initialize): Likewise.
2967 (vt_finalize): Likewise.
2969 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
2970 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
2971 (next_nonnote_nondebug_insn_bb): New.
2972 (prev_nonnote_nondebug_insn_bb): New.
2973 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
2974 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
2975 (prev_nonnote_nondebug_insn_bb): Declare.
2976 (next_nonnote_nondebug_insn_bb): Declare.
2977 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
2978 * cfgrtl.c (get_last_bb_insn): Likewise.
2979 * lra.c (push_insns): Likewise.
2981 2017-12-11 David Malcolm <dmalcolm@redhat.com>
2984 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
2986 * selftest.c (selftest::test_runner::test_runner): New ctor.
2987 (selftest::test_runner::~test_runner): New dtor.
2988 * selftest.h (class selftest::test_runner): New class.
2990 2017-12-11 Carl Love <cel@us.ibm.com>
2992 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
2993 vec_extract_fp32_from_shortl]): Add #defines.
2994 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
2995 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
2996 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
2997 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
2998 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
2999 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
3000 * doc/extend.texi: Add documentation for the added builtins.
3002 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
3004 PR rtl-optimization/80693
3005 PR rtl-optimization/81019
3006 PR rtl-optimization/81020
3007 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
3008 are not mentioned in i3. Place the REG_UNUSED note on i2,
3009 possibly modified to REG_DEAD, if it did not originate in i3.
3011 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3013 * recog.c (store_data_bypass_p_1): New function.
3014 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
3015 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
3017 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
3019 PR rtl-optimization/83361
3020 * ifcvt.c (if_convert): Call fixup_partitions.
3022 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
3024 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3025 early folding of splat_u{8,16,32}.
3027 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3029 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
3030 output_operand_lossage first argument with capital letter.
3031 (aarch64_override_options): Don't start error and sorry first argument
3032 with capital letter.
3034 2017-12-11 Andi Kleen <ak@linux.intel.com>
3036 PR gcov-profile/83355
3037 * auto-profile.c (string_table::get_index_by_decl): Don't
3038 recurse when abstract origin points to itself.
3040 2017-12-11 Bin Cheng <bin.cheng@arm.com>
3042 PR tree-optimization/83320
3043 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
3044 (prune_datarefs_not_in_loop): Ditto.
3046 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
3048 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
3050 2017-12-10 Jakub Jelinek <jakub@redhat.com>
3052 PR tree-optimization/83337
3053 * gimple-loop-interchange.cc (compute_access_stride): Handle
3054 bitfield DRs properly.
3056 2017-12-09 Jakub Jelinek <jakub@redhat.com>
3058 PR tree-optimization/83338
3059 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
3060 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
3061 vector of unsigned integers to vector of signed integers.
3063 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
3065 PR rtl-optimization/83317
3066 * lra-constraints.c (process_address_1): Add insn code check.
3068 2017-12-08 Michael Matz <matz@suse.de>
3070 Fix PR tree-optimization/83323
3071 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
3072 head-controlled loops and loop BBs.
3073 * common.opt (funroll-and-jam): Remove, instead ...
3074 (floop-unroll-and-jam): ... reuse this option.
3075 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
3076 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
3077 (-floop-unroll-and-jam): ... this option.
3079 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3081 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
3082 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
3083 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
3085 2017-12-08 Richard Biener <rguenther@suse.de>
3088 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
3089 handle accesses outside of zero-sized vars.
3091 2017-12-08 Martin Jambor <mjambor@suse.cz>
3093 PR tree-optimization/83141
3094 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
3095 test for MEM_REFs implicitely changing types with padding. Remove
3097 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
3099 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3101 * config/arc/arc.c (arc_attribute_table): Add exclusions to
3103 * config/avr/avr.c (avr_attribute_table): Likewise.
3104 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3105 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3106 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3107 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
3108 of struct attribute_spec.
3109 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3111 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
3114 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
3116 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
3119 * profile-count.c (profile_count::from_gcov_type): Move from
3120 profile-count.h; handle overflow.
3121 * profile-count.h (profile_count::from_gcov_type): Move offline.
3123 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
3125 PR rtl-optimization/83304
3126 * combine.c (move_deaths): If we do not know where a register died,
3129 2017-12-08 Richard Biener <rguenther@suse.de>
3131 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
3132 Provide -fopt-info-loop feedback when we interchange in a nest.
3134 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3136 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
3137 for armv6 ARM CPU IDs.
3139 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3141 * common/config/arm/arm-common.c: Include <algorithm>.
3142 (INCLUDE_VECTOR): Define.
3143 (compare_opt_names): New function.
3144 (arm_rewrite_selected_arch): Only strip out extensions that can be
3145 expressed through -mfpu. Sort the remaining extensions
3148 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3150 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
3151 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
3152 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
3153 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
3155 2017-06-08 Tristan Gingold <gindold@adacore.com>
3158 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
3159 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
3161 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3163 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
3164 of struct attribute_spec.
3166 2017-12-08 Julia Koval <julia.koval@intel.com>
3168 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
3169 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
3170 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
3171 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
3172 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
3173 _mm_maskz_dpwssds_epi32): Ditto.
3175 2017-12-08 Richard Biener <rguenther@suse.de>
3177 PR tree-optimization/81303
3178 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
3179 conditions try to create a comparison vector type matching
3180 the data vector type.
3181 (vectorizable_condition): Adjust.
3182 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
3183 Leave invariant conditions alone in case we can vectorize those.
3185 2017-12-08 Julia Koval <julia.koval@intel.com>
3187 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
3188 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
3189 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
3190 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
3191 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
3193 2017-12-08 Julia Koval <julia.koval@intel.com>
3195 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
3196 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
3197 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
3198 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
3199 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
3200 _mm_maskz_dpbusds_epi32): New intrinsics.
3202 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3204 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
3206 (ld<bh>io_signed): Likewise.
3207 (st<bhw>io): Likewise.
3208 * config/nios2/predicates.md (ldstio_memory_operand): Allow
3209 SMALL_INT12 constant integer operand.
3211 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3213 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
3215 (nios2_large_constant_allowed): ...this. Adjust uses.
3216 (nios2_plus_symbolic_constant_p): Rename to...
3217 (nios2_plus_large_constant_p): ...this. Adjust uses.
3218 (nios2_legitimate_address_p): Correct CONST_INT handling.
3219 (nios2_symbolic_memory_operand_p): Rename to...
3220 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
3221 (nios2_large_constant_p): Check for large constant integers too.
3222 (nios2_split_large_constant): Handle constant integers.
3223 (nios2_split_symbolic_memory_operand): Rename to...
3224 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
3225 (nios2_legitimize_constant_address): Handle constant integers.
3226 (r0rel_constant_p): Handle small constant integers.
3227 (nios2_print_operand_address): Handle r0-relative integer addresses.
3228 * config/nios2/nios2-protos.h: Adjust for renamed functions.
3229 * config/nios2/nios2.md: Adjust for renamed functions.
3231 2017-12-07 Andrew Waterman <andrew@sifive.com>
3233 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
3234 (TARGET_HAVE_SRODATA_SECTION): New define.
3235 (riscv_select_section): New function.
3237 2017-12-08 Joseph Myers <joseph@codesourcery.com>
3238 Alexander Monakov <amonakov@ispras.ru>
3239 Jakub Jelinek <jakub@redhat.com>
3242 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
3244 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3246 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
3247 a bool success value. Don't call output_operand_lossage here.
3248 (aarch64_print_ldpstp_address): Return a bool success value.
3249 (aarch64_print_operand_address): Call output_addr_const if
3250 aarch64_print_address_internal fails.
3251 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
3252 'y'; call output_operand_lossage instead. Call output_operand_lossage
3253 if aarch64_print_ldpstp_address fails.
3255 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3257 * tree-vector-builder.h
3258 (tree_vector_builder::binary_encoded_nelts): Declare.
3259 * tree-vector-builder.c
3260 (tree_vector_builder::binary_encoded_nelts): New function.
3261 * fold-const.c (negate_expr_p): Likewise.
3262 (operand_equal_p, fold_checksum_tree): Likewise.
3263 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
3264 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
3265 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
3266 (uniform_vector_p): Likewise.
3267 * varasm.c (const_hash_1, compare_constant): Likewise.
3268 * tree-ssa-ccp.c: Include tree-vector-builder.h.
3269 (valid_lattice_transition): Operate directly on the VECTOR_CST
3271 * ipa-icf.c: Include tree-vector-builder.h.
3272 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
3273 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
3275 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3277 * tree.c (build_vector): Delete.
3278 * tree.h (build_vector): Make static and move into the self-testing
3281 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3283 * vector-builder.h (vector_builder::derived): New const overload.
3284 (vector_builder::elt): New function.
3285 * tree-vector-builder.h (tree_vector_builder::type): New function.
3286 (tree_vector_builder::apply_step): Declare.
3287 * tree-vector-builder.c (tree_vector_builder::apply_step): New
3289 * gimple-fold.h (tree_vector_builder): Declare.
3290 (gimple_build_vector): Take a tree_vector_builder instead of a
3291 type and vector of elements.
3292 * gimple-fold.c (gimple_build_vector): Likewise.
3293 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
3295 (get_initial_defs_for_reduction): Likewise.
3296 (vectorizable_induction): Likewise.
3298 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3300 * tree-vector-builder.h
3301 (tree_vector_builder::new_binary_operation): Declare.
3302 * tree-vector-builder.c
3303 (tree_vector_builder::new_binary_operation): New function.
3304 * fold-const.c (fold_relational_const): Use it.
3305 (const_binop): Likewise. Check that both input vectors have
3306 the same number of elements, thus excluding things like WIDEN_SUM.
3307 Check whether it is possible to operate directly on the encodings
3310 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3312 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
3313 new_unary_operation, operating only on the encoded elements.
3314 (const_unop): Likewise.
3315 (exact_inverse): Likewise.
3316 (distributes_over_addition_p): New function.
3317 (const_binop): Use tree_vector_builder and new_unary_operation
3318 for combinations of VECTOR_CST and INTEGER_CST. Operate only
3319 on the encoded elements unless the encoding is strided and the
3320 operation does not distribute over addition.
3321 (fold_convert_const): Use tree_vector_builder and
3322 new_unary_operation. Operate only on the encoded elements
3323 for truncating integer conversions, or for non-stepped encodings.
3325 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3327 * config/sparc/sparc.c: Include tree-vector-builder.h.
3328 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
3329 * expmed.c: Include tree-vector-builder.h.
3330 (make_tree): Use tree_vector_builder instead of build_vector.
3331 * fold-const.c: Include tree-vector-builder.h.
3332 (const_binop): Use tree_vector_builder instead of build_vector.
3333 (const_unop): Likewise.
3334 (native_interpret_vector): Likewise.
3335 (fold_vec_perm): Likewise.
3336 (fold_ternary_loc): Likewise.
3337 * gimple-fold.c: Include tree-vector-builder.h.
3338 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
3340 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
3341 (simplify_vector_constructor): Use tree_vector_builder instead
3343 * tree-vect-generic.c: Include tree-vector-builder.h.
3344 (add_rshift): Use tree_vector_builder instead of build_vector.
3345 (expand_vector_divmod): Likewise.
3346 (optimize_vector_constructor): Likewise.
3347 * tree-vect-loop.c: Include tree-vector-builder.h.
3348 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
3349 of build_vector. Explicitly use a stepped encoding for
3351 * tree-vect-slp.c: Include tree-vector-builder.h.
3352 (vect_get_constant_vectors): Use tree_vector_builder instead
3354 (vect_transform_slp_perm_load): Likewise.
3355 (vect_schedule_slp_instance): Likewise.
3356 * tree-vect-stmts.c: Include tree-vector-builder.h.
3357 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
3358 (vect_gen_perm_mask_any): Likewise.
3359 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
3360 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
3362 (build_vector_from_val): Likewise. Explicitly use a duplicate
3365 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
3367 * doc/generic.texi (VECTOR_CST): Describe new representation of
3369 * vector-builder.h: New file.
3370 * tree-vector-builder.h: Likewise.
3371 * tree-vector-builder.c: Likewise.
3372 * Makefile.in (OBJS): Add tree-vector-builder.o.
3373 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
3374 * tree-core.h (tree_base): Add a vector_cst field to the u union.
3375 (tree_vector): Change the number of elements to
3376 vector_cst_encoded_nelts.
3377 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
3378 (VECTOR_CST_ELTS): Delete.
3379 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
3380 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
3381 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
3382 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
3383 (VECTOR_CST_ENCODED_ELT): Likewise.
3384 (vector_cst_encoded_nelts): New function.
3385 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3386 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3387 (vector_cst_int_elt, vector_cst_elt): Declare.
3388 * tree.c: Include tree-vector-builder.h.
3389 (tree_code_size): Abort if passed VECTOR_CST.
3390 (tree_size): Update for new VECTOR_CST layout.
3391 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3392 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3393 (build_vector): Use tree_vector_builder.
3394 (vector_cst_int_elt, vector_cst_elt): New functions.
3395 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
3396 encoded elements and then create the vector in the canonical form.
3397 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
3398 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
3399 (tree_c_tests): Call test_vector_cst_patterns.
3400 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
3402 (hash_tree): Likewise.
3403 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
3404 (streamer_write_tree_header): Likewise.
3405 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
3406 (streamer_alloc_tree): Likewise. Update call to make_vector.
3407 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
3409 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3411 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
3412 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
3413 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
3415 2017-12-07 Bin Cheng <bin.cheng@arm.com>
3416 Richard Biener <rguenther@suse.de>
3418 PR tree-optimization/81303
3419 * Makefile.in (gimple-loop-interchange.o): New object file.
3420 * common.opt (floop-interchange): Reuse the option from graphite.
3421 * doc/invoke.texi (-floop-interchange): Ditto. New document for
3422 -floop-interchange and mention it for -O3.
3423 * opts.c (default_options_table): Enable -floop-interchange at -O3.
3424 * gimple-loop-interchange.cc: New file.
3425 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
3426 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
3427 * passes.def (pass_linterchange): New pass.
3428 * timevar.def (TV_LINTERCHANGE): New time var.
3429 * tree-pass.h (make_pass_linterchange): New declaration.
3430 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
3431 interchange. Record IV before/after increment in new parameters.
3432 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
3433 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
3435 (check_reduction_path): ...New function here.
3436 * tree-vectorizer.h (check_reduction_path): New declaration.
3438 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
3441 PR rtl-optimization/80818
3442 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
3443 always early clobbered.
3444 * lra-lives.c (process_bb_lives): Check input hard regs for early
3445 clobbered non-operand hard reg.
3447 2017-12-07 Jakub Jelinek <jakub@redhat.com>
3450 * tree-cfg.c (verify_gimple_assign_binary): Don't require
3451 types_compatible_p, just that TYPE_MODE is the same.
3453 2017-12-07 Martin Sebor <msebor@redhat.com>
3456 * attribs.c (empty_attribute_table): Initialize new member of
3457 struct attribute_spec.
3458 (decl_attributes): Add argument. Handle mutually exclusive
3459 combinations of attributes.
3460 (selftests::test_attribute_exclusions): New function.
3461 (selftests::attribute_c_tests): Ditto.
3462 * attribs.h (decl_attributes): Add default argument.
3463 * selftest.h (attribute_c_tests): Declare.
3464 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
3465 * tree-core.h (attribute_spec::exclusions, exclude): New type and
3467 * doc/extend.texi (Common Function Attributes): Update const and pure.
3468 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
3469 of struct attribute_spec.
3470 * config/arc/arc.c (arc_attribute_table): Same.
3471 * config/arm/arm.c (arm_attribute_table): Same.
3472 * config/avr/avr.c ( avr_attribute_table): Same.
3473 * config/bfin/bfin.c (bfin_attribute_table): Same.
3474 * config/cr16/cr16.c (cr16_attribute_table): Same.
3475 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
3476 * config/h8300/h8300.c (h8300_attribute_table): Same.
3477 * config/i386/i386.c (ix86_attribute_table): Same.
3478 * config/ia64/ia64.c (ia64_attribute_table): Same.
3479 * config/m32c/m32c.c (m32c_attribute_table): Same.
3480 * config/m32r/m32r.c (m32r_attribute_table): Same.
3481 * config/m68k/m68k.c (m68k_attribute_table): Same.
3482 * config/mcore/mcore.c (mcore_attribute_table): Same.
3483 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
3484 * config/mips/mips.c (mips_attribute_table): Same.
3485 * config/msp430/msp430.c (msp430_attribute_table): Same.
3486 * config/nds32/nds32.c (nds32_attribute_table): Same.
3487 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
3488 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
3489 * config/rl78/rl78.c (rl78__attribute_table): Same.
3490 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
3491 * onfig/rx/rx.c (rx_attribute_table): Same.
3492 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
3493 * config/sh/sh.c (sh_attribute_table): Same.
3494 * config/sparc/sparc.c (sparc_attribute_table): Same.
3495 * config/spu/spu.c (spu_attribute_table): Same.
3496 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
3497 * config/v850/v850.c (v850_attribute_table): Same.
3498 * config/visium/visium.c (visium_attribute_table): Same.
3500 2017-12-07 Tamar Christina <tamar.christina@arm.com>
3503 * config/arm/arm.c (INCLUDE_STRING): Define.
3504 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
3505 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
3508 2017-12-07 Michael Matz <matz@suse.de>
3510 Add unroll and jam pass
3512 * gimple-loop-jam.c: New file.
3513 * Makefile.in (OBJS): Add gimple-loop-jam.o.
3514 * common.opt (funroll-and-jam): New option.
3515 * opts.c (default_options_table): Add unroll-and-jam at -O3.
3516 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
3517 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
3518 * passes.def: Add pass_loop_jam.
3519 * timevar.def (TV_LOOP_JAM): Add.
3520 * tree-pass.h (make_pass_loop_jam): Declare.
3521 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
3522 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
3523 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
3524 to flow_loop_tree_node_add.
3525 (duplicate_subloops, copy_loops_to): Append to sibling list.
3526 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
3527 * doc/invoke.texi (-funroll-and-jam): Document new option.
3528 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
3530 2017-12-07 Richard Biener <rguenther@suse.de>
3532 PR tree-optimization/83296
3533 PR tree-optimization/67769
3534 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
3535 flow sensitive info in an unrelated BB.
3536 (value_replacement): Use reset_flow_sensitive_info.
3537 (minmax_replacement): Reset flow sensitive info on the def
3538 we move. Do not reset flow sensitive info in the whole BB
3539 we move the stmt to.
3540 (abs_replacement): Likewise.
3542 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3545 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3546 rs6000_cpu to the given -mcpu=, or to the default processor.
3548 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3550 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
3551 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
3552 instead of rs6000_cpu_attr.
3553 (rs6000_adjust_cost): Ditto.
3554 (is_microcoded_insn): Ditto.
3555 (rs6000_adjust_priority): Ditto.
3556 (rs6000_issue_rate): Ditto.
3557 (rs6000_use_sched_lookahead): Ditto.
3558 (rs6000_use_sched_lookahead_guard): Ditto.
3559 (rs6000_sched_reorder): Ditto.
3560 (force_new_group): Ditto.
3561 * config/rs6000/rs6000.md (cpu attribute): Ditto.
3562 (group_ending_nop): Ditto.
3564 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3566 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
3567 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
3568 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
3569 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
3571 (rs6000_reassociation_width): Ditto.
3572 (rs6000_emit_epilogue): Ditto.
3573 (rs6000_adjust_cost): Ditto.
3574 (is_microcoded_insn): Ditto.
3575 (is_cracked_insn): Ditto.
3576 (rs6000_adjust_priority): Ditto.
3577 (rs6000_sched_reorder): Ditto.
3578 (rs6000_sched_reorder2): Ditto.
3579 (insn_must_be_first_in_group): Ditto.
3580 (insn_must_be_last_in_group): Ditto.
3581 (rs6000_register_move_cost): Ditto.
3582 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
3585 2017-12-07 Julia Koval <julia.koval@intel.com>
3587 * config.gcc: Add vaesintrin.h.
3588 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3590 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3591 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3593 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3594 * config/i386/immintrin.h: Include vaesintrin.h.
3595 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3596 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3597 _mm_aesdec_epi128): New intrinsics.
3599 2017-12-06 David Malcolm <dmalcolm@redhat.com>
3601 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
3602 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
3603 c-family/c-spellcheck.cc.
3604 (best_macro_match::best_macro_match): Likewise.
3606 (struct edit_distance_traits<cpp_hashnode *>): Move to
3607 c-family/c-spellcheck.h.
3608 (class best_macro_match): Likewise.
3610 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3612 PR tree-optimization/83293
3613 * gimple-ssa-strength-reduction.c (insert_initializers): Use
3614 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
3615 might insert into empty bb.
3618 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
3619 simplify for plus with :c added, and pointer_plus without that.
3620 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
3621 with undefined overflow and the conversion is not widening,
3622 perform negation in utype and only convert to type afterwards.
3623 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
3624 simplify for plus with :c added, and pointer_plus without that.
3625 If type is integral with undefined overflow and the conversion is
3626 not widening, perform minus in utype and only convert to type
3627 afterwards. Move the last pointer_diff_expr simplify into the
3630 2017-12-06 Martin Sebor <msebor@redhat.com>
3632 PR tree-optimization/82646
3633 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
3634 strncpy, not maxlen.
3636 2017-12-06 Martin Sebor <msebor@redhat.com>
3638 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
3641 PR tree-optimization/83075
3642 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
3643 strncat/strncpy don't change length of source string.
3645 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
3648 2017-11-29 Martin Aberg <maberg@gaisler.com>
3650 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3651 to prevent b2bst errata sequence.
3652 (sqrtdf2_fix): Likewise.
3654 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3656 PR tree-optimization/81945
3657 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
3658 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
3659 to dest_cfun has orig_loop_num set, either remap it to the new
3660 loop number if the loop got moved too, or clear it.
3662 2017-12-05 Steve Ellcey <sellcey@cavium.com>
3664 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
3666 (thunderx2t99_nothing): New insn reservation.
3667 (thunderx2t99_mrs): New insn reservation.
3668 (thunderx2t99_multiple): New insn reservation.
3669 (thunderx2t99_alu_basi): Add bfx to reservation.
3670 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
3672 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3675 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
3677 2017-12-05 Bin Cheng <bin.cheng@arm.com>
3679 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
3680 tree-ssa-pre.c::remove_dead_inserted_code.
3681 * tree-ssa-dce.h: New file.
3682 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
3683 (remove_dead_inserted_code): Move and rename to function
3684 tree-ssa-dce.c::simple_dce_from_worklist.
3685 (pass_pre::execute): Update use.
3687 2017-12-05 Richard Biener <rguenther@suse.de>
3689 PR tree-optimization/83277
3690 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
3691 to code-gen liveout vars.
3693 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
3695 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
3696 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
3697 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
3700 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
3702 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
3704 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
3705 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
3708 2017-12-05 Richard Biener <rguenther@suse.de>
3710 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
3711 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
3712 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
3713 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
3715 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3717 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
3719 2017-12-05 Martin Liska <mliska@suse.cz>
3720 Jakub Jelinek <jakub@redhat.com>
3722 * doc/invoke.texi: Document the options.
3723 * flag-types.h (enum sanitize_code): Add
3724 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3725 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
3726 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3727 * opts.c: Define new sanitizer options.
3728 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
3729 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
3731 2017-12-05 Julia Koval <julia.koval@intel.com>
3733 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
3734 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
3735 (ix86_handle_option): Handle -mavx512vnni.
3736 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
3737 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3738 * config/i386/i386-c (__AVX512VNNI__): New.
3739 * config/i386/i386.c (ix86_target_string): Handle new option.
3740 (ix86_valid_target_attribute_inner_p): Handle new option.
3741 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
3742 * config/i386/i386.opt (mavx512vnni): New option.
3744 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3747 * athlon.md: Disable for generic.
3748 * haswell.md: Enable for generic.
3749 * i386.c (ix86_sched_init_global): Add core hooks for generic.
3750 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
3752 (ix86_adjust_cost): Move generic to haswell path.
3754 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
3756 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
3757 instead of MEM_P in a couple more places. Fix formatting issues.
3759 2017-12-04 Jim Wilson <jimw@sifive.com>
3761 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
3762 instead of GP_REG_LAST-1.
3763 (riscv_adjust_libcall_cfi_prologue): Likewise.
3764 (riscv_adjust_libcall_cri_epilogue): Likewise.
3765 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
3768 2017-12-04 Luis Machado <luis.machado@linaro.org>
3770 * ipa-pure-const.c (check_decl): Add missing newline.
3771 (state_from_flags): Likewise.
3773 2017-12-04 Jeff Law <law@redhat.com>
3775 PR tree-optimizatin/78496
3776 * gimple-ssa-evrp-analyze.h
3777 (evrp_range_analyzer::get_vr_values): Simplify.
3778 * gimple-ssa-evrp-analyze.c: Corresponding changes.
3779 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
3780 and gimple-ssa-evrp-analyze.h.
3781 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
3782 (simplify_stmt_for_jump_threading): Copy a blob of code from
3783 tree-vrp.c to use ranges to simplify statements.
3784 (dom_opt_dom_walker::before_dom_children): Call
3785 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
3786 (dom_opt_dom_walker::after_dom_children): Similarly for
3787 evrp_range_analyzer::leave.
3788 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
3791 * gimple-ssa-evrp-analyze.c
3792 (evrp_range_analyzer::extract_range_from_stmt): Always use
3793 vr_values::update_value_range so preexisting range info is
3794 medged with new range info, even if the new range is VR_VARYING.
3796 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3798 * combine.c: Adjust comment.
3799 (use_crosses_set_p): Delete.
3800 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
3801 (try_combine): Ditto.
3803 2017-12-04 Richard Biener <rguenther@suse.de>
3805 PR tree-optimization/83255
3806 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
3807 Re-add zero-iteration check.
3809 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3811 PR rtl-optimization/83245
3812 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
3813 hard registers as earlyclobber, also if not in an asm.
3815 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3819 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3822 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3823 rs6000_cpu based on cpu_index, not tune_index.
3825 2017-12-04 Richard Biener <rguenther@suse.de>
3827 PR tree-optimization/83238
3828 * graphite-scop-detection.c (scop_detection::merge_sese): Make
3829 code match comment, rejecting invalid SESE regions.
3831 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
3833 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
3834 require base operand is a REG_POINTER prior to reload on targets
3835 with non-equivalent space registers.
3837 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3839 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
3840 (update_specialized_profile): Fix updating of counts.
3841 (perhaps_add_new_callers): Likewise.
3843 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3846 * x86-tune.def: Remove obsolette FIXMEs.
3847 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
3848 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
3849 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
3851 (X86_TUNE_PAD_RETURNS): Disable for generic.
3853 2017-12-02 Jakub Jelinek <jakub@redhat.com>
3855 PR tree-optimization/83170
3856 PR tree-optimization/83241
3857 * gimple-ssa-store-merging.c
3858 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
3859 gimple_vuse (ins_stmt) in case it has changed.
3860 (imm_store_chain_info::output_merged_store): Likewise.
3862 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
3866 * tree-cfg.c (pass_warn_function_return::execute): Handle
3867 __builtin_ubsan_handle_missing_return like __builtin_unreachable
3868 with BUILTINS_LOCATION.
3872 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
3873 is BLKmode for vector field with vector raw mode, use TYPE_MODE
3874 instead of DECL_MODE.
3876 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
3877 last argument to rtx pointer.
3878 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
3879 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
3880 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
3881 depending on the chosen ISAs.
3882 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
3883 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
3884 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
3886 * config/i386/sse.md (mov<mode>_internal): Likewise.
3887 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
3889 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3891 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
3893 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3895 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
3896 parameters from prototype.
3897 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
3898 parameters. Don't print a comment.
3899 (emit_fusion_gpr_load): Adjust.
3900 (emit_fusion_load_store): Adjust.
3901 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
3902 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
3903 comment on the second line.
3905 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3908 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3909 rs6000_cpu based on cpu_index, not tune_index.
3911 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3913 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
3914 which_alternative instead of which_alternative + 1.
3915 (output_asm_insn): Print an extra tab if the template is short.
3917 2017-12-01 Jim Wilson <jimw@sifive.com>
3919 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
3921 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
3922 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
3924 * doc/tm.texi: Regenerate.
3926 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
3929 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
3930 whether we can allocate pseudos before trying to fix an address.
3931 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
3932 memory address is indexed or indirect.
3933 (floatuns_<mode>si2_hw2): Likewise.
3935 2017-12-01 Jason Merrill <jason@redhat.com>
3937 * Makefile.in (TAGS): Add c-family/*.cc.
3939 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
3941 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
3942 (call_value_insn): Likewise.
3943 (sibcall_insn): Likewise.
3944 (sibcall_value_insn): Likewise.
3945 (movsi_aarch64): Likewise.
3946 (movdi_aarch64): Likewise.
3947 (add_losym_): Likewise.
3948 (ldr_got_small_): Likewise.
3949 (ldr_got_small_sidi): Likewise.
3950 (ldr_got_small_28k_): Likewise.
3951 (ldr_got_small_28k_sidi): Likewise.
3952 * config/aarch64/aarch64.c (aarch64_print_address_internal):
3953 Move output_addr_const to symbolic case. Add error check.
3955 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3957 * config/s390/predicates.md (plus16_Q_operand): New predicate.
3958 * config/s390/s390.md: Disable MVC merging peephole if it would
3959 disable operand forwarding.
3960 (new peephole2): Split MVCs if it would turn them into up to 2
3963 2017-12-01 Richard Biener <rguenther@suse.de>
3965 PR tree-optimization/83232
3966 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
3967 detection of same access. Instead of breaking the group here
3968 do not consider the duplicate. Add comment explaining real fix.
3970 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
3972 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
3974 2017-12-01 Sudakshina Das <sudi.das@arm.com>
3976 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
3978 2017-12-01 Jakub Jelinek <jakub@redhat.com>
3980 * function.h (struct function): Remove cilk_frame_decl,
3981 is_cilk_function and calls_cilk_spawn fields.
3982 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
3984 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
3985 cilk_elemental field.
3986 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
3987 * target.def: Adjust comment.
3988 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
3989 Don't test cilk_elemental.
3991 PR tree-optimization/83233
3992 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
3993 bswap_stat name for the struct.
3996 * tree.h (SWITCH_BREAK_LABEL_P): Define.
3997 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
3998 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
3999 SWITCH_BREAK_LABEL_P set on the label.
4000 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
4001 added for default case if it was missing and not all cases covered.
4002 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
4003 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
4005 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
4006 to avoid -Wimplicit-fallthrough warning.
4007 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
4008 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
4010 PR tree-optimization/83221
4011 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
4013 (init_reassoc): Formatting fix.
4016 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
4017 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
4021 * tree-cfg.c: Include asan.h.
4022 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
4024 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4026 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
4028 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4030 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
4031 vpcmpeqd instruction.
4033 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4035 * config/i386/i386.c (standard_sse_constant_opcode): Fix
4036 registers type for 128bit mode.
4038 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4040 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
4042 * gengtype-state.c (read_state_pair): Likewise.
4043 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
4044 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
4045 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
4046 * ifcvt.c (noce_try_store_flag_constants): Likewise.
4047 * tree-ssa-ccp.c (ccp_finalize): Likewise.
4048 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
4049 * builtins.c (fold_builtin_3): Likewise.
4050 * graphite-scop-detection.c
4051 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
4052 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
4055 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
4058 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
4059 parameter. Return true if ignore_decl_rtl_set_p is true and other
4060 conditions are satisfied.
4061 * asan.h (asan_protect_global): Add new parameter.
4062 * varasm.c (categorize_decl_for_section): Pass true as second parameter
4063 to asan_protect_global calls.
4065 2017-11-30 Jim Wilson <jimw@sifive.com>
4067 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
4068 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
4069 -mstrict-align, add info on default value. Delete redundant lines for
4070 -mabi. Add missing -mexplicit-relocs docs.
4072 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4074 * config/arc/arc.md (trap): New pattern.
4076 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4078 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
4079 instruction to end into a delay slot.
4080 * config/arc/arc.md (cond_delay_insn): Check if the instruction
4081 can be placed into a delay slot against reg_note.
4083 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4085 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
4086 labels number of usages.
4088 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4090 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
4092 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
4094 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4097 * internal-fn.c (expand_mul_overflow): Optimize unsigned
4098 multiplication by power of 2 constant into two shifts + comparison.
4100 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
4103 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
4105 2017-11-30 Richard Biener <rguenther@suse.de>
4107 PR tree-optimization/83202
4108 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
4109 allow_peel argument and guard peeling.
4110 (canonicalize_loop_induction_variables): Likewise.
4111 (canonicalize_induction_variables): Pass false.
4112 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
4113 peeling from cunrolli.
4115 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4117 * combine.c (try_combine): Print a message to dump file whenever
4118 I0, I1, or I2 cannot be combined into I3.
4120 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4122 PR rtl-optimization/83156
4123 PR rtl-optimization/82621
4124 * combine.c (try_combine): Don't split an I2 if one of the dests is
4125 set again before I3. Allow unused dests.
4127 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4129 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
4131 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
4133 PR rtl-optimization/80818
4134 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
4135 recursively. Use insn code for clobber.
4136 (lra_set_insn_recog_data): Pass the new arg to
4137 collect_non_operand_hard_regs.
4138 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
4140 (lra_update_insn_regno_info): Pass insn to
4141 add_regs_to_insn_regno_info.
4143 2017-11-29 Jim Wilson <jimw@sifive.com>
4144 Andrew Waterman <andrew@sifive.com>
4146 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
4147 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
4148 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
4149 use SINGLE_SHIFT_COST.
4150 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
4151 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
4153 2017-11-29 Julia Koval <julia.koval@intel.com>
4155 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
4156 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
4157 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
4158 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
4159 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
4160 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
4161 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
4162 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
4163 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
4164 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
4165 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
4166 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
4167 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
4168 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
4169 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
4170 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
4171 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
4172 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
4173 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
4174 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
4175 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
4176 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
4177 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
4178 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
4179 __builtin_ia32_vpshldv_v2di_mask,
4180 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
4181 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
4182 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
4184 2017-11-29 Julia Koval <julia.koval@intel.com>
4186 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
4187 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
4188 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
4189 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
4190 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
4191 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
4192 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
4193 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
4194 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
4195 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
4196 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
4197 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
4198 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
4199 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
4200 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
4201 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
4202 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
4203 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
4204 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4205 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
4206 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
4208 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4210 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
4211 movsi_pic_gotdata_op instruction as a load for the UT699 errata
4214 2017-11-29 Martin Aberg <maberg@gaisler.com>
4216 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4217 to prevent b2bst errata sequence.
4218 (sqrtdf2_fix): Likewise.
4220 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4222 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
4223 (div_sqrt_insn_p): New function.
4224 (sparc_do_work_around_errata): Insert NOP instructions to
4225 prevent sequences that could trigger the TN-0013 errata for
4226 certain LEON3 processors.
4227 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
4228 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
4229 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
4230 (in_branch_delay): Prevent div and sqrt in delay slot if
4232 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
4234 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4236 * config/sparc/sparc.c (atomic_insn_p): New function.
4237 (sparc_do_work_around_errata): Insert NOP instructions to
4238 prevent sequences that could trigger the TN-0010 errata for
4240 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
4241 instruction referable in atomic_insns_p.
4243 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4245 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
4246 (atomic_compare_and_swap_leon3_1): Likewise.
4249 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4251 * config/sparc/sparc.c (fpop_insn_p): New function.
4252 (sparc_do_work_around_errata): Insert NOP instructions to
4253 prevent sequences that could trigger the TN-0012 errata for
4255 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
4256 * config/sparc/sparc.md (fix_gr712rc): New attribute.
4257 (in_branch_annul_delay): Prevent floating-point instructions
4258 in delay slot of annulled integer branch.
4260 2017-11-29 Richard Biener <rguenther@suse.de>
4262 PR tree-optimization/83202
4263 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
4265 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
4266 nodes vectorized to the same stmts multiple times.
4267 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
4268 (vect_analyze_slp_instance): Adjust.
4269 (scalar_stmts_to_slp_tree_map_t): New typedef.
4270 (vect_schedule_slp_instance): Add a map recording the SLP node
4271 representing the vectorized stmts for a set of scalar stmts.
4272 Avoid code-generating redundancies.
4273 (vect_schedule_slp): Allocate map and pass it down.
4275 2017-11-29 Nathan Sidwell <nathan@acm.org>
4278 * tree.c (build_complex_type): Fix canonicalization. Only fill in
4281 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
4283 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
4284 cases for printing LDP/STP memory addresses.
4285 (aarch64_print_address_internal): Renamed from
4286 aarch64_print_operand_address, added parameter, add Pmode check.
4287 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
4288 (aarch64_print_operand_address): Indirect to
4289 aarch64_print_address_internal.
4290 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
4291 'y' operand output specifier.
4293 2017-11-29 Jakub Jelinek <jakub@redhat.com>
4296 * tree.c (build_simple_mem_ref_loc): Handle
4297 get_addr_base_and_unit_offset returning a MEM_REF.
4300 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
4303 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
4306 2017-11-28 David Malcolm <dmalcolm@redhat.com>
4308 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
4310 (layout::move_to_column): Likewise.
4312 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4315 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
4316 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
4318 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4319 Martin Jambor <mjambor@suse.cz>
4322 * tree.h (expr_type_first_operand_type_p): Declare
4323 * tree.c (expr_type_first_operand_type_p): New function.
4324 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
4325 (ipa_value_from_jfunc): Adjust declaration.
4326 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
4327 Use it as result type for arithmetics, unless it is NULL in which case
4328 be more conservative.
4329 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
4330 ipa_get_jf_pass_through_result.
4331 (propagate_vals_across_pass_through): Likewise.
4332 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
4333 is to propagate_vals_across_pass_through.
4334 (propagate_constants_across_call): Pass PARM_TYPE to
4335 propagate_scalar_across_jump_function.
4336 (find_more_scalar_values_for_callers_subset): Pass parameter type to
4337 ipa_value_from_jfunc.
4338 (cgraph_edge_brings_all_scalars_for_node): Likewise.
4339 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
4340 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
4341 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
4342 target_type, pass it to ipa_value_from_jfunc.
4343 (update_indirect_edges_after_inlining): Pass parameter type to
4344 try_make_edge_direct_simple_call.
4346 2017-11-28 Jeff Law <law@redhat.com>
4348 * gimple-ssa-evrp-analyze.c
4349 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
4350 refine ranges if scev_initialized_p returns true.
4351 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
4353 2017-11-28 Julia Koval <julia.koval@intel.com>
4355 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
4356 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
4357 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
4358 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
4359 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
4360 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
4361 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
4362 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
4363 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
4364 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
4365 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
4366 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
4367 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
4368 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
4369 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
4370 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
4371 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
4372 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
4373 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
4374 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
4375 __builtin_ia32_vpshrd_v2di_mask): New builtins.
4376 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
4378 2017-11-28 Julia Koval <julia.koval@intel.com>
4380 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
4381 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
4382 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
4383 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
4384 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
4385 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
4386 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
4387 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
4388 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
4389 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
4390 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
4391 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
4392 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
4393 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
4394 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
4395 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
4396 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
4397 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
4398 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
4399 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
4400 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
4401 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
4402 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
4403 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
4404 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
4405 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
4406 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
4407 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
4408 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
4409 __builtin_ia32_vpshld_v2di_mask): New builtins.
4410 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4411 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
4413 2017-11-28 Richard Biener <rguenther@suse.de>
4415 PR tree-optimization/80776
4416 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
4418 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
4420 (evrp_range_analyzer::record_ranges_from_incoming_edges):
4421 If the incoming edge is an effective fallthru because the other
4422 edge only reaches a __builtin_unreachable () then record ranges
4423 derived from the controlling condition in SSA info.
4424 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
4425 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4427 2017-11-28 Olivier Hainque <hainque@adacore.com>
4429 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
4430 on mingw build hosts.
4432 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4435 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
4437 2017-11-28 Richard Biener <rguenther@suse.de>
4440 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
4441 copies generated from memcpy use a character array as reference
4444 2017-11-28 Julia Koval <julia.koval@intel.com>
4445 Sebastian Peryt <sebastian.peryt@intel.com>
4447 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
4448 c-family/c-cilkplus.o, c-family/array-notation-common.o,
4449 cilk-common.o, cilk.h, cilk-common.c): Remove.
4451 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
4452 * builtins.c (is_builtin_name): Remove cilkplus condition.
4453 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
4454 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
4455 cilk-builtins.def, cilkplus.def): Remove.
4456 * cif-code.def (CILK_SPAWN): Remove.
4457 * cilk-builtins.def: Delete.
4458 * cilk-common.c: Ditto.
4460 * cilkplus.def: Ditto.
4461 * config/darwin.h (fcilkplus): Delete.
4462 * cppbuiltin.c: Ditto.
4463 * doc/extend.texi: Remove cilkplus doc.
4464 * doc/generic.texi: Ditto.
4465 * doc/invoke.texi: Ditto.
4466 * doc/passes.texi: Ditto.
4467 * gcc.c (fcilkplus): Remove.
4468 * gengtype.c (cilk.h): Remove.
4469 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
4471 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
4473 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
4475 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
4476 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
4477 cilkplus conditions.
4478 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
4479 inline_read_section): Ditto.
4480 * ipa-inline-analysis.c (cilk.h): Remove.
4481 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
4482 * lto-wrapper.c (merge_and_complain, append_compiler_options,
4483 append_linker_options): Remove condition for fcilkplus.
4484 * lto/lto-lang.c (cilk.h): Remove.
4485 (lto_init): Remove condition for fcilkplus.
4486 * omp-expand.c (expand_cilk_for_call): Delete.
4487 (expand_omp_taskreg, expand_omp_for_static_chunk,
4488 expand_omp_for): Remove cilkplus
4490 (expand_cilk_for): Delete.
4491 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
4492 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
4493 execute_lower_omp, diagnose_sb_0): Ditto.
4494 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
4495 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4496 * tree-nested.c: Ditto.
4497 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
4498 (dump_generic_node): Ditto.
4499 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4500 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
4502 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
4504 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4506 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
4507 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
4508 for vector mode and !TARGET_SIMD.
4510 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4512 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
4514 * tree.h (SWITCH_LABELS): Remove.
4515 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
4516 assert SWITCH_BODY is non-NULL.
4517 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
4519 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
4521 PR tree-optimization/80788
4522 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
4523 has TREE_OVERFLOW set, call drop_tree_overflow.
4525 2017-11-28 Richard Biener <rguenther@suse.de>
4527 PR tree-optimization/83158
4528 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
4530 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
4533 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
4534 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
4536 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
4538 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
4539 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
4541 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4544 * varasm.c (compare_constant): Take the mode of the constants into
4545 account when comparing floating point constants.
4547 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
4549 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
4550 from explicit instantiation of debug_helper.
4551 * vec.h (DEFINE_DEBUG_VEC): Ditto.
4553 2017-11-27 Richard Biener <rguenther@suse.de>
4555 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
4558 2017-11-27 Richard Biener <rguenther@suse.de>
4560 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
4561 (type_hash_canon): Also clear int_cst_hash_table entry for
4563 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
4565 2017-11-27 Tamar Christina <tamar.christina@arm.com>
4567 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4568 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4569 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4571 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4573 * hash-map.h (gt_cleare_cache): Avoid UB.
4575 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
4577 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
4578 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
4579 (decide_unroll_runtime_iterations): Likewise.
4580 (decide_unroll_stupid): Likewise.
4582 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4585 * config/i386/i386.md: Add a loop with incssp.
4587 2017-11-27 Martin Jambor <mjambor@suse.cz>
4589 PR tree-optimization/81248
4590 * tree-sra.c (splice_param_accesses): Remove size check.
4591 (decide_one_param_reduction): Fix size check.
4592 * gimple-pretty-print.c (dump_profile): Silence warning.
4593 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
4595 2017-11-27 Jakub Jelinek <jakub@redhat.com>
4598 * dbxout.c (lastlineno): New variable.
4599 (dbx_debug_hooks): Use dbxout_switch_text_section as
4600 switch_text_section debug hook.
4601 (dbxout_function_end): Switch to current_function_section
4602 rather than function_section. If crtl->has_bb_partition,
4603 output just one N_FUN, depending on in_cold_section_p.
4604 (dbxout_source_line): Remember last lineno in lastlineno.
4605 (dbxout_switch_text_section): New function.
4606 (dbxout_function_decl): Adjust dbxout_block caller.
4607 (dbx_block_with_cold_children): New function.
4608 (dbxout_block): Return true if any LBRAC/RBRAC have been
4609 emitted. Use dbx_block_with_cold_children at depth == 0
4610 in second partition. Add PARENT_BLOCKNUM argument, pass
4611 it optionally adjusted to children. Output LBRAC/RBRAC
4612 around recursive call only if the block is in the current
4613 partition, if not and anything was output, emit empty
4615 * final.c (final_scan_insn): Compute cold_function_name
4616 before calling switch_text_section debug hook. Call
4617 that hook even if dwarf2out_do_frame if not emitting
4621 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
4622 TREE_READONLY decls.
4624 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4626 PR rtl-optimization/82488
4627 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
4629 2017-11-26 Julia Koval <julia.koval@intel.com>
4631 * config/i386/i386.c (processor_target_table): Add skylake_cost for
4633 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
4636 2017-11-26 Julia Koval <julia.koval@intel.com>
4638 * config/i386/driver-i386.c (host_detect_local_cpu):
4639 Detect skylake-avx512.
4641 2017-11-26 Julia Koval <julia.koval@intel.com>
4643 * config.gcc: Add -march=cannonlake.
4644 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
4645 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
4646 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
4647 (PTA_CANNONLAKE): New.
4648 (processor_target_table): Add cannonlake.
4649 (ix86_option_override_internal): Ditto.
4650 (fold_builtin_cpu): Ditto.
4651 (get_builtin_code_for_version): Handle cannonlake.
4652 (M_INTEL_COREI7_CANNONLAKE): New.
4653 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
4654 * doc/invoke.texi: Add -march=cannonlake.
4656 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
4658 * plugin.c (add_new_plugin): Use platform-specific library extensions.
4659 (try_init_one_plugin): Alternative implementation for MinGW.
4660 * Makefile.in (plugin_implib): New.
4661 (gengtype-lex.c): Fix broken AIX workaround.
4662 * configure: Regenerate.
4663 * doc/plugins.texi: Document support for MinGW.
4665 2017-11-25 Jakub Jelinek <jakub@redhat.com>
4667 PR rtl-optimization/81553
4668 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
4669 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
4670 is a shift where C1 has different mode than the whole shift, use C1's
4671 mode for MULT rather than the shift's mode.
4674 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
4675 builtins not enabled in the currently selected ISA.
4677 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
4679 PR tree-optimization/71026
4680 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
4681 (insert_reciprocals): Change to insert reciprocals before a division
4682 by a square and to insert the square of a reciprocal.
4683 (execute_cse_reciprocals_1): Change to consider division by a square.
4684 (register_division_in): Add importance parameter.
4686 2017-11-24 Richard Biener <rguenther@suse.de>
4688 PR tree-optimization/82402
4689 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
4690 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
4692 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
4694 * match.pd (0-ptr): New transformation.
4696 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
4699 * ipa-inline.c (inline_small_functions): Set current badnes correctly
4700 when skipping checking.
4702 2017-11-24 Richard Biener <rguenther@suse.de>
4704 PR tree-optimization/83128
4705 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
4706 (vn_reference_lookup_3): Likewise.
4708 2017-11-24 Jakub Jelinek <jakub@redhat.com>
4711 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
4712 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
4715 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
4716 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
4717 check for the latter with gimple_call_builtin_p. Do not handle
4718 BUILT_IN_STPNCPY_CHK which is not a pass through call.
4720 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
4722 * config/arm/arm_neon.h: Fix pragma GCC push_options before
4725 2017-11-23 Julia Koval <julia.koval@intel.com>
4727 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
4728 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
4729 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
4730 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
4731 _mm512_maskz_expandloadu_epi16): New intrinsics.
4732 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
4733 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
4734 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
4735 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
4736 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
4737 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
4738 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
4739 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
4740 _mm256_maskz_expandloadu_epi8): New intrinsics.
4741 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
4742 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
4743 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
4744 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
4745 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
4746 * config/i386/sse.md (VI248_VLBW): New iterator.
4747 (expand<mode>_mask, expand<mode>_maskz): New patterns.
4749 2017-11-23 Julia Koval <julia.koval@intel.com>
4751 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
4752 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
4753 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
4754 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
4755 _mm512_mask_compressstoreu_epi16): New.
4756 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
4757 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
4758 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
4759 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
4760 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
4761 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
4762 _mm256_mask_compressstoreu_epi8): New.
4763 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
4764 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
4765 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
4766 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
4767 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
4768 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
4769 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
4770 __builtin_ia32_compresshi128_mask,
4771 __builtin_ia32_compressstoreuqi512_mask,
4772 __builtin_ia32_compressstoreuhi512_mask,
4773 __builtin_ia32_compressstoreuqi256_mask,
4774 __builtin_ia32_compressstoreuqi128_mask,
4775 __builtin_ia32_compressstoreuhi256_mask,
4776 __builtin_ia32_compressstoreuhi128_mask): New builtins.
4777 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
4779 (ix86_expand_special_args_builtin): Handle new types.
4780 (s4fma_expand): Handle new builtin array.
4781 * config/i386/immintrin.h: Include new headers.
4782 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
4783 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
4785 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4787 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
4788 fixes. Declare temp and g variables at the top in order to avoid
4789 {} in most of the cases.
4791 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
4793 * match.pd (ptr-0): New transformation.
4795 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4797 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
4798 (struct addr_mode_cost_table): New.
4799 (struct tune_params): Add field addr_mode_costs.
4800 * config/arm/arm.c (generic_addr_mode_costs): New.
4801 (arm_slowmul_tune): Initialise addr_mode_costs field.
4802 (arm_fastmul_tune): Likewise.
4803 (arm_strongarm_tune): Likewise.
4804 (arm_xscale_tune): Likewise.
4805 (arm_9e_tune): Likewise.
4806 (arm_marvell_pj4_tune): Likewise.
4807 (arm_v6t2_tune): Likewise.
4808 (arm_cortex_tune): Likewise.
4809 (arm_cortex_a8_tune): Likewise.
4810 (arm_cortex_a7_tune): Likewise.
4811 (arm_cortex_a15_tune): Likewise.
4812 (arm_cortex_a35_tune): Likewise.
4813 (arm_cortex_a53_tune): Likewise.
4814 (arm_cortex_a57_tune): Likewise.
4815 (arm_exynosm1_tune): Likewise.
4816 (arm_xgene1_tune): Likewise.
4817 (arm_cortex_a5_tune): Likewise.
4818 (arm_cortex_a9_tune): Likewise.
4819 (arm_cortex_a12_tune): Likewise.
4820 (arm_cortex_a73_tune): Likewise.
4821 (arm_v7m_tune): Likewise.
4822 (arm_cortex_m7_tune): Likewise.
4823 (arm_v6m_tune): Likewise.
4824 (arm_fa726te_tune): Likewise.
4825 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
4827 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4829 * config/arm/arm.c (arm_mem_costs): New function.
4830 (arm_rtx_costs_internal): Use arm_mem_costs.
4832 2017-11-23 Mark Wielaard <mark@klomp.org>
4834 * dwarf2out.c (init_sections_and_labels): Use generation to create
4835 unique ranges_section_label and ranges_base_label. Return generation.
4836 (output_rnglists): Add generation argument. Use generation to create
4837 unique ranges labels.
4838 (dwarf2out_finish): Get generation from init_sections_and_labels
4839 and pass generation to output_rnglists.
4841 2017-11-23 Mike Stump <mikestump@comcast.net>
4842 Eric Botcazou <ebotcazou@adacore.com>
4844 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
4845 * cfgloop.h (struct loop): Add unroll field.
4846 * function.h (struct function): Add has_unroll bitfield.
4847 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
4848 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
4849 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
4850 (pass_rtl_unroll_loops::gate): Likewise.
4851 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
4852 for which loop->unroll==1.
4853 (decide_unroll_constant_iterations): Use note for consistency and
4854 take loop->unroll into account. Return early if loop->unroll is set.
4855 Fix thinko in existing test.
4856 (decide_unroll_runtime_iterations): Use note for consistency and
4857 take loop->unroll into account.
4858 (decide_unroll_stupid): Likewise.
4859 * lto-streamer-in.c (input_cfg): Read loop->unroll.
4860 * lto-streamer-out.c (output_cfg): Write loop->unroll.
4861 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
4863 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
4864 (print_loop): Print loop->unroll if set.
4865 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
4866 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
4867 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
4869 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
4870 loop->unroll is set and smaller than the trip count. Otherwise bypass
4871 entirely the heuristics if loop->unroll is set. Remove dead note.
4872 Fix off-by-one bug in other note.
4873 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
4874 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
4876 (tree_unroll_loops_completely): Make static.
4877 (pass_complete_unroll::execute): Use correct type for variable.
4878 (pass_complete_unrolli::execute): Fix formatting.
4879 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
4881 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
4883 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
4884 enable when TARGET_PREFER_AVX128 is set.
4886 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4888 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
4890 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
4891 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
4893 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4895 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
4896 frequencies. Fix estimation of aggregate parameters.
4898 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4900 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
4903 2017-11-23 Tom de Vries <tom@codesourcery.com>
4905 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
4907 2017-11-23 Tom de Vries <tom@codesourcery.com>
4909 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
4912 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
4915 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
4916 sibcall_value_pcrel_fdpic): Use local variable instead of
4918 (calli_tbr_rel): Add missing operand 2.
4919 (call_valuei_tbr_rel): Add missing operand 3.
4921 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4924 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
4925 bitpos/bitsize covering the whole destination, use store_expr only if
4926 the complex mode is the same. Otherwise, use expand_normal and if
4927 it returns CONCAT, subreg each part separately instead of trying to
4928 subreg the whole result.
4930 2017-11-23 Richard Biener <rguenther@suse.de>
4932 PR tree-optimization/23094
4933 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
4934 come here from walking over backedges in the first iteration.
4935 (vn_reference_lookup_3): Skip clobbers that store the same value.
4937 2017-11-23 Richard Biener <rguenther@suse.de>
4939 PR tree-optimization/81403
4940 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
4941 a block we need a leader relative to.
4942 (phi_translate_1): For nary processing require a leader from
4943 get_representative_for given we run expression simplification
4944 using match-and-simplify. Remove previous fix.
4946 2017-11-22 Jeff Law <law@redhat.com>
4948 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
4949 Use new method allocate_value_range rather than accessing the
4950 vrp_value_range_pool data member directly.
4951 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
4952 to use extract_range_from_stmt method to avoid need for
4953 extract_range_from_assignment method.
4954 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
4955 method rather than setting values_propgated data member directly.
4956 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
4957 and values propagated data members and extract_range_from_assignment
4958 method. Reorder private data members to conform to standards.
4959 Add new methods set_lattice_propagation_complete and
4960 allocate_value_range.
4962 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
4964 PR rtl-optimization/83030
4965 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
4966 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
4967 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
4968 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
4969 CROSSING_JUMP_P flag is already set before setting it.
4970 * cfgrtl.c (fixup_partition_crossing): Likewise.
4971 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
4974 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4976 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
4977 Handle the case where both arguments are using gen_const_vec_series.
4979 2017-11-22 David Malcolm <dmalcolm@redhat.com>
4982 * pretty-print.c (pp_format): Move quoting implementation to
4983 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
4984 to pass address of "quote" local.
4985 (pp_begin_quote): New function.
4986 (pp_end_quote): New function.
4987 * pretty-print.h (printer_fn): Convert penultimate param from bool
4989 (pp_begin_quote): New decl.
4990 (pp_end_quote): New decl.
4991 * tree-diagnostic.c (default_tree_printer): Convert penultimate
4992 param from bool to bool *.
4993 * tree-diagnostic.h (default_tree_printer): Likewise.
4995 2017-11-22 Jeff Law <law@redhat.com>
4997 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
4998 Initialize vr_values.
4999 (evrp_range_analyzer::try_find_new_range): Call methods attached to
5000 vr_values via vr_values class instance rather than delegators.
5001 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
5002 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5003 (evrp_range_analyzer::push_value_range): Likewise.
5004 (evrp_range_analyzer::pop_value_range): Likewise.
5005 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
5006 most delegators. Those remaining are exposed as public interfaces.
5007 Make vr_values a pointer and private.
5008 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
5010 (evrp_range_analyzer::get_vr_value): New method.
5011 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
5012 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
5013 (class evrp_dom_walker): Attach evrp_folder class, initialize
5014 it in the ctor. Remove temporary delegators.
5015 (evrp_dom_walker::before_dom_children): Call methods in attached
5016 evrp_range_analyzer class via class instance pointer. Use
5017 free value_range_constant_singleton to remove need for
5018 op_with_constant_singleton_value delegator method. Do not
5019 create a vrp_prop class instance for every call! Narrow
5020 scope of a couple variables.
5021 (evrp_dom_walker::cleanup): Call methods in attached
5022 evrp_range_analyzer class via class instance pointer.
5023 * vr-values.h (class vr_values): Privatize many methods and
5026 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5028 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
5029 * cfgexpand.c (expand_debug_expr): Remove handling for them.
5030 * expr.c (expand_expr_real_2): Likewise.
5031 * fold-const.c (const_unop): Likewise.
5032 * optabs-tree.c (optab_for_tree_code): Likewise.
5033 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
5034 * tree-inline.c (estimate_operator_cost): Likewise.
5035 * tree-pretty-print.c (dump_generic_node): Likewise.
5036 (op_code_prio): Likewise.
5037 (op_symbol_code): Likewise.
5038 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
5039 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
5040 * internal-fn.c (direct_internal_fn_optab): New function.
5041 (direct_internal_fn_array, direct_internal_fn_supported_p
5042 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
5043 * fold-const-call.c (fold_const_reduction): New function.
5044 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
5046 * tree-vect-loop.c: Include internal-fn.h.
5047 (reduction_code_for_scalar_code): Rename to...
5048 (reduction_fn_for_scalar_code): ...this and return an internal
5050 (vect_model_reduction_cost): Take an internal_fn rather than
5052 (vect_create_epilog_for_reduction): Likewise. Build calls rather
5054 (vectorizable_reduction): Use internal functions rather than tree
5055 codes for the reduction operation. Update calls to the functions
5057 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
5058 Use calls to internal functions rather than REDUC tree codes.
5059 * config/aarch64/aarch64-simd.md: Update comment accordingly.
5061 2017-11-22 Olivier Hainque <hainque@adacore.com>
5063 * config/vxworks.c (vxworks_override_options): Pick default
5064 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
5065 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
5066 DWARF_GNAT_ENCODINGS_DEFAULT.
5067 * config/vxworksae.h: Likewise.
5069 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
5071 PR tree-optimization/83104
5072 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
5075 2017-11-22 Marek Polacek <polacek@redhat.com>
5076 H.J. Lu <hongjiu.lu@intel.com>
5077 Jason Merrill <jason@redhat.com>
5082 * calls.c (initialize_argument_information): Call
5083 warn_parameter_passing_abi target hook.
5084 (store_one_arg): Use 0 for empty record size. Don't push 0 size
5085 argument onto stack.
5086 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
5087 * common.opt: Update -fabi-version description.
5088 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
5089 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
5091 (ix86_is_empty_record): New function.
5092 (ix86_warn_parameter_passing_abi): New function.
5093 (TARGET_EMPTY_RECORD_P): Redefine.
5094 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
5095 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
5096 * doc/tm.texi: Regenerated.
5097 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
5098 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
5099 * dwarf2out.c (get_ultimate_context): Move to tree.c.
5100 * explow.c (hard_function_value): Call arg_int_size_in_bytes
5101 instead of int_size_in_bytes.
5102 * expr.c (copy_blkmode_to_reg): Likewise.
5103 * function.c (aggregate_value_p): Return 0 for empty types.
5104 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
5106 (locate_and_pad_parm): Call arg size_in_bytes instead
5108 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
5109 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
5110 * target.def (empty_record_p, warn_parameter_passing_abi): New target
5112 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
5113 (std_gimplify_va_arg_expr): Skip empty records. Call
5114 arg_size_in_bytes instead size_in_bytes.
5115 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
5116 * tree-core.h (tree_type_common): Add empty_flag.
5117 (tree_decl_common): Update comments.
5118 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
5120 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5121 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
5123 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5124 * tree.c (default_is_empty_type): New function.
5125 (default_is_empty_record): New function.
5126 (arg_int_size_in_bytes): New function.
5127 (arg_size_in_bytes): New function.
5128 (get_ultimate_context): New function.
5129 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
5130 TRANSLATION_UNIT_WARN_EMPTY_P.
5131 (default_is_empty_record, arg_int_size_in_bytes,
5132 arg_size_in_bytes, get_ultimate_context): Declare.
5134 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5136 * config/arm/arm.c (cmse_clear_registers): New function.
5137 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
5138 code by call to cmse_clear_registers.
5139 (cmse_nonsecure_entry_clear_before_return): Likewise.
5141 2017-11-22 Tamar Christina <tamar.christina@arm.com>
5143 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
5144 (vdot_s32, vdotq_s32): New.
5145 (vdot_lane_u32, vdotq_lane_u32): New.
5146 (vdot_lane_s32, vdotq_lane_s32): New.
5149 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5152 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
5153 for unsigned values with fewer HWIs than the precision.
5154 (test_overflow): New function.
5155 (wide_int_cc_tests): Call it.
5157 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5159 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
5160 has the same mode class as Pmode.
5162 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5164 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
5165 padding_bits_to_clear_ptr.
5166 (cmse_nonsecure_entry_clear_before_return): Likewise.
5168 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5170 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
5171 auto_sbitap instead of integer bitfield to control register needing
5174 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5176 PR tree-optimization/83044
5177 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
5178 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
5179 that need the upper bound. Subtract offset from
5180 get_addr_base_and_unit_offset only if positive and subtract it
5181 before division by eltsize rather than after it.
5184 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
5185 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
5186 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
5190 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
5191 expand_binop with *mul_widen_optab, make sure at least one of the
5192 operands doesn't have VOIDmode.
5195 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
5197 PR rtl-optimization/82044
5198 PR tree-optimization/82042
5199 * dse.c (record_store): Check for overflow.
5200 (check_mem_read_rtx): Properly check for overflow if width == -1, call
5201 add_wild_read instead of clear_rhs_from_active_local_stores on
5202 overflow and log it into dump_file.
5204 2017-11-22 Richard Biener <rguenther@suse.de>
5206 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
5207 fake edges to exit when looking for a place to insert.
5208 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
5210 (insert_into_preds_of_block): Commit edge insertion immediately,
5211 assert that doesn't require new BBs.
5212 (fini_pre): Release expressions.
5213 (pass_pre::execute): Shuffle things around a bit, if the fn
5214 is too large do not compute AVAIL either as this is really the
5217 2017-11-22 Richard Biener <rguenther@suse.de>
5219 PR tree-optimization/83089
5220 * tree-if-conv.c (pass_if_conversion::execute): If anything
5221 changed reset SCEV and free the number of iteration estimates.
5223 2017-11-21 Martin Sebor <msebor@redhat.com>
5225 PR tree-optimization/82945
5226 * calls.h (warn_nonstring_bound): Remove unused function.
5228 2017-11-21 Martin Sebor <msebor@redhat.com>
5230 PR tree-optimization/82945
5231 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
5232 * calls.h (maybe_warn_nonstring_arg): Declare new function.
5233 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
5235 (initialize_argument_information): Call maybe_warn_nonstring_arg.
5236 * calls.h (get_attr_nonstring_decl): Declare new function.
5237 * doc/extend.texi (attribute nonstring): Update.
5238 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
5239 get_attr_nonstring_decl and handle it.
5240 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
5241 detection of nul-termination.
5242 (strlen_to_stridx): Change to a pointer.
5243 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
5244 (pass_strlen::execute): Same.
5246 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
5248 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
5249 for the new option -mprefer-vector-width=[none|128|256|512].
5250 * config/i386/i386.c (ix86_target_string): remove old style options
5251 -mprefer-avx256 and make -mprefer-avx128 as alias.
5252 (ix86_option_override_internal): Apply defaults for the
5253 -mprefer-vector-width=[128|256] option.
5254 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
5255 Implement macros to work with -mprefer-vector-width=.
5256 * config/i386/i386.opt: Implemented option
5257 -mprefer-vector-width=[none|128|256|512].
5258 * doc/invoke.texi: Documentation for
5259 -mprefer-vector-width=[none|128|256|512].
5261 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
5263 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
5265 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
5267 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
5269 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
5271 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
5273 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
5274 * expr.c (expand_expr_real_2): Likewise.
5275 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
5276 fold_binary_loc): Likewise.
5277 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
5278 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
5279 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
5280 MINUS_EXPR transformations.
5281 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
5282 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
5283 * tree-inline.c (estimate_operator_cost): Likewise.
5284 * tree-pretty-print.c (dump_generic_node, op_code_prio,
5285 op_symbol_code): Likewise.
5286 * tree-vect-stmts.c (vectorizable_operation): Likewise.
5287 * vr-values.c (extract_range_from_binary_expr): Likewise.
5288 * varasm.c (initializer_constant_valid_p_1): Likewise.
5289 * tree.def: New tree code POINTER_DIFF_EXPR.
5291 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
5293 * config/i386/i386.md (*bswap<mode>2_movbe): Add
5294 integer suffix to movbe mnemonic.
5295 (*bswaphi2_movbe): Ditto.
5296 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
5298 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5301 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
5302 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
5303 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
5305 2017-11-21 Martin Liska <mliska@suse.cz>
5307 * tree-inline.c (expand_call_inline): Remove not needed
5310 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
5311 Jakub Jelinek <jakub@redhat.com>
5314 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
5315 Remove static keyword from f variable.
5317 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5319 PR tree-optimization/83086
5320 * gimple-ssa-store-merging.c
5321 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
5322 rather than n.base_addr.
5324 2017-11-21 Martin Liska <mliska@suse.cz>
5326 PR rtl-optimization/82044
5327 PR tree-optimization/82042
5328 * dse.c (check_mem_read_rtx): Check for overflow.
5330 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5332 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
5335 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5337 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
5338 and its corresponding call arg location note.
5340 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5341 Andrew Burgess <andrew.burgess@embecosm.com>
5343 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
5345 (arc_return_slot_offset): Likewise.
5346 (arc_eh_return_address_location): New declaration.
5347 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
5348 (MUST_SAVE_REGISTER): Add exception handler case.
5349 (MUST_SAVE_RETURN_ADDR): Likewise.
5350 (arc_frame_pointer_required): Likewise.
5351 (arc_frame_pointer_needed): New function.
5352 (arc_compute_frame_size): Changed.
5353 (arc_expand_prologue): Likewise.
5354 (arc_expand_epilogue): Likewise.
5355 (arc_initial_elimination_offset): Likewise.
5356 (arc_return_slot_offset): Delete.
5357 (arc_eh_return_address_location): New function.
5358 (arc_builtin_setjmp_frame_value): Likewise.
5359 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
5360 (EH_RETURN_STACKADJ_RTX): Define.
5361 (EH_RETURN_HANDLER_RTX): Likewise.
5362 * config/arc/arc.md (eh_return): Delete.
5364 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5366 * print-tree.h (debug_vec_tree): Remove prototype.
5367 * gdbinit.in (pvt): Remove macro.
5369 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5371 PR tree-optimization/83047
5372 * gimple-ssa-store-merging.c
5373 (imm_store_chain_info::output_merged_store): If the loads with the
5374 same vuse are in different basic blocks, for load_gsi pick a load
5375 location that is dominated by the other loads.
5378 * config/i386/i386.c (ix86_memmodel_check): Start
5379 -Winvalid-memory-model diagnostics with lowercase letter.
5382 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
5383 set in_cold_section_p to the partition containing loc_list->first.
5384 When seeing loc_list->last_before_switch node, update secname and
5385 perform range_across_switch second partition handling only after that.
5388 * run-rtl-passes.c: Include debug.h.
5389 (run_rtl_passes): Call debug_hooks->assembly_start.
5390 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
5394 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
5395 OPTAB_DIRECT in calls to expand_simple_binop.
5397 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5400 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
5402 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5405 * spellcheck.h (best_match::blithely_get_best_candidate): New
5408 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5410 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
5411 with lower case letter.
5413 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
5415 * config/i386/i386.md (bswaphi2): New expander.
5416 (*bswaphi2_movbe): New insn pattern.
5417 (bswaphi -> rorhi pepehole2): New peephole pattern.
5419 2017-11-20 Jeff Law <law@redhat.com>
5421 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
5422 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5423 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
5424 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
5425 gimple-ssa-evrp-analyze.h.
5427 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
5428 set BB_VISITED here.
5429 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
5431 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5433 * tree-predcom.c: Add general comment on Store-Store chains.
5434 (split_data_refs_to_components): Postpone clearing eliminate_store_p
5436 (get_chain_last_ref_at): Rename into...
5437 (get_chain_last_write_at): ...this.
5438 (get_chain_last_write_before_load): New function.
5439 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
5440 CT_STORE_STORE when write reference is added.
5441 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
5442 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
5443 (initialize_root_vars_store_elim_1): Ditto.
5444 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
5445 definition is created.
5446 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
5447 chain by replacing it with dominant stored value.
5449 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5451 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
5453 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
5455 * vr-values.c (extract_range_from_binary_expr): Use a full range
5458 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
5460 * config/arm/arm.md (R4_REGNUM): Define constant.
5461 (nonsecure_call_internal): Remove r4 clobber.
5462 (nonsecure_call_value_internal): Likewise.
5463 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
5464 clobber and resequence match_operands.
5465 (nonsecure_call_value_reg_thumb1_v5): Likewise.
5466 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
5467 (nonsecure_call_value_reg_thumb2): Likewise.
5469 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5471 PR tree-optimization/78821
5472 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
5473 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
5474 to the address of the base rather than the base itself.
5475 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
5476 (find_bswap_or_nop_finalize): New function.
5477 (find_bswap_or_nop): Use it.
5478 (bswap_replace): Return a tree rather than bool, change first
5479 argument from gimple * to gimple_stmt_iterator, allow inserting
5480 into an empty sequence, allow ins_stmt to be NULL - then emit
5481 all stmts into gsi. Fix up MEM_REF address gimplification.
5482 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
5483 (struct store_immediate_info): Add N and INS_STMT non-static
5485 (store_immediate_info::store_immediate_info): Initialize them
5486 from newly added ctor args.
5487 (merged_store_group::apply_stores): Formatting fixes. Sort by
5489 (stmts_may_clobber_ref_p): For stores call also
5490 refs_anti_dependent_p.
5491 (gather_bswap_load_refs): New function.
5492 (imm_store_chain_info::try_coalesce_bswap): New method.
5493 (imm_store_chain_info::coalesce_immediate_stores): Use it.
5494 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
5495 (imm_store_chain_info::output_merged_store): Fail if number of
5496 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
5497 and NOP_EXPR rhs_code.
5498 (pass_store_merging::process_store): Compute n and ins_stmt, if
5499 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
5500 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
5502 (pass_store_merging::execute): Calculate dominators.
5504 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
5505 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
5506 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
5507 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5508 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
5509 class pass_optimize_bswap, bswap_replace,
5510 pass_optimize_bswap::execute): Moved to ...
5511 * gimple-ssa-store-merging.c: ... this file.
5512 Include optabs-tree.h.
5513 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
5514 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5515 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
5516 anonymous namespace, remove static keywords.
5517 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
5518 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
5520 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
5523 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
5525 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
5527 * vec.h (debug_helper): New function.
5528 (DEFINE_DEBUG_VEC): New macro.
5529 * hash-set.h (debug_helper): New function.
5530 (DEFINE_DEBUG_HASH_SET): New macro.
5531 * cfg.c (debug_slim (edge)): New function.
5532 Call DEFINE_DEBUG_VEC for edges.
5533 Call DEFINE_DEBUG_HASH_SET for edges.
5534 * cfghooks.c (debug_slim (basic_block)): New function.
5535 Call DEFINE_DEBUG_VEC for basic blocks.
5536 Call DEFINE_DEBUG_HASH_SET for basic blocks.
5537 * print-tree.c (debug_slim): New function to handle trees.
5538 Call DEFINE_DEBUG_VEC for trees.
5539 Call DEFINE_DEBUG_HASH_SET for trees.
5540 (debug (vec<tree, va_gc>) &): Remove.
5541 (debug (<vec<tree, va_gc>) *): Remove.
5542 * print-rtl.c (debug_slim): New function to handle const_rtx.
5543 Call DEFINE_DEBUG_VEC for rtx_def.
5544 Call DEFINE_DEBUG_VEC for rtx_insn.
5545 Call DEFINE_DEBUG_HASH_SET for rtx_def.
5546 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
5547 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
5548 (debug (vec<rtx_insn *> *ptr): Remove.
5549 (debug_insn_vector): Remove.
5550 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
5552 2017-11-20 Tom de Vries <tom@codesourcery.com>
5554 PR rtl-optimization/82020
5555 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
5556 IF_THEN_ELSE condition.
5558 2017-11-19 Jeff Law <law@redhat.com>
5560 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
5561 of degenerates resulting from ignoring an edge.
5563 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5566 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
5568 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5571 * profile-count.c (profile_count::to_sreal_scale): Fix return value
5572 for uninitialied counts.
5574 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5577 * tree-inline.c (estimate_num_insns): Set to 1 at least.
5579 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5582 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
5584 2017-11-19 Tom de Vries <tom@codesourcery.com>
5586 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
5589 2017-11-19 Tom de Vries <tom@codesourcery.com>
5591 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
5594 2017-11-19 Tom de Vries <tom@codesourcery.com>
5596 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5599 2017-11-19 Tom de Vries <tom@codesourcery.com>
5601 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
5603 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
5605 2017-11-19 Tom de Vries <tom@codesourcery.com>
5607 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
5609 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
5610 semicolon after MCORE_EXPORT_NAME call.
5612 2017-11-19 Tom de Vries <tom@codesourcery.com>
5615 * vmsdbgout.c (vmsdbgout_early_finish): New function.
5616 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
5618 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
5620 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
5621 (cgraph_edge::clone): Cleanup updating of profile.
5622 * ipa-cp.c (update_profiling_info): Likewise.
5623 * ipa-inline-transform.c (inline_transform): Likewise.
5624 * ipa-inline.c (inline_small_functions): Add missing space to dump.
5625 * ipa-split.c (execute_split_functions): Do not split when function
5627 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
5628 * profile-count.c (profile_count::dump): Add global0.
5629 (profile_count::to_cgraph_frequency): Do not ICE when entry is
5631 (profile_count::to_sreal_scale): Likewise.
5632 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
5633 (profile_count::combine_with_ipa_count): New function.
5634 * profile-count.h (profile_guessed_global0adjusted): New.
5635 (profile_count::adjusted_zero): New.
5636 (profile_count::global0adjusted): New.
5637 (profile_count::combine_with_ipa_count): New.
5638 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
5639 correct profile of return block of split functions.
5640 (copy_cfg_body): Remove unused profile_count.
5641 (copy_body): Likewise.
5642 (expand_call_inline): Update.
5643 (tree_function_versioning): Update.
5645 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
5647 * hash-set.h (hash_set::empty): New.
5648 * tree-ssa-threadbackward.h: Delete.
5649 * tree-ssa-threadbackward.c (class thread_jumps): New.
5650 Move max_threaded_paths into class.
5651 (fsm_find_thread_path): Remove arguments that are now in class.
5652 (profitable_jump_thread_path): Rename to...
5653 (thread_jumps::profitable_jump_thread_path): ...this.
5654 (convert_and_register_jump_thread_path): Rename to...
5655 (thread_jumps::convert_and_register_current_path): ...this.
5656 (check_subpath_and_update_thread_path): Rename to...
5657 (thread_jumps::check_subpath_and_update_thread_path): ...this.
5658 (register_jump_thread_path_if_profitable): Rename to...
5659 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
5660 (handle_phi): Rename to...
5661 (thread_jumps::handle_phi): ...this.
5662 (handle_assignment): Rename to...
5663 (thread_jumps::handle_assignment): ...this.
5664 (fsm_find_control_statement_thread_paths): Rename to...
5665 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
5666 (find_jump_threads_backwards): Rename to...
5667 (thread_jumps::find_jump_threads_backwards): ...this.
5668 Initialize path local data.
5669 (pass_thread_jumps::execute): Call find_jump_threads_backwards
5670 from within thread_jumps class.
5671 (pass_early_thread_jumps::execute): Same.
5673 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5675 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
5677 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5679 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
5680 to canonical location.
5682 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5685 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
5687 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
5689 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
5691 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
5692 rcrt1.o%s/grcrt1.o%s for -static-pie.
5694 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5696 * i386.c (ix86_multiplication_cost, ix86_division_cost,
5697 ix86_shift_rotate_cost): Break out from ...
5698 (ix86_rtx_costs): ... here.
5699 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
5702 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5704 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
5705 when entry block was promoted unlikely.
5706 (estimate_bb_frequencies): Increase frequency scale.
5707 * profile-count.h (profile_count): Export precision info.
5709 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5711 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
5712 disturbin profile of entry block.
5714 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5716 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
5719 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5721 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
5722 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
5724 2017-11-17 Jeff Law <law@redhat.com>
5726 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
5727 from evrp_dom_walker class. Various methods moved into new class.
5728 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
5729 (evrp_range_analyzer::enter): New method.
5730 (evrp_range_analyzer::leave): New method.
5731 (evrp_dom_walker): Remove delegators no longer needed by this class.
5732 Replace vr_values data member with evrp_range_analyzer
5734 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
5735 method extracted from evrp_dom_walker::before_dom_children.
5736 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
5737 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
5739 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
5740 Add private copy constructor and move assignment operators.
5741 Privatize methods and class data where trivially possible.
5742 (evrp_dom_walker::cleanup): New function, extracted from
5743 execute_early_vrp. Simplify access to class data.
5745 * vr-values.h (get_output_for_vrp): Prototype.
5746 * vr-values.c (get_output_for_vrp): New function extracted from
5747 vrp_visit_assignment_or_call and extract_range_from_stmt.
5748 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
5750 2017-11-17 Luis Machado <luis.machado@linaro.org>
5752 * config/aarch64/aarch64.c
5753 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
5754 (qdf24xx_tunings) <autoprefetcher_model>: Set to
5755 tune_params::AUTOPREFETCHER_WEAK.
5757 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5760 * config/arm/arm.c (arm_valid_target_attribute_rec):
5761 Parse "arch=" and "+<ext>".
5762 (arm_valid_target_attribute_tree): Re-init global options.
5763 (arm_option_override): Make non-static.
5764 (arm_options_perform_arch_sanity_checks): Make errors fatal.
5765 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
5766 (__ARM_FEATURE_CRC32): Support undef.
5767 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
5768 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
5770 2017-11-17 David Malcolm <dmalcolm@redhat.com>
5772 * gdbinit.in (break-on-diagnostic): New command.
5774 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5776 * config/i386/i386.c (ix86_expand_epilogue): Change simple
5777 return to indirect jump for EH return if control-flow
5778 protection is enabled. Change explicit 'false' argument in
5779 pro_epilogue_adjust_stack with a value of
5781 * config/i386/i386.md (simple_return_indirect_internal):
5782 Remove SImode restriction to support 64-bit.
5784 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5786 * combine.c (added_notes_insn): New.
5787 (try_combine): Handle added_notes_insn like added_links_insn.
5788 Rewrite return value code.
5789 (distribute_notes): Set added_notes_insn to the earliest insn we added
5792 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5794 PR rtl-optimization/82621
5795 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
5796 dest of one of those SETs is unused.
5798 2017-11-17 Richard Biener <rguenther@suse.de>
5801 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
5802 * tree-pretty-print.c (dump_generic_node): Handle
5803 annot_expr_parallel_kind.
5804 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
5805 * gimplify.c (gimple_boolify): Likewise.
5807 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5809 * config.gcc (extra_headers): Add cet.h for x86 targets.
5810 * config/i386/cet.h: New file.
5811 * doc/install.texi: Add --enable-cet/--disable-cet.
5813 2017-11-17 Richard Biener <rguenther@suse.de>
5815 PR tree-optimization/83017
5816 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
5817 (gen_parallel_loop): Properly count iterations.
5818 (parallelize_loops): Handle loop->can_be_parallel independent
5819 of flag_loop_parallelize_all. Make static profitability test match
5821 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
5822 * invoke.texi (parloops-min-per-thread): Document.
5824 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
5826 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
5827 upstreaming review comments.
5829 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5831 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
5832 with fast unaligned access.
5833 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
5835 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
5837 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
5838 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
5840 2017-11-17 Richard Biener <rguenther@suse.de>
5842 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
5843 folding of references.
5845 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
5848 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
5849 of replacing call to strncmp with corresponding call to strcmp when
5852 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
5854 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
5855 option prefer-avx256 for skylake-avx512 configuration.
5856 * config/i386/i386.c (ix86_option_override_internal): Ditto.
5857 (get_builtin_code_for_version): Ditto.
5859 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5860 Monk Chiang <sh.chiang04@gmail.com>
5862 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
5863 (FIXED_REGISTERS): Reserve more register numbers.
5864 (CALL_USED_REGISTERS): Likewise.
5865 (REG_ALLOC_ORDER): Likewise.
5866 (REG_CLASS_CONTENTS): Likewise.
5867 (REGISTER_NAMES): Likewise.
5869 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5870 Kito Cheng <kito.cheng@gmail.com>
5872 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
5874 * config/nds32/iterators.md: Add vector mode iterators and attributes.
5876 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
5878 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
5879 parameter list for vec_splats.
5881 2017-11-16 Joseph Myers <joseph@codesourcery.com>
5883 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
5885 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
5887 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
5890 * doc/install.texi: Document incompatibility of Perl >=5.6.26
5891 with the required version of automake 1.11.6.
5893 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
5895 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
5897 (DU_C2_3_power9): Correct reservation combinations.
5898 (FP_DIV_power9, VEC_DIV_power9): New.
5899 (power9-alu): Split out rotate/shift...
5900 (power9-rot): ...to here, correct dispatch resource.
5901 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
5903 (power9-fp): Correct latency.
5904 (power9-sdiv): Add div/sqrt resource.
5905 (power9-ddiv): Correct latency, add div/sqrt resource.
5906 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
5907 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
5909 (power9-qpdiv, power9-qpmul): Adjust resource usage.
5911 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
5913 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
5914 switch statement mapping KF built-ins to TF built-ins if we don't
5915 have the proper ISA 3.0 assembler support.
5917 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5919 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
5920 (lower_emutls_function_body): Do not compute it.
5922 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5924 * ipa-split.c (split_bb_info): Turn time to sreal.
5925 (split_point): Likewise.
5926 (dump_split_point): Likewise.
5927 (fine_split_points): Likewise.
5928 (execute_split_functions): Only zero split_bbs; turn time to sreals.
5930 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5932 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
5935 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5937 * predict.c (combine_predictions_for_bb): Preserve zero predicted
5939 (expensive_function_p): Remove useless assert.
5940 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
5942 2017-11-16 Martin Sebor <msebor@redhat.com>
5944 PR tree-optimization/82588
5945 PR tree-optimization/82583
5946 * tree-vrp.c (check_array_ref): Handle flexible array members,
5947 string literals, and inner indices.
5948 (search_for_addr_array): Add detail to diagnostics.
5950 2017-11-16 Nathan Sidwell <nathan@acm.org>
5954 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
5955 (SET_DECL_ASSEMBLER_NAME): Forward to
5956 overwrite_decl_assembler_name.
5957 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
5958 (overwrite_decl_assembler_name): Declare.
5959 * tree.c (overwrite_decl_assembler_name): New.
5960 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
5961 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
5962 (LANG_HOOKS_INITIALIZER): Add it.
5963 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
5964 * langhooks.c (lhd_set_decl_assembler_name): Use
5965 SET_DECL_ASSEMBLER_NAME.
5966 (lhd_overwrite_decl_assembler_name): Default implementation.
5968 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5969 Jackson Woodruff <jackson.woodruff@arm.com>
5971 PR tree-optimization/71026
5972 * match.pd: Canonicalize constant multiplies in division.
5974 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5976 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
5977 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
5978 Remove OPT_fomit_frame_pointer entry.
5979 * common/config/arc/arc-common.c: Likewise.
5980 * common/config/arm/arm-common.c: Likewise.
5981 * common/config/avr/avr-common.c: Likewise.
5982 * common/config/c6x/c6x-common.c: Likewise.
5983 * common/config/cr16/cr16-common.c: Likewise.
5984 * common/config/cris/cris-common.c: Likewise.
5985 * common/config/epiphany/epiphany-common.c: Likewise.
5986 * common/config/fr30/fr30-common.c: Likewise.
5987 * common/config/frv/frv-common.c: Likewise.
5988 * common/config/ia64/ia64-common.c: Likewise.
5989 * common/config/iq2000/iq2000-common.c: Likewise.
5990 * common/config/lm32/lm32-common.c: Likewise.
5991 * common/config/m32r/m32r-common.c: Likewise.
5992 * common/config/mcore/mcore-common.c: Likewise.
5993 * common/config/microblaze/microblaze-common.c: Likewise.
5994 * common/config/mips/mips-common.c: Likewise.
5995 * common/config/mmix/mmix-common.c: Likewise.
5996 * common/config/mn10300/mn10300-common.c: Likewise.
5997 * common/config/nios2/nios2-common.c: Likewise.
5998 * common/config/pa/pa-common.c: Likewise.
5999 * common/config/pdp11/pdp11-common.c: Likewise.
6000 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
6001 * common/config/riscv/riscv-common.c: Likewise.
6002 * common/config/rs6000/rs6000-common.c: Likewise.
6003 * common/config/rx/rx-common.c: Likewise.
6004 * common/config/s390/s390-common.c: Likewise.
6005 * common/config/sh/sh-common.c: Likewise.
6006 * common/config/sparc/sparc-common.c: Likewise.
6007 * common/config/tilegx/tilegx-common.c: Likewise.
6008 * common/config/tilepro/tilepro-common.c: Likewise.
6009 * common/config/v850/v850-common.c: Likewise.
6010 * common/config/visium/visium-common.c: Likewise.
6011 * common/config/xstormy16/xstormy16-common.c: Likewise.
6012 * common/config/xtensa/xtensa-common.c: Likewise.
6013 * invoke.texi (-fomit-frame-pointer): Update documentation.
6015 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6017 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
6019 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6021 * cfg.c (scale_bbs_frequencies_int,
6022 cale_bbs_frequencies_gcov_type): Remove.
6023 * cfg.h (scale_bbs_frequencies_int,
6024 cale_bbs_frequencies_gcov_type): Remove.
6026 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6028 * tree-ssa-loop-manip.c
6029 (scale_dominated_blocks_in_loop): Update to profile counts.
6030 (tree_transform_and_unroll_loop): Likewise.
6032 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6034 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
6035 scale_bbs_frequencies_int.
6037 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6039 * final.c (compute_alignments): Use counts rather than frequencies.
6041 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6043 * cfgloopanal.c: Include sreal.h
6044 (average_num_loop_insns): Use counts and sreal for accounting.
6046 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6048 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
6051 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6052 Kito Cheng <kito.cheng@gmail.com>
6054 * config/nds32/constraints.md: Provide more constraints.
6055 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
6056 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
6057 support constraints usage.
6059 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6061 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
6063 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6064 Kito Cheng <kito.cheng@gmail.com>
6066 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
6067 * config/nds32/nds32.opt: Refine the layout.
6068 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
6069 TARGET_EXT_STRING): Support new options.
6070 * config/nds32/nds32.h: Likewise.
6071 * config/nds32/nds32.md: Likewise.
6072 * config/nds32/nds32-predicates.c: Likewise.
6073 * config/nds32/constraints.md: Likewise.
6074 * common/config/nds32/nds32-common.c: Likewise.
6076 2017-11-16 Julia Koval <julia.koval@intel.com>
6079 * config/i386/gfniintrin.h: Add sse check.
6080 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
6082 2017-11-16 Julia Koval <julia.koval@intel.com>
6084 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
6085 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
6086 (ix86_handle_option): Handle -mavx512vbmi2.
6087 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
6088 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
6089 * config/i386/i386-c.c (__AVX512VBMI2__): New.
6090 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
6091 (ix86_valid_target_attribute_inner_p): Ditto.
6092 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
6093 * config/i386/i386.opt (mavx512vbmi2): New option.
6094 * doc/invoke.texi: Add new option.
6096 2017-11-16 Julia Koval <julia.koval@intel.com>
6098 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
6099 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
6100 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
6101 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
6102 _mm512_gf2p8mul_epi8): New intrinsics.
6103 * config/i386/i386-builtin-types.def
6104 (V64QI_FTYPE_V64QI_V64QI): New type.
6105 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
6106 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
6107 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
6108 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
6109 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
6110 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
6112 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
6114 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
6115 explicitly as a stream of bytes.
6117 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6119 * config/rs6000/altivec.h (vec_xst_be): New #define.
6120 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
6121 and externalize from *altivec_vperm_<mode>_internal.
6122 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
6124 (XL_BE_V8HI): Likewise.
6125 (XL_BE_V4SI): Likewise.
6126 (XL_BE_V4SI): Likewise.
6127 (XL_BE_V2DI): Likewise.
6128 (XL_BE_V4SF): Likewise.
6129 (XL_BE_V2DF): Likewise.
6130 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
6131 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
6132 all array entries with these keys: VSX_BUILTIN_VEC_XL,
6133 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
6134 VSX_BUILTIN_VEC_XST_BE.
6135 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
6136 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
6138 (altivec_init_builtins): Replace conditional calls to def_builtin
6139 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
6140 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
6141 with unconditional calls. Remove calls to def_builtin for
6142 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
6143 __builtin_vec_xst_be.
6144 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
6145 to define_expand, and add alternate RTL generation for P8.
6146 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
6147 vsx_ld_elemrev_v8hi.
6148 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
6149 add alternate RTL generation for P8.
6150 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
6151 vsx_ld_elemrev_v16qi.
6152 (vsx_st_elemrev_v8hi): Convert define_insn
6153 to define_expand, and add alternate RTL generation for P8.
6154 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
6155 vsx_st_elemrev_v8hi.
6156 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
6157 add alternate RTL generation for P8.
6158 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
6159 vsx_st_elemrev_v16qi.
6161 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
6164 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
6165 TARGET_AVX512ER check.
6166 (ix86_option_override_internal): Set MASK_VZEROUPPER if
6167 neither -mzeroupper nor -mno-zeroupper is used and
6168 TARGET_EMIT_VZEROUPPER is set.
6169 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
6170 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
6172 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6174 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6175 folding of vector compares.
6176 (fold_build_vec_cmp): New helper function.
6177 (fold_compare_helper): New helper function.
6178 (builtin_function_type): Add compare builtins to the list of functions
6179 having unsigned arguments. Cosmetic updates to comment indentation.
6180 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
6181 the not+eq combination.
6183 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6185 PR tree-optimization/82726
6186 PR tree-optimization/70754
6187 * tree-predcom.c (order_drefs_by_pos): New function.
6188 (combine_chains): Move code setting has_max_use_after to...
6189 (try_combine_chains): ...here. New parameter. Sort combined chains
6190 according to position information.
6191 (tree_predictive_commoning_loop): Update call to above function.
6192 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
6194 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6196 PR tree-optimization/82726
6198 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6200 PR tree-optimization/70754
6201 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6202 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6203 combined stmt before it if not NULL.
6204 (combine_chains): Process refs reversely and compute dominance point
6208 2017-02-23 Bin Cheng <bin.cheng@arm.com>
6210 PR tree-optimization/79663
6211 * tree-predcom.c (combine_chains): Process refs in reverse order
6212 only for ZERO length chains, and add explaining comment.
6214 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6216 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
6217 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
6218 * doc/invoke.texi (ARM Options): Add armv8.3-a.
6220 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6222 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
6224 2017-11-15 Martin Liska <mliska@suse.cz>
6226 * tree-cfg.c (pass_warn_function_return::execute):
6227 Compare warn_return_type for greater than zero.
6229 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
6233 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
6234 to return true on Xeon and not on Xeon Phi.
6235 (ix86_check_avx256_register): Changed to ...
6236 (ix86_check_avx_upper_register): ... this. Add extra check for
6237 VALID_AVX512F_REG_OR_XI_MODE.
6238 (ix86_avx_u128_mode_needed): Changed
6239 ix86_check_avx256_register to ix86_check_avx_upper_register.
6240 (ix86_check_avx256_stores): Changed to ...
6241 (ix86_check_avx_upper_stores): ... this. Changed
6242 ix86_check_avx256_register to ix86_check_avx_upper_register.
6243 (ix86_avx_u128_mode_after): Changed
6244 avx_reg256_found to avx_upper_reg_found. Changed
6245 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
6246 (ix86_avx_u128_mode_entry): Changed
6247 ix86_check_avx256_register to ix86_check_avx_upper_register.
6248 (ix86_avx_u128_mode_exit): Ditto.
6249 * config/i386/i386.h: (host_detect_local_cpu): New define.
6251 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6253 * config/arm/xgene1.md (xgene1): Split into automatons
6254 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
6255 (xgene1_f_load): Adjust reservations and/or types.
6256 (xgene1_f_store): Likewise.
6257 (xgene1_load_pair): Likewise.
6258 (xgene1_store_pair): Likewise.
6259 (xgene1_fp_load1): Likewise.
6260 (xgene1_load1): Likewise.
6261 (xgene1_store1): Likewise.
6262 (xgene1_move): Likewise.
6263 (xgene1_alu): Likewise.
6264 (xgene1_simd): Likewise.
6265 (xgene1_bfm): Likewise.
6266 (xgene1_neon_load1): Likewise.
6267 (xgene1_neon_store1): Likewise.
6268 (xgene1_neon_logic): Likewise.
6269 (xgene1_neon_st1): Likewise.
6270 (xgene1_neon_ld1r): Likewise.
6271 (xgene1_alu_cond): Added.
6272 (xgene1_shift_reg): Likwise.
6273 (xgene1_bfx): Likewise.
6274 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
6276 2017-11-15 Jakub Jelinek <jakub@redhat.com>
6279 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
6281 (can_widen_mult_without_libcall): New function.
6282 (expand_mul_overflow): If only checking unsigned mul overflow,
6283 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
6284 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
6285 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
6286 (expand_DIVMOD): Formatting fix.
6287 * expmed.h (expand_mult): Add NO_LIBCALL argument.
6288 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
6289 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
6291 PR tree-optimization/82977
6292 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
6293 constructed temporary to strlen_to_stridx.put.
6295 2017-11-15 Martin Liska <mliska@suse.cz>
6297 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
6298 * configure: Regenerate.
6300 2017-11-15 Martin Liska <mliska@suse.cz>
6303 * config/sh/sh-mem.cc: Use proper probability for
6306 2017-11-14 Jeff Law <law@redhat.com>
6308 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
6309 the red zone for stack_clash_protection_final_dynamic_probe targets
6310 when the total dynamic stack size is zero bytes.
6312 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
6313 blocks is post order.
6315 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
6317 * dumpfile.h (TDF_COMPARE_DEBUG): New.
6318 * final.c (rest_of_clean_state): Set it for the
6319 -fcompare-debug dump.
6320 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
6321 class when TDF_COMPARE_DEBUG is set.
6323 * dwarf2out.c (gen_producer_string): Discard
6326 2017-11-15 Joseph Myers <joseph@codesourcery.com>
6329 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
6330 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
6331 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
6332 (__TGMATH_REAL_2_3): Remove macros.
6333 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
6334 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
6336 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
6338 (remquo): Define using __TGMATH_REAL_3.
6340 2017-11-14 Jeff Law <law@redhat.com>
6342 * vr-values.c: New file with contents extracted from tree-vrp.c.
6343 * Makefile.in (OBJS): Add vr-values.o
6344 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
6345 (set_value_range, set_and_canonicalize_value_range): Likewise.
6346 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
6347 (value_range_constant_singleton, symbolic_range_p): Likewise.
6348 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
6349 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
6350 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
6351 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
6352 (find_case_label_range, find_case_label_index): Likewise.
6353 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
6354 (range_int_cst_singleton_p, value_inside_range): Likewise.
6355 (get_single_symbol): Likewise.
6356 (switch_update): Move structure definition here.
6357 (to_remove_edges, to_update_switch_stmts): Provide externs.
6358 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
6359 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
6360 (vrp_val_is_min, set_value_range): Likewise.
6361 (set_and_canonicalize_value_range, copy_value_range): Likewise.
6362 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
6363 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
6364 (range_is_nonnull, range_int_cst_p): Likewwise.
6365 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
6366 (get_single_symbol, operand_less_p): Likewise
6367 (compare_values_warnv, compare_values): Likewise.
6368 (value_inside_range, value_range_constant_singleton): Likewise.
6369 (zero_nonzero_bitgs_from_vr): Likewise.
6370 (extract_range_from_binary_expr_1): Likewise.
6371 (overflow_comparison_p): Likewise.
6372 (to_remove_edges, to_update_switch_stmts): Likewise.
6373 (find_case_label-index, find_case_label_range): Likewise.
6374 (switch_update, set_value_range_to_nonnegative): Remove.
6375 (set_value_range_to_truthvalue): Likewise.
6376 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
6377 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
6378 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
6379 (find_case_label_ranges, test_for_singularity): Likewise.
6380 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
6381 (x_vr_values): Move to its remaining use site.
6383 2017-11-10 Jeff Law <law@redhat.com>
6385 * vr-values.h (VR_INITIALIZER): Move #define here.
6386 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
6387 * Makefile.in (OBJS): Add tree-evrp.o
6388 * tree-vrp.h (assert_info): Move structure definition here.
6389 (set_value_range_to_varying): Prototype.
6390 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6391 (infer_value_range, register_edge_assert_for): Likewise.
6392 (stmt_interesting_for_vrp): Likewise.
6393 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
6394 (set_value_range_to_varying): No longer static.
6395 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6396 (infer_value_range, register_edge_assert_for): Likewise.
6398 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6400 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
6401 generate the XXBRD instruction.
6403 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
6404 (rs6000_builtin_type_compatible): Treat _Float128 and long double
6405 as being compatible if -mabi=ieeelongdouble.
6406 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
6407 to setup float128 built-ins with hardware support.
6408 (BU_FLOAT128_HW_2): Likewise.
6409 (BU_FLOAT128_HW_3): Likewise.
6410 (BU_FLOAT128_HW_VSX_1): Likewise.
6411 (BU_FLOAT128_HW_VSX_2): Likewise.
6412 (scalar_extract_expq): Change float128 built-in functions to
6413 accommodate having both KFmode and TFmode functions. Use the
6414 KFmode variant as the default.
6415 (scalar_extract_sigq): Likewise.
6416 (scalar_test_neg_qp): Likewise.
6417 (scalar_insert_exp_q): Likewise.
6418 (scalar_insert_exp_qp): Likewise.
6419 (scalar_test_data_class_qp): Likewise.
6420 (sqrtf128_round_to_odd): Delete processing the round to odd
6421 built-in functions as special built-in functions, and define them
6422 as float128 built-ins. Use the KFmode variant as the default.
6423 (truncf128_round_to_odd): Likewise.
6424 (addf128_round_to_odd): Likewise.
6425 (subf128_round_to_odd): Likewise.
6426 (mulf128_round_to_odd): Likewise.
6427 (divf128_round_to_odd): Likewise.
6428 (fmaf128_round_to_odd): Likewise.
6429 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6430 support for KFmode and TFmode xststdcqp calls.
6431 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
6432 point, switch the built-in handlers for the get/set float128
6433 exponent, get float128 mantissa, float128 test built-ins, and the
6434 float128 round to odd built-in functions. Eliminate creating the
6435 float128 round to odd built-in functions as special built-ins.
6436 (rs6000_init_builtins): Eliminate special creation of the float128
6437 round to odd built-in functions.
6438 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
6439 function insns to support both TFmode and KFmode variants.
6440 (xsxsigqp_<mode>): Likewise.
6441 (xsiexpqpf_<mode>): Likewise.
6442 (xsiexpqp_<mode>): Likewise.
6443 (xststdcqp_<mode>): Likewise.
6444 (xststdcnegqp_<mode>): Likewise.
6445 (xststdcqp_<mode>): Likewise.
6447 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6449 * tree-ssa-threadupdate.c (compute_path_counts): Remove
6450 unused path_in_freq_ptr parameter.
6451 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
6453 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6455 * ipa-inline.c (edge_badness): Dump sreal frequency.
6456 (compute_inlined_call_time): Match natural implementaiton ...
6457 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
6458 forgotten division by CGRAPH_FREQ_BASE.
6460 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6462 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
6463 Solaris 11. Update comment.
6464 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
6466 * config/sol2.h (STARTFILE_SPEC): Likewise.
6467 * configure: Regenerate.
6469 2017-11-14 Carl Love <cel@us.ibm.com>
6471 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
6472 le_ and be_ prefixes to swap* variables. Remove
6473 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
6476 2017-11-14 Jason Merrill <jason@redhat.com>
6478 Support GTY((cache)) on hash_map.
6479 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
6480 (ggc_cache_remove): Override it instead of ggc_mx.
6481 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
6482 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
6483 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
6484 (simple_cache_map_traits): Override maybe_mx.
6485 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
6486 (hash_map): Friend gt_cleare_cache.
6487 (gt_cleare_cache): New.
6488 * tree.h (tree_cache_traits): New hash_map traits class.
6489 (tree_cache_map): New typedef.
6491 2017-11-14 Richard Biener <rguenther@suse.de>
6493 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
6494 paramter and handling.
6495 (cleanup_control_flow_bb): Likewise.
6496 (cleanup_control_flow_pre): New helper performing a DFS walk
6497 to call cleanup_control_flow_bb in PRE order.
6498 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
6499 via cleanup_control_flow_pre.
6501 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
6503 * config/aarch64/aarch64-simd.md
6504 (aarch64_simd_bsl<mode>_internal): Remove DImode.
6505 (*aarch64_simd_bsl<mode>_alt): Likewise.
6506 (aarch64_simd_bsldi_internal): New.
6507 (aarch64_simd_bsldi_alt): Likewise.
6509 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6511 * tracer.c (better_p): Do not compare frequencies.
6512 * reg-stack.c (better_edge): Likewise.
6513 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
6516 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6518 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
6519 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
6520 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
6521 * lto-streamer-in.c (input_function): Use update_max_bb_count.
6522 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
6523 * predict.c (maybe_hot_frequency_p): Inline to ...
6524 (maybe_hot_count_p): ... here; rewrite to counts.
6525 (counts_to_freqs): Rename to ...
6526 (update_max_bb_count): ... this one.
6527 (expensive_function_p): Use counts.
6528 (estimate_bb_frequencies): Update.
6529 (rebuild_frequencies): Update.
6530 * predict.h (counts_to_freqs): Rename to ...
6531 (update_max_bb_count): ... this one.
6532 * profile.c (compute_branch_probabilities): Add debug info
6533 * tree-inline.c (expand_call_inline): Update debug info.
6534 (optimize_inline_calls): Use update_max_bb_count..
6535 (tree_function_versioning): Use update_max_bb_count..
6536 * value-prof.c (gimple_value_profile_transformations):
6537 Do not use update_max_bb_count.
6539 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6541 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
6542 always use frequencies.
6544 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6546 * bb-reorder.c: Remove frequencies from comments.
6547 (better_edge_p): Use profile counts.
6548 (find_traces): Dump profile counts.
6549 (rotate_loop): Use profile counts.
6550 (find_traces_1_round): Likewise.
6551 (connect_better_edge_p): Use counts instead of probabilities for
6553 (copy_bb_p): Drop early check for non-0 frequency.
6554 (sanitize_hot_paths): Update comments.
6556 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6558 * ipa-split.c (struct split_point): Add count.
6559 (consider_split): Do not compute incoming frequency; compute incoming
6560 count and store it to split_point.
6561 (split_function): Set count of the call to split part correctly.
6563 2017-11-13 Carl Love <cel@us.ibm.com>
6565 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
6567 2017-11-13 Tom Tromey <tom@tromey.com>
6569 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
6571 2017-11-13 Carl Love <cel@us.ibm.com>
6573 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
6574 Add support for builtins:
6575 unsigned int vec_first_{,miss}_match_{,or_eos}index,
6576 vector {un,}signed {char,int,short},
6577 vector {un,}signed {char,int,short}) arguments.
6578 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
6579 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
6580 Add BU_P9V_AV_2 expansions for the builtins.
6581 * config/rs6000/altivec.h (vec_first_match_index,
6582 vec_first_mismatch_index, vec_first_match_or_eos_index,
6583 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
6584 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
6585 new extern declaration.
6586 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
6587 * config/rs6000/vsx.md (first_match_index_<mode>,
6588 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
6589 first_mismatch_or_eos_index_<mode>): Add define expand.
6590 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
6591 * doc/extend.texi: Update the built-in documenation file for the new
6594 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
6596 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
6597 into the min/max operations for _Float<N> and _Float<N>X types.
6599 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6602 * dwarf2out.c (do_eh_frame): New static variable.
6603 (dwarf2out_begin_prologue): Set it.
6604 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
6606 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6608 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
6610 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
6612 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6615 * debug.h (dwarf2out_do_eh_frame): Declare.
6616 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
6617 (dwarf2out_do_frame): Use it.
6618 (dwarf2out_do_cfi_asm): Likewise.
6619 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
6620 (dwarf2out_assembly_start): Likewise.
6621 (dwarf2out_begin_prologue): Fix comment.
6622 * toplev.c (compile_file): Always call dwarf2out_frame_finish
6623 if the target needs either debug or unwind DWARF2 info.
6624 * lto-opts.c (lto_write_options): Do not save -fexceptions,
6625 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
6626 -ftrapping-math, -ftrapv and -fwrapv.
6628 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6630 * cgraph.c (cgraph_edge::sreal_frequency): New function.
6631 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
6632 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
6633 (estimate_edge_size_and_time): Likewise.
6634 (ipa_merge_fn_summary_after_inlining): Likewise.
6635 * ipa-inline.c (cgraph_freq_base_rec): Remove.
6636 (compute_uninlined_call_time): Use sreal_frequency.
6637 (compute_inlined_call_time): Likewise.
6638 (ipa_inline): Do not initialize cgraph_freq_base_rec.
6639 * profile-count.c: Include sreal.h.
6640 (profile_count::to_sreal_scale): New.
6641 * profile-count.h: Forward declare sreal.
6642 (profile_count::to_sreal_scale): Declare.
6644 2017-11-13 Nathan Sidwell <nathan@acm.org>
6646 * diagnostic.c (maybe_line_and_column): New.
6647 (diagnostic_get_location_text): Use it.
6648 (diagnostic_report_current_module): Likewise.
6649 (test_diagnostic_get_location_text): Add tests.
6651 2017-11-13 Luis Machado <luis.machado@linaro.org>
6653 * doc/md.texi (Specifying processor pipeline description): Fix
6654 incorrect latency for the div instruction example.
6656 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6658 PR tree-optimization/78821
6659 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
6660 that bit_not_p is the same.
6661 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
6662 (split_group): Count precisely bit_not_p bits in each statement.
6663 (invert_op): New function.
6664 (imm_store_chain_info::output_merged_store): Use invert_op to
6665 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
6666 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
6668 2017-11-13 Martin Liska <mliska@suse.cz>
6670 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
6671 (struct source_info): Likewise.
6672 (add_branch_counts): Likewise.
6673 (add_line_counts): Likewise.
6674 (function_summary): Likewise.
6675 (output_intermediate_line): Likewise.
6676 (generate_results): Likewise.
6678 2017-11-13 Martin Liska <mliska@suse.cz>
6680 * gcov.c (struct block_info): Remove typedef for block_t.
6681 (struct line_info): Likewise.
6682 (line_info::has_block): Likewise.
6683 (EXIT_BLOCK): Likewise.
6684 (unblock): Likewise.
6685 (circuit): Likewise.
6686 (get_cycles_count): Likewise.
6687 (process_file): Likewise.
6688 (read_graph_file): Likewise.
6689 (solve_flow_graph): Likewise.
6690 (find_exception_blocks): Likewise.
6691 (add_line_counts): Likewise.
6692 (accumulate_line_info): Likewise.
6693 (output_line_details): Likewise.
6695 2017-11-13 Martin Liska <mliska@suse.cz>
6697 * gcov.c (struct arc_info): Remove typedef for arc_t.
6698 (struct line_info): Likewise.
6699 (add_branch_counts): Likewise.
6700 (output_branch_count): Likewise.
6701 (function_info::~function_info): Likewise.
6702 (circuit): Likewise.
6703 (output_intermediate_line): Likewise.
6704 (read_graph_file): Likewise.
6705 (solve_flow_graph): Likewise.
6706 (find_exception_blocks): Likewise.
6707 (add_line_counts): Likewise.
6708 (accumulate_line_info): Likewise.
6709 (output_line_details): Likewise.
6710 (output_function_details): Likewise.
6712 2017-11-13 Martin Liska <mliska@suse.cz>
6714 * gcov.c (struct function_info): Remove typedef for function_t.
6715 (struct source_info): Likewise.
6716 (source_info::get_functions_at_location): Likewise.
6717 (solve_flow_graph): Likewise.
6718 (find_exception_blocks): Likewise.
6719 (add_line_counts): Likewise.
6720 (output_intermediate_file): Likewise.
6721 (process_file): Likewise.
6722 (generate_results): Likewise.
6723 (release_structures): Likewise.
6724 (read_graph_file): Likewise.
6725 (read_count_file): Likewise.
6726 (accumulate_line_counts): Likewise.
6727 (output_lines): Likewise.
6729 2017-11-13 Martin Liska <mliska@suse.cz>
6731 * gcov.c (function_info::function_info): Remove num_counts
6732 and add vector<gcov_type>.
6733 (function_info::~function_info): Use the vector.
6734 (process_file): Likewise.
6735 (read_graph_file): Likewise.
6736 (read_count_file): Likewise.
6737 (solve_flow_graph): Likewise.
6739 2017-11-13 Martin Liska <mliska@suse.cz>
6741 * gcov.c (function_info::is_artificial): New function.
6742 (process_file): Erase all artificial early.
6743 (generate_results): Skip as all artificial are already
6746 2017-11-13 Martin Liska <mliska@suse.cz>
6748 * gcov.c (read_graph_file): Store to global vector of functions.
6749 (read_count_file): Iterate the vector.
6750 (process_file): Likewise.
6751 (generate_results): Likewise.
6752 (release_structures): Likewise.
6754 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6756 PR tree-optimization/82954
6757 * gimple-ssa-store-merging.c
6758 (imm_store_chain_info::coalesce_immediate_stores): If
6759 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
6761 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
6763 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6764 Upddate call to ENDIAN_LANE_N.
6765 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
6766 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
6767 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
6768 and use aarch64_endian_lane_rtx.
6769 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
6771 2017-11-12 Tom de Vries <tom@codesourcery.com>
6773 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
6775 2017-11-12 Tom de Vries <tom@codesourcery.com>
6777 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
6779 2017-11-12 Tom de Vries <tom@codesourcery.com>
6781 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
6783 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6784 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6785 * defaults.h (ASM_OUTPUT_LABELREF): Same.
6787 2017-11-11 Martin Sebor <msebor@redhat.com>
6790 * doc/extend.texi (attribute nonstring): Remove spurious argument.
6793 * prefic.c (translate_name): Replace strncpy with memcpy to
6794 avoid -Wstringop-truncation.
6796 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6798 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
6800 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6802 * predict.c (maybe_hot_frequency_p): Do not use cfun.
6804 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6806 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
6809 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
6811 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
6812 enable generating XXBRH if the value is in a vector register.
6813 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
6814 value is in a vector register.
6815 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
6816 register to register bswap64's instead of doing the GPR sequence
6817 used on previous machines.
6818 (bswapdi2_xxbrd): New insn.
6819 (bswapdi2_reg): Disallow on ISA 3.0.
6820 (register to register bswap64 splitter): Do not split the insn on
6821 ISA 3.0 systems that use XXBRD.
6823 2017-11-10 Martin Sebor <msebor@redhat.com>
6826 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
6828 (find_subframework_file): Same.
6830 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6832 * auto-profile.c (afdo_indirect_call): Drop frequency.
6833 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
6834 (cgraph_node::create_edge): Drop frequency argument.
6835 (cgraph_node::create_indirect_edge): Drop frequency argument.
6836 (cgraph_edge::make_speculative): Drop frequency arguments.
6837 (cgraph_edge::resolve_speculation): Do not update frequencies
6838 (cgraph_edge::dump_edge_flags): Do not dump frequency.
6839 (cgraph_node::dump): Check consistency in IPA mode.
6840 (cgraph_edge::maybe_hot_p): Use IPA counter.
6841 (cgraph_edge::verify_count_and_frequency): Rename to ...
6842 (cgraph_edge::verify_count): ... this one; drop frequency checking.
6843 (cgraph_node::verify_node): Update.
6844 * cgraph.h (struct cgraph_edge): Drop frequency.
6845 (cgraph_edge::frequency): New function.
6846 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
6848 (cgraph_edge::rebuild_edges): Likewise.
6849 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
6850 (duplicate_thunk_for_node): Do not pass frequency.
6851 (cgraph_node::create_clone): Scale only counts.
6852 (cgraph_node::create_virtual_clone): Do not pass frequency.
6853 (cgraph_node::create_edge_including_clones): Do not pass frequency.
6854 (cgraph_node::create_version_clone): Do not pass frequency.
6855 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
6856 (cgraph_node::expand_thunk): Do not pass frequency.
6857 (cgraph_node::create_wrapper): Do not pass frequency.
6858 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
6860 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
6861 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
6862 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
6863 (gather_caller_stats): Use frequency function.
6864 (ipcp_cloning_candidate_p): Use frequency function.
6865 (ipcp_propagate_stage): Use frequency function.
6866 (get_info_about_necessary_edges): Use frequency function.
6867 (update_profiling_info): Update only IPA profile.
6868 (update_specialized_profile): Use frequency functoin.
6869 (perhaps_add_new_callers): Update only IPA profile.
6870 * ipa-devirt.c (ipa_devirt): Use IPA profile.
6871 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
6872 (dump_ipa_call_summary): Use frequency function.
6873 (estimate_edge_size_and_time): Use frequency function.
6874 (ipa_merge_fn_summary_after_inlining): Use frequency function.
6875 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
6876 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
6877 (update_noncloned_counts): ... ths one; scale counts only.
6878 (clone_inlined_nodes): Do not scale frequency.
6879 (inline_call): Do not pass frequency.
6880 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
6881 (compute_inlined_call_time): Use IPA profile.
6882 (want_inline_small_function_p): Use IPA profile.
6883 (want_inline_self_recursive_call_p): Use IPA profile.
6884 (edge_badness): Use IPA profile.
6885 (lookup_recursive_calls): Use IPA profile.
6886 (recursive_inlining): Do not pass frequency.
6887 (resolve_noninline_speculation): Do not update frequency.
6888 (inline_small_functions): Collect max of IPA profile.
6889 (dump_overall_stats): Dump IPA porfile.
6890 (dump_inline_stats): Dump IPA porfile.
6891 (ipa_inline): Collect IPA stats.
6892 * ipa-inline.h (clone_inlined_nodes): Update prototype.
6893 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
6894 (ipa_propagate_frequency): Use frequency function.
6895 (ipa_profile): Cleanup.
6896 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
6897 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
6898 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
6899 (input_node): Do not stream frequency.
6900 (input_edge): Do not stream frequency.
6901 (merge_profile_summaries): Scale only IPA profiles.
6902 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
6903 * predict.c (drop_profile): Do not recompute frequency.
6904 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
6905 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
6906 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
6907 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
6908 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
6909 * tree-inline.c (copy_bb): Do not scale frequency.
6910 (expand_call_inline): Do not scale frequency.
6911 (tree_function_versioning): Do not scale frequency.
6912 * ubsan.c (ubsan_create_edge): Do not pass frequency.
6914 2017-11-10 Julia Koval <julia.koval@intel.com>
6916 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
6917 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
6918 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
6919 (_mm256_maskz_gf2p8affine_epi64_epi8)
6920 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
6921 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
6922 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
6923 (__builtin_ia32_vgf2p8affineqb_v32qi)
6924 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
6925 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
6927 2017-11-10 Tamar Christina <tamar.christina@arm.com>
6931 (arm_option_override): Refactor.
6932 (arm_option_reconfigure_globals): New.
6933 (arm_options_perform_arch_sanity_checks): New.
6934 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
6936 (arm_options_perform_arch_sanity_checks): Likewise
6938 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
6940 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
6941 (power9-qpmul): New.
6942 * rs6000/rs6000.md ("type" attr): Add qmul.
6943 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
6944 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
6945 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
6947 2017-11-10 Martin Sebor <msebor@redhat.com>
6950 * builtins.c (compute_objsize): Handle arrays that
6951 compute_builtin_object_size likes to fail for. Make extern.
6952 * builtins.h (compute_objsize): Declare.
6953 (check_strncpy_sizes): New function.
6954 (expand_builtin_strncpy): Call check_strncpy_sizes.
6955 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
6956 -Wstringop-truncation.
6957 (gimple_fold_builtin_strncat): Same.
6958 * gimple.c (gimple_build_call_from_tree): Set call location.
6959 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
6960 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
6961 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
6962 (handle_builtin_strlen): Use strlen_to_stridx.
6963 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
6965 Use strlen_to_stridx.
6966 (pass_strlen::execute): Release strlen_to_stridx.
6967 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
6968 (-Wstringop-truncation): Document new option.
6970 2017-11-10 Martin Liska <mliska@suse.cz>
6972 PR gcov-profile/82702
6973 * gcov.c (main): Handle intermediate files in a different
6975 (get_gcov_intermediate_filename): New function.
6976 (output_gcov_file): Remove support of intermediate files.
6977 (generate_results): Allocate intermediate file.
6978 (release_structures): Clean-up properly fn_end.
6979 (output_intermediate_file): Start iterating with line 1.
6981 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6983 PR tree-optimization/82929
6984 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6985 ops_swapped_p non-static data member.
6986 (store_immediate_info::store_immediate_info): Clear it.
6987 (imm_store_chain_info::coalesce_immediate_stores): If swapping
6988 ops set ops_swapped_p.
6989 (count_multiple_uses): Handle ops_swapped_p.
6991 2017-11-10 Martin Liska <mliska@suse.cz>
6993 * coverage.c (coverage_init): Stream information about
6994 support of has_unexecuted_blocks.
6995 * doc/gcov.texi: Document that.
6996 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
6997 * gcov.c (read_graph_file): Likewise.
6998 (output_line_beginning): Fix a small issue with
7001 2017-11-10 Bin Cheng <bin.cheng@arm.com>
7003 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
7004 reference of trivial component.
7006 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7009 * gimple-ssa-store-merging.c
7010 (pass_store_merging::terminate_all_aliasing_chains): For
7011 gimple_store_p stmts also call refs_output_dependent_p.
7013 PR rtl-optimization/82913
7014 * compare-elim.c (try_merge_compare): Punt if def_insn is not
7017 2017-11-09 Jeff Law <law@redhat.com>
7019 * vr-values.h: New file with vr_values class.
7020 * tree-vrp.c: Include vr-values.h
7021 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
7022 data objects into the vr_values class.
7023 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
7024 (get_value_range): Make it a member function within vr_values class.
7025 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
7026 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
7027 (op_with_constant_singleton_value_range): Likewise.
7028 (extract_range_for_var_from_comparison_expr): Likewise.
7029 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
7030 (extract_range_from_binary_expr): Likewise.
7031 (extract_range_from_unary_expr): Likewise.
7032 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
7033 (check_for_binary_op_overflow, extract_range_basic): Likewise.
7034 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
7035 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
7036 (compare_name_with_value, compare_names): Likewise.
7037 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
7038 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
7039 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
7040 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
7041 (extract_range_from_phi_node): Likewise.
7042 (simplify_truth_ops_using_ranges): Likewise.
7043 (simplify_div_or_mod_using_ranges): Likewise.
7044 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
7045 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
7046 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
7047 (simplify_float_conversion_using_ranges): Likewise.
7048 (simplify_internal_call_using_ranges): Likewise.
7049 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
7050 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
7051 poitner via x_vr_values for calls into gimple folder.
7052 (vrp_initialize_lattice): Make this the vr_values ctor.
7053 (vrp_free_lattice): Make this the vr_values dtor.
7054 (set_vr_value): New function.
7055 (class vrp_prop): Add vr_values data member. Add various member
7056 functions as well as member functions that delegate to vr_values.
7057 (check_array_ref): Make a member function within vrp_prop class.
7058 (search_for_addr_array, vrp_initialize): Likewise.
7059 (vrp_finalize): Likewise. Revamp to avoid direct access to
7060 vr_value, values_propagated, etc.
7061 (check_array_bounds): Extract vrp_prop class instance pointer from
7062 walk info structure. Use it to call member functions.
7063 (check_all_array_refs): Make a member function within vrp_prop class.
7064 Smuggle class instance pointer via walk info structure.
7065 (x_vr_values): New local static.
7066 (vrp_valueize): Use x_vr_values to get class instance.
7067 (vr_valueize_1): Likewise.
7068 (class vrp_folder): Add vr_values data member. Add various member
7069 functions as well as member functions that delegate to vr_values.
7070 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
7071 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
7072 class instance from vr_values. Use it to call member functions.
7073 (vrp_dom_walker): Add vr_values data member.
7074 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
7075 instance via x_vr_values.
7076 (identify_jump_threads): Accept vr_values as argument. Store
7077 it into the walker structure.
7078 (evrp_dom_walker): Add vr_values class data member. Add various
7080 (evrp_dom_walker::try_find_new_range): Use vr_values data
7081 member to access the memory allocator.
7082 (evrp_dom_walker::before_dom_children): Store vr_values class
7083 instance into the vrp_folder class.
7084 (evrp_dom_walker::push_value_range): Rework to avoid direct
7085 access to num_vr_values and vr_value.
7086 (evrp_dom_walker::pop_value_range): Likewise.
7087 (execute_early_vrp): Remove call to vrp_initialize_lattice.
7088 Use vr_values to get to dump_all_value_ranges member function.
7089 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
7090 and simplify_cond_using_ranges_2 via vrp_prop class instance.
7091 Pass vr_values class instance down to identify_jump_threads.
7092 Remove call to vrp_free_lattice.
7093 (debug_all_value_ranges): Remove.
7095 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
7096 (vrp_folder): Likewise.
7098 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
7099 Get it from the existing bitmap instead.
7100 (vrp_intersect_ranges_1): Likewise.
7102 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7104 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7106 (store_immediate_info::store_immediate_info): Add bitnotp argument,
7107 set bit_not_p to it.
7108 (imm_store_chain_info::coalesce_immediate_stores): Break group
7109 if bit_not_p is different.
7110 (count_multiple_uses, split_group,
7111 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
7112 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
7113 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
7114 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
7115 to store_immediate_info ctor.
7117 2017-11-09 Jim Wilson <jimw@sifive.com>
7119 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
7120 (scan_prog_file): Likewise.
7122 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7124 * bb-reorder.c (max_entry_frequency): Remove.
7125 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
7126 connect_traces, push_to_next_round_p): Remove prototypes.
7127 (find_traces_1_round): Use counts only.
7128 (push_to_next_round_p): Likewise.
7129 (find_traces): Likewise.
7130 (rotate_loop): Likewise.
7131 (find_traces_1_round): Likewise.
7132 (connect_traces): Likewise.
7133 (edge_order): Likewise.
7135 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
7137 * config/arm/arm.c (output_return_instruction): Add comments to
7138 indicate requirement for cmse_nonsecure_entry return to account
7139 for the size of clearing instruction output here.
7140 (thumb_exit): Likewise.
7141 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
7142 return in hardfloat mode.
7144 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
7146 * config/rs6000/rs6000.c (machine_function): Add a bool,
7147 "toc_is_wrapped_separately".
7148 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
7149 if it wasn't explicitly set or unset, we are optimizing for speed, and
7150 doing separate shrink-wrapping.
7151 (rs6000_get_separate_components): Enable the TOC component if
7152 saving the TOC register in the prologue.
7153 (rs6000_components_for_bb): Handle the TOC component.
7154 (rs6000_emit_prologue_components): Store the TOC register where needed.
7155 (rs6000_set_handled_components): Mark TOC as handled, if handled.
7156 (rs6000_emit_prologue): Don't save the TOC if that is already done.
7158 2017-11-09 Martin Jambor <mjambor@suse.cz>
7160 * ipa-param-manipulation.c: New file.
7161 * ipa-param-manipulation.h: Likewise.
7162 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
7163 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
7164 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
7165 (ipa_parm_adjustment): Likewise.
7166 (ipa_parm_adjustment_vec): Likewise.
7167 (ipa_get_vector_of_formal_parms): Moved declaration to
7168 ipa-param-manipulation.h.
7169 (ipa_get_vector_of_formal_parm_types): Likewise.
7170 (ipa_modify_formal_parameters): Likewise.
7171 (ipa_modify_call_arguments): Likewise.
7172 (ipa_combine_adjustments): Likewise.
7173 (ipa_dump_param_adjustments): Likewise.
7174 (ipa_modify_expr): Likewise.
7175 (ipa_get_adjustment_candidate): Likewise.
7176 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
7177 ipa-param-manipulation.c.
7178 (ipa_get_vector_of_formal_parm_types): Likewise.
7179 (ipa_modify_formal_parameters): Likewise.
7180 (ipa_modify_call_arguments): Likewise.
7181 (ipa_modify_expr): Likewise.
7182 (get_ssa_base_param): Likewise.
7183 (ipa_get_adjustment_candidate): Likewise.
7184 (index_in_adjustments_multiple_times_p): Likewise.
7185 (ipa_combine_adjustments): Likewise.
7186 (ipa_dump_param_adjustments): Likewise.
7187 * tree-sra.c: Also include ipa-param-manipulation.h
7188 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
7191 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7192 Alan Hayward <alan.hayward@arm.com>
7193 David Sherwood <david.sherwood@arm.com>
7195 * doc/sourcebuild.texi (vect_masked_store): Document.
7197 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7198 Alan Hayward <alan.hayward@arm.com>
7199 David Sherwood <david.sherwood@arm.com>
7201 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
7203 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7204 Alan Hayward <alan.hayward@arm.com>
7205 David Sherwood <david.sherwood@arm.com>
7207 * doc/sourcebuild.texi (vect_variable_length): Document.
7209 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7210 Alan Hayward <alan.hayward@arm.com>
7211 David Sherwood <david.sherwood@arm.com>
7213 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
7215 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7216 Alan Hayward <alan.hayward@arm.com>
7217 David Sherwood <david.sherwood@arm.com>
7219 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
7221 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7222 Alan Hayward <alan.hayward@arm.com>
7223 David Sherwood <david.sherwood@arm.com>
7225 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
7226 previously undocumented selectors.
7227 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
7229 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7231 * doc/rtl.texi (const_vector): Say that elements can be
7232 const_wide_ints too.
7233 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
7234 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
7235 (gen_vec_duplicate): Use it instead of CONSTANT_P.
7236 * optabs.c (expand_vector_broadcast): Likewise.
7238 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7239 Alan Hayward <alan.hayward@arm.com>
7240 David Sherwood <david.sherwood@arm.com>
7242 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
7243 scaled index even if the unscaled address was invalid.
7244 Don't increase the complexity of using a scale in that case.
7246 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7247 Alan Hayward <alan.hayward@arm.com>
7248 David Sherwood <david.sherwood@arm.com>
7250 * doc/rtl.texi: Rewrite the subreg rules so that they partition
7251 the inner register into REGMODE_NATURAL_SIZE bytes rather than
7252 UNITS_PER_WORD bytes.
7253 * emit-rtl.c (validate_subreg): Divide subregs into blocks
7254 based on REGMODE_NATURAL_SIZE of the inner mode.
7255 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
7256 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
7257 * expmed.c (lowpart_bit_field_p): Divide the value up into
7258 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
7259 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
7260 whether something is likely to occupy more than one register.
7262 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7265 * ipa-inline-transform.c (update_noncloned_frequencies): Use
7266 profile_count::adjust_for_ipa_scaling.
7267 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
7268 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
7270 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
7272 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7274 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
7275 (split_group): Add total_orig and total_new arguments, estimate the
7276 number of statements related to the store group without store merging
7277 and with store merging.
7278 (imm_store_chain_info::output_merged_store): Adjust split_group
7279 callers, punt if estimated number of statements with store merging
7280 is not smaller than estimated number of statements without it.
7282 (handled_load): Remove has_single_use checks.
7283 (pass_store_merging::process_store): Likewise.
7285 2017-11-09 Richard Biener <rguenther@suse.de>
7287 PR tree-optimization/82902
7288 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
7290 2017-11-09 Martin Liska <mliska@suse.cz>
7293 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
7296 2017-11-09 Martin Liska <mliska@suse.cz>
7298 PR tree-optimization/82669
7299 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
7301 2017-11-09 Martin Liska <mliska@suse.cz>
7303 PR gcov-profile/48463
7304 * coverage.c (coverage_begin_function): Output also end locus
7305 of a function and information whether the function is
7307 * gcov-dump.c (tag_function): Parse and print the information.
7308 * gcov.c (INCLUDE_MAP): Add include.
7309 (INCLUDE_SET): Likewise.
7310 (struct line_info): Move earlier in the source file because
7311 of vector<line_info> in function_info structure.
7312 (line_info::line_info): Likewise.
7313 (line_info::has_block): Likewise.
7314 (struct source_info): Add new member index.
7315 (source_info::get_functions_at_location): New function.
7316 (function_info::group_line_p): New function.
7317 (output_intermediate_line): New function.
7318 (output_intermediate_file): Use the mentioned function.
7319 (struct function_start): New.
7320 (struct function_start_pair_hash): Likewise.
7321 (process_file): Add code that identifies group functions.
7322 Assign lines either to global or function scope.
7323 (generate_results): Skip artificial functions.
7324 (find_source): Assign index for each source file.
7325 (read_graph_file): Read new flag artificial and end_line.
7326 (add_line_counts): Assign it either to global of function scope.
7327 (accumulate_line_counts): Isolate core of the function to
7328 accumulate_line_info and call it for both function and global
7330 (accumulate_line_info): New function.
7331 (output_line_beginning): Fix GNU coding style.
7332 (print_source_line): New function.
7333 (output_line_details): Likewise.
7334 (output_function_details): Likewise.
7335 (output_lines): Iterate both source (global) scope and function
7337 (struct function_line_start_cmp): New class.
7338 * doc/gcov.texi: Reflect changes in documentation.
7340 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7343 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
7344 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
7345 and similarly for not instead of neg.
7347 2017-11-08 Andi Kleen <ak@linux.intel.com>
7349 * config/i386/i386.opt: Add -mforce-indirect-call.
7350 * config/i386/predicates.md: Check for flag_force_indirect_call.
7351 * doc/invoke.texi: Document -mforce-indirect-call
7353 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
7355 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
7357 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
7358 config/riscv/riscv.c (predict.h): New include.
7359 (riscv_slow_unaligned_access_p): No longer static.
7360 (riscv_block_move_straight): Add require.
7361 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7363 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7366 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
7367 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
7368 nonimmediate_operand predicate for operand 1 instead of
7371 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7373 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
7375 * config/aarch64/constraints.md (Uml): New constraint.
7376 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
7379 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7381 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
7382 of two vec_duplicates into a vec_concat.
7384 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7386 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7387 Simplify vec_merge of vec_duplicate and vec_concat.
7388 * config/aarch64/constraints.md (Utq): New constraint.
7389 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
7392 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7394 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7395 Simplify vec_merge of vec_duplicate and const_vector.
7396 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
7398 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
7399 mode iterator. Update predicate on operand 1 to
7400 handle non-const_vec constants. Delete constraints.
7401 (*aarch64_combinez_be<mode>): Likewise for operand 2.
7403 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7405 PR tree-optimization/78821
7406 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
7408 (store_operand_info::store_operand_info): Initialize it to false.
7409 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
7410 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
7411 store in the group, and if chain_info is non-NULL, to ignore altogether
7413 (compatible_load_p): Fail if bit_not_p does not match.
7414 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
7415 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
7416 (pass_store_merging::process_store): Adjust
7417 terminate_all_aliasing_chains calls to pass NULL in all current spots,
7418 call terminate_all_aliasing_chains newly when adding a store into
7419 a chain with non-NULL chain_info.
7421 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7423 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
7425 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7427 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7429 (aarch64_layout_frame): Initialise emit_frame_chain.
7430 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
7431 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
7433 2017-11-08 Martin Liska <mliska@suse.cz>
7435 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
7436 of gimple_call_internal_p.
7438 2017-11-07 Tom de Vries <tom@codesourcery.com>
7440 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
7442 2017-11-07 Tom de Vries <tom@codesourcery.com>
7444 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
7447 2017-11-08 Martin Liska <mliska@suse.cz>
7450 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
7452 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7454 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
7456 2017-11-07 Jakub Jelinek <jakub@redhat.com>
7459 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
7460 (*cmp<mode>_ccz_1): New insn with $k alternative.
7463 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
7464 RTX_COMM_COMPARE as commutative as well.
7465 (ix86_binary_operator_ok): Formatting fix.
7466 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
7467 *<code><mode>3<mask_name><round_saeonly_name>,
7468 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
7469 *<s>mul<mode>3_highpart<mask_name>,
7470 *vec_widen_umult_even_v16si<mask_name>,
7471 *vec_widen_umult_even_v8si<mask_name>,
7472 *vec_widen_umult_even_v4si<mask_name>,
7473 *vec_widen_smult_even_v16si<mask_name>,
7474 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
7475 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
7476 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
7477 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
7478 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
7479 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
7480 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
7481 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
7482 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
7483 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
7484 ix86_binary_operator_ok. Formatting fixes.
7485 (*<plusminus_insn><mode>3<mask_name><round_name>,
7486 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
7489 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
7491 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
7492 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
7493 eq and ne if TARGET_ISEL.
7494 (cmp): New code_iterator.
7495 (UNS, UNSU_, UNSIK): New code_attrs.
7496 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
7497 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
7499 ("eq<mode>3"): ... this.
7500 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
7502 ("ne<mode>3"): ... this.
7504 2017-11-07 Julia Koval <julia.koval@intel.com>
7507 * common/config/i386/i386-common.c
7508 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
7509 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
7510 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
7511 * config/i386/i386.opt: Ditto.
7512 * config/i386/i386.c (ix86_target_string): Ditto.
7513 (ix86_option_override_internal): Ditto.
7514 (ix86_init_mpx_builtins): Move MPX to args2.
7515 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
7516 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7517 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7518 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7519 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7520 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7521 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
7523 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
7526 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
7527 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
7528 (zero-extendsidi peephole2): Remove peephole.
7530 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7533 * stor-layout.c: Include attribs.h.
7534 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
7535 explicit lookup of "aligned" attribute.
7537 2017-11-07 Andrew Waterman <andrew@sifive.com>
7539 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
7540 (riscv_expand_block_move): Likewise.
7541 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
7543 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
7544 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
7545 * config/riscv/riscv.c (riscv_block_move_straight): New function.
7546 (riscv_adjust_block_mem): Likewise.
7547 (riscv_block_move_loop): Likewise.
7548 (riscv_expand_block_move): Likewise.
7549 * config/riscv/riscv.md (movmemsi): New pattern.
7551 2017-11-07 Michael Clark <michaeljclark@mac.com>
7553 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
7554 (MUSL_DYNAMIC_LINKER): Likewise.
7556 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
7558 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
7559 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
7562 2017-11-07 Richard Biener <rguenther@suse.de>
7564 * match.pd: Fix build.
7566 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
7567 Jackson Woodruff <jackson.woodruff@arm.com>
7569 PR tree-optimization/71026
7570 * match.pd: Canonicalize negate in division.
7572 2017-11-07 Sudakshina Das <sudi.das@arm.com>
7575 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
7577 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7579 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
7580 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
7582 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7584 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
7585 non-scalar integral types.
7586 * match.pd (negate_expr_p): Handle MINUS_EXPR.
7587 (-(A-B), -(~A)): New transformations.
7589 2017-11-07 Tom de Vries <tom@codesourcery.com>
7591 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
7592 semicolon after "do {} while (0)".
7593 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7594 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7595 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7596 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7597 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7599 2017-11-07 Tom de Vries <tom@codesourcery.com>
7601 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
7602 after "do {} while (0)".
7603 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7604 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7605 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7606 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7607 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7609 2017-11-07 Tom de Vries <tom@codesourcery.com>
7612 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
7614 (arm_rtx_costs_internal): Add missing semicolon after
7615 HANDLE_NARROW_SHIFT_ARITH call.
7617 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7619 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7620 disable isel if it was not set explicitly.
7622 2017-11-06 James Bowman <james.bowman@ftdichip.com>
7624 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
7625 (add_type_attribute) likewise.
7627 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
7629 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
7630 of struct ix86_frame.
7631 (ix86_initial_elimination_offset): Likewise.
7632 (ix86_expand_split_stack_prologue): Likewise.
7634 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
7636 * tree-vrp.h (enum value_range_type): Update stale comment.
7638 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7639 Alan Hayward <alan.hayward@arm.com>
7640 David Sherwood <david.sherwood@arm.com>
7642 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
7643 (aarch64_expand_vec_perm_const): Take the number of units too.
7644 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
7645 (aarch64_expand_vec_perm_const): Likewise.
7646 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
7647 (vec_perm<mode>): Update accordingly.
7649 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7650 Alan Hayward <alan.hayward@arm.com>
7651 David Sherwood <david.sherwood@arm.com>
7653 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
7654 Take the number of units too.
7655 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
7656 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
7657 but check for a vector mode before rather than after the call.
7658 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7659 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
7660 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
7661 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
7662 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
7663 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
7664 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
7665 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
7666 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
7667 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
7668 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
7669 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
7670 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
7671 (aarch64_sqdmull2_n<mode>): Update accordingly.
7673 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7674 Alan Hayward <alan.hayward@arm.com>
7675 David Sherwood <david.sherwood@arm.com>
7677 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
7678 the number of units too.
7679 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
7680 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
7681 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
7682 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
7683 (vec_store_lanesxi<mode>): Update accordingly.
7685 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7686 Alan Hayward <alan.hayward@arm.com>
7687 David Sherwood <david.sherwood@arm.com>
7689 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
7690 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
7691 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
7692 of units rather than the mode.
7693 * config/aarch64/iterators.md (nunits): New mode attribute.
7694 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
7695 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
7696 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
7697 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
7698 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
7699 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
7700 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
7701 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
7702 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
7703 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
7704 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
7705 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
7706 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
7707 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
7708 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
7709 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
7710 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
7711 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
7712 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
7713 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
7714 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
7715 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
7716 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
7717 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
7718 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
7719 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
7720 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
7721 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
7722 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
7723 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
7724 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
7725 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
7726 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
7727 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
7728 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
7729 (aarch64_simd_vec_setv2di): Likewise.
7731 2017-11-06 Carl Love <cel@us.ibm.com>
7733 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
7735 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
7736 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
7738 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
7740 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
7741 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
7742 Add power 8 macro expansions.
7743 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
7744 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
7745 power 8 instructions. (VSX_XXBR): Add iterator.
7747 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
7749 * config/arm/arm.md (predicable_short_it): Change default to "no",
7750 improve documentation, remove uses that are identical to the default.
7751 (enabled_for_depr_it): Rename to enabled_for_short_it.
7752 * gcc/config/arm/arm-fixed.md (predicable_short_it):
7753 Remove default uses.
7754 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
7755 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
7756 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
7757 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
7759 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
7762 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
7763 float128 helper macros, which are no longer used after deleting
7764 the old 'q' built-in functions, and moving the round to odd
7765 built-in functions to being special built-in functions.
7766 (BU_FLOAT128_2): Likewise.
7767 (BU_FLOAT128_1_HW): Likewise.
7768 (BU_FLOAT128_2_HW): Likewise.
7769 (BU_FLOAT128_3_HW): Likewise.
7770 (FABSQ): Delete old 'q' built-in functions.
7771 (COPYSIGNQ): Likewise.
7772 (SQRTF128_ODD): Move round to odd built-in functions to be
7773 special built-in functions, so that we can handle
7774 -mabi=ieeelongdouble.
7775 (TRUNCF128_ODD): Likewise.
7776 (ADDF128_ODD): Likewise.
7777 (SUBF128_ODD): Likewise.
7778 (MULF128_ODD): Likewise.
7779 (DIVF128_ODD): Likewise.
7780 (FMAF128_ODD): Likewise.
7781 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
7782 built-in names to 'f128'.
7783 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
7784 old 'q' built-in functions, as the machine independent code for
7785 'f128' built-in functions handles this.
7786 (rs6000_expand_builtin): Add expansion for float128 round to odd
7787 functions, keying off on -mabi=ieeelongdouble of whether to use
7788 the KFmode or TFmode variant.
7789 (rs6000_init_builtins): Initialize the _Float128 round to odd
7791 * doc/extend.texi (PowerPC Built-in Functions): Document the old
7792 _Float128 'q' built-in functions are now mapped into the new
7793 'f128' built-in functions.
7795 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
7797 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
7798 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
7800 2017-11-06 David Malcolm <dmalcolm@redhat.com>
7803 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
7804 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
7805 * toplev.c: Include "ipa-fnsummary.h".
7806 (toplev::finalize): Call ipa_fnsummary_c_finalize.
7808 2017-11-06 Jakub Jelinek <jakub@redhat.com>
7810 PR tree-optimization/82838
7811 * gimple-ssa-store-merging.c
7812 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
7813 on a separate gimple_seq which is then appended to seq.
7815 2017-11-06 Jeff Law <law@redhat.com>
7818 * config/i386/i386.c (PROBE_INTERVAL): Remove.
7819 (get_probe_interval): New functions.
7820 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
7821 (ix86_adjust_stack_and_probe): Likewise.
7822 (output_adjust_stack_and_probe): Likewise.
7823 (ix86_emit_probe_stack_range): Likewise.
7824 (ix86_expand_prologue): Likewise.
7826 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7828 PR tree-optimization/82816
7829 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
7830 if the modes of the two types are the same.
7831 (convert_plusminus_to_widen): Likewise.
7833 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7835 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
7837 (usadv16qi): New define_expand.
7838 (usadv8hi): New define_expand.
7840 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
7843 * ipa-inline-transform.c (update_noncloned_frequencies): Always
7845 (inline_transform): Likewise.
7846 * predict.c (counts_to_freqs): Remove useless conditional.
7847 * profile-count.h (profile_count::apply_scale): Move sanity check.
7848 * tree-inline.c (copy_bb): Always scale.
7849 (copy_cfg_body): Likewise.
7851 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
7854 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
7856 (*cmp_ite0): Add enabled_for_depr_it attribute.
7857 (*cmp_ite1): Likewise.
7859 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7861 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
7864 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7866 * tree-vrp.c (vrp_int_const_binop): Return true on success and
7867 return the value by pointer.
7868 (extract_range_from_multiplicative_op_1): Update accordingly.
7869 Return as soon as an operation fails.
7871 2017-11-05 Tom de Vries <tom@codesourcery.com>
7874 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
7875 (DEF_SANITIZER_BUILTIN): ... here.
7876 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
7877 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
7879 2017-11-05 Tom de Vries <tom@codesourcery.com>
7882 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
7884 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
7885 ASM_OUTPUT_BEFORE_CASE_LABEL call.
7886 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
7888 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7889 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7890 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7892 2017-11-05 Tom de Vries <tom@codesourcery.com>
7895 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
7898 2017-11-04 Michael Clark <michaeljclark@mac.com>
7900 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
7901 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
7903 (*addsi3_extended): Likewise.
7904 (*addsi3_extended2): Likewise.
7905 (<optab>si3): Likewise.
7906 (<optab>di3): Likewise.
7907 (<optab><mode>3): Likewise.
7908 (<*optabe>si3_internal): Likewise.
7909 (zero_extendqi<SUPERQI:mode>2): Likewise.
7910 (*add<mode>hi3): Likewise.
7911 (*xor<mode>hi3): Likewise.
7912 (<optab>di3): Likewise.
7913 (*<optab>si3_extend): Likewise.
7914 (*sge<u>_<X:mode><GPR:mode>): Likewise.
7915 (*slt<u>_<X:mode><GPR:mode>): Likewise.
7916 (*sle<u>_<X:mode><GPR:mode>): Likewise.
7918 2017-11-04 Andrew Waterman <andrew@sifive.com>
7920 * config/riscv/riscv.c (riscv_option_override): Conditionally set
7921 TARGET_STRICT_ALIGN based upon -mtune argument.
7923 2017-11-04 Andrew Waterman <andrew@sifive.com>
7925 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
7927 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
7929 * config/i386/i386.c (choose_basereg): Use optional scratch
7930 register and add assertion.
7931 (x86_emit_outlined_ms2sysv_save): Use scratch register when
7932 needed, and don't allocate stack.
7933 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
7934 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
7935 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
7937 2017-11-03 Jeff Law <law@redhat.com>
7939 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
7940 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
7941 to probe at the start of a noreturn function.
7943 2017-11-03 Jakub Jelinek <jakub@redhat.com>
7945 PR tree-optimization/78821
7946 * gimple-ssa-store-merging.c: Update the file comment.
7947 (MAX_STORE_ALIAS_CHECKS): Define.
7948 (struct store_operand_info): New type.
7949 (store_operand_info::store_operand_info): New constructor.
7950 (struct store_immediate_info): Add rhs_code and ops data members.
7951 (store_immediate_info::store_immediate_info): Add rhscode, op0r
7952 and op1r arguments to the ctor, initialize corresponding data members.
7953 (struct merged_store_group): Add load_align_base and load_align
7955 (merged_store_group::merged_store_group): Initialize them.
7956 (merged_store_group::do_merge): Update them.
7957 (merged_store_group::apply_stores): Pick the constant for
7958 encode_tree_to_bitpos from one of the two operands, or skip
7959 encode_tree_to_bitpos if neither operand is a constant.
7960 (class pass_store_merging): Add process_store method decl. Remove
7961 bool argument from terminate_all_aliasing_chains method decl.
7962 (pass_store_merging::terminate_all_aliasing_chains): Remove
7963 var_offset_p argument and corresponding handling.
7964 (stmts_may_clobber_ref_p): New function.
7965 (compatible_load_p): New function.
7966 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
7967 if there is overlap and rhs_code is not INTEGER_CST. For
7968 non-overlapping stores terminate group if rhs is not mergeable.
7969 (get_alias_type_for_stmts): Change first argument from
7970 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
7971 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
7972 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
7974 (get_location_for_stmts): Change first argument from
7975 auto_vec<gimple *> & to vec<gimple *> &.
7976 (struct split_store): Remove orig_stmts data member, add orig_stores.
7977 (split_store::split_store): Create orig_stores rather than orig_stmts.
7978 (find_constituent_stmts): Renamed to ...
7979 (find_constituent_stores): ... this. Change second argument from
7980 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
7981 to info structures rather than the statements.
7982 (split_group): Rename ALLOW_UNALIGNED argument to
7983 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
7984 it. Adjust find_constituent_stores caller.
7985 (imm_store_chain_info::output_merged_store): Handle rhs_code other
7986 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
7987 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
7988 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
7989 (mem_valid_for_store_merging): New function.
7990 (handled_load): New function.
7991 (pass_store_merging::process_store): New method.
7992 (pass_store_merging::execute): Use process_store method. Adjust
7993 terminate_all_aliasing_chains caller.
7995 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7997 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
7998 Return true for more constants, symbols and label references.
7999 (aarch64_valid_floating_const): Remove unused function.
8001 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8004 * config/aarch64/aarch64.c (aarch64_layout_frame):
8005 Undo forcing of LR at bottom of frame.
8007 2017-11-03 Jeff Law <law@redhat.com>
8010 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
8011 for int_registers_saved.
8013 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
8014 extracted from tree-ssa-dom.c.
8015 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
8016 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
8017 (record_equivalences_from_incoming_edge): Add additional argument
8018 to single_pred_edge_ignoring_loop_edges call.
8019 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
8020 (uncprop_dom_walker::before_dom_children): Add additional argument
8021 to single_pred_edge_ignoring_loop_edges call.
8022 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
8023 single_pred_edge_ignoring_loop_edges rather than open coding.
8024 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
8026 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
8028 * match.pd (-(-A)): Rewrite.
8030 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8032 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
8033 (rs6000_emit_int_cmove): New declaration.
8034 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
8035 (rs6000_emit_sISEL): Delete.
8036 (rs6000_emit_int_cmove): Make non-static.
8037 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
8038 instead of rs6000_emit_sISEL.
8040 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
8042 * asan.c (create_cond_insert_point): Maintain profile.
8043 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
8045 * basic-block.h (struct basic_block_def): Remove frequency.
8046 (EDGE_FREQUENCY): Use to_frequency
8047 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
8049 (find_traces): Update to use to_frequency.
8050 (find_traces_1_round): Likewise; use only IPA counts.
8051 (bb_to_key): Likewise.
8052 (connect_traces): Use IPA counts only.
8053 (copy_bb_p): Update to use to_frequency.
8054 (fix_up_crossing_landing_pad): Likewise.
8055 (sanitize_hot_paths): Likewise.
8056 * bt-load.c (basic_block_freq): Likewise.
8057 * cfg.c (init_flow): Set count_max to uninitialized.
8058 (check_bb_profile): Remove frequencies; check counts.
8059 (dump_bb_info): Do not dump frequencies.
8060 (update_bb_profile_for_threading): Update counts only.
8061 (scale_bbs_frequencies_int): Likewise.
8062 (MAX_SAFE_MULTIPLIER): Remove.
8063 (scale_bbs_frequencies_gcov_type): Update counts only.
8064 (scale_bbs_frequencies_profile_count): Update counts only.
8065 (scale_bbs_frequencies): Update counts only.
8066 * cfg.h (struct control_flow_graph): Add count-max.
8067 (update_bb_profile_for_threading): Update prototype.
8068 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
8069 (find_many_sub_basic_blocks): Likewise.
8070 * cfgcleanup.c (try_forward_edges): Likewise.
8071 (try_crossjump_to_edge): Likewise.
8072 * cfgexpand.c (expand_gimple_cond): Likewise.
8073 (expand_gimple_tailcall): Likewise.
8074 (construct_init_block): Likewise.
8075 (construct_exit_block): Likewise.
8076 * cfghooks.c (verify_flow_info): Check consistency of counts.
8077 (dump_bb_for_graph): Do not dump frequencies.
8078 (split_block_1): Do not update frequencies.
8079 (split_edge): Do not update frequencies.
8080 (make_forwarder_block): Do not update frequencies.
8081 (duplicate_block): Do not update frequencies.
8082 (account_profile_record): Do not update frequencies.
8083 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
8084 for global heuristics.
8085 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
8086 (expected_loop_iterations_unbounded): Use counts only.
8087 * cfgloopmanip.c (scale_loop_profile): Simplify.
8088 (create_empty_loop_on_edge): Simplify
8090 (duplicate_loop_to_header_edge): Simplify
8091 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
8092 (update_br_prob_note): Take care of removing note when profile
8094 (relink_block_chain): Do not dump frequency.
8095 (rtl_account_profile_record): Use to_frequency.
8096 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
8097 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
8098 (cgraph_update_edges_for_call_stmt_node): Likewise.
8099 (cgraph_edge::verify_count_and_frequency): Update.
8100 (cgraph_node::verify_node): Temporarily disable frequency verification.
8101 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
8102 to_cgraph_frequency.
8103 (cgraph_edge::rebuild_edges): Convert to ipa counts.
8104 * cgraphunit.c (init_lowered_empty_function): Do not initialize
8106 (cgraph_node::expand_thunk): Update profile.
8107 * except.c (dw2_build_landing_pads): Do not update frequency.
8108 * final.c (compute_alignments): Use to_frequency.
8109 (dump_basic_block_info): Do not dump frequency.
8110 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
8111 (dump_gimple_bb_header): Do not dump frequency.
8112 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
8114 * gimple-streamer-in.c (input_bb): Do not stream frequency.
8115 * gimple-streamer-out.c (output_bb): Do not stream frequency.
8116 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
8117 (init_before_recovery): Do not update frequency.
8118 (sched_create_recovery_edges): Do not update frequency.
8119 * hsa-gen.c (convert_switch_statements): Do not update frequency.
8120 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
8121 (ipa_cp_c_finalize): Set max_count to uninitialized.
8122 * ipa-fnsummary.c (get_minimal_bb): Use counts.
8123 (param_change_prob): Use counts.
8124 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
8126 * ipa-split.c (consider_split): Use to_frequency.
8127 (split_function): Use to_frequency.
8128 * ira-build.c (loop_compare_func): Likewise.
8129 (mark_loops_for_removal): Likewise.
8130 (mark_all_loops_for_removal): Likewise.
8131 * loop-doloop.c (doloop_modify): Do not update frequency.
8132 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
8134 * lto-streamer-in.c (input_function): Update count_max.
8135 * omp-expand.c (expand_omp_taskreg): Update count_max.
8136 * omp-simd-clone.c (simd_clone_adjust): Update profile.
8137 * predict.c (maybe_hot_frequency_p): Use to_frequency.
8138 (maybe_hot_count_p): Use ipa counts only.
8139 (maybe_hot_bb_p): Simplify.
8140 (maybe_hot_edge_p): Simplify.
8141 (probably_never_executed): Do not take frequency argument.
8142 (probably_never_executed_bb_p): Do not pass frequency.
8143 (probably_never_executed_edge_p): Likewise.
8144 (combine_predictions_for_bb): Check that profile is nonzero.
8145 (propagate_freq): Do not set frequency.
8146 (drop_profile): Simplify.
8147 (counts_to_freqs): Simplify.
8148 (expensive_function_p): Use to_frequency.
8149 (propagate_unlikely_bbs_forward): Simplify.
8150 (determine_unlikely_bbs): Simplify.
8151 (estimate_bb_frequencies): Add hack to silence graphite issues.
8152 (compute_function_frequency): Use ipa counts.
8153 (pass_profile::execute): Update.
8154 (rebuild_frequencies): Use counts only.
8155 (force_edge_cold): Use counts only.
8156 * profile-count.c (profile_count::dump): Dump new count types.
8157 (profile_count::differs_from_p): Check compatiblity.
8158 (profile_count::to_frequency): New function.
8159 (profile_count::to_cgraph_frequency): New function.
8160 * profile-count.h (struct function): Declare.
8161 (enum profile_quality): Add profile_guessed_local and
8162 profile_guessed_global0.
8163 (class profile_proability): Decrease number of bits to 29;
8164 update from_reg_br_prob_note and to_reg_br_prob_note.
8165 (class profile_count: Update comment; decrease number of bits
8166 to 61. Check compatibility.
8167 (profile_count::compatible_p): New private member function.
8168 (profile_count::ipa_p): New member function.
8169 (profile_count::operator<): Handle global zero correctly.
8170 (profile_count::operator>): Handle global zero correctly.
8171 (profile_count::operator<=): Handle global zero correctly.
8172 (profile_count::operator>=): Handle global zero correctly.
8173 (profile_count::nonzero_p): New member function.
8174 (profile_count::force_nonzero): New member function.
8175 (profile_count::max): New member function.
8176 (profile_count::apply_scale): Handle IPA scalling.
8177 (profile_count::guessed_local): New member function.
8178 (profile_count::global0): New member function.
8179 (profile_count::ipa): New member function.
8180 (profile_count::to_frequency): Declare.
8181 (profile_count::to_cgraph_frequency): Declare.
8182 * profile.c (OVERLAP_BASE): Delete.
8183 (compute_frequency_overlap): Delete.
8184 (compute_branch_probabilities): Do not use compute_frequency_overlap.
8185 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
8186 * sched-ebb.c (rank): Use counts only.
8187 * shrink-wrap.c (handle_simple_exit): Use counts only.
8188 (try_shrink_wrapping): Use counts only.
8189 (place_prologue_for_one_component): Use counts only.
8190 * tracer.c (find_best_predecessor): Use to_frequency.
8191 (find_trace): Use to_frequency.
8192 (tail_duplicate): Use to_frequency.
8193 * trans-mem.c (expand_transaction): Do not update frequency.
8194 * tree-call-cdce.c: Do not update frequency.
8195 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
8196 (gimple_merge_blocks): Likewise.
8197 (gimple_split_edge): Likewise.
8198 (gimple_duplicate_sese_region): Likewise.
8199 (gimple_duplicate_sese_tail): Likewise.
8200 (move_sese_region_to_fn): Likewise.
8201 (gimple_account_profile_record): Likewise.
8202 (insert_cond_bb): Likewise.
8203 * tree-complex.c (expand_complex_div_wide): Likewise.
8204 * tree-eh.c (lower_resx): Update profile.
8205 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
8207 (initialize_cfun): Do not initialize frequencies
8208 (freqs_to_counts): Delete.
8209 (copy_cfg_body): Ignore count parameter.
8210 (copy_body): Update.
8211 (expand_call_inline): Update count_max.
8212 (optimize_inline_calls): Update count_max.
8213 (tree_function_versioning): Update count_max.
8214 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
8215 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
8217 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
8218 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
8219 (try_peel_loop): Likewise.
8220 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
8222 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
8223 (tree_transform_and_unroll_loop): Do not use frequencies
8224 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
8225 Use reliable prediction only.
8226 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
8227 * tree-ssa-sink.c (select_best_block): Use to_frequency.
8228 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
8229 probability scaling.
8230 * tree-ssa-threadupdate.c (create_block_for_threading): Do
8231 not update frequency
8232 (any_remaining_duplicated_blocks): Likewise.
8233 (update_profile): Likewise.
8234 (estimated_freqs_path): Delete.
8235 (freqs_to_counts_path): Delete.
8236 (clear_counts_path): Delete.
8237 (ssa_fix_duplicate_block_edges): Likewise.
8238 (duplicate_thread_path): Likewise.
8239 * tree-switch-conversion.c (gen_inbound_check): Use counts.
8240 * tree-tailcall.c (decrease_profile): Do not update frequency.
8241 (eliminate_tail_call): Likewise.
8242 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
8243 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
8244 (optimize_mask_stores): Likewise.
8245 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8246 * ubsan.c (ubsan_expand_null_ifn): Update profile.
8247 (ubsan_expand_ptr_ifn): Update profile.
8248 * value-prof.c (gimple_ic): Simplify.
8249 * value-prof.h (gimple_ic): Update prototype.
8250 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
8251 * ipa-inline.c (compute_uninlined_call_time): Be sure that
8253 (want_inline_self_recursive_call_p): Likewise.
8254 (resolve_noninline_speculation): Only cummulate defined counts.
8255 (inline_small_functions): Use nonzero_p.
8256 (ipa_inline): Do not access freed node.
8258 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8260 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8261 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
8263 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
8265 * config/riscv/riscv.c (riscv_legitimize_move): Handle
8266 non-legitimate address.
8268 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8270 * config/rs6000/rs6000.md (*lt0_disi): Delete.
8271 (*lt0_<mode>di, *lt0_<mode>si): New.
8273 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8275 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
8276 TARGET_PAIRED_FLOAT.
8278 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
8279 Jim Wilson <jim.wilson@linaro.org>
8281 * config/aarch64/aarch64-cores.def (saphira): New CPU.
8282 * config/aarch64/aarch64-tune.md: Regenerated.
8283 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
8284 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
8286 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
8288 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
8289 (arc_expand_prologue): Restore blink for millicode.
8290 * config/arc/linux.h (LINK_EH_SPEC): Defined.
8292 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
8295 * config/i386/i386.c (ix86_vector_duplicate_value): Use
8296 gen_vec_duplicate after forcing the scalar into a register.
8298 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
8300 * combine (try_combine): Print the insns input to try_combine to the
8303 2017-11-02 Steve Ellcey <sellcey@cavium.com>
8306 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
8307 Remove second argument from aarch64_process_target_attr call.
8308 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
8310 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
8312 (aarch64_handle_attr_arch): Remove second argument.
8313 (aarch64_handle_attr_cpu): Ditto.
8314 (aarch64_handle_attr_tune): Ditto.
8315 (aarch64_handle_attr_isa_flags): Ditto.
8316 (aarch64_process_one_target_attr): Ditto.
8317 (aarch64_process_target_attr): Ditto.
8318 (aarch64_option_valid_attribute_p): Remove second argument.
8319 on aarch64_process_target_attr call.
8321 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8323 * diagnostic.c: Include "selftest-diagnostic.h".
8324 (selftest::assert_location_text): New function.
8325 (selftest::test_diagnostic_get_location_text): New function.
8326 (selftest::diagnostic_c_tests): Call it.
8328 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8330 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
8331 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
8332 (class selftest::test_diagnostic_context): Move to...
8333 * selftest-diagnostic.c: New file.
8334 * selftest-diagnostic.h: New file.
8336 2017-11-02 James Bowman <james.bowman@ftdichip.com>
8338 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
8339 offset range for FT32B.
8340 * config/ft32/ft32.h: option "mcompress" enables relaxation.
8341 * config/ft32/ft32.md: Add TARGET_NOPM.
8342 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
8343 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
8345 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
8347 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
8349 2017-11-02 Jeff Law <law@redhat.com>
8351 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
8352 virtual keyword on FINAL OVERRIDE members.
8354 * tree-ssa-propagate.h (ssa_propagation_engine): Group
8355 virtuals together. Add virtual destructor.
8356 (substitute_and_fold_engine): Similarly.
8358 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
8360 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
8362 2017-11-02 Richard Biener <rguenther@suse.de>
8364 PR tree-optimization/82795
8365 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
8367 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8369 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
8371 (gcc_GAS_CHECK_FEATURE): Remove.
8372 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
8373 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
8374 * configure: Regenerate.
8376 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
8378 * config/arc/arc.c (hwloop_optimize): Account for empty
8381 2017-11-02 Richard Biener <rguenther@suse.de>
8384 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
8385 Truncate ARRAY_REF index and element size.
8387 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8389 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
8391 2017-11-01 Jeff Law <law@redhat.com>
8393 * tree-ssa-ccp.c (ccp_folder): New class derived from
8394 substitute_and_fold_engine.
8395 (ccp_folder::get_value): New member function.
8396 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
8397 (ccp_fold_stmt): Remove prototype.
8398 (ccp_finalize): Call substitute_and_fold from the ccp_class.
8399 * tree-ssa-copy.c (copy_folder): New class derived from
8400 substitute_and_fold_engine.
8401 (copy_folder::get_value): Renamed from get_value.
8402 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
8403 * tree-vrp.c (vrp_folder): New class derived from
8404 substitute_and_fold_engine.
8405 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
8406 (vrp_folder::get_value): New member function.
8407 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
8408 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
8409 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
8410 provide a class interface to folder/substitute routines.
8411 (ssa_prop_fold_stmt_fn): Remove typedef.
8412 (ssa_prop_get_value_fn): Likewise.
8413 (subsitute_and_fold): Remove prototype.
8414 (replace_uses_in): Likewise.
8415 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
8416 Renamed from replace_uses_in. Call the virtual member function
8417 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
8418 (substitute_and_fold_dom_walker): Remove initialization of
8419 data member entries for calbacks. Add substitute_and_fold_engine
8420 member and initialize it.
8421 (substitute_and_fold_dom_walker::before_dom_children0: Use the
8422 member functions for get_value, replace_phi_args_in c
8423 replace_uses_in, and fold_stmt calls.
8424 (substitute_and_fold_engine::substitute_and_fold): Renamed from
8425 substitute_and_fold. Remove assert. Update ctor call.
8427 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
8428 (ssa_prop_visit_phi_fn): Likewise.
8429 (class ssa_propagation_engine): New class to provide an interface
8431 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
8433 (ssa_prop_visit_phi): Likewise.
8434 (ssa_propagation_engine::simulate_stmt): Moved into class.
8435 Call visit_phi/visit_stmt from the class rather than via
8436 file scoped static variables.
8437 (ssa_propagation_engine::simulate_block): Moved into class.
8438 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
8439 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
8440 set file scoped statics for the visit_stmt/visit_phi callbacks.
8441 * tree-complex.c (complex_propagate): New class derived from
8442 ssa_propagation_engine.
8443 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
8444 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
8445 (tree_lower_complex): Call ssa_propagate via the complex_propagate
8447 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
8448 ssa_propagation_engine.
8449 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
8450 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
8451 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
8452 * tree-ssa-copy.c (copy_prop): New class derived from
8453 ssa_propagation_engine.
8454 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
8455 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
8456 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
8457 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
8458 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
8459 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
8460 (execute_vrp): Call ssa_propagate from the vrp_prop class.
8462 2017-11-01 Jakub Jelinek <jakub@redhat.com>
8464 PR rtl-optimization/82778
8465 PR rtl-optimization/82597
8466 * compare-elim.c (struct comparison): Add in_a_setter field.
8467 (find_comparison_dom_walker::before_dom_children): Remove killed
8468 bitmap and df_simulate_find_defs call, instead walk the defs.
8469 Compute last_setter and initialize in_a_setter. Merge definitions
8470 with first initialization for a few variables.
8471 (try_validate_parallel): Use insn_invalid_p instead of
8472 recog_memoized. Return insn rather than just the pattern.
8473 (try_merge_compare): Fix up comment. Don't uselessly test if
8474 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
8476 (execute_compare_elim_after_reload): Remove df_chain_add_problem
8479 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8480 Alan Hayward <alan.hayward@arm.com>
8481 David Sherwood <david.sherwood@arm.com>
8483 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
8484 aarch64_hard_regno_nregs to get the number of registers
8487 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8488 Alan Hayward <alan.hayward@arm.com>
8489 David Sherwood <david.sherwood@arm.com>
8491 * config/aarch64/constraints.md (Upl): Rename to...
8493 * config/aarch64/aarch64.md
8494 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
8497 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8498 Alan Hayward <alan.hayward@arm.com>
8499 David Sherwood <david.sherwood@arm.com>
8501 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
8502 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
8505 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8506 Alan Hayward <alan.hayward@arm.com>
8507 David Sherwood <david.sherwood@arm.com>
8509 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
8510 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
8511 (aarch64_evpc_dup): Generate rtl direcly, rather than using
8513 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
8514 of a single element.
8515 * config/aarch64/iterators.md: Add a comment above the permute
8516 unspecs to say that they are generated directly by
8517 aarch64_expand_vec_perm_const.
8518 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
8520 2017-11-01 Nathan Sidwell <nathan@acm.org>
8522 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
8524 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8526 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
8527 and medany code models, and describe what they do.
8529 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8531 Revert accidental duplicate:
8533 * combine.c (can_change_dest_mode): Reject changes in
8534 REGMODE_NATURAL_SIZE.
8536 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
8538 PR rtl-optimization/64682
8539 PR rtl-optimization/69567
8540 PR rtl-optimization/69737
8541 PR rtl-optimization/82683
8542 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
8543 register mentioned in the note, drop the note, unless it came from I3,
8544 in which case it should go to I3 again.
8546 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8548 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
8549 and return false if not.
8550 (clear_bytes_written_by, live_bytes_read): Update accordingly.
8552 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8554 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
8555 range is known to be empty.
8557 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8558 Alan Hayward <alan.hayward@arm.com>
8559 David Sherwood <david.sherwood@arm.com>
8561 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
8562 and CONST_VECTOR_NUNITS instead of computing the number of units from
8563 the byte sizes of the vector and element.
8564 (simplify_binary_operation_1): Likewise.
8565 (simplify_const_binary_operation): Likewise.
8566 (simplify_ternary_operation): Likewise.
8568 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8569 Alan Hayward <alan.hayward@arm.com>
8570 David Sherwood <david.sherwood@arm.com>
8572 * var-tracking.c (INT_MEM_OFFSET): Replace with...
8573 (int_mem_offset): ...this new function.
8574 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
8575 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
8576 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
8579 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8580 Alan Hayward <alan.hayward@arm.com>
8581 David Sherwood <david.sherwood@arm.com>
8583 * lower-subreg.c (interesting_mode_p): New function.
8584 (compute_costs, find_decomposable_subregs, decompose_register)
8585 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
8586 (resolve_clobber, dump_choices): Use it.
8588 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8589 Alan Hayward <alan.hayward@arm.com>
8590 David Sherwood <david.sherwood@arm.com>
8592 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
8594 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8595 Alan Hayward <alan.hayward@arm.com>
8596 David Sherwood <david.sherwood@arm.com>
8598 * alias.c (find_base_value, find_base_term): Only process integer
8599 truncations. Check the precision rather than the size.
8601 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8602 Alan Hayward <alan.hayward@arm.com>
8603 David Sherwood <david.sherwood@arm.com>
8605 * machmode.h (is_narrower_int_mode): New function
8606 * optabs.c (expand_float, expand_fix): Use it.
8607 * dwarf2out.c (rotate_loc_descriptor): Likewise.
8609 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8610 Alan Hayward <alan.hayward@arm.com>
8611 David Sherwood <david.sherwood@arm.com>
8613 * rtl.h (narrower_subreg_mode): New function.
8614 * ira-color.c (update_costs_from_allocno): Use it.
8616 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8617 Alan Hayward <alan.hayward@arm.com>
8618 David Sherwood <david.sherwood@arm.com>
8620 * optabs-query.h (convert_optab_p): New function, split out from...
8621 (convert_optab_handler): ...here.
8622 (widening_optab_handler): Delete.
8623 (find_widening_optab_handler): Remove permit_non_widening parameter.
8624 (find_widening_optab_handler_and_mode): Likewise. Provide an
8625 override that operates on mode class wrappers.
8626 * optabs-query.c (widening_optab_handler): Delete.
8627 (find_widening_optab_handler_and_mode): Remove permit_non_widening
8628 parameter. Assert that the two modes are the same class and that
8629 the "from" mode is narrower than the "to" mode. Use
8630 convert_optab_handler instead of widening_optab_handler.
8631 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
8632 instead of widening_optab_handler.
8633 * expr.c (expand_expr_real_2): Update calls to
8634 find_widening_optab_handler.
8635 * optabs.c (expand_widen_pattern_expr): Likewise.
8636 (expand_binop_directly): Take the insn_code as a parameter.
8637 (expand_binop): Only call find_widening_optab_handler for
8638 conversion optabs; use optab_handler otherwise. Update calls
8639 to find_widening_optab_handler and expand_binop_directly.
8640 Use convert_optab_handler instead of widening_optab_handler.
8641 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
8642 find_widening_optab_handler and use scalar_mode rather than
8644 (convert_plusminus_to_widen): Likewise.
8646 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8647 Alan Hayward <alan.hayward@arm.com>
8648 David Sherwood <david.sherwood@arm.com>
8650 * machmode.h (fixed_size_mode): New class.
8651 * rtl.h (get_pool_mode): Return fixed_size_mode.
8652 * gengtype.c (main): Add fixed_size_mode.
8653 * target.def (get_raw_result_mode): Return a fixed_size_mode.
8654 (get_raw_arg_mode): Likewise.
8655 * doc/tm.texi: Regenerate.
8656 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
8657 * targhooks.c (default_get_reg_raw_mode): Likewise.
8658 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
8659 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
8660 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
8661 (msp430_get_raw_result_mode): Likewise.
8662 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
8663 * dbxout.c (dbxout_parms): Require fixed-size modes.
8664 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
8665 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8666 * omp-low.c (lower_oacc_reductions): Likewise.
8667 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
8668 (simplify_subreg): Update accordingly.
8669 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
8670 (force_const_mem): Update accordingly. Return NULL_RTX for modes
8671 that aren't fixed-size.
8672 (get_pool_mode): Return a fixed_size_mode.
8673 (output_constant_pool_2): Take a fixed_size_mode.
8675 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8676 Alan Hayward <alan.hayward@arm.com>
8677 David Sherwood <david.sherwood@arm.com>
8679 * doc/rtl.texi (vec_series): Document.
8680 (const): Say that the operand can be a vec_series.
8681 * rtl.def (VEC_SERIES): New rtx code.
8682 * rtl.h (const_vec_series_p_1): Declare.
8683 (const_vec_series_p): New function.
8684 * emit-rtl.h (gen_const_vec_series): Declare.
8685 (gen_vec_series): Likewise.
8686 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
8687 (gen_vec_series): Likewise.
8688 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
8689 * simplify-rtx.c (simplify_unary_operation): Handle negations
8691 (simplify_binary_operation_series): New function.
8692 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
8693 (test_vector_ops_series): New function.
8694 (test_vector_ops): Call it.
8695 * config/powerpcspe/altivec.md (altivec_lvsl): Use
8696 gen_const_vec_series.
8697 (altivec_lvsr): Likewise.
8698 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
8700 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8701 Alan Hayward <alan.hayward@arm.com>
8702 David Sherwood <david.sherwood@arm.com>
8704 * doc/rtl.texi (const): Update description of address constants.
8705 Say that vector constants are allowed too.
8706 * common.md (E, F): Use CONSTANT_P instead of checking for
8708 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
8709 checking for CONST_VECTOR.
8710 * expmed.c (make_tree): Use build_vector_from_val for a CONST
8712 * expr.c (expand_expr_real_2): Check for vector modes instead
8713 of checking for CONST_VECTOR.
8714 * rtl.h (const_vec_p): New function.
8715 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
8716 (unwrap_const_vec_duplicate): Handle them here too.
8718 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8719 David Malcolm <dmalcolm@redhat.com>
8720 Alan Hayward <alan.hayward@arm.com>
8721 David Sherwood <david.sherwood@arm.com>
8723 * rtl.h (vec_duplicate_p): New function.
8724 * selftest-rtl.c (assert_rtx_eq_at): New function.
8725 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
8726 (assert_rtx_eq_at): Declare.
8727 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
8728 * selftest-run-tests.c (selftest::run_tests): Call it.
8729 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
8730 (simplify_unary_operation_1): Recursively handle vector duplicates.
8731 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
8733 (simplify_subreg): Handle subregs of vector duplicates.
8734 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
8735 (selftest::simplify_rtx_c_tests): New functions.
8737 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8738 Alan Hayward <alan.hayward@arm.com>
8739 David Sherwood <david.sherwood@arm.com>
8741 * emit-rtl.h (gen_const_vec_duplicate): Declare.
8742 (gen_vec_duplicate): Likewise.
8743 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
8745 (gen_const_vector): ...here.
8746 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
8747 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
8748 whose elements are all equal.
8749 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
8750 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
8751 (simplify_relational_operation): Likewise.
8752 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
8754 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
8755 (aarch64_expand_vector_init): Likewise.
8756 * config/arm/arm.c (neon_vdup_constant): Likewise.
8757 (neon_expand_vector_init): Likewise.
8758 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
8759 (arm_block_set_unaligned_vect): Likewise.
8760 (arm_block_set_aligned_vect): Likewise.
8761 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
8762 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
8763 (expand_vec_perm_even_odd_pack): Likewise.
8764 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
8765 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
8766 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
8767 gen_const_vec_duplicate.
8768 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
8769 * config/mips/mips.c (mips_gen_const_int_vector): Use
8770 gen_const_vec_duplicate.
8771 (mips_expand_vector_init): Use CONST0_RTX.
8772 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
8773 (define_split): Use gen_const_vec_duplicate.
8774 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
8775 (define_split): Use gen_const_vec_duplicate.
8776 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
8777 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
8778 * config/spu/spu.c (spu_const): Likewise.
8780 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8781 Alan Hayward <alan.hayward@arm.com>
8782 David Sherwood <david.sherwood@arm.com>
8784 * combine.c (can_change_dest_mode): Reject changes in
8785 REGMODE_NATURAL_SIZE.
8787 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
8789 * configure.ac (--enable-libssp): New.
8790 (gcc_cv_libc_provides_ssp): Check for explicit setting before
8791 trying to determine target-specific default. Adjust indentation.
8792 * configure: Regenerated.
8793 * doc/install.texi (Configuration): Expand --disable-libssp
8796 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
8798 config/i386/i386.c (ix86_expand_epilogue): Correct stack
8801 2017-10-31 Martin Jambor <mjambor@suse.cz>
8804 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
8806 2017-10-31 David Malcolm <dmalcolm@redhat.com>
8808 * auto-profile.c (autofdo_source_profile::read): Use
8809 UNKNOWN_LOCATION rather than 0.
8810 * diagnostic-core.h (warning_at_rich_loc): Rename to...
8811 (warning_at): ...this overload.
8812 (warning_at_rich_loc_n): Rename to...
8813 (warning_n): ...this overload.
8814 (error_at_rich_loc): Rename to...
8815 (error_at): ...this overload.
8816 (pedwarn_at_rich_loc): Rename to...
8817 (pedwarn): ...this overload.
8818 (permerror_at_rich_loc): Rename to...
8819 (permerror): ...this overload.
8820 (inform_at_rich_loc): Rename to...
8821 (inform): ...this overload.
8822 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
8823 (diagnostic_n_impl_richloc): Rename to...
8824 (diagnostic_n_impl): ...this rich_location *-based decl.
8825 (inform_at_rich_loc): Rename to...
8826 (inform): ...this, and add an assertion.
8827 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
8828 (warning_at_rich_loc): Rename to...
8829 (warning_at): ...this, and add an assertion.
8830 (warning_at_rich_loc_n): Rename to...
8831 (warning_n): ...this, and add an assertion.
8832 (warning_n): Update location_t-based implementation for removal of
8833 location_t-based diagnostic_n_impl.
8834 (pedwarn_at_rich_loc): Rename to...
8835 (pedwarn): ...this, and add an assertion.
8836 (permerror_at_rich_loc): Rename to...
8837 (permerror): ...this, and add an assertion.
8838 (error_n): Update for removal of location_t-based diagnostic_n_impl.
8839 (error_at_rich_loc): Rename to...
8840 (error_at): ...this, and add an assertion.
8841 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
8842 (driver::do_spec_on_infiles): Likewise.
8843 * substring-locations.c (format_warning_va): Update for renaming
8844 of inform_at_rich_loc.
8846 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
8848 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
8849 _Float<N>X built-in functions so that the variant without the
8850 "__builtin_" prefix is only enabled for the GNU C and Objective C
8851 languages when they are in non-strict ANSI/ISO mode.
8852 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
8853 * target.def (floatn_builtin_p): Add a target hook to control
8854 whether _Float<N> and _Float<N>X built-in functions without the
8855 "__builtin_" prefix are enabled, and return true for C and
8856 Objective C in the default hook. Include langhooks.h in
8858 * targhooks.h (default_floatn_builtin_p): Likewise.
8859 * targhooks.c (default_floatn_builtin_p): Likewise.
8860 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
8861 floatn_builtin_p target hook.
8862 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
8864 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
8865 Eric Botcazou <ebotcazou@adacore.com>
8867 PR rtl-optimization/81803
8868 * lra-constraints.c (curr_insn_transform): Also reload the whole
8869 register for a strict subreg no wider than a word if this is for
8870 a WORD_REGISTER_OPERATIONS target.
8872 2017-10-31 Jason Merrill <jason@redhat.com>
8874 * gdbinit.in: Skip over inlines from timevar.h.
8876 2017-10-31 Martin Liska <mliska@suse.cz>
8878 * doc/gcov.texi: Document new option.
8879 * gcov.c (print_usage): Likewise print it.
8880 (process_args): Support the argument.
8881 (format_count): New function.
8882 (format_gcov): Use the function.
8884 2017-10-31 Martin Liska <mliska@suse.cz>
8886 * gcov.c (struct name_map): do not use typedef.
8887 Define operator== and operator<.
8888 (name_search): Remove.
8889 (name_sort): Remove.
8890 (main): Do not allocate names.
8891 (process_file): Add vertical space.
8892 (generate_results): Use std::find.
8893 (release_structures): Do not release memory.
8894 (find_source): Use std::find.
8896 2017-10-31 Martin Liska <mliska@suse.cz>
8898 * gcov.c (struct line_info): Remove it's typedef.
8899 (line_info::line_info): Add proper ctor.
8900 (line_info::has_block): Do not use a typedef.
8901 (struct source_info): Do not use typedef.
8902 (circuit): Likewise.
8903 (get_cycles_count): Likewise.
8904 (output_intermediate_file): Iterate via vector iterator.
8905 (add_line_counts): Use std::vector methods.
8906 (accumulate_line_counts): Likewise.
8907 (output_lines): Likewise.
8909 2017-10-31 Martin Liska <mliska@suse.cz>
8911 * gcov.c (struct source_info): Remove typedef.
8912 (source_info::source_info): Add proper ctor.
8913 (accumulate_line_counts): Use struct, not it's typedef.
8914 (output_gcov_file): Likewise.
8915 (output_lines): Likewise.
8916 (main): Do not allocate an array.
8917 (output_intermediate_file): Use size of vector container.
8918 (process_file): Resize the vector.
8919 (generate_results): Do not preallocate, use newly added vector
8921 (release_structures): Do not release sources.
8922 (find_source): Use vector methods.
8923 (add_line_counts): Do not use typedef.
8925 2017-10-31 Martin Liska <mliska@suse.cz>
8927 * doc/gcov.texi: Document that.
8928 * gcov.c (add_line_counts): Mark lines with a non-executed
8930 (output_line_beginning): Handle such lines.
8931 (output_lines): Pass new argument.
8932 (output_intermediate_file): Print it in intermediate format.
8934 2017-10-31 Martin Liska <mliska@suse.cz>
8936 * color-macros.h: New file.
8937 * diagnostic-color.c: Factor out color related to macros to
8939 * doc/gcov.texi: Document -k option.
8940 * gcov.c (INCLUDE_STRING): Include string.h.
8941 (print_usage): Add -k option.
8942 (process_args): Parse it.
8943 (pad_count_string): New function.
8944 (output_line_beginning): Likewise.
8945 (DEFAULT_LINE_START): New macro.
8946 (output_lines): Support color output.
8948 2017-10-31 Martin Liska <mliska@suse.cz>
8950 PR gcov-profile/82633
8951 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
8952 their interaction with GCOV infrastructure.
8953 * configure.ac: Add -fkeep-{inline,static}-functions to
8955 * configure: Regenerate.
8957 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
8960 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
8962 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
8965 * config/rs6000/rs6000.md (allocate_stack): Force update interval
8966 into a register if it does not fit into an immediate offset field.
8968 2017-10-31 Olivier Hainque <hainque@adacore.com>
8970 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
8972 2017-10-31 Julia Koval <julia.koval@intel.com>
8974 * config.gcc: Add gfniintrin.h.
8975 * config/i386/gfniintrin.h: New.
8976 * config/i386/i386-builtin-types.def
8977 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8978 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8979 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8980 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8981 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8982 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
8983 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
8984 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
8985 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
8986 V64QI_FTYPE_V64QI_V64QI_INT): New types.
8987 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
8988 * config/i386/immintrin.h: Include gfniintrin.h.
8989 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
8991 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
8993 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
8995 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8997 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
8998 (arm_ashldi3_1bit): Remove pattern.
8999 (ashrdi3): Remove shift by 1 expansion.
9000 (arm_ashrdi3_1bit): Remove pattern.
9001 (lshrdi3): Remove shift by 1 expansion.
9002 (arm_lshrdi3_1bit): Remove pattern.
9003 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
9004 cost of ashldi3 by 1.
9005 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
9006 (<shift>di3_neon): Likewise.
9008 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9010 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
9011 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
9012 and (*aarch64_simd_mov<VQ:mode>).
9013 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
9014 pattern alternative.
9015 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
9016 attributes to match pattern alternative.
9018 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
9020 * config.gcc (powerpc*-*-*): Add emmintrin.h.
9021 * config/rs6000/emmintrin.h: New file.
9022 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
9024 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9026 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
9027 * (movdi_vfp_cortexa8): Remove pattern.
9029 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9031 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
9033 (Specific, alpha*-dec-osf5.1): Remove.
9034 (Specific, mips-sgi-irix5): Remove.
9035 (Specific, mips-sgi-irix6): Remove.
9037 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9040 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
9041 arguments to clear_bit_region_be.
9043 2017-10-30 Jim Wilson <wilson@tuliptree.org>
9045 * gimplify.c: Include memmodel.h.
9047 2017-10-30 Martin Jambor <mjambor@suse.cz>
9049 * omp-grid.c (grid_attempt_target_gridification): Also insert a
9050 condition whether loop should be executed at all.
9052 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
9054 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
9055 gimple folding of vec_madd() intrinsics.
9056 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
9057 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
9058 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
9060 2017-10-30 Richard Biener <rguenther@suse.de>
9062 PR tree-optimization/82762
9064 2017-10-23 Richard Biener <rguenther@suse.de>
9066 PR tree-optimization/82129
9068 2017-08-01 Richard Biener <rguenther@suse.de>
9070 PR tree-optimization/81181
9071 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9072 (compute_antic): ... end of iteration here.
9074 2017-10-30 Joseph Myers <joseph@codesourcery.com>
9076 * doc/invoke.texi (C Dialect Options): Document -std=c17,
9077 -std=iso9899:2017 and -std=gnu17.
9078 * doc/standards.texi (C Language): Document C17 support.
9079 * doc/cpp.texi (Overview): Mention -std=c17.
9080 (Standard Predefined Macros): Document C11 and C17 values of
9081 __STDC_VERSION__. Do not refer to C99 support as incomplete.
9082 * doc/extend.texi (Inline): Do not list individual options for
9083 standards newer than C99.
9084 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
9086 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
9089 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
9091 * asan.c (asan_finish_file): Align asan globals array by shadow
9094 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9097 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
9098 (struct store_immediate_info): Add bitregion_start and bitregion_end
9100 (store_immediate_info::store_immediate_info): Add brs and bre
9101 arguments and initialize bitregion_{start,end} from those.
9102 (struct merged_store_group): Add bitregion_start, bitregion_end,
9103 align_base and mask fields. Drop unnecessary struct keyword from
9104 struct store_immediate_info. Add do_merge method.
9105 (clear_bit_region_be): Use memset instead of loop storing zeros.
9106 (merged_store_group::do_merge): New method.
9107 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
9108 stores as long as the surrounding bitregions have no gaps.
9109 (merged_store_group::merge_overlapping): Use do_merge.
9110 (merged_store_group::apply_stores): Test that bitregion_{start,end}
9111 is byte aligned, rather than requiring that start and width are
9112 byte aligned. Drop unnecessary struct keyword from
9113 struct store_immediate_info. Allocate and populate also mask array.
9114 Make start of the arrays relative to bitregion_start rather than
9115 start and size them according to bitregion_{end,start} difference.
9116 (struct imm_store_chain_info): Drop unnecessary struct keyword from
9117 struct store_immediate_info.
9118 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
9119 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
9120 struct keyword from struct store_immediate_info.
9121 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
9122 between stores as long as the surrounding bitregions have no gaps.
9124 (struct split_store): Add orig non-static data member.
9125 (split_store::split_store): Initialize orig to false.
9126 (find_constituent_stmts): Return store_immediate_info *, non-NULL
9127 if there is exactly a single original stmt. Change stmts argument
9128 to pointer from reference, if NULL, don't push anything to it. Add
9129 first argument, use it to optimize skipping over orig stmts that
9130 are known to be before bitpos already. Simplify.
9131 (split_group): Return unsigned int count how many stores are or
9132 would be needed rather than a bool. Add allow_unaligned argument.
9133 Change split_stores argument from reference to pointer, if NULL,
9134 only do a dry run computing how many stores would be produced.
9135 Rewritten algorithm to use both alignment and misalign if
9136 !allow_unaligned and handle bitfield stores with gaps.
9137 (imm_store_chain_info::output_merged_store): Set start_byte_pos
9138 from bitregion_start instead of start. Compute allow_unaligned
9139 here, if true, do 2 split_group dry runs to compute which one
9140 produces fewer stores and prefer aligned if equal. Punt if
9141 new count is bigger or equal than original before emitting any
9142 statements, rather than during that. Remove no longer needed
9143 new_ssa_names tracking. Replace num_stmts with
9144 split_stores.length (). Use 32-bit stack allocated entries
9145 in split_stores auto_vec. Try to reuse original store lhs/rhs1
9146 if possible. Handle bitfields with gaps.
9147 (pass_store_merging::execute): Ignore bitsize == 0 stores.
9148 Compute bitregion_{start,end} for the stores and construct
9149 store_immediate_info with that. Formatting fixes.
9151 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
9154 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
9155 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
9157 2017-10-29 Jim Wilson <wilson@tuliptree.org>
9159 * gimplify.c: Include tm_p.h.
9161 * common.opt (gcoff): Re-add as ignored option.
9162 (gcoff1, gcoff2, gcoff3): Likewise.
9164 * Makefile.in (OBJS): Delete sdbout.o.
9165 (GTFILES): Delete $(srcdir)/sdbout.c.
9166 * debug.h: Delete sdb_debug_hooks.
9167 * final.c: Delete sdbout.h include.
9168 (final_scan_insn): Delete SDB_DEBUG check.
9169 (rest_of_clean_state): Likewise.
9170 * output.h: Delete sdb_begin_function_line.
9173 * toplev.c: Delete sdbout.h include.
9174 (process_options): Delete SDB_DEBUG check.
9175 * tree-core.h (tree_type_common): Delete pointer field of
9177 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
9178 TYPE_SYMTAB_POINTER.
9179 * tree.h (TYPE_SYMTAB_POINTER): Delete.
9180 (TYPE_SYMTAB_IS_POINTER): Delete.
9181 (TYPE_SYMTAB_IS_DIE): Renumber.
9182 * xcoffout.c: Refer to former sdbout.c file.
9183 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
9185 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
9186 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
9187 (Debugging Options): Delete -gcoff.
9188 (-gstabs): Delete SDB reference.
9190 (-gcoff@var{level}): Delete.
9191 * doc/passes.texi (Debugging information output): Delete SDB and
9192 sdbout.c references.
9193 * doc/tm.texi: Regenerate.
9194 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
9195 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
9197 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
9198 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
9199 (SDB_DEBUGGING_INFO): Delete.
9200 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
9201 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
9202 * target.def (output_source_filename): Delete COFF reference.
9204 * common.opt (gcoff): Delete.
9205 (gxcoff+): Update Negative chain.
9206 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
9208 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
9209 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
9210 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
9212 (expand_function_start): Change sdb reference to past tense.
9213 (expand_function_end): Change sdb reference to past tense.
9214 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
9215 * opts.c (debug_type_names): Delete coff entry.
9216 (common_handle_option): Delete OPT_gcoff case.
9217 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
9219 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
9220 * config/cris/cris.h: Delete SDB reference in comment.
9221 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
9222 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
9223 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
9224 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
9226 (ix86_expand_prologue): Likewise.
9227 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
9228 * config/ia64/ia64.h: Likewise.
9229 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
9230 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
9232 * config/mmix/mmix.h: Likewise.
9233 * config/nds32/nds32.c: Likewise.
9234 * config/stormy/storym16.h: Likewise.
9235 * config/visium/visium.h: Likewise.
9236 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
9238 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
9240 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
9241 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
9242 FRAME_POINTER_REGNUM point at high end of local var area.
9244 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
9246 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
9247 Move comment around. Do not reset best_edge for a copiable
9248 destination if the copy would cause a partition change.
9249 (better_edge_p): Remove redundant check.
9251 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9253 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
9255 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
9257 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
9258 for math functions that have _Float<N> and _Float<N>X variants.
9259 (mathfn_built_in_2): Add support for math functions that have
9260 _Float<N> and _Float<N>X variants.
9261 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
9262 (expand_builtin_mathfn_ternary): Add support for fma with
9263 _Float<N> and _Float<N>X variants.
9264 (expand_builtin): Likewise.
9265 (fold_builtin_3): Likewise.
9266 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
9267 create math function _Float<N> and _Float<N>X variants as external
9269 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
9270 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
9271 the __builtin_ prefix and if not strict ansi, without the prefix.
9272 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
9273 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
9274 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
9275 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
9276 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
9277 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
9278 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
9279 function signatures for fma _Float<N> and _Float<N>X variants.
9280 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
9281 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
9282 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
9283 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
9284 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
9285 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
9286 * gencfn-macros.c (print_case_cfn): Add support for math functions
9287 that have _Float<N> and _Float<N>X variants.
9288 (print_define_operator_list): Likewise.
9289 (fltfn_suffixes): Likewise.
9291 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
9292 for math functions that have _Float<N> and _Float<N>X variants.
9293 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
9294 and _Float<N>X variants.
9295 (COPYSIGN): Likewise.
9298 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
9299 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
9301 (integer_valued_read_call_p): Likewise.
9302 * fold-const-call.c (fold_const_call_ss): Likewise.
9303 (fold_const_call_sss): Add support for copysign, fmin, and fmax
9304 _Float<N> and _Float<N>X variants.
9305 (fold_const_call_ssss): Add support for fma _Float<N> and
9306 _Float<N>X variants.
9307 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
9308 support for copysign and fma _Float<N> and _Float<N>X variants.
9309 (backprop::process_builtin_call_use): Likewise.
9310 * tree-call-cdce.c (can_test_argument_range); Add support for
9311 sqrt _Float<N> and _Float<N>X variants.
9312 (edom_only_function): Likewise.
9313 (get_no_error_domain): Likewise.
9314 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
9315 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
9316 copysign _Float<N> and _Float<N>X variants.
9317 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
9318 handled by machine independent code.
9319 (FMAF128): Likewise.
9320 * doc/cpp.texi (Common Predefined Macros): Document defining
9321 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
9322 fma _Float<N> and _Float<N>X variants.
9324 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9327 * config/i386/i386-modes.def (CCFPU): Remove definition.
9328 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
9329 (ix86_cc_modes_compatible): Ditto.
9330 (ix86_expand_carry_flag_compare): Ditto.
9331 (ix86_expand_int_movcc): Ditto.
9332 (ix86_expand_int_addcc): Ditto.
9333 (ix86_reverse_condition): Ditto.
9334 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
9335 Return true/false for unordered/ordered fp comparisons.
9336 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
9337 (ix86_prepare_fp_compare_args): Update for rename.
9338 (ix86_expand_fp_compare): Update for rename. Generate unordered
9339 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
9340 (ix86_expand_sse_compare_and_jump): Ditto.
9341 * config/i386/predicates.md (fcmov_comparison_operator):
9342 Remove CCFPU mode handling.
9343 (ix86_comparison_operator): Ditto.
9344 (ix86_carry_flag_operator): Ditto.
9345 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
9346 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
9347 (*cmpu<mode>_cc_i387): Ditto.
9348 (FPCMP): Remove mode iterator.
9349 (unord): Remove mode attribute.
9350 (unord_subst): New define_subst transformation
9351 (unord): New define_subst attribute.
9353 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
9354 (*cmpi<unord>xf_i387): Ditto.
9355 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
9356 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
9357 using unord_subst transformation.
9358 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
9359 (round_saeonly): Also handle CCFP mode.
9360 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
9361 Remove UNSPEC_SAHF unspec handling.
9363 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9365 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
9367 2017-10-27 Jeff Law <law@redhat.com>
9369 * gimple-ssa-sprintf.c: Include domwalk.h.
9370 (class sprintf_dom_walker): New class, derived from dom_walker.
9371 (sprintf_dom_walker::before_dom_children): New function.
9372 (struct call_info): Moved into sprintf_dom_walker class
9373 (compute_formath_length, handle_gimple_call): Likewise.
9374 (sprintf_length::execute): Call the dominator walker rather
9375 than walking the statements.
9377 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
9378 gimple statement locations.
9379 (check_array_bounds): Corresponding changes. Get the statement's
9380 location directly from wi->stmt.
9382 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
9385 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
9387 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9389 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
9390 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
9392 2017-10-27 Jakub Jelinek <jakub@redhat.com>
9395 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
9396 * config/i386/i386.c (maybe_get_pool_constant): Removed.
9397 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
9398 maybe_get_pool_constant.
9399 * config/i386/predicates.md (zero_extended_scalar_load_operand):
9402 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9404 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
9405 2.26 caveat. Update gas and gld versions.
9406 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
9409 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9411 * cgraph.h (set_malloc_flag): Declare.
9412 * cgraph.c (set_malloc_flag_1): New function.
9413 (set_malloc_flag): Likewise.
9414 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
9415 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
9417 (read_ipa_call_summary): Add support for reading is_return_callee.
9418 (write_ipa_call_summary): Stream is_return_callee.
9419 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
9420 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
9421 ipa-prop.h, ipa-fnsummary.h.
9422 (pure_const_names): Change to static.
9423 (malloc_state_e): Define.
9424 (malloc_state_names): Define.
9425 (funct_state_d): Add field malloc_state.
9426 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
9427 (check_retval_uses): New function.
9428 (malloc_candidate_p): Likewise.
9429 (analyze_function): Add support for malloc attribute.
9430 (pure_const_write_summary): Stream malloc_state.
9431 (pure_const_read_summary): Add support for reading malloc_state.
9432 (dump_malloc_lattice): New function.
9433 (propagate_malloc): New function.
9434 (warn_function_malloc): New function.
9435 (ipa_pure_const::execute): Call propagate_malloc and
9436 ipa_free_fn_summary.
9437 (pass_local_pure_const::execute): Add support for malloc attribute.
9438 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
9439 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
9441 2017-10-27 Martin Liska <mliska@suse.cz>
9443 PR gcov-profile/82457
9444 * doc/invoke.texi: Document that one needs a non-strict ISO mode
9445 for fork-like functions to be properly instrumented.
9447 2017-10-27 Richard Biener <rguenther@suse.de>
9450 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
9451 info when we redirected EH.
9453 2017-10-26 Michael Collison <michael.collison@arm.com>
9455 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
9457 (<optab>_trunchf<GPI:mode>2: New pattern.
9458 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
9459 * config/aarch64/iterators.md (wv): New mode attribute.
9460 (vf, VF): New mode attributes.
9461 (vgp, VGP): New mode attributes.
9462 (s): Update attribute with SImode and DImode prefixes.
9464 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9466 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
9467 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
9468 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
9469 (nios2_option_overide): Initialize it. Don't allow R0-relative
9470 addressing with PIC.
9471 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
9472 (nios2_symbolic_constant_p): Likewise.
9473 (nios2_legitimate_address_p): Likewise.
9474 (nios2_r0rel_section_name_p): New.
9475 (nios2_symbol_ref_in_r0rel_data_p): New.
9476 (nios2_emit_move_sequence): Handle r0rel_constant_p.
9477 (r0rel_constant_p): New.
9478 (nios2_print_operand_address): Handle r0rel_constant_p.
9479 (nios2_cdx_narrow_form_p): Likewise.
9480 * config/nios2/nios2.opt (mr0rel-sec=): New option.
9481 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
9482 (Nios II Options): Document -mr0rel-sec.
9484 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9486 * config/nios2/nios2.c: Include xregex.h.
9487 (nios2_gprel_sec_regex): New.
9488 (nios2_option_overide): Initialize it. Don't allow GP-relative
9489 addressing with PIC.
9490 (nios2_small_section_name_p): Check for regex match.
9491 * config/nios2/nios2.opt (mgprel-sec=): New option.
9492 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
9493 (Nios II Options): Document -mgprel-sec.
9495 2017-10-26 Jim Wilson <wilson@tuliptree.org>
9497 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
9499 2017-10-26 Tom de Vries <tom@codesourcery.com>
9501 PR tree-optimization/82707
9502 * gimple.c (gimple_copy): Fix unsharing of
9503 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
9505 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
9507 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
9508 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
9509 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
9510 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
9511 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
9512 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
9513 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
9514 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
9515 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
9516 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
9517 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
9518 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
9519 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
9520 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
9521 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
9522 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
9523 _mm512_mask_cmpunord_ps_mask): New intrinsics.
9525 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9527 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
9529 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
9530 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
9531 warning to rs6000.c. Remove the Undocumented flag, since it has
9533 (-mabi=ibmlongdouble): Likewise.
9534 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
9535 already set, set the default format for long double.
9536 (rs6000_debug_reg_global): Print whether long double is IBM or
9538 (rs6000_option_override_internal): Rework setting long double
9539 format. Only warn if the user is changing the long double default
9540 and they did not use -Wno-psabi.
9541 * doc/invoke.texi (PowerPC options): Update the documentation for
9542 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
9544 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9545 Alan Hayward <alan.hayward@arm.com>
9546 David Sherwood <david.sherwood@arm.com>
9548 * rtl.h (wider_subreg_mode): New function.
9549 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
9550 rather than an unsigned int *.
9551 * ira-color.c (regno_max_ref_width): Replace with...
9552 (regno_max_ref_mode): ...this new variable.
9553 (coalesced_pseudo_reg_slot_compare): Update accordingly.
9554 Use wider_subreg_mode.
9555 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
9556 rather than an unsigned int *.
9557 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
9558 (process_alt_operands): Likewise.
9559 (invariant_p): Likewise.
9560 * lra-spills.c (assign_mem_slot): Likewise.
9561 (add_pseudo_to_slot): Likewise.
9562 * lra.c (collect_non_operand_hard_regs): Likewise.
9563 (add_regs_to_insn_regno_info): Likewise.
9564 * reload1.c (regno_max_ref_width): Replace with...
9565 (regno_max_ref_mode): ...this new variable.
9566 (reload): Update accordingly. Update call to
9567 ira_sort_regnos_for_alter_reg.
9568 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
9569 (init_eliminable_invariants): Update to use regno_max_ref_mode.
9570 (scan_paradoxical_subregs): Likewise.
9572 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9574 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
9575 (aarch64_frame): Add emit_frame_chain boolean.
9576 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9577 Move eh_return case to aarch64_layout_frame.
9578 (aarch64_layout_frame): Initialize emit_frame_chain.
9579 (aarch64_expand_prologue): Use emit_frame_chain.
9581 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9583 * config/aarch64/aarch64.c (aarch64_layout_frame):
9584 Ensure LR is always stored at the bottom of the callee-saves.
9585 Remove rarely used frame layout which saves callee-saves at top of
9586 frame, so the store of LR can be used as a valid probe in all cases.
9588 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9590 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
9591 Improve unaligned TImode/TFmode base/offset split.
9593 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9594 Alan Hayward <alan.hayward@arm.com>
9595 David Sherwood <david.sherwood@arm.com>
9597 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
9598 * combine.c (find_single_use_1): Likewise.
9599 (expand_field_assignment): Likewise.
9600 (move_deaths): Likewise.
9601 * lra-constraints.c (simplify_operand_subreg): Likewise.
9602 (curr_insn_transform): Likewise.
9603 * lra.c (collect_non_operand_hard_regs): Likewise.
9604 (add_regs_to_insn_regno_info): Likewise.
9605 * rtlanal.c (reg_referenced_p): Likewise.
9606 (covers_regno_no_parallel_p): Likewise.
9608 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9610 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
9611 Don't print any bits outside the precision of the value.
9612 * wide-int.cc (test_printing): Add some new tests.
9614 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9616 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
9617 supports -xbrace_comment option.
9618 * configure: Regenerate.
9619 * config.in: Regenerate.
9620 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
9621 (ASM_CPU_SPEC): Use it.
9623 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9625 * target.def (static_rtx_alignment): New hook.
9626 * targhooks.h (default_static_rtx_alignment): Declare.
9627 * targhooks.c (default_static_rtx_alignment): New function.
9628 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
9629 * doc/tm.texi: Regenerate.
9630 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
9631 instead of targetm.constant_alignment. Remove call to
9633 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9634 (cris_preferred_mininum_alignment): New function, split out from...
9635 (cris_constant_alignment): ...here.
9636 (cris_static_rtx_alignment): New function.
9637 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
9639 (ix86_constant_alignment): ...here.
9640 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9641 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9642 (mmix_static_rtx_alignment): New function.
9643 * config/spu/spu.c (spu_static_rtx_alignment): New function.
9644 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9646 2017-10-26 Tamar Christina <tamar.christina@arm.com>
9649 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
9650 Add flag_trapping_math and flag_fp_int_builtin_inexact.
9652 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
9654 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
9655 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
9656 mark as a sign-extending load.
9657 (local_pic_load_u): Define.
9659 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
9662 * fold-const.c (operand_equal_for_comparison_p): Also return true
9663 if ARG0 is a simple variant of ARG1 with narrower precision.
9664 (fold_ternary_loc): Always pass unstripped operands to the predicate.
9666 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
9668 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
9670 * i386.h (processor_costs): Add gather_static, gather_per_elt,
9671 scatter_static, scatter_per_elt.
9672 * x86-tune-costs.h: Add new cost entries.
9674 2017-10-25 Richard Biener <rguenther@suse.de>
9676 * tree-ssa-sccvn.h (vn_eliminate): Declare.
9677 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
9678 class pass_fre): Move to ...
9679 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
9680 class pass_fre): ... here and adjust for statistics.
9682 2017-10-25 Jakub Jelinek <jakub@redhat.com>
9685 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
9686 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
9687 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
9688 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
9691 2017-10-25 Richard Biener <rguenther@suse.de>
9693 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
9694 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
9695 eliminate_push_avail, eliminate_insert): Move inside...
9696 (class eliminate_dom_walker): ... this class in preparation
9698 (fini_eliminate): Remove by merging with ...
9699 (eliminate): ... this function. Adjust for class changes.
9700 (pass_pre::execute): Remove fini_eliminate call.
9701 (pass_fre::execute): Likewise.
9703 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9706 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
9707 (VPERMI2, VPERMI2I): New mode iterators.
9708 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
9709 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
9711 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
9712 mode iterator. Remove 3 old define_insn patterns.
9713 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
9714 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
9715 VPERMI2 mode iterator, remove the other two expanders.
9716 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
9717 to use VPERMI2 mode iterator, add another alternative for vpermi2*
9718 instructions, remove the other two patterns.
9719 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
9720 mode iterator, remove the other two patterns.
9721 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
9722 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
9723 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
9724 and adjust argument order accordingly.
9725 (ix86_expand_vec_perm): Adjust caller.
9726 (expand_vec_perm_1): Likewise.
9727 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
9728 (expand_vec_perm_vpermt2_vpshub2): ... this.
9729 (ix86_expand_vec_perm_const_1): Adjust caller.
9730 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
9733 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
9734 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
9735 (vec_shl_<mode>): Remove unused expander.
9736 (avx512bw_<shift_insn><mode>3): New define_insn.
9737 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
9738 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
9740 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
9743 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
9746 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9748 PR rtl-optimization/82396
9749 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
9750 (autopref_multipass_init): Simplify initialization.
9751 (autopref_rank_data): Simplify sort order.
9752 * gcc/sched-int.h (autopref_multipass_data_): Remove
9753 multi_mem_insn_p, min_offset and max_offset.
9755 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9758 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9759 Check special value of flag_omit_frame_pointer.
9760 (aarch64_can_eliminate): Likewise.
9761 (aarch64_override_options_after_change_1): Simplify handling of
9762 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
9764 2017-10-24 Richard Biener <rguenther@suse.de>
9766 PR tree-optimization/82697
9767 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
9768 zero for conditional load and unconditional store.
9770 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9772 * doc/install.texi: Document bootstrap-cet.
9774 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9777 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
9778 ENDBR instruction at function entrance if function is only
9781 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9784 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
9785 patterns to better describe from which operation the CF is computed.
9786 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
9787 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
9788 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
9789 is 0, use _0 suffixed expanders instead of emitting a comparison
9792 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
9794 * config/i386/i386.md(*movsf_internal, *movdf_internal):
9795 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
9797 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
9800 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
9801 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
9802 * loop-iv.c (iv_get_reaching_def): Likewise.
9803 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
9804 variable is promoted and the partition contains undefined values.
9806 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9808 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
9810 (nios2_address_cost): Define.
9811 (nios2_legitimize_address): Recognize (exp + constant) directly.
9812 (TARGET_ADDRESS_COST): Define.
9814 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9816 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
9817 (nios2_symbolic_memory_operand_p): Declare.
9818 (nios2_split_large_constant): Declare.
9819 (nios2_split_symbolic_memory_operand): Declare.
9820 * config/nios2/nios2.c: Adjust includes.
9821 (nios2_symbolic_constant_allowed): New.
9822 (nios2_symbolic_constant_p): New.
9823 (nios2_plus_symbolic_constant_p): New.
9824 (nios2_valid_addr_expr_p): Recognize addresses involving
9826 (nios2_legitimate_address_p): Likewise, also LO_SUM.
9827 (nios2_symbolic_memory_operand_p): New.
9828 (nios2_large_constant_p): New.
9829 (nios2_split_large_constant): New.
9830 (nios2_split_plus_large_constant): New.
9831 (nios2_split_symbolic_memory_operand): New.
9832 (nios2_legitimize_address): Code refactoring. Handle addresses
9833 involving symbolic constants.
9834 (nios2_emit_move_sequence): Likewise.
9835 (nios2_print_operand): Improve error output.
9836 (nios2_print_operand_address): Handle LO_SUM.
9837 (nios2_cdx_narrow_form_p): Likewise.
9838 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
9839 operands involving symbolic constants.
9840 (movhi_internal, movsi_internal): Likewise.
9841 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
9842 (extendhisi2, extendqi<mode>2): Likewise.
9844 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9846 * tree-pass.h (PROP_rtl_split_insns): Define.
9847 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
9849 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9851 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
9853 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9856 * target.def (const_not_ok_for_debug_p): Default to
9857 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
9858 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
9859 * targhooks.c (default_const_not_ok_for_debug_p): New function.
9860 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
9861 which targetm.const_not_ok_for_debug_p returned true.
9862 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
9864 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
9866 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
9867 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
9868 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
9870 (ix86_const_not_ok_for_debug_p): New function.
9871 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
9872 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
9874 2017-10-23 David Malcolm <dmalcolm@redhat.com>
9877 * system.h: Conditionally include "unique-ptr.h" if
9878 INCLUDE_UNIQUE_PTR is defined.
9879 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
9880 of defining INCLUDE_UNIQUE_PTR before including "system.h".
9882 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
9884 * config/rl78/rl78.md: New define_expand "subdi3".
9886 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
9889 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
9890 DF_REF_INSN if DF_REF_INSN_INFO is false.
9892 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
9894 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
9895 xmm_move instead of sse_move.
9896 (sse_store_index): New function.
9897 (ix86_register_move_cost): Be more sensible about mismatch stall;
9898 model AVX moves correctly; make difference between sse->integer and
9900 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
9901 moves; make difference between SSE and AVX.
9902 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
9903 and zmm_move. Increase size of sse load and store tables;
9904 add unaligned load and store tables; add ssemmx_to_integer.
9905 * x86-tune-costs.h: Update all entries according to real
9906 move latencies from Agner Fog's manual and chip documentation.
9908 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9911 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
9912 * config/i386/constraints.md (Wf): New constraint.
9913 * config/i386/i386.md (UNSPEC_SBB): New unspec.
9914 (cmp<dwi>_doubleword): Removed.
9915 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
9916 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
9917 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
9918 expand with cmp<dwi>_doubleword. For LTU and GEU use
9919 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
9921 * common.opt (gcolumn-info): Enable by default.
9922 * doc/invoke.texi (gcolumn-info): Document new default.
9924 2017-10-23 Richard Biener <rguenther@suse.de>
9926 PR tree-optimization/82672
9927 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
9928 Fold the stmt if we propagated into it.
9930 2017-10-23 Richard Biener <rguenther@suse.de>
9932 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
9933 (bitmap_remove_expr_from_set): ... this. All callers call this
9934 for non-constant values.
9935 (bitmap_set_subtract): Rename to...
9936 (bitmap_set_subtract_expressions): ... this. Adjust and
9938 (bitmap_set_contains_value): Remove superfluous check.
9939 (bitmap_set_replace_value): Inline into single caller ...
9940 (bitmap_value_replace_in_set): ... here and simplify.
9941 (dependent_clean): Merge into ...
9942 (clean): ... this using an overload. Adjust.
9943 (prune_clobbered_mems): Adjust.
9944 (compute_antic_aux): Likewise.
9945 (compute_partial_antic_aux): Likewise.
9947 2017-10-23 Richard Biener <rguenther@suse.de>
9949 PR tree-optimization/82129
9951 2017-08-01 Richard Biener <rguenther@suse.de>
9953 PR tree-optimization/81181
9954 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9955 (compute_antic): ... end of iteration here.
9957 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9959 * target.def (starting_frame_offset): New hook.
9960 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
9961 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
9962 * doc/tm.texi.in: Regenerate.
9963 * hooks.h (hook_hwi_void_0): Declare.
9964 * hooks.c (hook_hwi_void_0): New function.
9965 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
9966 STARTING_FRAME_OFFSET.
9967 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
9968 * reload1.c (reload): Likewise.
9969 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
9970 instead of STARTING_FRAME_OFFSET.
9971 * function.c (try_fit_stack_local): Likewise.
9972 (assign_stack_local_1): Likewise
9973 (instantiate_virtual_regs): Likewise.
9974 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
9975 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
9976 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
9977 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
9978 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
9979 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
9980 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
9981 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
9982 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
9983 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
9984 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
9985 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
9986 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
9987 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
9988 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
9989 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
9990 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
9991 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
9992 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
9993 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
9994 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
9995 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
9996 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
9997 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
9998 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
9999 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
10000 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
10001 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
10002 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
10003 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
10004 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
10005 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
10006 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
10007 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
10008 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
10009 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
10010 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
10011 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
10012 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
10013 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
10014 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
10015 * config/avr/avr.c (avr_starting_frame_offset): Make static and
10016 return a HOST_WIDE_INT.
10017 (avr_builtin_setjmp_frame_value): Use it instead of
10018 STARTING_FRAME_OFFSET.
10019 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10020 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
10021 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
10023 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10024 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
10025 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
10026 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10027 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
10028 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
10029 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10030 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
10031 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
10032 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10033 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
10034 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
10036 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10037 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
10038 * config/mips/mips.c (mips_compute_frame_info): Refer to
10039 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10040 (mips_starting_frame_offset): New function.
10041 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10042 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
10043 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
10044 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
10045 and return a HOST_WIDE_INT.
10046 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10047 (mmix_initial_elimination_offset): Refer to
10048 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10049 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
10050 * config/pa/pa.c (pa_starting_frame_offset): New function.
10051 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
10052 (pa_expand_prologue): Likewise.
10053 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10054 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
10055 !FRAME_GROWS_DOWNWARD handling to...
10056 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10057 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
10058 !FRAME_GROWS_DOWNWARD handling to...
10059 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10060 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
10061 !FRAME_GROWS_DOWNWARD handling to...
10062 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10063 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
10065 (rs6000_starting_frame_offset): New function.
10066 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
10067 !FRAME_GROWS_DOWNWARD handling to...
10068 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10069 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
10070 !FRAME_GROWS_DOWNWARD handling to...
10071 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10072 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
10073 !FRAME_GROWS_DOWNWARD handling to...
10074 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10075 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
10076 (rs6000_starting_frame_offset): New function.
10077 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
10078 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
10079 * config/vax/vax.c (vax_starting_frame_offset): New function.
10080 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
10081 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10082 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
10083 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
10084 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10085 * system.h (STARTING_FRAME_OFFSET): Poison.
10087 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10089 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10090 SCALAR_TYPE_MODE instead of TYPE_MODE.
10092 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10093 Alan Hayward <alan.hayward@arm.com>
10094 David Sherwood <david.sherwood@arm.com>
10096 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
10098 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10099 Alan Hayward <alan.hayward@arm.com>
10100 David Sherwood <david.sherwood@arm.com>
10102 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
10104 2017-10-23 Richard Biener <rguenther@suse.de>
10106 PR tree-optimization/82129
10107 * tree-ssa-pre.c (bitmap_set_and): Remove.
10108 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
10109 canonicalizing expressions in the set to those with lowest
10110 ID rather than taking that from the first edge.
10112 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10114 * combine.c (rtx_equal_for_field_assignment_p): Use
10115 byte_lowpart_offset.
10117 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10118 Alan Hayward <alan.hayward@arm.com>
10119 David Sherwood <david.sherwood@arm.com>
10121 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
10122 to a SUBREG_PROMOTED_VAR.
10124 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10125 Alan Hayward <alan.hayward@arm.com>
10126 David Sherwood <david.sherwood@arm.com>
10128 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
10129 (expand_debug_source_expr): Likewise.
10130 * combine.c (combine_simplify_rtx): Likewise.
10131 * cse.c (fold_rtx): Likewise.
10132 * optabs.c (expand_float): Likewise.
10133 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10134 (simplify_binary_operation_1): Likewise.
10136 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10137 Alan Hayward <alan.hayward@arm.com>
10138 David Sherwood <david.sherwood@arm.com>
10140 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
10141 (record_promoted_value): Likewise.
10142 * expr.c (expand_expr_real_2): Likewise.
10143 * ree.c (update_reg_equal_equiv_notes): Likewise.
10144 (combine_set_extension): Likewise.
10145 * rtlanal.c (low_bitmask_len): Likewise.
10146 * simplify-rtx.c (neg_const_int): Likewise.
10147 (simplify_binary_operation_1): Likewise.
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 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
10154 * regcprop.c (maybe_mode_change): Likewise.
10155 * reload1.c (alter_reg): Likewise.
10157 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10159 * inchash.h (inchash::hash::add_wide_int): New function.
10160 * lto-streamer-out.c (hash_tree): Use it.
10162 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10164 * inchash.h (inchash::hash::add_wide_int): Rename to...
10165 (inchash::hash::add_hwi): ...this.
10166 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
10167 (polymorphic_call_target_hasher::hash): Likewise.
10168 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
10169 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
10170 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
10171 * lto-streamer-out.c (hash_tree): Likewise.
10172 * optc-save-gen.awk: Likewise.
10173 * tree.c (add_expr): Likewise.
10175 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10178 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
10179 for ordered inequality comparisons even with TARGET_IEEE_FP.
10181 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10184 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
10185 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10186 Expand with cmp<dwi>_doubleword.
10188 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10190 * extend.texi: Add x86 specific to 'nocf_check' attribute.
10191 List CET intrinsics.
10192 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
10193 specific to -fcf-protection option.
10195 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10197 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
10198 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
10199 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
10200 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
10201 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
10202 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
10203 (extra_objs): Add cet.o for Linux/x86 targets.
10204 (tmake_file): Add i386/t-cet for Linux/x86 targets.
10205 * config/i386/cet.c: New file.
10206 * config/i386/cetintrin.h: Likewise.
10207 * config/i386/t-cet: Likewise.
10208 * config/i386/cpuid.h (bit_SHSTK): New.
10209 (bit_IBT): Likewise.
10210 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
10211 pass IBT and SHSTK bits.
10212 * config/i386/i386-builtin-types.def
10213 (VOID_FTYPE_UNSIGNED_PVOID): New.
10214 (VOID_FTYPE_UINT64_PVOID): Likewise.
10215 * config/i386/i386-builtin.def: Add CET intrinsics.
10216 * config/i386/i386-c.c (ix86_target_macros_internal): Add
10217 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
10218 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
10219 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
10221 * config/i386/i386.c (rest_of_insert_endbranch): New.
10222 (pass_data_insert_endbranch): Likewise.
10223 (pass_insert_endbranch): Likewise.
10224 (make_pass_insert_endbranch): Likewise.
10225 (ix86_notrack_prefixed_insn_p): Likewise.
10226 (ix86_target_string): Add -mibt, -mshstk flags.
10227 (ix86_option_override_internal): Add flag_cf_protection
10229 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
10230 (ix86_print_operand): Add 'notrack' prefix output.
10231 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
10232 (ix86_expand_builtin): Expand CET intrinsics.
10233 (x86_output_mi_thunk): Add 'endbranch' instruction.
10234 * config/i386/i386.h (TARGET_IBT): New.
10235 (TARGET_IBT_P): Likewise.
10236 (TARGET_SHSTK): Likewise.
10237 (TARGET_SHSTK_P): Likewise.
10238 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
10239 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
10240 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
10241 (builtin_setjmp_setup): New pattern.
10242 (builtin_longjmp): Likewise.
10243 (rdssp<mode>): Likewise.
10244 (incssp<mode>): Likewise.
10245 (saveprevssp): Likewise.
10246 (rstorssp): Likewise.
10247 (wrss<mode>): Likewise.
10248 (wruss<mode>): Likewise.
10249 (setssbsy): Likewise.
10250 (clrssbsy): Likewise.
10251 (nop_endbr): Likewise.
10252 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
10254 * config/i386/immintrin.h: Include <cetintrin.h>.
10255 * config/i386/linux-common.h
10256 (file_end_indicate_exec_stack_and_cet): New prototype.
10257 (TARGET_ASM_FILE_END): New.
10259 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10261 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
10262 latencies instead of having separate table; make difference between
10263 integer and float costs.
10264 * i386.h (processor_costs): Remove scalar_stmt_cost,
10265 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
10266 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
10268 * x86-tune-costs.h: Remove entries which has been removed in
10269 procesor_costs from all tables; make cond_taken_branch_cost
10270 and cond_not_taken_branch_cost COST_N_INSNS based.
10272 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10274 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
10276 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10278 * config/i386/i386.md (isa): Remove fma_avx512f.
10279 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
10280 <avx512>_fmadd_<mode>_mask3<round_name>,
10281 <avx512>_fmsub_<mode>_mask<round_name>,
10282 <avx512>_fmsub_<mode>_mask3<round_name>,
10283 <avx512>_fnmadd_<mode>_mask<round_name>,
10284 <avx512>_fnmadd_<mode>_mask3<round_name>,
10285 <avx512>_fnmsub_<mode>_mask<round_name>,
10286 <avx512>_fnmsub_<mode>_mask3<round_name>,
10287 <avx512>_fmaddsub_<mode>_mask<round_name>,
10288 <avx512>_fmaddsub_<mode>_mask3<round_name>,
10289 <avx512>_fmsubadd_<mode>_mask<round_name>,
10290 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
10291 (*vec_widen_umult_even_v16si<mask_name>,
10292 *vec_widen_smult_even_v16si<mask_name>): Likewise.
10293 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
10295 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10297 * extend.texi: Add 'nocf_check' documentation.
10298 * gimple.texi: Add second parameter to
10299 gimple_build_call_from_tree.
10300 * invoke.texi: Add -fcf-protection documentation.
10301 * rtl.texi: Add REG_CALL_NOTRACK documenation.
10303 2017-10-20 Richard Biener <rguenther@suse.de>
10305 PR tree-optimization/82473
10306 * tree-vect-loop.c (vectorizable_reduction): Properly get at
10307 the largest input type.
10309 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10311 * c-attribs.c (handle_nocf_check_attribute): New function.
10312 (c_common_attribute_table): Add 'nocf_check' handling.
10313 * gimple-parser.c: Add second argument NULL to
10314 gimple_build_call_from_tree.
10315 * attrib.c (comp_type_attributes): Check nocf_check attribute.
10316 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
10318 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
10319 * common.opt: Add fcf-protection flag.
10320 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
10321 * flag-types.h: Add enum cf_protection_level.
10322 * gimple.c (gimple_build_call_from_tree): Add second parameter.
10323 Add 'nocf_check' attribute propagation to gimple call.
10324 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
10325 (gimple_build_call_from_tree): Update prototype.
10326 (gimple_call_nocf_check_p): New function.
10327 (gimple_call_set_nocf_check): Likewise.
10328 * gimplify.c: Add second argument to gimple_build_call_from_tree.
10329 * ipa-icf.c: Add nocf_check attribute in statement hash.
10330 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
10331 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
10332 * toplev.c (process_options): Add flag_cf_protection handling.
10334 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10336 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
10338 2017-10-20 Richard Biener <rguenther@suse.de>
10340 PR tree-optimization/82603
10341 * tree-if-conv.c (predicate_mem_writes): Make sure to only
10342 remove false predicated stores.
10344 2017-10-20 Richard Biener <rguenther@suse.de>
10346 * graphite-isl-ast-to-gimple.c
10347 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
10348 Remove return value and simplify, dump copied stmt after lhs
10350 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
10351 Reduce dump verbosity.
10352 (gsi_insert_earliest): Likewise.
10353 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
10354 * graphite.c (print_global_statistics): Adjust dumping.
10355 (print_graphite_scop_statistics): Likewise.
10356 (print_graphite_statistics): Do not dump loops here.
10357 (graphite_transform_loops): But here.
10359 2017-10-20 Nicolas Roche <roche@adacore.com>
10361 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
10362 * configure: Regenerate.
10364 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10367 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
10368 functions when optimizing replace GIMPLE_RETURN stmts with
10369 calls to __builtin_unreachable ().
10372 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
10373 for -fsanitize=thread link of executables.
10374 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
10375 link of executables.
10378 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
10379 New mode iterators.
10380 (<shift_insn><mode>3<mask_name>): Change the last of the 3
10381 define_insns for logical vector shifts to use VI248_AVX512BW
10382 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
10383 condition, useless isa and prefix attributes. Change the first
10384 2 of these define_insns to ...
10385 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
10386 define_insn for avx512vl.
10387 (<shift_insn><mode>3): ... and this, new define_insn without
10388 masking for non-avx512vl.
10391 * config/i386/sse.md (*andnot<mode>3,
10392 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
10393 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
10394 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
10395 not applied use empty suffix even for TARGET_AVX512VL.
10396 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
10397 is applied, supply evex,evex or evex,evex,evex instead of just
10400 2017-10-20 Julia Koval <julia.koval@intel.com>
10402 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
10403 (OPTION_MASK_ISA_GFNI_UNSET): New.
10404 (ix86_handle_option): Handle OPT_mgfni.
10405 * config/i386/cpuid.h (bit_GFNI): New.
10406 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
10407 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
10408 * config/i386/i386.c (ix86_target_string): Add -mgfni.
10409 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
10410 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
10411 * config/i386/i386.opt: Add mgfni.
10413 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
10415 * config/msp430/msp430.c (msp430_option_override): Disable
10416 -fdelete-null-pointer-checks.
10417 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
10419 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10421 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
10422 of x87 and SSE instructions.
10424 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10426 * asan.c (create_cond_insert_point): Do not update edge count.
10427 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
10428 (afdo_propagate_circuit): Likewise.
10429 (afdo_calculate_branch_prob): Likewise.
10430 (afdo_annotate_cfg): Likewise.
10431 * basic-block.h (struct edge_def): Remove count.
10432 (edge_def::count): New accessor.
10433 * bb-reorder.c (rotate_loop): Update.
10434 (find_traces_1_round): Update.
10435 (connect_traces): Update.
10436 (sanitize_hot_paths): Update.
10437 * cfg.c (unchecked_make_edge): Update.
10438 (make_single_succ_edge): Update.
10439 (check_bb_profile): Update.
10440 (dump_edge_info): Update.
10441 (update_bb_profile_for_threading): Update.
10442 (scale_bbs_frequencies_int): Update.
10443 (scale_bbs_frequencies_gcov_type): Update.
10444 (scale_bbs_frequencies_profile_count): Update.
10445 (scale_bbs_frequencies): Update.
10446 * cfganal.c (connect_infinite_loops_to_exit): Update.
10447 * cfgbuild.c (compute_outgoing_frequencies): Update.
10448 (find_many_sub_basic_blocks): Update.
10449 * cfgcleanup.c (try_forward_edges): Update.
10450 (try_crossjump_to_edge): Update
10451 * cfgexpand.c (expand_gimple_cond): Update
10452 (expand_gimple_tailcall): Update
10453 (construct_exit_block): Update
10454 * cfghooks.c (verify_flow_info): Update
10455 (redirect_edge_succ_nodup): Update
10456 (split_edge): Update
10457 (make_forwarder_block): Update
10458 (duplicate_block): Update
10459 (account_profile_record): Update
10460 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10461 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10462 * cfgloopmanip.c (scale_loop_profile): Update.
10464 (lv_adjust_loop_entry_edge): Update.
10465 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10466 (force_nonfallthru_and_redirect): Update.
10467 (purge_dead_edges): Update.
10468 (rtl_flow_call_edges_add): Update.
10469 * cgraphunit.c (init_lowered_empty_function): Update.
10470 (cgraph_node::expand_thunk): Update.
10471 * gimple-pretty-print.c (dump_probability): Update.
10472 (dump_edge_probability): Update.
10473 * gimple-ssa-isolate-paths.c (isolate_path): Update.
10474 * haifa-sched.c (sched_create_recovery_edges): Update.
10475 * hsa-gen.c (convert_switch_statements): Update.
10476 * ifcvt.c (dead_or_predicable): Update.
10477 * ipa-inline-transform.c (inline_transform): Update.
10478 * ipa-split.c (split_function): Update.
10479 * ipa-utils.c (ipa_merge_profiles): Update.
10480 * loop-doloop.c (add_test): Update.
10481 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
10482 * lto-streamer-in.c (input_cfg): Update.
10483 (input_function): Update.
10484 * lto-streamer-out.c (output_cfg): Update.
10485 * modulo-sched.c (sms_schedule): Update.
10486 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10487 * predict.c (maybe_hot_edge_p): Update.
10488 (unlikely_executed_edge_p): Update.
10489 (probably_never_executed_edge_p): Update.
10490 (dump_prediction): Update.
10491 (drop_profile): Update.
10492 (propagate_unlikely_bbs_forward): Update.
10493 (determine_unlikely_bbs): Update.
10494 (force_edge_cold): Update.
10495 * profile.c (compute_branch_probabilities): Update.
10496 * reg-stack.c (better_edge): Update.
10497 * shrink-wrap.c (handle_simple_exit): Update.
10498 * tracer.c (better_p): Update.
10499 * trans-mem.c (expand_transaction): Update.
10500 (split_bb_make_tm_edge): Update.
10501 * tree-call-cdce.c: Update.
10502 * tree-cfg.c (gimple_find_sub_bbs): Update.
10503 (gimple_split_edge): Update.
10504 (gimple_duplicate_sese_region): Update.
10505 (gimple_duplicate_sese_tail): Update.
10506 (gimple_flow_call_edges_add): Update.
10507 (insert_cond_bb): Update.
10508 (execute_fixup_cfg): Update.
10509 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
10510 * tree-complex.c (expand_complex_div_wide): Update.
10511 * tree-eh.c (lower_resx): Update.
10512 (unsplit_eh): Update.
10513 (cleanup_empty_eh_move_lp): Update.
10514 * tree-inline.c (copy_edges_for_bb): Update.
10515 (freqs_to_counts): Update.
10516 (copy_cfg_body): Update.
10517 * tree-ssa-dce.c (remove_dead_stmt): Update.
10518 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10519 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10520 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10521 (unloop_loops): Update.
10522 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10523 * tree-ssa-loop-split.c (connect_loops): Update.
10524 (split_loop): Update.
10525 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10526 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10527 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10528 * tree-ssa-reassoc.c (branch_fixup): Update.
10529 * tree-ssa-tail-merge.c (replace_block_by): Update.
10530 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10531 (compute_path_counts): Update.
10532 (update_profile): Update.
10533 (recompute_probabilities): Update.
10534 (update_joiner_offpath_counts): Update.
10535 (estimated_freqs_path): Update.
10536 (freqs_to_counts_path): Update.
10537 (clear_counts_path): Update.
10538 (ssa_fix_duplicate_block_edges): Update.
10539 (duplicate_thread_path): Update.
10540 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10541 (case_bit_test_cmp): Update.
10542 (collect_switch_conv_info): Update.
10543 (gen_inbound_check): Update.
10544 (do_jump_if_equal): Update.
10545 (emit_cmp_and_jump_insns): Update.
10546 * tree-tailcall.c (decrease_profile): Update.
10547 (eliminate_tail_call): Update.
10548 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10549 (vect_do_peeling): Update.
10550 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10551 * ubsan.c (ubsan_expand_null_ifn): Update.
10552 (ubsan_expand_ptr_ifn): Update.
10553 * value-prof.c (gimple_divmod_fixed_value): Update.
10554 (gimple_mod_pow2): Update.
10555 (gimple_mod_subtract): Update.
10556 (gimple_ic): Update.
10557 (gimple_stringop_fixed_value): Update.
10559 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10562 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
10564 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
10566 PR rtl-optimization/82395
10567 * ira-color.c (allocno_priority_compare_func): Fix comparison step
10568 based on non_spilled_static_chain_regno_p.
10570 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10572 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
10573 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
10574 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
10576 2017-10-19 Martin Sebor <msebor@redhat.com>
10578 PR tree-optimization/82596
10579 * tree.c (array_at_struct_end_p): Handle STRING_CST.
10581 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10583 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
10584 (get_mem_refs_of_builtin_call): Likewise.
10585 * builtins.c (expand_builtin_apply): Adjust call to
10586 allocate_dynamic_stack_space.
10587 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
10588 the third argument to allocate_dynamic_stack_space, otherwise -1.
10589 (expand_builtin): Deal with all alloca variants.
10590 (is_inexpensive_builtin): Likewise.
10591 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
10592 * calls.c (special_function_p): Deal with all alloca variants.
10593 (initialize_argument_information): Adjust call to
10594 allocate_dynamic_stack_space.
10595 (expand_call): Likewise.
10596 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
10597 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
10598 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
10599 use it for the stack usage computation.
10600 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
10601 * function.c (gimplify_parameters): Call build_alloca_call_expr.
10602 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
10603 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
10604 (in_loop_p): Remove first argument and useless check.
10605 (pass_walloca::execute): Remove useless test and adjust call to above.
10606 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
10607 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
10608 (gimplify_call_expr): Deal with all alloca variants.
10609 * hsa-gen.c (gen_hsa_alloca): Likewise.
10610 (gen_hsa_insns_for_call): Likewise.
10611 * ipa-pure-const.c (special_builtin_state): Likewise.
10612 * tree-chkp.c (chkp_build_returned_bound): Likewise.
10613 * tree-object-size.c (alloc_object_size): Likewise.
10614 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
10615 (call_may_clobber_ref_p_1): Likewise.
10616 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
10617 (ccp_fold_stmt): Likewise.
10618 (optimize_stack_restore): Likewise.
10619 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
10620 (mark_all_reaching_defs_necessary_1): Likewise.
10621 (propagate_necessity): Likewise.
10622 (eliminate_unnecessary_stmts): Likewise.
10623 * tree.c (build_common_builtin_nodes): Build
10624 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
10625 (build_alloca_call_expr): New function.
10626 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
10627 (CASE_BUILT_IN_ALLOCA): Likewise.
10628 (build_alloca_call_expr): Declare.
10629 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
10631 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10634 * dwarf2out.c (new_die_raw): New static inline function.
10635 (new_die): Use it to create the DIE.
10636 (add_AT_external_die_ref): Likewise.
10637 (clone_die): Likewise.
10638 (clone_as_declaration): Likewise.
10639 (dwarf2out_vms_debug_main_pointer): Likewise.
10640 (base_type_die): Likewise. Remove early return for corner cases.
10641 Do not call add_pubtype on the DIE here.
10642 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
10643 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
10644 typedefs for base types with DW_AT_endianity. Make sure a DIE with
10645 native order exists for base types, attach the DIE manually and call
10646 add_pubtype on it. Do not equate a reverse order DIE to the type.
10648 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
10650 * config/arm/arm.c (align_ok_ldrd_strd): New function.
10651 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
10653 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
10655 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10657 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
10658 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
10659 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
10660 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
10661 * opts.c (sanitizer_opts): Add builtin.
10662 * ubsan.c (instrument_builtin): New function.
10663 (pass_ubsan::execute): Call it.
10664 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
10665 * doc/invoke.texi: Document -fsanitize=builtin.
10667 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
10668 builtins, store max (log2 (align), 0) into uchar field instead of
10669 align into uptr field.
10670 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
10671 store uchar 0 field instead of uptr 0 field.
10672 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
10673 instead of passing one address of struct with 2 locations pass
10674 two addresses of structs with 1 location each.
10675 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
10676 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
10677 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
10678 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
10679 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
10680 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
10681 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
10682 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
10684 2017-10-19 Martin Liska <mliska@suse.cz>
10687 * file-find.c (remove_prefix): Remove.
10688 * file-find.h (remove_prefix): Likewise.
10689 * gcc-ar.c: Remove smartness of lookup.
10691 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
10693 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
10694 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
10695 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
10697 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10700 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
10701 (setcc + and to xor + setcc): New peephole2.
10703 2017-10-19 Tom de Vries <tom@codesourcery.com>
10705 * doc/sourcebuild.texi (Test Directives, Variants of
10706 dg-require-support): Add dg-require-stack-size.
10708 2017-10-19 Martin Liska <mliska@suse.cz>
10711 * gimplify.c (gimplify_decl_expr): Do not instrument variables
10712 that have a large alignment.
10713 (gimplify_target_expr): Likewise.
10715 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
10717 PR rtl-optimization/82602
10718 * ira.c (rtx_moveable_p): Return false for volatile asm.
10720 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
10723 * config/i386/i386-modes.def (CCGZ): New CC mode.
10724 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
10725 * config/i386/predicates.md (ix86_comparison_operator):
10727 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10728 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
10729 with double-word subtraction.
10730 (put_condition_code): Handle CCGZmode.
10732 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
10734 * wide-int.cc (debug (const wide_int &)): New.
10735 (debug (const wide_int *)): New.
10736 (debug (const widest_int &)): New.
10737 (debug (const widest_int *)): New.
10739 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
10741 PR middle-end/82556
10742 * lra-constraints.c (curr_insn_transform): Use non-input operand
10743 instead of output one for matched reload.
10745 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10747 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
10748 (tree-ssa-loop-ivopts.h): New header file.
10749 (struct builtin_info): New fields.
10750 (classify_builtin_1): Compute and record base and offset parts for
10751 memset builtin partition by calling strip_offset.
10752 (offset_cmp, fuse_memset_builtins): New functions.
10753 (finalize_partitions): Fuse adjacent memset partitions by calling
10755 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
10756 Expose the interface.
10757 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
10759 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10761 PR tree-optimization/82574
10762 * tree-loop-distribution.c (find_single_drs): New parameter. Check
10763 that data reference must be executed exactly once per iteration
10764 against the outermost loop in nest.
10765 (classify_partition): Update call to above function.
10767 2017-10-18 Richard Biener <rguenther@suse.de>
10769 PR tree-optimization/82591
10770 * graphite.c (graphite_transform_loops): Move code gen message
10772 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10773 Here. Handle scop_to_isl_ast failing.
10774 (scop_to_isl_ast): Limit the number of ISL operations.
10776 2017-10-18 Richard Biener <rguenther@suse.de>
10778 * graphite-isl-ast-to-gimple.c
10779 (translate_isl_ast_to_gimple::set_rename): Simplify.
10780 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
10781 (graphite_copy_stmts_from_block): ... here.
10782 (copy_bb_and_scalar_dependences): Simplify.
10783 (add_parameters_to_ivs_params): Canonicalize.
10784 (generate_entry_out_of_ssa_copies): Simplify.
10785 * graphite-sese-to-poly.c (extract_affine_name): Simplify
10786 by passing in ISL dimension.
10787 (parameter_index_in_region_1): Rename to ...
10788 (parameter_index_in_region): ... this.
10789 (extract_affine): Adjust assert, pass down parameter index.
10790 (add_param_constraints): Use range-info when available.
10791 (build_scop_context): Adjust.
10792 * sese.c (new_sese_info): Adjust.
10793 (free_sese_info): Likewise.
10794 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
10795 Remove unused typedefs.
10796 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
10798 2017-10-18 Martin Liska <mliska@suse.cz>
10800 * combine.c (simplify_compare_const): Add gcc_fallthrough.
10802 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10804 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
10805 (s390_sched_init): Do not reset s390_sched_state if we entered the
10806 current basic block via a fallthru edge and all others are unlikely.
10808 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10810 * config/s390/s390.c (NUM_SIDES): New variable.
10811 (LONGRUNNING_THRESHOLD): New variable.
10812 (LATENCY_FACTOR): New variable.
10813 (s390_sched_score): Decrease score for long-running instructions on
10815 (s390_sched_variable_issue): Perform bookkeeping for long-running
10818 2017-10-18 Richard Biener <rguenther@suse.de>
10820 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
10821 Simplify with removal of the parameter rename map.
10822 (set_rename): Likewise.
10823 (should_copy_to_new_region): Likewise.
10824 (graphite_copy_stmts_from_block): Likewise.
10825 (copy_bb_and_scalar_dependences): Remove initialization of
10826 unused copied_bb_map.
10827 (copy_def): Remove.
10828 (copy_internal_parameters): Likewise.
10829 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
10830 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
10831 Use INTEGRAL_TYPE_P.
10832 (parameter_index_in_region_1): Rename to ...
10833 (assign_parameter_index_in_region): ... this. Assert we have
10834 a parameter we handle.
10835 (scan_tree_for_params): Adjust.
10836 * sese.h (parameter_rename_map_t): Remove.
10837 (struct sese_info_t): Remove unused parameter_rename_map and
10838 copied_bb_map members.
10839 * sese.c (new_sese_info): Adjust.
10840 (free_sese_info): Likewise.
10842 2017-10-18 Martin Liska <mliska@suse.cz>
10845 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
10846 on an abnormal edge.
10848 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
10850 * doc/invoke.texi (ffunction-sections and fdata-sections):
10853 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10855 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
10856 the use statement can throw internally.
10858 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10860 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
10861 any RTX present on the RHS of a SET.
10862 * compare-elim.c (try_eliminate_compare): Restore comment.
10864 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10866 * langhooks.h (struct lang_hooks): Document that tree_size langhook
10867 may be also called on tcc_type nodes.
10868 * langhooks.c (lhd_tree_size): Likewise.
10870 2017-10-17 David Malcolm <dmalcolm@redhat.com>
10872 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
10873 format_warning_at_substring.
10874 (maybe_warn): Convert source_range * param to a location_t. Pass
10875 UNKNOWN_LOCATION rather than NULL to fmtwarn.
10876 (format_directive): Remove code to extract source_ranges and
10877 source_range * in favor of just a location_t.
10878 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
10880 * substring-locations.c (format_warning_va): Convert
10881 source_range * param to a location_t.
10882 (format_warning_at_substring): Likewise.
10883 * substring-locations.h (format_warning_va): Likewise.
10884 (format_warning_at_substring): Likewise.
10886 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
10888 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
10890 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
10891 and scatter/gather ops.
10893 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
10894 vec_gather_load and vec_scatter_store.
10895 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
10896 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
10897 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
10898 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
10899 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
10900 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
10902 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
10904 * reg-stack.c (compare_for_stack_reg): Add bool argument.
10905 Detect FTST instruction and handle its register pops. Only pop
10906 second operand if can_pop_second_op is true.
10907 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
10908 set can_pop_second_op to false in the compare_for_stack_reg call.
10910 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
10911 output_fp_compare for stack register operands.
10912 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
10913 instructions here. Do not emit stack register pops here. Assert
10914 that FCOMPP pops next to top stack register. Rewrite function.
10916 2017-10-17 Nathan Sidwell <nathan@acm.org>
10918 PR middle-end/82577
10919 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
10920 use DECL_ASSEMBLER_NAME_RAW.
10922 PR middle-end/82546
10923 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
10926 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
10927 Wilco Dijkstra <wilco.dijkstra@arm.com>
10929 * builtins.c (expand_builtin_update_setjmp_buf): Add a
10930 converstion to Pmode from the buf_addr.
10932 2017-10-17 Richard Biener <rguenther@suse.de>
10934 * graphite-dependences.c (scop_get_reads_and_writes): Change
10935 output parameters to references.
10937 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
10939 PR 71026/tree-optimization
10940 * fold-const.c (distribute_real_division): Removed.
10941 (fold_binary_loc): Remove calls to distribute_real_divison.
10943 2017-10-17 Richard Biener <rguenther@suse.de>
10945 * graphite-scop-detection.c
10946 (scop_detection::stmt_has_simple_data_refs_p): Always use
10947 the full nest as region.
10948 (try_generate_gimple_bb): Likewise.
10949 * sese.c (scalar_evolution_in_region): Simplify now that
10950 SCEV can handle instantiation in regions.
10951 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
10952 in the non-loop part of a function if requested.
10954 2017-10-17 Richard Biener <rguenther@suse.de>
10956 PR tree-optimization/82563
10957 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
10959 (graphite_regenerate_ast_isl): Call it.
10960 * graphite-scop-detection.c (build_scops): Remove entry edge split.
10962 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10964 PR tree-optimization/82549
10965 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
10966 Formatting fixes. Instead of calling make_bit_field_ref with negative
10969 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
10971 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
10972 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
10973 _mm_maskz_reduce_ss): New.
10974 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
10975 __builtin_ia32_reducess_mask): Ditto..
10976 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
10977 * config/i386/sse.md (reduces<mode>): Renamed to ...
10978 (reduces<mode><mask_scalar_name>): ... this.
10979 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
10981 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
10982 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
10984 2017-10-16 David Malcolm <dmalcolm@redhat.com>
10986 * Makefile.in (OBJS): Add unique-ptr-tests.o.
10987 * selftest-run-tests.c (selftest::run_tests): Call
10988 selftest::unique_ptr_tests_cc_tests.
10989 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
10990 * unique-ptr-tests.cc: New file.
10992 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
10995 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10997 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10998 (make_hard_regno_born, make_hard_regno_dead): Update
10999 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
11001 2017-10-16 Jeff Law <law@redhat.com>
11003 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11005 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
11007 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
11009 2017-10-16 Olivier Hainque <hainque@adacore.com>
11011 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
11012 with_cpu if we were configured for an e500v2 target cpu name.
11014 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
11016 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
11017 * doc/invoke.texi: Document +nodsp as a valid extension for
11020 2017-10-16 Martin Liska <mliska@suse.cz>
11022 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
11023 (test_set_range): Likewise.
11024 (test_range_functions): Rename to ...
11025 (test_bit_in_range): ... this.
11026 (sbitmap_c_tests): Add new test.
11028 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11030 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
11032 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
11033 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
11035 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11037 * config/aarch64/aarch64-builtins.c
11038 (aarch64_types_quadopu_lane_qualifiers): New.
11039 (TYPES_QUADOPU_LANE): New.
11040 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
11041 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
11042 (aarch64_<sur>dot_laneq<vsi2qi>): New.
11043 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
11044 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
11045 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
11046 (Vdottype, DOTPROD): New.
11047 (sur): Add SDOT and UDOT.
11049 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11051 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
11052 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
11053 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
11054 Add TARGET_DOTPROD.
11055 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
11056 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
11057 Enable TARGET_DOTPROD.
11058 (cortex-a75.cortex-a55): Likewise.
11059 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
11061 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11063 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
11064 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
11066 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
11068 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
11069 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
11070 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
11071 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
11072 * config/arm/types.md (neon_dot, neon_dot_q): New.
11073 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
11075 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11077 * config/arm/arm.h (TARGET_DOTPROD): New.
11078 * config/arm/arm.c (arm_arch_dotprod): New.
11079 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
11080 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
11081 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
11082 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
11083 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
11084 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
11085 * doc/invoke.texi (armv8.2-a): Document dotprod
11087 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11089 * i386.c (ix86_vec_cost): New function.
11090 (ix86_rtx_costs): Handle vector operations better.
11091 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
11092 * x86-tune-costs.h: Add new costs to all tables.
11094 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11096 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
11098 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
11099 divsd, sqrtss and sqrtsd
11100 * x86-tune-costs.h: Add new entries to all costs.
11101 (znver1_cost): Fix to match real instruction latencies.
11103 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11104 Michael Collison <michael.collison@arm.com>
11106 * compare-elim.c: Include emit-rtl.h.
11107 (can_merge_compare_into_arith): New function.
11108 (try_validate_parallel): Likewise.
11109 (try_merge_compare): Likewise.
11110 (try_eliminate_compare): Call the above when no previous clobber
11112 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
11115 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11117 PR middle-end/62263
11118 PR middle-end/82498
11119 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
11120 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
11122 PR middle-end/62263
11123 PR middle-end/82498
11124 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
11125 to be any operand_equal_p operands. For & (B - 1) require
11126 B to be power of 2. Recognize
11127 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
11129 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11132 * optabs.c (expand_memory_blockage): Fix call of
11133 targetm.have_memory_blockage.
11135 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11138 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
11139 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
11140 objects to extra_objs instead of overwriting it.
11142 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11144 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
11145 Use any_fp_register_operand as operand[3] predicate. Simplify
11146 equality test for operands[2] and operands[4] memory location.
11147 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
11148 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
11149 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
11150 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
11151 any_fp_register_operand as operand[1] predicate. Simplify
11152 equality test for operands[0] and operands[3] memory location.
11153 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
11154 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
11155 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
11157 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11159 * target-insns.def: Add memory_blockage.
11160 * optabs.c (expand_memory_blockage): New function.
11161 (expand_asm_memory_barrier): Rename ...
11162 (expand_asm_memory_blockage): ... to this.
11163 (expand_mem_thread_fence): Call expand_memory_blockage
11164 instead of expand_asm_memory_barrier.
11165 (expand_mem_singnal_fence): Ditto.
11166 (expand_atomic_load): Ditto.
11167 (expand_atomic_store): Ditto.
11168 * doc/md.texi (Standard Pattern Names For Generation):
11169 Document memory_blockage instruction pattern.
11171 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
11173 * config/rl78/rl78.c (rl78_emit_libcall): New function.
11174 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
11175 * config/rl78/rl78.md: New define_expand "adddi3".
11177 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11179 * cfghooks.c (verify_flow_info): Disable check that all probabilities
11182 2017-10-13 Jeff Law <law@redhat.com>
11184 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
11186 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11189 * internal-fn.c (expand_mul_overflow): If both operands have
11190 the same highpart of -1 or 0 and the topmost bit of lowpart
11191 is different, overflow is if res <= 0 rather than res < 0.
11193 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
11195 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
11196 TARGET_P9_VECTOR code for unaligned_load case.
11198 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11200 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
11202 2017-10-13 Nathan Sidwell <nathan@acm.org>
11204 * tree-core.h (tree_contains_struct): Make bool.
11205 * tree.c (tree_contains_struct): Likewise.
11206 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
11207 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
11208 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
11209 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
11210 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
11212 2017-10-13 Richard Biener <rguenther@suse.de>
11214 * graphite-isl-ast-to-gimple.c
11215 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
11216 parameters and dominance check.
11217 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
11218 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
11219 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
11220 Do not update SSA form here or do intermediate IL verification.
11221 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
11222 (graphite_initialize): Remove check on the number of loops in
11223 the function and inline into graphite_transform_loops.
11224 (graphite_finalize): Inline into graphite_transform_loops.
11225 (graphite_transform_loops): Perform SSA update and IL verification
11227 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
11229 2017-10-13 Richard Biener <rguenther@suse.de>
11231 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
11232 graphite_expression_type_precision): Avoid global constructor
11234 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
11235 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
11236 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
11237 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
11238 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
11239 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
11241 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
11244 * config/i386/i386.h (ix86_red_zone_size): New.
11245 * config/i386/i386.md (push peephole2s): Replace
11246 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
11248 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11249 Alan Hayward <alan.hayward@arm.com>
11250 David Sherwood <david.sherwood@arm.com>
11252 * combine.c (can_change_dest_mode): Reject changes in
11253 REGMODE_NATURAL_SIZE.
11255 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11256 Alan Hayward <alan.hayward@arm.com>
11257 David Sherwood <david.sherwood@arm.com>
11259 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
11260 (expand_debug_source_expr): Likewise.
11261 * combine.c (combine_simplify_rtx): Likewise.
11262 * cse.c (fold_rtx): Likewise.
11263 * fwprop.c (canonicalize_address): Likewise.
11264 * targhooks.c (default_shift_truncation_mask): Likewise.
11266 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11267 Alan Hayward <alan.hayward@arm.com>
11268 David Sherwood <david.sherwood@arm.com>
11270 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
11271 (widened_mode): Likewise.
11272 (expand_unop): Likewise.
11273 * ree.c (transform_ifelse): Likewise.
11274 (merge_def_and_ext): Likewise.
11275 (combine_reaching_defs): Likewise.
11276 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11278 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11279 Alan Hayward <alan.hayward@arm.com>
11280 David Sherwood <david.sherwood@arm.com>
11282 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
11283 * combine.c (gen_lowpart_for_combine): Likewise.
11284 * dwarf2out.c (rtl_for_decl_location): Likewise.
11285 * final.c (alter_subreg): Likewise.
11286 * rtlhooks.c (gen_lowpart_general): Likewise.
11287 (gen_lowpart_if_possible): Likewise.
11289 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11290 Alan Hayward <alan.hayward@arm.com>
11291 David Sherwood <david.sherwood@arm.com>
11293 * calls.c (expand_call): Use subreg_lowpart_offset.
11294 * cse.c (cse_insn): Likewise.
11295 * regcprop.c (copy_value): Likewise.
11296 (copyprop_hardreg_forward_1): Likewise.
11298 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11301 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
11302 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
11303 =Q constraints to +Q and into insn condition add check
11304 that operands[0] and operands[1] are equal.
11305 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
11306 =Q constraints to +Q and into insn condition add check
11307 that operands[0] is equal to either operands[1] or operands[2].
11310 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
11311 instead of handling MINUS_EXPR twice (once for each argument),
11312 canonicalize operand order and handle just once, use rtype where
11313 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
11316 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
11317 any values of __C while still being pattern recognizable as a simple
11318 rotate instruction.
11320 2017-10-13 Richard Biener <rguenther@suse.de>
11322 PR tree-optimization/82451
11324 2017-10-02 Richard Biener <rguenther@suse.de>
11326 PR tree-optimization/82355
11327 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11328 a mapping for the enclosing loop but avoid generating one for
11329 the loop tree root.
11330 (copy_bb_and_scalar_dependences): Remove premature codegen
11331 error on PHIs in blocks duplicated into multiple places.
11332 * graphite-scop-detection.c
11333 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11334 in the region use it as loop and nest to analyze the DR in.
11335 (try_generate_gimple_bb): Likewise.
11336 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11337 (add_loop_constraints): For blocks in a loop not in the region
11338 create a dimension with a single iteration.
11339 * sese.h (gbb_loop_at_index): Remove assert.
11341 * cfgloop.c (loop_preheader_edge): For the loop tree root
11342 return the single successor of the entry block.
11343 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11344 Reset the SCEV hashtable and niters.
11345 * graphite-scop-detection.c
11346 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
11347 assert that we only have POLYNOMIAL_CHREC that vary in loops
11348 contained in the region.
11349 (scop_detection::graphite_can_represent_expr): Adjust.
11350 (scop_detection::stmt_has_simple_data_refs_p): For loops
11351 not in the region set loop to NULL. The nest is now the
11352 entry edge to the region.
11353 (try_generate_gimple_bb): Likewise.
11354 * sese.c (scalar_evolution_in_region): Adjust for
11355 instantiate_scev change.
11356 * tree-data-ref.h (graphite_find_data_references_in_stmt):
11357 Make nest parameter the edge into the region.
11358 (create_data_ref): Likewise.
11359 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
11360 entry edge into a region and adjust instantiate_scev calls.
11361 (create_data_ref): Likewise.
11362 (graphite_find_data_references_in_stmt): Likewise.
11363 (find_data_references_in_stmt): Pass the loop preheader edge
11364 from the nest argument.
11365 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
11366 parameter the edge into the region.
11367 (instantiate_parameters): Use the loop preheader edge as entry.
11368 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
11370 (get_instantiated_value_entry): Make instantiate_below parameter
11371 the edge into the region.
11372 (instantiate_scev_name): Likewise. Adjust dominance checks,
11373 when we cannot use loop-based instantiation instantiate by
11374 walking use-def chains.
11375 (instantiate_scev_poly): Adjust.
11376 (instantiate_scev_binary): Likewise.
11377 (instantiate_scev_convert): Likewise.
11378 (instantiate_scev_not): Likewise.
11379 (instantiate_array_ref): Remove.
11380 (instantiate_scev_3): Likewise.
11381 (instantiate_scev_2): Likewise.
11382 (instantiate_scev_1): Likewise.
11383 (instantiate_scev_r): Do not blindly handle N-operand trees.
11384 Do not instantiate array-refs. Handle all constants and invariants.
11385 (instantiate_scev): Make instantiate_below parameter
11386 the edge into the region.
11387 (resolve_mixers): Use the loop preheader edge for the region
11388 parameter to instantiate_scev_r.
11389 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
11391 2017-10-13 Richard Biener <rguenther@suse.de>
11393 PR tree-optimization/82525
11394 * graphite-isl-ast-to-gimple.c
11395 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
11397 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
11398 Fail code generation when we cannot represent the isl integer.
11399 (binary_op_to_tree): Elide modulo operations that are no-ops
11400 in the type we code generate. Remove now superfluous code
11403 2017-10-13 Richard Biener <rguenther@suse.de>
11405 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
11406 (scop_detection::harmful_loop_in_region): Remove premature
11407 IV type restriction.
11408 (scop_detection::graphite_can_represent_scev): We can handle
11409 pointer IVs just fine.
11411 2017-10-13 Alan Modra <amodra@gmail.com>
11413 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
11414 "Clobbers and Scratch Registers". Add paragraph on
11415 alternative to clobbers for scratch registers and OpenBLAS
11418 2017-10-13 Alan Modra <amodra@gmail.com>
11420 * doc/extend.texi (Clobbers): Correct vax example. Delete old
11421 example of a memory input for a string of known length. Move
11422 commentary out of table. Add a number of new examples
11423 covering array memory inputs.
11425 2017-10-12 Martin Liska <mliska@suse.cz>
11427 PR tree-optimization/82493
11428 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
11429 (test_range_functions): New function.
11430 (sbitmap_c_tests): Likewise.
11431 * selftest-run-tests.c (selftest::run_tests): Run new tests.
11432 * selftest.h (sbitmap_c_tests): New function.
11434 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11436 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11438 * config/rs6000/amo.h: Fix spacing issue.
11440 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11443 * config/i386/i386.md (*ashl<mode>3_mask_1,
11444 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
11445 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
11448 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11450 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
11451 (profile_probability): Set max_probability
11452 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
11454 * profile-count.c (profile_probability::differs_from_p): Do not
11455 rely on max_probaiblity == 10000
11457 2017-10-12 Jeff Law <law@redhat.com>
11459 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
11462 2017-10-12 Martin Sebor <msebor@redhat.com>
11466 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
11467 (handle_alias_pairs): Call it.
11468 * common.opt (-Wattribute-alias): New option.
11469 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
11470 * doc/invoke.texi (-Wattribute-alias): Document.
11472 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
11475 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11477 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11479 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11480 (make_hard_regno_born, make_hard_regno_dead): Update
11481 bb_killed_pseudos and bb_gen_pseudos.
11483 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11485 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
11487 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
11489 * config/alpha/alpha.c (alpha_split_conditional_move):
11490 Use std::swap instead of manually swapping.
11491 (alpha_stdarg_optimize_hook): Ditto.
11492 (alpha_canonicalize_comparison): Ditto.
11494 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11496 * tree-loop-distribution.c (struct builtin_info): New struct.
11497 (struct partition): Refactor fields into struct builtin_info.
11498 (partition_free): Free struct builtin_info.
11499 (build_size_arg_loc, build_addr_arg_loc): Delete.
11500 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
11501 information from struct builtin_info.
11502 (find_single_drs): New function refactored from classify_partition.
11503 Also moved builtin validity checks to this function.
11504 (compute_access_range, alloc_builtin): New functions.
11505 (classify_builtin_st, classify_builtin_ldst): New functions.
11506 (classify_partition): Refactor code into functions find_single_drs,
11507 classify_builtin_st and classify_builtin_ldst.
11508 (distribute_loop): Don't do runtime alias check when distributing
11510 (find_seed_stmts_for_distribution): New function.
11511 (pass_loop_distribution::execute): Refactor code finding seed
11512 stmts into above function. Support distribution for the innermost
11513 two-level loop nest. Adjust dump information.
11515 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11517 * tree-loop-distribution.c: Adjust the general comment.
11518 (NUM_PARTITION_THRESHOLD): New macro.
11519 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
11520 (classify_partition): Skip builtin pattern of loop nest's inner loop.
11521 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
11522 in call to build_partition_graph.
11523 (finalize_partitions): New parameter. Make loop distribution more
11524 conservative by fusing more partitions.
11525 (distribute_loop): Don't do runtime alias check in case of loop nest
11527 (find_seed_stmts_for_distribution): New function.
11528 (prepare_perfect_loop_nest): New function.
11529 (pass_loop_distribution::execute): Refactor code finding seed stmts
11530 and loop nest into above functions. Support loop nest distribution.
11531 Adjust dump information accordingly.
11533 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11535 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
11536 and set PTYPE_SEQUENTIAL for merged partition.
11538 2017-10-12 Richard Biener <rguenther@suse.de>
11540 PR tree-optimization/69728
11542 2017-09-19 Richard Biener <rguenther@suse.de>
11544 PR tree-optimization/69728
11545 * graphite-sese-to-poly.c (schedule_error): New global.
11546 (add_loop_schedule): Handle empty domain by failing the
11548 (build_original_schedule): Handle schedule_error.
11550 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
11551 domain by returning an unchanged schedule.
11553 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11555 * genrecog.c (validate_pattern): For VEC_SELECT verify that
11556 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
11558 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
11560 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
11563 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11566 * expr.c (store_field): Don't optimize away bitsize == 0 store
11567 from CALL_EXPR with addressable return type.
11569 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11571 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
11572 * config/rs6000/rs6000.md (sel): Delete mode attribute.
11573 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
11574 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
11575 TARGET_ISEL instead of TARGET_ISEL<sel>.
11577 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
11579 * config/rs6000/rs6000.c
11580 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
11582 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11584 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
11586 (reg_or_cint_operand): Fix comment.
11587 (reg_or_zero_operand): New predicate.
11588 * config/rs6000/rs6000-protos.h (output_isel): Delete.
11589 * config/rs6000/rs6000.c (output_isel): Delete.
11590 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
11591 instead of reg_or_cint_operand. Output instruction directly (not via
11593 (isel_unsigned_<mode>): Ditto.
11594 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
11595 gpc_reg_operand. Add an instruction alternative for this. Output
11596 instruction directly.
11597 (*isel_reversed_unsigned_<mode>): Ditto.
11599 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11601 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
11602 (TARGET_CANONICALIZE_COMPARISON): Define.
11604 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
11607 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
11608 Check whether the dest is REG before adding REG_EQUIV note.
11610 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11613 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11615 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11616 (make_hard_regno_born, make_hard_regno_dead): Update
11617 bb_killed_pseudos and bb_gen_pseudos.
11619 2017-10-11 Nathan Sidwell <nathan@acm.org>
11621 * incpath.h (enum incpath_kind): Name enum, prefix values.
11622 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
11623 * incpath.c (heads, tails): Use INC_MAX.
11624 (add_env_var_paths, add_standard_paths): Use incpath_kind.
11625 (merge_include_chains, split_quote_chain,
11626 register_include_chains): Update incpath_kind names.
11627 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
11628 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
11630 (add_framework_path, darwin_register_objc_includes): Likewise.
11631 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
11633 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11635 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
11636 Do not use float_operator operator predicate.
11637 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
11638 * config/i386/predicates.md (float_operator): Remove predicate.
11640 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11642 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
11643 (*jccxf_i387): Ditto.
11644 (*jcc<mode>_i387): Ditto.
11645 (*jccu<mode>_i387): Ditto.
11646 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
11647 (*jcc_*_i387 splitters): Remove.
11648 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
11649 * config/i386/i386.c (ix86_split_fp_branch): Remove.
11650 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
11653 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11655 * profile-count.h (slow_safe_scale_64bit): New function.
11656 (safe_scale_64bit): New inline.
11657 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
11658 * profile-count.c: Include wide-int.h
11659 (slow_safe_scale_64bit): New.
11661 2017-10-11 Nathan Sidwell <nathan@acm.org>
11663 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
11664 HAS_DECL_ASSEMBLER_NAME_P.
11665 * gimple-expr.c (gimple_decl_printable_name: Check
11666 HAS_DECL_ASSEMBLER_NAME_P too.
11667 * ipa-utils.h (type_in_anonymous_namespace_p): Check
11668 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
11669 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
11670 * passes.c (rest_of_decl_compilation): Check
11671 HAS_DECL_ASSEMBLER_NAME_P too.
11672 * recog.c (verify_changes): Likewise.
11673 * tree-pretty-print.c (dump_decl_name): Likewise.
11674 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
11676 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
11677 (SET_DECL_ASSEMBLER_NAME): Use it.
11678 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
11679 (COPY_DECL_ASSEMBLER_NAME): Likewise.
11680 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
11682 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11684 * config.gcc (i386, x86_64): Add extra objects.
11685 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
11686 (ix86_min_insn_size): Declare.
11687 (ix86_issue_rate): Declare.
11688 (ix86_adjust_cost): Declare.
11689 (ia32_multipass_dfa_lookahead): Declare.
11690 (ix86_macro_fusion_p): Declare.
11691 (ix86_macro_fusion_pair_p): Declare.
11692 (ix86_bd_has_dispatch): Declare.
11693 (ix86_bd_do_dispatch): Declare.
11694 (ix86_core2i7_init_hooks): Declare.
11695 (ix86_atom_sched_reorder): Declare.
11696 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
11697 (COSTS_N_BYTES): Move to x86-tune-costs.h.
11698 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
11699 (rip_relative_addr_p): Rename to ...
11700 (ix86_rip_relative_addr_p): ... this one; export.
11701 (memory_address_length): Update.
11702 (ix86_issue_rate): Move to x86-tune-sched.c.
11703 (ix86_flags_dependent): Move to x86-tune-sched.c.
11704 (ix86_agi_dependent): Move to x86-tune-sched.c.
11705 (exact_dependency_1): Move to x86-tune-sched.c.
11706 (exact_store_load_dependency): Move to x86-tune-sched.c.
11707 (ix86_adjust_cost): Move to x86-tune-sched.c.
11708 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
11709 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
11710 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
11711 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
11712 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
11713 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
11714 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
11715 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
11716 (min_insn_size): Rename to ...
11717 (ix86_min_insn_size): ... this one; export.
11718 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
11719 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
11720 (core2i7_first_cycle_multipass_backtrack): Move to
11721 x86-tune-sched-core.c.
11722 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
11723 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
11724 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
11725 (ix86_avoid_jump_mispredicts): Update.
11726 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
11727 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
11728 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
11729 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
11730 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
11731 (MAX_INSN): Move to ix86-tune-sched-bd.c.
11732 (MAX_IMM): Move to ix86-tune-sched-bd.c.
11733 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
11734 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
11735 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
11736 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
11737 (MAX_STORE): Move to ix86-tune-sched-bd.c.
11738 (BIG): Move to ix86-tune-sched-bd.c.
11739 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
11740 (enum insn_path): Move to ix86-tune-sched-bd.c.
11741 (get_mem_group): Move to ix86-tune-sched-bd.c.
11742 (is_cmp): Move to ix86-tune-sched-bd.c.
11743 (dispatch_violation): Move to ix86-tune-sched-bd.c.
11744 (is_branch): Move to ix86-tune-sched-bd.c.
11745 (is_prefetch): Move to ix86-tune-sched-bd.c.
11746 (init_window): Move to ix86-tune-sched-bd.c.
11747 (allocate_window): Move to ix86-tune-sched-bd.c.
11748 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
11749 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
11750 (process_end_window): Move to ix86-tune-sched-bd.c.
11751 (allocate_next_window): Move to ix86-tune-sched-bd.c.
11752 (find_constant): Move to ix86-tune-sched-bd.c.
11753 (get_num_immediates): Move to ix86-tune-sched-bd.c.
11754 (has_immediate): Move to ix86-tune-sched-bd.c.
11755 (get_insn_path): Move to ix86-tune-sched-bd.c.
11756 (get_insn_group): Move to ix86-tune-sched-bd.c.
11757 (count_num_restricted): Move to ix86-tune-sched-bd.c.
11758 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
11759 (add_insn_window): Move to ix86-tune-sched-bd.c.
11760 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
11761 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
11762 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
11763 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
11764 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
11765 (do_dispatch): Move to ix86-tune-sched-bd.c.
11766 (has_dispatch): Move to ix86-tune-sched-bd.c.
11767 * config/i386/t-i386: Add new object files.
11768 * config/i386/x86-tune-costs.h: New file.
11769 * config/i386/x86-tune-sched-atom.c: New file.
11770 * config/i386/x86-tune-sched-bd.c: New file.
11771 * config/i386/x86-tune-sched-core.c: New file.
11772 * config/i386/x86-tune-sched.c: New file.
11774 2017-10-11 Liu Hao <lh_mouse@126.com>
11776 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
11777 the generic version below instead.
11778 (should_colorize): Recognize Windows consoles as terminals
11780 * pretty-print.c [__MINGW32__] (write_all): New function.
11781 [__MINGW32__] (find_esc_head): Likewise.
11782 [__MINGW32__] (find_esc_terminator): Likewise.
11783 [__MINGW32__] (eat_esc_sequence): Likewise.
11784 [__MINGW32__] (mingw_ansi_fputs): New function that handles
11786 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
11789 2017-10-11 Richard Biener <rguenther@suse.de>
11791 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
11792 Properly call analyze_scalar_evolution with the loop of the stmt.
11794 2017-10-11 Richard Biener <rguenther@suse.de>
11796 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
11797 * tree-core.h (tree_base): Add chrec_var union member.
11798 * tree.h (CHREC_VAR): Remove.
11799 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
11800 * tree-chrec.h (build_polynomial_chrec): Adjust.
11801 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
11802 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
11804 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
11806 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
11807 * match.pd: ... here.
11808 ((T) X == (T) Y): Relax condition.
11810 2017-10-11 Bin Cheng <bin.cheng@arm.com>
11812 PR tree-optimization/82472
11813 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
11815 (break_alias_scc_partitions): Update postorder number.
11817 2017-10-11 Martin Liska <mliska@suse.cz>
11820 * opts.c (parse_no_sanitize_attribute): Do not use error_value
11822 * opts.h (parse_no_sanitize_attribute): Remove last argument.
11824 2017-10-11 Martin Liska <mliska@suse.cz>
11826 * print-rtl.c (print_insn): Move declaration of idbuf
11827 to same scope as name.
11829 2017-10-11 Martin Liska <mliska@suse.cz>
11834 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11835 volatile arguments.
11837 2017-10-11 Martin Liska <mliska@suse.cz>
11840 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11841 volatile arguments.
11843 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
11845 * config.gcc (default_gnu_indirect_function): Default to yes for
11846 arm*-*-linux* with glibc.
11848 2017-10-11 Richard Biener <rguenther@suse.de>
11850 * tree-scalar-evolution.c (get_scalar_evolution): Handle
11851 default-defs and types we do not want to analyze.
11852 (interpret_loop_phi): Replace unreachable code with an assert.
11853 (compute_scalar_evolution_in_loop): Remove and inline ...
11854 (analyze_scalar_evolution_1): ... here, replacing condition with
11855 what makes the intent clearer. Remove handling of cases
11856 get_scalar_evolution now handles.
11858 2017-10-10 Jim Wilson <wilson@tuliptree.org>
11860 PR rtl-optimization/81434
11861 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
11862 comment for main loop. In sched_group_found if, also add checks for
11863 pass and min_cost_group.
11865 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
11867 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
11868 (rs6000_insn_cost): New function.
11869 * config/rs6000/rs6000.md (cost): New attribute.
11871 2017-10-10 Jakub Jelinek <jakub@redhat.com>
11872 H.J. Lu <hongjiu.lu@intel.com>
11876 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
11877 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
11878 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
11879 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
11880 to OPTION_MASK_ISA_AVX512VL - builtins that have both
11881 OPTION_MASK_ISA_MMX and some other bit set require both
11882 mmx and the ISAs without the mmx bit.
11883 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
11884 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
11885 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
11886 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
11887 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
11888 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
11889 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
11890 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
11891 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
11892 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
11893 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
11894 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
11895 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
11896 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
11897 Add OPTION_MASK_ISA_MMX.
11899 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
11901 * config.gcc (armv7*-*-freebsd*): New target.
11902 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
11904 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
11906 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
11907 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
11910 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
11912 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
11913 a template parameter.
11914 (WIDE_INT_REF_FOR): Update accordingly.
11915 * tree.h (wi::int_traits <const_tree>): Delete.
11916 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
11917 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
11918 (wi::tree_to_wide_ref): New typedef.
11919 (wi::to_wide): New function.
11920 * calls.c (get_size_range): Use wi::to_wide when operating on
11921 trees as wide_ints.
11922 * cgraph.c (cgraph_node::create_thunk): Likewise.
11923 * config/i386/i386.c (ix86_data_alignment): Likewise.
11924 (ix86_local_alignment): Likewise.
11925 * dbxout.c (stabstr_O): Likewise.
11926 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
11927 * expr.c (const_vector_from_tree): Likewise.
11928 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
11929 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
11930 (fold_negate_expr_1, int_const_binop_1, const_binop)
11931 (fold_convert_const_int_from_real, optimize_bit_field_compare)
11932 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
11933 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
11934 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
11935 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
11936 (fold_not_const, round_up_loc): Likewise.
11937 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
11938 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
11939 (alloca_call_type): Likewise.
11940 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
11941 * godump.c (go_output_typedef): Likewise.
11942 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
11943 * internal-fn.c (get_min_precision): Likewise.
11944 * ipa-cp.c (ipcp_store_vr_results): Likewise.
11945 * ipa-polymorphic-call.c
11946 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
11947 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
11948 (ipa_modify_call_arguments): Likewise.
11949 * match.pd: Likewise.
11950 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
11951 * print-tree.c (print_node_brief, print_node): Likewise.
11952 * stmt.c (expand_case): Likewise.
11953 * stor-layout.c (layout_type): Likewise.
11954 * tree-affine.c (tree_to_aff_combination): Likewise.
11955 * tree-cfg.c (group_case_labels_stmt): Likewise.
11956 * tree-data-ref.c (dr_analyze_indices): Likewise.
11957 (prune_runtime_alias_test_list): Likewise.
11958 * tree-dump.c (dequeue_and_dump): Likewise.
11959 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
11960 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
11961 * tree-pretty-print.c (dump_generic_node): Likewise.
11962 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
11963 (simple_iv_with_niters): Likewise.
11964 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
11965 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
11966 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
11967 * tree-ssa-loop-niter.c (split_to_var_and_offset)
11968 (refine_value_range_using_guard, number_of_iterations_ne_max)
11969 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
11970 (get_cst_init_from_scev, record_nonwrapping_iv)
11971 (scev_var_range_cant_overflow): Likewise.
11972 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
11973 * tree-ssa-pre.c (compute_avail): Likewise.
11974 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
11975 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
11976 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
11977 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
11978 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
11979 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
11980 (dump_case_nodes, try_switch_expansion): Likewise.
11981 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
11982 (vect_do_peeling): Likewise.
11983 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
11984 * tree-vect-stmts.c (vectorizable_load): Likewise.
11985 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
11986 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
11987 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
11988 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
11989 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
11990 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
11991 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
11992 (evrp_dom_walker::before_dom_children): Likewise.
11993 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
11994 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
11995 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
11996 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
11997 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
11998 (get_range_pos_neg): Likewise.
11999 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
12000 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
12001 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
12002 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
12003 * config/avr/avr.c (avr_fold_builtin): Likewise.
12004 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
12005 * config/msp430/msp430.c (msp430_attr): Likewise.
12006 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
12007 * config/powerpcspe/powerpcspe-c.c
12008 (altivec_resolve_overloaded_builtin): Likewise.
12009 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
12010 (rs6000_expand_ternop_builtin): Likewise.
12011 * config/rs6000/rs6000-c.c
12012 (altivec_resolve_overloaded_builtin): Likewise.
12013 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
12014 (rs6000_expand_ternop_builtin): Likewise.
12015 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
12017 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12019 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
12020 when copying loop nest with only one inner loop.
12022 2017-10-10 Richard Biener <rguenther@suse.de>
12024 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
12025 blocks if SCEV is active.
12026 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
12028 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
12029 (scev_initialize): Assert we are not yet initialized.
12031 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12033 * tree-loop-distribution.c (generate_loops_for_partition): Remove
12034 inner loop's exit stmt by making it always exit the loop, otherwise
12035 we would generate an infinite empty loop.
12037 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12039 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
12040 renaming variables in new preheader if it's deleted.
12042 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12044 * tree-loop-distribution.c (struct partition): Remove unused field
12045 loops of the structure.
12046 (partition_alloc, partition_free): Ditto.
12047 (build_rdg_partition_for_vertex): Ditto.
12049 2017-10-09 Jeff Law <law@redhat.com>
12051 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
12052 return type to match prototype and documentation.
12054 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12056 * config/rs6000/rs6000.c (processor_costs): Move to ...
12057 * config/rs6000/rs6000.h: ... here.
12058 (rs6000_cost): Declare.
12060 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
12062 * except.c (setjmp_fn): New global variable.
12063 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
12064 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
12065 if DONT_USE_BUILTIN_SETJMP is defined.
12067 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12069 * target.def (insn_cost): New hook.
12070 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
12071 * doc/tm.texi: Regenerate.
12072 * rtlanal.c (insn_cost): Use the new hook.
12074 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12076 * combine.c (combine_validate_cost): Compute the new insn_cost,
12077 not just pattern_cost.
12078 (try_combine): Adjust comment.
12080 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12082 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
12084 * combine.c (uid_insn_cost): Adjust comment.
12085 (combine_validate_cost): Adjust comment. Use pattern_cost instead
12087 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
12088 * dse.c (find_shift_sequence): Ditto.
12089 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
12090 (bb_valid_for_noce_process_p): Use pattern_cost.
12091 * rtl.h (insn_rtx_cost): Delete.
12092 (pattern_cost): New prototype.
12093 (insn_cost): New prototype.
12094 * rtlanal.c (insn_rtx_cost): Rename to...
12095 (pattern_cost): ... this.
12098 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
12100 * config/i386/i386.md (*jcc_2): Remove insn pattern.
12101 (*jcc<mode>_0_r_i387): Ditto.
12102 (*jccxf_r_i387): Ditto.
12103 (*jcc<mode>_r_i387): Ditto.
12104 (*jccu<mode>_r_i387): Ditto.
12105 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
12106 (*jcc): Rename from *jcc_1.
12108 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12110 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
12111 deferred rescans after the lvx/stvx recombination pre-pass.
12113 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
12115 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
12116 memory operation instruction support.
12117 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
12118 (rs6000-ibm-aix[789]*): Likewise.
12119 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
12120 Document new functions.
12122 2017-10-09 Richard Biener <rguenther@suse.de>
12124 PR tree-optimization/82397
12125 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
12126 equality only for semantically equal trees.
12128 2017-10-09 Richard Biener <rguenther@suse.de>
12130 PR tree-optimization/82449
12131 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
12132 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
12133 allow constant addresses.
12134 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
12137 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12139 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
12142 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12145 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
12148 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12151 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
12153 2017-10-09 Jakub Jelinek <jakub@redhat.com>
12156 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
12157 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
12159 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
12161 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
12162 (WI_BINARY_PREDICATE_RESULT): Likewise.
12163 (wi::binary_traits::operator_result): New type.
12164 (wi::binary_traits::predicate_result): Likewise.
12165 (generic_wide_int::operator~, unary generic_wide_int::operator-)
12166 (generic_wide_int::operator==, generic_wide_int::operator!=)
12167 (generic_wide_int::operator&, generic_wide_int::and_not)
12168 (generic_wide_int::operator|, generic_wide_int::or_not)
12169 (generic_wide_int::operator^, generic_wide_int::operator+
12170 (binary generic_wide_int::operator-, generic_wide_int::operator*):
12172 (operator~, unary operator-, operator==, operator!=, operator&)
12173 (operator|, operator^, operator+, binary operator-, operator*): New
12175 * expr.c (get_inner_reference): Use wi::bit_and_not.
12176 * fold-const.c (fold_binary_loc): Likewise.
12177 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
12178 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
12179 (bit_value_binop): Likewise.
12180 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
12181 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
12182 (extract_range_from_binary_expr_1): Likewise.
12183 (masked_increment): Likewise.
12184 (simplify_bit_ops_using_ranges): Likewise.
12186 2017-10-09 Martin Jambor <mjambor@suse.cz>
12189 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
12190 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
12191 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
12193 (hsa_fixup_mov_insn_type): New function.
12194 (hsa_op_with_type::get_in_type): Use it.
12195 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
12196 immediates in an assert.
12197 (hsa_op_with_type::extend_int_to_32bit): New method.
12198 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
12199 types. Convert to dest type if necessary.
12200 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
12201 (reg_for_gimple_ssa): Pass false as min32int to
12202 hsa_type_for_scalar_tree_type.
12203 (gen_hsa_addr): Fixup type when creating addresable temporary.
12204 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
12205 (gen_hsa_unary_operation): Extend operands and convert to dest type if
12206 necessary. Call hsa_fixup_mov_insn_type.
12207 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
12208 extend operands and convert to dest type if necessary.
12209 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
12210 to dest type if necessary.
12211 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
12212 if conversion nt necessary and size matches.
12213 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
12214 to dest type if necessary.
12215 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
12216 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
12218 (gen_hsa_clrsb): Likewise.
12219 (gen_hsa_ffs): Likewise.
12220 (gen_hsa_divmod): Extend operands and convert to dest type if
12222 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
12224 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12226 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
12227 Remove empty default arguments. Use a brace block as output
12229 (conditional return): Ditto.
12231 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
12232 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
12234 (group_ending_nop): Ditto.
12235 (doloop_end): Ditto.
12236 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
12237 (splitters for those): Ditto.
12239 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12241 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
12242 a conditional jump (and the compare for it) so that pc_rtx is the
12244 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
12245 for the deleted and renamed ctr<mode>_internal[234] patterns.
12246 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
12247 Delete second conditional return pattern.
12248 (ctr<mode>_internal2): Delete this second bdnz pattern.
12249 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
12250 (ctr<mode>_internal4): Delete this second bdz pattern.
12252 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12254 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
12255 (always_initialized_rtx_for_ssa_name_p): New predicate.
12256 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
12257 (finish_out_of_ssa): Free new field of SA.
12258 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
12259 * tree-ssa-coalesce.c: Include tree-ssa.h.
12260 (get_parm_default_def_partitions): Remove extern keyword.
12261 (get_undefined_value_partitions): New function.
12262 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
12263 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
12264 uninitialized bits.
12265 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
12267 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12269 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
12271 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12273 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
12274 for targets that preffer 128bit.
12276 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12278 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
12280 2017-10-08 Olivier Hainque <hainque@adacore.com>
12282 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
12283 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
12285 (thumb_set_return_address): Likewise.
12287 2017-10-08 Olivier Hainque <hainque@adacore.com>
12289 * common/config/arm/arm-common.c (arm_except_unwind_info):
12290 Handle DWARF2_UNWIND_INFO.
12292 2017-10-07 Michael Collison <michael.collison@arm.com>
12294 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
12297 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
12299 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
12300 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
12301 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
12302 defined, force the creation of a new block for a dispatch label.
12304 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
12306 * invoke.texi (Wsuggest-attribute=cold): Document.
12307 * common.opt (Wsuggest-attribute=cold): New
12308 * ipa-pure-const.c (warn_function_cold): New function.
12309 * predict.c (compute_function_frequency): Use it.
12310 * predict.h (warn_function_cold): Declare.
12312 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
12314 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
12317 2017-10-06 Martin Liska <mliska@suse.cz>
12319 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
12320 keyword for member functions.
12321 (struct sanopt_tree_couple): New struct.
12322 (struct sanopt_tree_couple_hash): New function.
12323 (struct sanopt_ctx): Add new hash_map.
12324 (has_dominating_ubsan_ptr_check): New function.
12325 (record_ubsan_ptr_check_stmt): Likewise.
12326 (maybe_optimize_ubsan_ptr_ifn): Likewise.
12327 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
12328 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
12330 2017-10-06 Sudakshina Das <sudi.das@arm.com>
12333 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
12334 aarch64_simd_valid_immediate on CONST_VECTORs.
12335 (aarch64_reg_or_bic_imm): Likewise.
12337 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
12339 PR rtl-optimization/82396
12340 * haifa-sched.c (ready_sort_real): Disable qsort checking.
12342 2017-10-06 Sebastian Pop <sebpop@gmail.com>
12344 * graphite-dependences.c (scop_get_reads): Move code to...
12345 (scop_get_must_writes): Move code to...
12346 (scop_get_may_writes): Move code to...
12347 (scop_get_reads_and_writes): ... here.
12348 (scop_get_dependences): Call scop_get_reads_and_writes.
12350 2017-10-06 Jakub Jelinek <jakub@redhat.com>
12352 PR tree-optimization/82434
12353 * fold-const.h (can_native_encode_type_p,
12354 can_native_encode_string_p): Remove.
12355 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
12356 don't encode anything, just return what would be otherwise returned.
12357 (native_encode_fixed, native_encode_complex, native_encode_vector):
12359 (native_encode_string): Likewise. Inline by hand
12360 can_native_encode_string_p.
12361 (can_native_encode_type_p): Remove.
12362 (can_native_encode_string_p): Remove.
12363 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
12364 STRING_CSTs using can_native_encode_string_p, test all
12365 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
12366 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
12367 argument from native_encode_expr.
12368 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
12369 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
12372 2017-10-06 Richard Biener <rguenther@suse.de>
12374 PR tree-optimization/82397
12375 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
12376 operand_equal_p but rely on data_ref_compare_tree for detecting
12378 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
12379 to match up with dr_group_sort_cmp.
12381 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12384 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
12386 * config/s390/s390-builtin-types.def: Regenerate.
12388 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12391 * config/s390/s390-builtin-types.def: Regenerate.
12392 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
12393 Change flag from B_VXE to B_VX.
12394 (s390_vec_min_dbl): Remove B_VXE flag.
12396 2017-10-06 Richard Biener <rguenther@suse.de>
12398 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
12399 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
12400 translate_isl_ast_to_gimple::is_valid_rename,
12401 translate_isl_ast_to_gimple::get_rename,
12402 translate_isl_ast_to_gimple::get_def_bb_for_const,
12403 translate_isl_ast_to_gimple::get_new_name,
12404 translate_isl_ast_to_gimple::collect_all_ssa_names,
12405 translate_isl_ast_to_gimple::copy_loop_phi_args,
12406 translate_isl_ast_to_gimple::collect_all_ssa_names,
12407 translate_isl_ast_to_gimple::copy_loop_phi_args,
12408 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
12409 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
12410 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
12411 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
12412 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
12413 translate_isl_ast_to_gimple::copy_cond_phi_args,
12414 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
12415 translate_isl_ast_to_gimple::edge_for_new_close_phis,
12416 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
12417 translate_isl_ast_to_gimple::rename_uses,
12418 translate_isl_ast_to_gimple::rename_all_uses): Remove.
12419 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
12420 (set_rename_for_each_def): Likewise.
12421 (graphite_copy_stmts_from_block): Handle debug stmt resetting
12422 here. Handle rewriting SCEV analyzable uses here.
12423 (copy_bb_and_scalar_dependences): Generate code for PHI
12425 (graphite_regenerate_ast_isl): Adjust.
12426 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
12427 (add_write, add_read): New functions.
12428 (build_cross_bb_scalars_def): Use it and simplify.
12429 (build_cross_bb_scalars_use): Likewise.
12430 (graphite_find_cross_bb_scalar_vars): Inline into...
12431 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
12432 simulating out-of-SSA. Compute liveout and add dependencies.
12433 (build_scops): Force an empty entry block.
12434 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
12436 (sese_build_liveouts): Declare.
12437 (sese_trivially_empty_bb_p): Likewise.
12438 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
12439 compute liveout and debug_liveout.
12440 (sese_bad_liveouts_use): Remove.
12441 (sese_reset_debug_liveouts_bb): Likewise.
12442 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
12443 (sese_build_liveouts): Build liveout and debug_liveout and store
12445 (new_sese_info): Adjust.
12446 (free_sese_info): Likewise.
12447 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
12448 do not build liveout here.
12449 (move_sese_in_condition): Adjust region entry.
12450 (scev_analyzable_p): Match up with chrec_apply requirements.
12451 (sese_trivially_empty_bb_p): New.
12452 * tree-into-ssa.c (get_reaching_def): Properly support generating
12453 default-defs for incremental rewrite of anonymous names.
12455 2017-10-06 Richard Biener <rguenther@suse.de>
12457 * graphite-sese-to-poly.c (extract_affine): For casts increasing
12458 precision do not perform modulo reduction.
12460 2017-10-06 Richard Biener <rguenther@suse.de>
12462 PR tree-optimization/82436
12463 * tree-vect-slp.c (vect_supported_load_permutation_p): More
12464 conservatively choose the vectorization factor when checking
12465 whether we can perform the required load permutation.
12466 (vect_transform_slp_perm_load): Assert when we may not fail.
12468 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
12470 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
12471 message for incompatible -msdata=* and -mcall-* options.
12473 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12475 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
12476 rate for post-reload scheduling.
12478 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12480 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
12482 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12484 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
12485 to improve monte carlo in scimark.
12487 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12489 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
12490 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
12491 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
12492 pentium4_cost, nocona_cost): Set reassociation width to 1.
12493 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
12494 width to 2 for fp operations and 1 otherwise.
12495 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
12497 (atom_cost): Set reassociation width to 2.
12498 (slm_cost, generic_cost): Set fp reassociation width
12499 to 2 and 1 otherwise.
12500 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
12501 (core_cost): Set fp reassociation width to 4 and vector to 2.
12502 (ix86_reassociation_width): Rewrite using cost table; special case
12503 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
12504 and TARGET_AVX128_OPTIMAL.
12505 * config/i386/i386.h (processor_costs): Add
12506 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
12507 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
12508 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
12509 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
12510 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
12511 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
12513 2017-10-05 Nathan Sidwell <nathan@acm.org>
12515 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
12517 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12519 * config/arm/arm.c (arm_test_fpu_data): New.
12520 (arm_run_selftests): Call arm_test_fpu_data.
12522 2017-10-04 Nathan Sidwell <nathan@acm.org>
12524 * toplev.c (toplev::main): Remove excess parens on pretty_printer
12526 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
12528 2017-10-04 Sudakshina Das <sudi.das@arm.com>
12530 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
12531 check type for aarch64_simd_valid_immediate.
12532 (aarch64_output_simd_mov_immediate): Update prototype.
12533 (aarch64_simd_valid_immediate): Update prototype.
12534 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
12535 support for ORR-immediate.
12536 (and<mode>3): modified pattern to add support for BIC-immediate.
12537 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
12538 now checks for valid immediate for BIC and ORR based on new enum
12540 (aarch64_output_simd_mov_immediate): Function now used to output
12541 BIC/ORR imm as well based on new enum argument.
12542 * config/aarch64/constraints.md (Do): New vector immediate constraint.
12544 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
12545 (aarch64_reg_or_bic_imm): Likewise.
12547 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12549 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
12550 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
12552 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12556 PR rtl-optimization/82396
12557 * haifa-sched.c (autopref_multipass_init): Simplify
12559 (autopref_rank_data): Simplify sort order.
12560 * sched-int.h (autopref_multipass_data_): Remove
12561 multi_mem_insn_p, min_offset and max_offset.
12563 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12565 * doc/sourcebuild.texi: Document vect_peeling_profitable.
12567 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12569 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
12570 vect_doubleint_cvt.
12572 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12574 * doc/sourcebuild.texi: Document vect_long_mult.
12576 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
12578 PR tree-optimization/82413
12579 * fold-const.c (build_range_check): Use widest_int when comparing
12580 the maximum ETYPE value with HIGH.
12582 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12584 PR rtl-optimization/82396
12585 * haifa-sched.c (autopref_multipass_init): Simplify
12587 (autopref_rank_data): Simplify sort order.
12588 * sched-int.h (autopref_multipass_data_): Remove
12589 multi_mem_insn_p, min_offset and max_offset.
12591 2017-10-04 Jakub Jelinek <jakub@redhat.com>
12593 PR tree-optimization/82381
12594 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
12595 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
12598 PR tree-optimization/82374
12599 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
12600 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
12601 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
12602 current_function_decl to the new decl.
12604 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
12606 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
12607 helper macro for IEEE float128 hardware built-in functions.
12608 (SQRTF128_ODD): Add built-in functions with the round-to-odd
12610 (TRUNCF128_ODD): Likewise.
12611 (ADDF128_ODD): Likewise.
12612 (SUBF128_ODD): Likewise.
12613 (MULF128_ODD): Likewise.
12614 (DIVF128_ODD): Likewise.
12615 (FMAF128_ODD): Likewise.
12616 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
12617 UNSPEC_TRUNC_ROUND_TO_ODD.
12618 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
12619 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
12620 floating point round to odd instructions.
12621 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
12622 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
12623 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
12624 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
12625 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
12626 (trunc<mode>sf2_hw): Change the truncate with round to odd
12627 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
12628 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
12629 to odd hardware instructions.
12630 (sub<mode>3_odd): Likewise.
12631 (mul<mode>3_odd): Likewise.
12632 (div<mode>3_odd): Likewise.
12633 (sqrt<mode>2_odd): Likewise.
12634 (fma<mode>4_odd): Likewise.
12635 (fms<mode>4_odd): Likewise.
12636 (nfma<mode>4_odd): Likewise.
12637 (nfms<mode>4_odd): Likewise.
12638 (trunc<mode>df2_odd): Change the truncate with round to odd
12639 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
12641 * doc/extend.texi (PowerPC built-in functions): Update documentation
12642 for existing IEEE float128-bit built-in functions. Add built-in
12643 functions that generate the IEEE 128-bit floating point round to
12646 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
12648 PR rtl-optimization/77729
12649 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
12650 to (X&(C1&~C2))|C2 transformations.
12652 2017-10-03 Martin Jambor <mjambor@suse.cz>
12654 PR tree-optimization/82363
12655 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
12656 mismatch, mark lacc written regardless of racc.
12658 2017-10-03 Jakub Jelinek <jakub@redhat.com>
12660 PR tree-optimization/82381
12661 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
12662 stmt_to_insert nor wheather SSA_NAMEs are default defs.
12663 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
12664 fallthrough into reassoc_stmt_dominates_stmt_p.
12667 * combine.c (combine_instructions): Don't combine in unreachable
12670 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
12673 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
12674 function to not use the have_cpu variable. Do not set cpu_index,
12675 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
12676 or the default cpu.
12677 (rs6000_valid_attribute_p): Remove duplicate initializations of
12678 old_optimize and func_optimize.
12679 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
12680 (rs6000_activate_target_options): Make global.
12681 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
12684 2017-10-02 Jakub Jelinek <jakub@redhat.com>
12686 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
12687 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
12688 Set *poffset to 0 and *psize and *pmax_size to -1 if
12689 *poffset + *psize overflows in HOST_WIDE_INT.
12691 PR tree-optimization/82387
12692 PR tree-optimization/82388
12693 PR tree-optimization/82389
12694 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
12695 instead of live_bytes non-NULL.
12697 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
12700 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
12703 2017-10-02 Richard Biener <rguenther@suse.de>
12705 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
12706 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
12707 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
12709 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12711 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
12712 requested precision matches the type's.
12713 * calls.c (alloc_max_size): Calculate the new candidate size as
12714 a widest_int and use wi::to_widest when comparing it with the
12715 current candidate size.
12716 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
12717 zero rather than integer_zero_node.
12718 * match.pd: Check for a no-op conversion before using wi::add
12719 rather than after. Use tree_to_uhwi when summing small shift
12720 counts into an unsigned int.
12722 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12723 Alan Hayward <alan.hayward@arm.com>
12724 David Sherwood <david.sherwood@arm.com>
12727 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
12728 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
12729 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
12730 POINTER_AND_FP_REGS.
12732 2017-10-02 Richard Biener <rguenther@suse.de>
12734 PR tree-optimization/82355
12735 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12736 a mapping for the enclosing loop but avoid generating one for
12737 the loop tree root.
12738 (copy_bb_and_scalar_dependences): Remove premature codegen
12739 error on PHIs in blocks duplicated into multiple places.
12740 * graphite-scop-detection.c
12741 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12742 in the region use it as loop and nest to analyze the DR in.
12743 (try_generate_gimple_bb): Likewise.
12744 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12745 (add_loop_constraints): For blocks in a loop not in the region
12746 create a dimension with a single iteration.
12747 * sese.h (gbb_loop_at_index): Remove assert.
12749 2017-10-01 Kevin Buettner <kevinb@redhat.com>
12751 * omp-expand.c (adjust_context_scope): New function.
12752 (expand_parallel_call): Call adjust_context_scope.
12754 2017-10-01 Jeff Law <law@redhat.com>
12756 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
12757 dom_opt_dom_walker class with direct access to private members.
12758 Add comments. Call test_for_singularity.
12759 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
12760 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
12761 m_dummy_cond anymore.
12762 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
12764 (pass_dominator:execute): Build the dummy_cond here and pass it
12765 to the dom_opt_dom_walker ctor.
12766 (test_for_singularity): New function.
12768 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
12769 Maya Rashish <coypu@sdf.org>
12771 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
12772 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
12773 (alpha*-*-netbsd*) Use nbsd_tm_file.
12774 (arm*-*-netbsdelf*) Likewise.
12775 (i[34567]86-*-netbsdelf*) Likewise.
12776 (x86_64-*-netbsd*) Likewise.
12777 (mips*-*-netbsd*) Likewise.
12778 (powerpc-*-netbsd*) Likewise.
12779 (sh*-*-netbsd*) Likewise.
12780 (sparc-*-netbsdelf*) Likewise.
12781 (sparc64-*-netbsd*) Likewise.
12782 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
12784 (vax-*-netbsdelf*) Likewise.
12785 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
12786 (UINT_FAST8_TYPE) Likewise.
12787 (INT_FAST16_TYPE) Check CHAR_FAST16.
12788 (UINT_FAST16_TYPE) Likewise.
12790 2017-09-30 Jakub Jelinek <jakub@redhat.com>
12793 * config/i386/i386.md
12794 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
12795 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
12796 *divmodsi4_zext_2): New define_insn_and_split.
12797 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
12798 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
12799 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
12800 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
12801 New define_insn_and_split.
12802 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
12803 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
12804 operands[1] having DImode when mode is SImode.
12806 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
12807 always SImode for DIV and MOD in REG_EQUAL notes.
12809 2017-09-29 Yury Gribov <tetra2005@gmail.com>
12811 PR middle-end/82319
12812 * match.pd: Fix handling of NaNs in pattern.
12814 2017-09-29 Jeff Law <law@redhat.com>
12816 * sbitmap.c (bitmap_bit_in_range_p): New function.
12817 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
12818 * tree-ssa-dse.c (live_bytes_read): New function.
12819 (dse_classify_store): Ignore reads of dead bytes.
12821 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
12822 typos and whitespace errors.
12823 * config/i386/predicates.md (address_no_seg_operand): Likewise.
12824 * config/s390/s390.c (s390_emit_prologue): Likewise.
12826 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12829 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
12830 with a symbol for LRA.
12832 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12834 PR rtl-optimization/82338
12835 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
12837 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
12839 * genmodes.c (calc_wider_mode): Suppress qsort macro.
12840 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
12841 (qsort_chk): Declare.
12842 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
12843 (qsort_chk): New function.
12845 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12847 PR tree-optimization/82337
12848 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12849 phi definition if the PHI result appears in an abnormal PHI.
12850 (find_basis_for_base_expr): Don't record a basis if the LHS of the
12851 basis appears in an abnormal PHI.
12853 2017-09-29 Richard Biener <rguenther@suse.de>
12855 * graphite-isl-ast-to-gimple.c
12856 (translate_isl_ast_to_gimple::set_codegen_error): New function.
12857 (binary_op_to_tree): Use it.
12858 (get_rename_from_scev): Likewise.
12859 (copy_loop_phi_nodes): Likewise.
12860 (copy_bb_and_scalar_dependences): Likewise.
12861 (translate_pending_phi_nodes): Likewise.
12863 2017-09-29 Jakub Jelinek <jakub@redhat.com>
12866 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
12867 for movabsq $(i32 << shift), r64.
12869 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
12871 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
12872 index when encoding %esp as %rsp to avoid 0x67 prefix.
12874 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
12876 * config/i386/i386.md (*movsf_internal, *movdf_internal):
12877 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
12879 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
12881 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
12882 Extensions with more than 16 double VFP registers.
12883 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
12884 to_clear_mask and all code related to it. Replace the remaining
12885 entry by a sbitmap and adapt code accordingly.
12887 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
12889 * brig-builtins.def: Change pure attributes to const.
12891 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12893 * config.gcc (default_gnu_indirect_function): Default to yes for
12894 sparc*-*-linux* with glibc.
12896 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12898 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
12899 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
12900 when creating .init_array and .fini_array sections with priority
12903 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
12906 * config/aarch64/aarch64.c
12907 (aarch64_builtin_support_vector_misalignment): Always return false
12908 when misalignment is unknown.
12910 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12912 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
12913 this function to return false if the definition used by the swap
12914 instruction is artificial, or if the memory address from which the
12915 constant value is loaded is not represented by a base address held
12916 in a register or if the base address register is a frame or stack
12917 pointer. Additionally, return false if the base address of the
12918 loaded constant is a SYMBOL_REF but is not considered to be a
12920 (replace_swapped_load_constant): New function.
12921 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
12922 loaded constant vector with a load of a swapped constant vector.
12924 2017-09-27 Carl Love <cel@us.ibm.com>
12926 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
12927 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
12928 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
12931 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
12933 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
12934 first, always call autopref_rank_data otherwise.
12936 2017-09-27 Richard Biener <rguenther@suse.de>
12938 * graphite-scop-detection.c (find_scop_parameters): Move
12939 loop bound handling ...
12940 (gather_bbs::before_dom_children): ... here, avoiding the need
12941 to build scop_info->loop_nest.
12942 (record_loop_in_sese): Remove.
12943 * sese.h (sese_info_t::loop_nest): Remove.
12944 * sese.c (new_sese_info): Do not allocate loop_nest.
12945 (free_sese_info): Do not free loop_nest.
12947 2017-09-27 Jakub Jelinek <jakub@redhat.com>
12950 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
12951 lhs from calls if the lhs has addressable type.
12953 2017-09-27 Richard Biener <rguenther@suse.de>
12955 * graphite.h (scop::max_alias_set): New member.
12956 * graphite-scop-detection.c: Remove references to non-existing
12957 --param in comments.
12958 (build_alias_sets): Record the maximum alias set used for drs.
12959 (build_scops): Support zero as unlimited for
12960 --param graphite-max-arrays-per-scop.
12961 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
12962 and inline into ...
12963 (build_poly_sr_1): ... here. Compute alias set based on the
12964 maximum alias set used for drs rather than
12965 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
12967 2017-09-27 Richard Biener <rguenther@suse.de>
12969 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
12970 --param loop-block-tile-size=0 to disable tiling.
12972 2017-09-27 Richard Biener <rguenther@suse.de>
12974 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
12975 (graphite-max-nb-scop-params): Document special value zero.
12976 * domwalk.h (dom_walker::STOP): New symbolical constant.
12977 (dom_walker::dom_walker): Add optional parameter for bb to
12979 (dom_walker::~dom_walker): Declare.
12980 (dom_walker::before_dom_children): Document STOP return value.
12981 (dom_walker::m_user_bb_to_rpo): New member.
12982 (dom_walker::m_bb_to_rpo): Likewise.
12983 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
12984 mapping here if not provided by the user.
12985 (dom_walker::~dom_walker): Free bb to RPO mapping if not
12986 provided by the user.
12987 (dom_walker::STOP): Define.
12988 (dom_walker::walk): Do not compute bb to RPO mapping here.
12989 Support STOP return value from before_dom_children to stop
12991 * graphite-optimize-isl.c (optimize_isl): If the schedule
12992 is the same still generate code if -fgraphite-identity
12993 or -floop-parallelize-all are given.
12994 * graphite-scop-detection.c: Include cfganal.h.
12995 (gather_bbs::gather_bbs): Get and pass through bb to RPO
12997 (gather_bbs::before_dom_children): Return STOP for BBs
12999 (build_scops): Compute bb to RPO mapping and pass it to
13000 the domwalk. Treat --param graphite-max-nb-scop-params=0
13001 as not limiting the number of params.
13002 * graphite.c (graphite_initialize): Remove limit on the
13003 number of basic-blocks in a function.
13004 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
13005 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
13006 default value of 10.
13008 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13010 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
13011 Adjust code to eliminate needing to do the shift right 32-bits
13012 operation after XSCVDPSPN.
13014 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13016 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
13017 ((X / Y) != 0 -> X >= Y): Likewise.
13019 2017-09-26 Carl Love <cel@us.ibm.com>
13021 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
13022 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
13023 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
13024 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
13025 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
13026 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
13027 definitions and overloading.
13028 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
13029 statement for P9V_BUILTIN_XST_LEN_R.
13030 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
13031 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
13032 define_expand and define_insn for the instructions and builtins.
13033 * doc/extend.texi: Update the built-in documentation file for the new
13034 built-in functions.
13035 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
13036 define_insn for the instructions
13038 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
13041 * gcc/config/netbsd-protos.h: New file.
13042 * gcc/config/netbsd.c: New file.
13043 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
13044 * gcc/config/t-netbsd: New file.
13045 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
13046 (tmake_file) Add t-netbsd.
13047 (extra_objs) Add netbsd.o.
13049 2017-09-26 Janus Weil <janus@gcc.gnu.org>
13053 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
13055 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13057 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
13058 sign extension from a vector register to a GPR by doing a 32-bit
13059 direct move and then an EXTSW.
13060 (extendsi<mode>2 splitter): Likewise.
13061 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
13062 right or vector extract after doing XSCVDPSPN. Use
13063 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
13065 (movdi_from_sf_zero_ext): Likewise.
13066 (reload_gpr_from_vsxsf): Likewise.
13067 (p8_mfvsrd_4_disf): Delete, no longer used.
13068 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
13069 then needing to move the SFmode to a GPR to use the XSCVDPSP
13070 instruction instead of FRSP and XSCVDPSPN.
13071 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
13072 it is adjacent to the other XSCVSPDP insns.
13073 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
13074 SFmode to be in traditional Altivec registers.
13075 (vsx_xscvdpspn): Eliminate useless alternative constraint.
13076 (vsx_xscvspdpn): Likewise.
13077 (vsx_xscvspdpn_scalar): Likewise.
13079 2017-09-26 Martin Jambor <mjambor@suse.cz>
13081 * tree-sra.c (compare_access_positions): Put integral types first,
13082 stabilize sorting of integral types, remove conditions putting
13083 non-full-precision integers last.
13084 (sort_and_splice_var_accesses): Disable scalarization if a
13085 non-integert would be represented by a non-full-precision integer.
13087 2017-09-26 Joseph Myers <joseph@codesourcery.com>
13089 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
13090 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
13091 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
13092 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
13093 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
13094 conditionals inside the function instead of around it. Call
13095 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
13096 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
13098 2017-09-26 Richard Biener <rguenther@suse.de>
13100 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
13102 (scop_detection::build_scop_breadth): ... this. Removed.
13103 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
13104 (scop_detection::harmful_stmt_in_bb): Likewise.
13105 (scop_detection::graphite_can_represent_stmt): Likewise.
13106 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
13107 (scop_detection::can_represent_loop): Remove recursion, fold in ...
13108 (scop_detection::can_represent_loop_1): ... this. Removed.
13109 (scop_detection::harmful_loop_in_region): Simplify after inlining
13110 the above and remove more quadraticness.
13111 (build_scops): Adjust.
13112 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
13115 2017-09-26 Jakub Jelinek <jakub@redhat.com>
13118 * config/i386/i386.c (ix86_print_operand_address_as): Only test
13119 REGNO (base) == SP_REG if base is a REG.
13121 PR middle-end/35691
13122 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
13123 if it is different SSA_NAME.
13124 (optimize_range_tests_cmp_bitwise): New function.
13125 (optimize_range_tests): Call it.
13127 2017-09-26 Richard Biener <rguenther@suse.de>
13129 PR tree-optimization/82321
13130 * graphite.c (canonicalize_loop_closed_ssa): Properly check
13131 for the def being inside the loop.
13133 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13135 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
13137 * config/s390/s390-builtins.def: Fix constraint on op4.
13139 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13141 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
13142 independent expanders.
13143 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
13144 ("vec_ordered", "vec_unordered"): New expanders.
13146 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13148 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
13151 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13153 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
13154 vec_unpacks_lo_v16qi.
13155 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
13157 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13159 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
13160 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
13161 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
13163 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13165 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
13167 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
13169 ("*vec_slb<mode>"): New insn pattern.
13170 ("vec_shr_<mode>"): New expander.
13171 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
13172 and force the shift count operand to V16QImode.
13173 ("vec_srb<mode>"): Set shift count mode to V16QI.
13175 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13177 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
13178 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
13179 ("vec_widen_smult_hi_<mode>"): New expander definitions.
13181 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
13184 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
13186 2017-09-26 Richard Biener <rguenther@suse.de>
13188 PR tree-optimization/82320
13189 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
13192 2017-09-25 Jeff Law <law@redhat.com>
13194 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
13195 prototype for new argument.
13196 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
13197 mostly extracted from rs6000_emit_allocate_stack.
13198 (rs6000_emit_probe_stack_range_stack_clash): New function.
13199 (rs6000_emit_allocate_stack): Call
13200 rs6000_emit_probe_stack_range_stack_clash as needed.
13201 (rs6000_emit_probe_stack_range): Add additional argument
13202 to call to gen_probe_stack_range{si,di}.
13203 (output_probe_stack_range): New.
13204 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
13205 (output_probe_stack_range_stack_clash): New.
13206 (rs6000_emit_prologue): Emit notes into dump file as requested.
13207 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
13208 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
13209 Add additional operand and pass it to output_probe_stack_range.
13211 2017-09-25 Bin Cheng <bin.cheng@arm.com>
13213 PR tree-optimization/82163
13214 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
13215 (checking_verify_loop_closed_ssa): New parameter.
13216 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
13217 (check_loop_closed_ssa_stmt): Delete.
13218 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
13219 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
13220 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
13223 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
13225 * brig-builtins.def: Treat HSAIL barrier builtins as
13226 setjmp/longjump style functions.
13228 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13230 * target.def (constant_alignment): New hook.
13231 * defaults.h (CONSTANT_ALIGNMENT): Delete.
13232 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
13233 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
13234 * doc/tm.texi: Regenerate.
13235 * targhooks.h (default_constant_alignment): Declare.
13236 (constant_alignment_word_strings): Likewise.
13237 * targhooks.c (default_constant_alignment): New function.
13238 (constant_alignment_word_strings): Likewise.
13239 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
13240 instead of CONSTANT_ALIGNMENT.
13241 * varasm.c (align_variable, get_variable_align, build_constant_desc)
13242 (force_const_mem): Likewise.
13243 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
13244 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
13245 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
13246 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13247 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
13249 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
13250 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13251 constant_alignment_word_strings.
13252 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
13253 (CONSTANT_ALIGNMENT): Likewise.
13254 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13255 (arm_constant_alignment): New function.
13256 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
13257 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13258 constant_alignment_word_strings.
13259 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
13260 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13261 constant_alignment_word_strings.
13262 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
13263 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13264 (cris_constant_alignment): New function.
13265 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
13266 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13267 (epiphany_constant_alignment): New function.
13268 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
13269 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13270 constant_alignment_word_strings.
13271 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
13272 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13273 constant_alignment_word_strings.
13274 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
13275 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13276 constant_alignment_word_strings.
13277 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
13278 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
13279 * config/i386/i386.c (ix86_constant_alignment): Make static.
13280 Use the same interface as the target hook.
13281 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13282 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
13283 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13284 constant_alignment_word_strings.
13285 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
13286 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
13287 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13288 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
13289 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13290 constant_alignment_word_strings.
13291 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
13292 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13293 constant_alignment_word_strings.
13294 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
13295 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13296 constant_alignment_word_strings.
13297 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
13298 * config/microblaze/microblaze.c (microblaze_constant_alignment):
13300 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13301 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
13302 * config/mips/mips.c (mips_constant_alignment): New function.
13303 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13304 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
13305 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
13306 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13307 (mmix_constant_alignment): Make static. Use the same interface
13308 as the target hook.
13309 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
13310 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13311 constant_alignment_word_strings.
13312 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
13313 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13314 constant_alignment_word_strings.
13315 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
13316 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13317 constant_alignment_word_strings.
13318 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
13319 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13320 (rs6000_constant_alignment): New function.
13321 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
13322 * config/riscv/riscv.c (riscv_constant_alignment): New function.
13323 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13324 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
13325 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13326 (rs6000_constant_alignment): New function.
13327 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
13328 * config/s390/s390.c (s390_constant_alignment): New function.
13329 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13330 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
13331 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13332 constant_alignment_word_strings.
13333 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
13334 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13335 (sparc_constant_alignment): New function.
13336 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
13337 * config/spu/spu.c (spu_constant_alignment): New function.
13338 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13339 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
13340 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13341 constant_alignment_word_strings.
13342 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
13343 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13344 constant_alignment_word_strings.
13345 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
13346 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13347 constant_alignment_word_strings.
13348 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
13349 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13350 (visium_constant_alignment): New function.
13351 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
13352 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13353 (xtensa_constant_alignment): New function.
13354 * system.h (CONSTANT_ALIGNMENT): Poison.
13356 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
13358 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13359 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
13360 (rs6000_builtin_valid_without_lhs): New helper function.
13361 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13362 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
13364 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13366 * target.h (vec_perm_indices): Use unsigned short rather than
13368 (auto_vec_perm_indices): Likewise.
13369 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
13370 Use unsigned int rather than unsigned char.
13371 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
13373 2017-09-25 Richard Biener <rguenther@suse.de>
13375 * cfgloop.h (sort_sibling_loops): Declare.
13376 * cfgloop.c (sort_sibling_loops_cmp): New helper.
13377 (sort_sibling_loops): New function sorting the sibling loop list
13379 * graphite.c (graphite_transform_loops): Sort sibling loops.
13381 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
13383 * target.def (vec_perm_const_ok): Change sel parameter to
13385 * optabs-query.c (can_vec_perm_p): Update accordingly.
13386 * doc/tm.texi: Regenerate.
13387 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
13388 auto_vec_perm_indices and remove separate nelt field.
13389 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
13390 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
13391 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
13392 (aarch64_expand_vec_perm_const): Update accordingly.
13393 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
13394 to vec_perm_indices.
13395 * config/arm/arm.c (expand_vec_perm_d): Change perm to
13396 auto_vec_perm_indices and remove separate nelt field.
13397 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
13398 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
13399 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
13401 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
13402 to vec_perm_indices.
13403 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
13404 sel to vec_perm_indices.
13405 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
13406 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
13407 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
13409 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
13412 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
13415 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
13416 on the FUNCTION_DECL function context if it has a DIE that is a
13419 2017-09-25 Richard Biener <rguenther@suse.de>
13421 PR tree-optimization/82285
13422 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
13425 2017-09-25 Tom de Vries <tom@codesourcery.com>
13429 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
13432 2017-09-25 Richard Biener <rguenther@suse.de>
13434 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
13435 ISL errors other than isl_error_quota happen. Dump if the
13436 schedule is the same.
13437 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
13438 errors instead of aborting inside ISL.
13440 2017-09-25 Iain Sandoe <iain@codesourcery.com>
13443 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
13444 of libgcc_eh for m64.
13445 * config/i386/darwin64.h: Likewise.
13447 2017-09-25 Richard Biener <rguenther@suse.de>
13449 PR middle-end/82144
13450 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
13451 attribute for incomplete types nor twice for complete ones.
13453 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
13456 * config/i386/i386.c (ix86_print_operand_address_as): Encode
13457 %esp as %rsp to avoid 0x67 prefix if there is no index or base
13460 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
13463 * config/i386/i386.opt (mprefer-avx256): Use
13464 ix86_target_flags variable.
13465 * config/i386/i386.c (ix86_target_string): Move
13466 -mprefer-avx256 to flag2_opts.
13468 2017-09-22 Jakub Jelinek <jakub@redhat.com>
13470 PR middle-end/35691
13471 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
13472 and x != -1 | y != -1 into (x & y) != -1.
13474 2017-09-22 Steve Ellcey <sellcey@cavium.com>
13476 * config.gcc: Add new case statement to set
13477 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
13478 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
13479 s390x-*-linux* case statements. Added aarch64 to the list of
13480 supported architectures.
13482 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13484 PR tree-optimization/82289
13485 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
13486 STMT_VINFO_RELEVANT_P.
13488 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13489 Alan Hayward <alan.hayward@arm.com>
13490 David Sherwood <david.sherwood@arm.com>
13492 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
13493 for VR_RANGE only; don't allow VR_ANTI_RANGE.
13494 (extract_range_from_binary_expr_1): Don't call
13495 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
13497 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13498 Alan Hayward <alan.hayward@arm.com>
13499 David Sherwood <david.sherwood@arm.com>
13501 * target.def (preferred_vector_alignment): New hook.
13502 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
13504 * doc/tm.texi: Regenerate.
13505 * targhooks.h (default_preferred_vector_alignment): Declare.
13506 * targhooks.c (default_preferred_vector_alignment): New function.
13507 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
13509 (DR_TARGET_ALIGNMENT): New macro.
13510 (aligned_access_p): Update commentary.
13511 (vect_known_alignment_in_bytes): New function.
13512 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
13514 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
13515 Calculate the misalignment based on the target alignment rather than
13517 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
13518 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
13519 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
13520 the target alignment, rather than masking the element misalignment
13521 with the number of elements in a vector. Also use the target
13522 alignment when calculating the maximum number of peels.
13523 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
13524 instead of TYPE_ALIGN_UNIT.
13525 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
13526 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
13527 (vect_create_addr_base_for_vector_ref): Update call accordingly.
13528 (vect_create_data_ref_ptr): Likewise.
13529 (vect_setup_realignment): Realign by ANDing with
13530 -DR_TARGET_MISALIGNMENT.
13531 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
13532 the number of peels based on DR_TARGET_ALIGNMENT.
13533 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
13534 with the guaranteed alignment boundary when deciding whether
13536 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
13537 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
13538 (ensure_base_align): Remove stmt_info parameter. Get the
13539 target base alignment from DR_TARGET_ALIGNMENT.
13540 (vectorizable_store): Update call accordingly. Interpret
13541 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
13543 (vectorizable_load): Likewise.
13545 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13546 Alan Hayward <alan.hayward@arm.com>
13547 David Sherwood <david.sherwood@arm.com>
13549 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
13550 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
13551 (vect_enhance_data_refs_alignment): Likewise.
13553 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13555 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13556 error. Only quit immediately if parsing is complete.
13557 (BEGIN): Initialize fatal_err and parse_done.
13558 (begin fpu, end fpu): Check number of arguments.
13559 (begin arch, end arch): Likewise.
13560 (begin cpu, end cpu): Likewise.
13561 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13562 (optalias): Likewise.
13564 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13566 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13567 * config/arm/arm-isa.h: Delete. Move definitions to ...
13568 * arm-cpus.in: ... here. Use new feature and fgroup values.
13569 * config/arm/arm.c (arm_option_override): Use lower case for feature
13571 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13572 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13573 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13575 (print_isa_bits_for): New function.
13576 (gen_isa): New function.
13577 (gen_comm_data): Use print_isa_bits_for.
13578 (define feature): New keyword.
13579 (define fgroup): New keyword.
13580 * config/arm/t-arm (TM_H): Remove.
13581 (GTM_H): Add arm-isa.h.
13582 (arm-isa.h): Add rule to generate file.
13583 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13584 case for feature bit names.
13586 2017-09-22 Richard Biener <rguenther@suse.de>
13588 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
13590 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
13591 print of no dependency loops ...
13592 * graphite.c (graphite_transform_loops): ... here.
13593 (canonicalize_loop_closed_ssa_form): Work from inner to outer
13595 (same_close_phi_node, remove_duplicate_close_phi,
13596 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
13597 (canonicalize_loop_closed_ssa): ... here and simplify.
13598 * graphite-optimize-isl.c: Include tree-vectorizer.h.
13599 (optimize_isl): Use dump_printf_loc to tell when we stopped
13600 optimizing because of an ISL timeout.
13602 2017-09-22 Richard Biener <rguenther@suse.de>
13604 PR tree-optimization/82291
13605 * tree-if-conv.c (predicate_mem_writes): Make sure to
13606 remove writes in blocks predicated with false.
13608 2017-09-22 Richard Biener <rguenther@suse.de>
13610 * sese.c: Include cfganal.h.
13611 (if_region_set_false_region): Remove.
13612 (create_if_region_on_edge): Likewise.
13613 (move_sese_in_condition): Re-implement without destroying
13616 2017-09-22 Richard Biener <rguenther@suse.de>
13618 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
13619 Verify both BBs contain loop PHI nodes before dispatching to
13620 copy_loop_phi_args.
13621 (graphite_regenerate_ast_isl): Do not recompute dominators,
13622 do not verify three times. Restructure for clarity.
13623 * graphite-scop-detection.c (same_close_phi_node,
13624 remove_duplicate_close_phi, make_close_phi_nodes_unique,
13625 defined_in_loop_p, canonicalize_loop_closed_ssa,
13626 canonicalize_loop_closed_ssa_form): Simplify, remove excess
13627 checking and SSA rewrite, move to ...
13628 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
13629 (graphite_initialize): Do not pass in ctx, do not reset the
13630 SCEV cache, compute only dominators.
13631 (graphite_transform_loops): Allocate ISL ctx after
13632 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
13633 Maintain post-dominators only around build_scops.
13634 * sese.c (if_region_set_false_region): Make static. Free
13635 and recompute dominators.
13636 (move_sese_in_condition): Assert we don't get called with
13637 post-dominators computed.
13638 * sese.h (if_region_set_false_region): Remove.
13640 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
13642 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
13643 mode attribute for TARGET_AVX512VL.
13645 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
13647 * config/i386/i386.opt (mprefer-avx256): New option.
13648 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
13650 (ix86_preferred_simd_mode): Return 256-bit AVX modes
13651 for TARGET_PREFER_AVX256.
13652 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
13654 2017-09-21 Jeff Law <law@redhat.com>
13656 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
13657 Fix dump output if the only stack space is for pushed registers.
13659 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13661 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
13664 2017-09-21 Martin Sebor <msebor@redhat.com>
13667 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
13668 code (in C++) or code that triggers warnings.
13670 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
13672 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
13674 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
13676 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
13677 * sched-rgn.c: Ditto.
13678 * sel-sched-ir.c: Ditto.
13680 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
13682 * toplev.h (set_random_seed): Adjust return type.
13683 * toplev.c (init_local_tick): Move eager initialization of random_seed
13684 to get_random_seed. Adjust comment.
13685 (init_random_seed): Inline to get_random_seed, delete.
13686 (get_random_seed): Initialize random_seed lazily.
13687 (set_random_seed): Do not return previous value.
13688 (print_switch_value): Do not call get_random_seed.
13690 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
13692 * cgraph.c (delete_function_version): New, broken out from...
13693 (cgraph_node::delete_function_version): ...here. Rename to
13694 cgraph_node::delete_function_version_by_decl. Update all uses.
13695 (cgraph_node::remove): Call delete_function_version.
13697 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13700 * tree-inline.c (expand_call_inline): Emit clobber stmts for
13701 VAR_DECLs to which addressable non-volatile parameters are mapped
13702 and for id->retvar after the return value assignment. Clear
13703 id->retval and id->retbnd after inlining.
13705 2017-09-21 Richard Biener <rguenther@suse.de>
13707 PR tree-optimization/82276
13708 PR tree-optimization/82244
13709 * tree-vrp.c (build_assert_expr_for): Set
13710 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
13712 (remove_range_assertions): Revert earlier change.
13714 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
13717 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
13719 2017-09-21 Richard Biener <rguenther@suse.de>
13721 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
13722 Restore valid IL after code generation errors.
13723 * graphite.c (graphite_transform_loops): Diagnose code
13724 generation issues as MSG_MISSED_OPTIMIZATION and continue
13725 with processing SCOPs.
13727 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13728 Alan Hayward <alan.hayward@arm.com>
13729 David Sherwood <david.sherwood@arm.com>
13731 * calls.c (compute_argument_addresses): Use simplify_gen_binary
13732 rather than choosing between plus_constant and gen_rtx_<CODE>.
13733 * expr.c (emit_push_insn): Likewise.
13734 (expand_expr_real_2): Likewise.
13736 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13737 Alan Hayward <alan.hayward@arm.com>
13738 David Sherwood <david.sherwood@arm.com>
13740 * loop-unroll.c (split_iv): Call copy_rtx on the step.
13742 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13743 Alan Hayward <alan.hayward@arm.com>
13744 David Sherwood <david.sherwood@arm.com>
13746 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
13747 calling tree_to_uhwi.
13749 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13750 Alan Hayward <alan.hayward@arm.com>
13751 David Sherwood <david.sherwood@arm.com>
13753 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
13754 INTEGER_CST rather than a negative test for ADDR_EXPR.
13756 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13757 Alan Hayward <alan.hayward@arm.com>
13758 David Sherwood <david.sherwood@arm.com>
13760 * tree-vrp.c (extract_range_from_binary_expr_1): Check
13761 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
13763 2017-09-21 Richard Biener <rguenther@suse.de>
13765 PR tree-optimization/71351
13766 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
13767 graphite_create_new_loop_guard): Remove, fold remaining parts
13769 (translate_isl_ast_node_for): ... here and simplify.
13771 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13774 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
13775 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
13776 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
13777 alternative always use QI mode, for -Os imov (=R,R) alternative
13778 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
13781 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13782 Jeff Law <law@redhat.com>
13784 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
13785 (allocate_stack_space): New function, partially extracted from
13786 s390_emit_prologue.
13787 (s390_emit_prologue): Track offset to most recent stack probe.
13788 Code to allocate space moved into allocate_stack_space.
13789 Dump actions when no stack is allocated.
13790 (s390_prologue_plus_offset): New function.
13791 (s390_emit_stack_probe): Likewise.
13793 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
13795 * common.opt (Wa, Wl, Wp, g, gz=): Add
13797 (gno-column-info): Remove.
13798 (gcolumn-info): Drop RejectNegative.
13799 (gno-): New prefix.
13800 (gno-record-gcc-switches): Remove.
13801 (grecord-gcc-switches): Drop RejectNegative.
13802 (gno-split-dwarf): Remove.
13803 (gsplit-dwarf): Drop RejectNegative.
13804 (gno-strict-dwarf): Remove.
13805 (gstrict-dwarf): Drop RejectNegative.
13806 * config/darwin.opt (gfull, gused): Add RejectNegative.
13807 * dwarf2out.c (gen_producer_string): Drop
13808 gno-record-gcc-switches handler.
13809 * optc-gen.awk: Add g to prefixes with negative forms.
13810 * opts-common.c (remapping_prefix_p): New.
13811 (find_opt): Check it.
13812 (generate_canonical_option): Test g prefix.
13813 (option_map): Add -gno- mapping.
13814 (add_misspelling_candidates): Check remapping_prefix_p.
13816 2017-09-20 Jeff Law <law@redhat.com>
13818 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
13819 thinko in stack clash protection support.
13821 * explow.c (compute_stack_clash_protection_loop_data): Use
13822 CONST_INT_P instead of explicit test. Verify object is a
13823 CONST_INT_P before looking at INTVAL.
13824 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
13825 instead of explicit test.
13827 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
13830 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
13831 address instead of to r1 and r11.
13833 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
13835 * config.gcc: Support "knm".
13836 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
13837 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13839 * config/i386/i386.c (m_KNM): Define.
13840 (processor_target_table): Add "knm".
13842 (ix86_option_override_internal): Add "knm".
13843 (ix86_issue_rate): Add PROCESSOR_KNM.
13844 (ix86_adjust_cost): Ditto.
13845 (ia32_multipass_dfa_lookahead): Ditto.
13846 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
13847 (fold_builtin_cpu): Add M_INTEL_KNM.
13848 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
13849 (processor_type): Add PROCESSOR_KNM.
13850 * config/i386/x86-tune.def: Add m_KNM.
13851 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
13853 2017-09-20 Richard Biener <rguenther@suse.de>
13855 PR tree-optimization/80213
13856 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
13857 are allowed in empty BBs as well.
13858 (canonicalize_loop_closed_ssa): Also look for other complex
13860 (scop_detection::get_sese): Include the loop-closed PHI block
13862 (scop_detection::merge_sese): Remove code adding extra blocks.
13863 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
13864 (build_scops): Assert the final returned scop is invalid.
13866 2017-09-20 Richard Biener <rguenther@suse.de>
13868 PR tree-optimization/82264
13869 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
13871 (vn_phi_lookup): Likewise.
13872 (vn_phi_insert): Likewise.
13874 2017-09-20 Jakub Jelinek <jakub@redhat.com>
13876 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
13877 that fits into uhwi or shwi, add DW_AT_const_value regardless
13878 of early_dwarf without going through RTL, using add_AT_unsigned
13881 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
13882 (DEBUG_LTO_ABBREV_SECTION): Likewise.
13883 (DEBUG_LTO_MACINFO_SECTION): Likewise.
13884 (DEBUG_MACRO_SECTION): Likewise.
13885 (DEBUG_LTO_MACRO_SECTION): Likewise.
13886 (DEBUG_STR_DWO_SECTION): Likewise.
13887 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
13888 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
13889 (DEBUG_LTO_DWO_LINE_SECTION): Define.
13890 (DEBUG_LTO_LINE_STR_SECTION): Define.
13891 (init_sections_and_labels): Initialize debug_line_str_section
13892 variable. Initialize debug_loc_section for -gdwarf-5 to
13893 DEBUG_LOCLISTS_SECTION. Formatting fixes.
13895 2017-09-20 Richard Biener <rguenther@suse.de>
13897 * graphite-sese-to-poly.c (extract_affine): Properly handle
13898 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
13900 2017-09-20 Richard Biener <rguenther@suse.de>
13902 PR tree-optimization/81373
13903 * graphite-scop-detection.c (build_cross_bb_scalars_def):
13904 Force SESE live-out defs to be handled even if they are
13907 2017-09-19 Jeff Law <law@redhat.com>
13909 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
13910 nothing for stack adjustments with REG_STACK_CHECK.
13911 * sched-deps.c (parse_add_or_inc): Reject insns with
13912 REG_STACK_CHECK from dependency breaking.
13913 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
13914 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
13915 * reg-notes.def (STACK_CHECK): New note.
13917 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
13918 (ix86_expand_prologue): Dump stack clash info as needed.
13919 Call ix86_adjust_stack_and_probe_stack_clash as needed.
13921 * function.c (dump_stack_clash_frame_info): New function.
13922 * function.h (dump_stack_clash_frame_info): Prototype.
13923 (enum stack_clash_probes): New enum.
13925 * config/alpha/alpha.c (alpha_expand_prologue): Also check
13926 flag_stack_clash_protection.
13927 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
13928 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
13929 (arm_frame_pointer_required): Likewise.
13930 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
13931 (ia64_expand_prologue): Likewise.
13932 * config/mips/mips.c (mips_expand_prologue): Likewise.
13933 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
13934 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13935 (sparc_flat_expand_prologue): Likewise.
13936 * config/spu/spu.c (spu_expand_prologue): Likewise.
13938 * explow.c: Include "params.h".
13939 (anti_adjust_stack_and_probe_stack_clash): New function.
13940 (get_stack_check_protect): Likewise.
13941 (compute_stack_clash_protection_loop_data): Likewise.
13942 (emit_stack_clash_protection_loop_start): Likewise.
13943 (emit_stack_clash_protection_loop_end): Likewise.
13944 (allocate_dynamic_stack_space): Use get_stack_check_protect.
13945 Use anti_adjust_stack_and_probe_stack_clash.
13946 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
13947 (emit_stack_clash_protection_loop_start): Likewise.
13948 (emit_stack_clash_protection_loop_end): Likewise.
13949 * rtl.h (get_stack_check_protect): Prototype.
13950 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
13951 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
13952 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
13954 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
13956 * doc/tm.texi: Rebuilt.
13957 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
13958 get_stack_check_protect.
13959 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
13960 * config/arm/arm.c (arm_expand_prologue): Likewise.
13961 (arm_frame_pointer_required): Likewise.
13962 * config/i386/i386.c (ix86_expand_prologue): Likewise.
13963 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
13964 * config/mips/mips.c (mips_expand_prologue): Likewise.
13965 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
13966 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
13967 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13968 (sparc_flat_expand_prologue): Likewise.
13970 * common.opt (-fstack-clash-protection): New option.
13971 * flag-types.h (enum stack_check_type): Note difference between
13972 -fstack-check= and -fstack-clash-protection.
13973 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
13974 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
13975 * toplev.c (process_options): Issue warnings/errors for cases
13976 not handled with -fstack-clash-protection.
13977 * doc/invoke.texi (-fstack-clash-protection): Document new option.
13978 (-fstack-check): Note additional problem with -fstack-check=generic.
13979 Note that -fstack-check is primarily for Ada and refer users
13980 to -fstack-clash-protection for stack-clash-protection.
13981 Document new params for stack clash protection.
13983 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
13985 * config/i386/i386.c (ix86_split_long_move): Do not handle
13986 address used for LEA in a special way.
13988 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
13990 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
13992 2017-09-19 Martin Sebor <msebor@redhat.com>
13995 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
13996 of incompatible types.
13998 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
14000 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14001 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
14002 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14003 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
14005 2017-09-19 Richard Biener <rguenther@suse.de>
14007 PR tree-optimization/82244
14008 * tree-vrp.c (remove_range_assertions): Do not propagate
14009 a constant to abnormals but replace the assert with a copy.
14011 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
14013 PR rtl-optimization/57878
14014 PR rtl-optimization/68988
14015 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
14016 avoidance test involving non_reload_pseudos. Move frequency test
14017 below the general fragmentation avoidance test.
14019 2017-09-19 Richard Biener <rguenther@suse.de>
14021 PR tree-optimization/69728
14022 * graphite-sese-to-poly.c (schedule_error): New global.
14023 (add_loop_schedule): Handle empty domain by failing the
14025 (build_original_schedule): Handle schedule_error.
14027 2017-09-19 Richard Biener <rguenther@suse.de>
14029 * graphite-scop-detection.c (scop_detection::can_represent_loop):
14030 Do not iterate to sibling loops but only to siblings of inner
14033 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
14036 * config/m68k/m68k.md (moveq feeding equality comparison): Check
14037 that the registers are different.
14039 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
14041 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
14042 to processor_model and "amdfam17h" to arch_names_table.
14043 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
14045 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14048 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
14050 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14051 Alan Hayward <alan.hayward@arm.com>
14052 David Sherwood <david.sherwood@arm.com>
14054 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
14056 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
14058 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
14059 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
14060 vect_schedule_slp_instance.
14061 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
14062 Update call to vect_slp_analyze_node_operations. Simplify return
14064 (vect_slp_analyze_bb_1): Update call accordingly.
14065 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
14066 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
14067 (vect_schedule_slp): Update call accordingly.
14069 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14070 Alan Hayward <alan.hayward@arm.com>
14071 David Sherwood <david.sherwood@arm.com>
14073 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
14074 with types that aren't in fact scalar.
14076 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14078 * tree-vect-slp.c (vect_record_max_nunits): New function,
14080 (vect_build_slp_tree_1): ...here.
14081 (vect_build_slp_tree_2): Call it for phis too.
14083 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14085 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
14086 to vect_get_vec_def_for_operand when getting the mask operand.
14088 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14089 Alan Hayward <alan.hayward@arm.com>
14090 David Sherwood <david.sherwood@arm.com>
14092 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
14095 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14096 Alan Hayward <alan.hayward@arm.com>
14097 David Sherwood <david.sherwood@arm.com>
14099 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
14100 calculation for vector booleans.
14102 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14103 Alan Hayward <alan.hayward@arm.com>
14104 David Sherwood <david.sherwood@arm.com>
14106 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
14108 (vect_transform_stmt): ...here.
14109 (vect_analyze_stmt): Use it instead of calling
14110 vectorizable_live_operation directly.
14112 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
14114 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
14115 non-SIMT targets in acc vector loops.
14117 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
14119 * configure.ac: Add arc and check if assembler supports gdwarf2.
14120 * configure: Regenerate.
14122 2017-09-18 Richard Biener <rguenther@suse.de>
14124 PR tree-optimization/82220
14125 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
14126 epilogue niters from the min_profitable_iters compute.
14128 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14131 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
14132 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
14133 (ix86_init_pic_reg): Revert 2017-09-01 changes.
14135 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
14138 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
14139 switching to a new text section.
14141 2017-09-18 Richard Biener <rguenther@suse.de>
14143 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
14145 (build_alias_set): Reject aliases with no access function.
14147 2017-09-18 Richard Biener <rguenther@suse.de>
14149 PR tree-optimization/79622
14150 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
14152 (build_cross_bb_scalars_use): Likewise.
14154 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
14156 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
14158 2017-09-18 Alan Modra <amodra@gmail.com>
14161 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
14162 stack_pointer_rtx for count 0. Update comments. Break up
14163 large rtl expression.
14165 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
14167 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
14168 Increase to 20 bytes.
14169 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
14170 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
14171 or avx version of the stub.
14173 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
14176 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
14177 compute the minimum stack alignment. Also update preferred stack
14178 boundary for leaf functions.
14180 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
14182 PR tree-optimization/82228
14183 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
14186 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14188 * common/config/nds32/nds32-common.c
14189 (nds32_option_optimization_table): Refine formatting.
14190 (nds32_option_optimization_table): Use -fsched-pressure and
14191 -fomit-frame-pointer for specific optimization level.
14193 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14195 * config/nds32/nds32.c: Refine formatting and comments.
14196 * config/nds32/nds32.h: Likewise.
14197 * config/nds32/nds32.md: Likewise.
14198 * config/nds32/nds32-cost.c: Likewise.
14199 * config/nds32/nds32-isr.c: Likewise.
14200 * config/nds32/nds32-md-auxiliary.c: Likewise.
14201 * config/nds32/nds32-multiple.md: Likewise.
14202 * config/nds32/nds32-predicates.c: Likewise.
14204 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
14205 Jakub Jelinek <jakub@redhat.com>
14207 Add support for -std=c++2a.
14208 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
14210 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
14212 2017-09-15 Steve Ellcey <sellcey@cavium.com>
14215 * doc/extend.texi (Common Function Attributes): Add
14216 references to ARM, AArch64, and S/390 specific attributes.
14217 (Function Specific Option Pragmas): Add AArch64 and S/390
14218 to list of back ends that support the target pragma.
14220 2017-09-15 Nathan Sidwell <nathan@acm.org>
14222 * doc/standards.texi: Fix C++17 description. Update URLs for
14225 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
14227 * common.opt (Wcast-align=strict): New warning option.
14228 * doc/invoke.texi: Document -Wcast-align=strict.
14230 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
14232 * cgraph.h (cgraph_thunk_info): Add comments.
14233 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
14234 assert for VIRTUAL_* arguments stricter.
14236 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
14238 PR tree-optimization/71026
14239 * match.pd: Move RDIV patterns from fold-const.c
14240 * fold-const.c (distribute_real_division): Removed.
14241 (fold_binary_loc): Remove calls to distribute_real_divison.
14243 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14245 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
14246 c++1z and gnu++1z as deprecated. Change other references to
14247 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
14248 Change -Wc++1z-compat to -Wc++17-compat.
14249 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
14250 * dwarf2out.c (highest_c_language): Handle C++17.
14251 (gen_compile_unit_die): Likewise.
14253 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14255 PR rtl-optimization/82192
14256 * combine.c (make_extraction): Don't look through non-paradoxical
14257 SUBREGs or TRUNCATE if pos + len is or might be bigger than
14260 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14261 Alan Hayward <alan.hayward@arm.com>
14262 David Sherwood <david.sherwood@arm.com>
14264 * target.def (function_arg_offset): New hook.
14265 * targhooks.h (default_function_arg_offset): Declare.
14266 * targhooks.c (default_function_arg_offset): New function.
14267 * function.c (locate_and_pad_parm): Use
14268 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
14269 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
14270 (TARGET_FUNCTION_ARG_OFFSET): ...this.
14271 * doc/tm.texi: Regenerate.
14272 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
14273 * config/spu/spu.c (spu_function_arg_offset): New function.
14274 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
14275 * system.h (FUNCTION_ARG_OFFSET): Poison.
14277 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14278 Alan Hayard <alan.hayward@arm.com>
14279 David Sherwood <david.sherwood@arm.com>
14281 * target.def (truly_noop_truncation): New hook.
14282 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
14283 than TRULY_NOOP_TRUNCATION.
14284 * hooks.h (hook_bool_uint_uint_true): Declare.
14285 * hooks.c (hook_bool_uint_uint_true): New function.
14286 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
14287 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
14288 * doc/tm.texi: Regenerate.
14289 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
14290 rather than TRULY_NOOP_TRUNCATION in comments.
14291 (simplify_comparison): Likewise.
14292 (record_truncated_value): Likewise.
14293 * expmed.c (extract_bit_field_1): Likewise.
14294 (extract_split_bit_field): Likewise.
14295 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
14296 instead of TRULY_NOOP_TRUNCATION.
14297 * function.c (assign_parm_setup_block): Likewise.
14298 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
14299 * rtlhooks.c: Include target.h.
14300 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
14301 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
14302 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
14303 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
14304 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
14305 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
14306 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
14307 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
14308 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
14309 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
14310 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
14311 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
14312 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
14313 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
14314 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
14315 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
14316 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
14317 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
14318 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
14319 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
14320 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
14321 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
14322 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
14323 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
14324 * config/mips/mips.c (mips_truly_noop_truncation): New function.
14325 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14326 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14327 rather than TRULY_NOOP_TRUNCATION in comments.
14328 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
14329 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
14330 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
14331 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
14332 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
14333 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
14334 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
14335 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
14336 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
14337 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
14338 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
14339 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14340 rather than TRULY_NOOP_TRUNCATION in comments.
14341 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
14342 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
14343 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
14344 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
14345 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
14346 TRULY_NOOP_TRUNCATION condition.
14347 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
14348 (TRULY_NOOP_TRUNCATION): Delete.
14349 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
14350 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
14351 * config/spu/spu.c (spu_truly_noop_truncation): New function.
14352 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14353 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
14354 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
14355 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
14356 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14357 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14358 rather than TRULY_NOOP_TRUNCATION in comments.
14359 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
14360 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
14361 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
14362 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
14363 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
14364 * system.h (TRULY_NOOP_TRUNCATION): Poison.
14366 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
14369 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
14370 (*cmp_ior): Likewise.
14371 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
14372 (*ior_scc_scc_cmp): Likewise.
14373 (*and_scc_scc): Likewise.
14374 (*and_scc_scc_cmp): Likewise.
14376 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14377 Alan Hayard <alan.hayward@arm.com>
14378 David Sherwood <david.sherwood@arm.com>
14380 * target.def (can_change_mode_class): New hook.
14381 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
14382 (hard_regno_nregs): Likewise.
14383 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
14384 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
14385 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
14386 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
14387 (LOAD_EXTEND_OP): Update accordingly.
14388 * doc/tm.texi: Regenerate.
14389 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
14390 CANNOT_CHANGE_MODE_CLASS.
14391 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
14392 (REG_CAN_CHANGE_MODE_P): ...this new macro.
14393 * combine.c (simplify_set): Update accordingly.
14394 * emit-rtl.c (validate_subreg): Likewise.
14395 * recog.c (general_operand): Likewise.
14396 * regcprop.c (mode_change_ok): Likewise.
14397 * reload1.c (choose_reload_regs): Likewise.
14398 (inherit_piecemeal_p): Likewise.
14399 * rtlanal.c (simplify_subreg_regno): Likewise.
14400 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
14401 instead of CANNOT_CHANGE_MODE_CLASS.
14402 (reload_cse_simplify_operands): Likewise.
14403 * reload.c (push_reload): Use targetm.can_change_mode_class
14404 instead of CANNOT_CHANGE_MODE_CLASS.
14405 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
14406 REG_CANNOT_CHANGE_MODE_P.
14407 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14408 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
14409 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14410 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14411 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14412 (arm_can_change_mode_class): New function.
14413 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14414 than CANNOT_CHANGE_MODE_CLASS in comments.
14415 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14416 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
14417 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
14418 (ix86_can_change_mode_class): ...this new function, inverting the
14419 sense of the return value.
14420 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14421 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14422 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14423 (ia64_can_change_mode_class): New function.
14424 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14425 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
14426 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
14427 (m32c_can_change_mode_class): ...this new function, inverting the
14428 sense of the return value.
14429 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14430 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14431 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
14432 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
14433 (mips_can_change_mode_class): ...this new function, inverting the
14434 sense of the return value.
14435 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14436 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14437 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14438 (msp430_can_change_mode_class): New function.
14439 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14440 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
14441 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14442 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14443 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14444 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
14445 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14446 (pa_cannot_change_mode_class): Replace with...
14447 (pa_can_change_mode_class): ...this new function, inverting the
14448 sense of the return value.
14449 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14450 than CANNOT_CHANGE_MODE_CLASS in comments.
14451 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14452 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
14453 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14454 (pdp11_cannot_change_mode_class): Replace with...
14455 (pdp11_can_change_mode_class): ...this new function, inverting the
14456 sense of the return value.
14457 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14458 * config/powerpcspe/powerpcspe-protos.h
14459 (rs6000_cannot_change_mode_class_ptr): Delete.
14460 * config/powerpcspe/powerpcspe.c
14461 (rs6000_cannot_change_mode_class_ptr): Delete.
14462 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14463 (rs6000_option_override_internal): Assign to
14464 targetm.can_change_mode_class instead of
14465 rs6000_cannot_change_mode_class_ptr.
14466 (rs6000_cannot_change_mode_class): Replace with...
14467 (rs6000_can_change_mode_class): ...this new function, inverting the
14468 sense of the return value.
14469 (rs6000_debug_cannot_change_mode_class): Replace with...
14470 (rs6000_debug_can_change_mode_class): ...this new function.
14471 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14472 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
14473 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14474 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14475 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
14477 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
14478 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14479 (rs6000_option_override_internal): Assign to
14480 targetm.can_change_mode_class instead of
14481 rs6000_cannot_change_mode_class_ptr.
14482 (rs6000_cannot_change_mode_class): Replace with...
14483 (rs6000_can_change_mode_class): ...this new function, inverting the
14484 sense of the return value.
14485 (rs6000_debug_cannot_change_mode_class): Replace with...
14486 (rs6000_debug_can_change_mode_class): ...this new function.
14487 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14488 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
14489 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
14490 (s390_can_change_mode_class): ...this new function, inverting the
14491 sense of the return value.
14492 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14493 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14494 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
14495 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14496 (sh_cannot_change_mode_class): Replace with...
14497 (sh_can_change_mode_class): ...this new function, inverting the
14498 sense of the return value.
14499 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14500 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14501 (sparc_can_change_mode_class): New function.
14502 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14503 * config/spu/spu.c (spu_can_change_mode_class): New function.
14504 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14505 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14506 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14507 (visium_can_change_mode_class): New function.
14508 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
14510 2017-09-15 Richard Biener <rguenther@suse.de>
14512 PR tree-optimization/82217
14513 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
14514 but not undefined case.
14516 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14519 * postreload.c (reload_cse_simplify_operands): Skip
14520 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
14522 2017-09-15 Richard Biener <rguenther@suse.de>
14524 PR tree-optimization/68823
14525 * graphite-scop-detection.c (build_alias_set): If we have a
14526 possible dependence check whether we can handle them by just
14527 looking at the DRs DR_ACCESS_FNs.
14528 (build_scops): If build_alias_set fails, fail the SCOP.
14530 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
14532 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
14533 to support float128 built-in functions that require the ISA 3.0
14535 (BU_FLOAT128_3_HW): Likewise.
14536 (SQRTF128): Add support for the IEEE 128-bit square root and fma
14537 built-in functions.
14538 (FMAF128): Likewise.
14540 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14541 support for built-in functions that need the ISA 3.0 IEEE 128-bit
14542 floating point instructions.
14543 (rs6000_invalid_builtin): Likewise.
14544 (rs6000_builtin_mask_names): Likewise.
14545 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
14546 (RS6000_BTM_FLOAT128_HW): Likewise.
14547 (RS6000_BTM_COMMON): Likewise.
14548 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
14550 * doc/extend.texi (RS/6000 built-in functions): Document the
14551 IEEE 128-bit floating point square root and fused multiply-add
14552 built-in functions.
14554 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
14556 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
14557 reg (r2) isn't in the set of registers defined in the prologue.
14559 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14560 Alan Hayward <alan.hayward@arm.com>
14561 David Sherwood <david.sherwood@arm.com>
14563 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
14564 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
14565 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
14566 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
14567 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
14569 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
14570 max_vectorization_factor.
14571 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
14573 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14574 Alan Hayward <alan.hayward@arm.com>
14575 David Sherwood <david.sherwood@arm.com>
14577 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
14578 (vect_worthwhile_without_simd_p): Declare.
14579 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
14580 (vectorizable_reduction): Use it.
14581 * tree-vect-stmts.c (vectorizable_shift): Likewise.
14582 (vectorizable_operation): Likewise.
14584 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14585 Alan Hayward <alan.hayward@arm.com>
14586 David Sherwood <david.sherwood@arm.com>
14588 * tree-vectorizer.h (vect_get_num_copies): New function.
14589 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
14590 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14591 (vectorizable_induction): Likewise.
14592 (vectorizable_live_operation): Likewise.
14593 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14594 (vectorizable_bswap): Likewise.
14595 (vectorizable_call): Likewise.
14596 (vectorizable_conversion): Likewise.
14597 (vectorizable_assignment): Likewise.
14598 (vectorizable_shift): Likewise.
14599 (vectorizable_operation): Likewise.
14600 (vectorizable_store): Likewise.
14601 (vectorizable_load): Likewise.
14602 (vectorizable_condition): Likewise.
14603 (vectorizable_comparison): Likewise.
14604 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
14606 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14607 Alan Hayward <alan.hayward@arm.com>
14608 David Sherwood <david.sherwood@arm.com>
14610 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
14611 of vect_init_vector.
14613 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14614 Alan Hayward <alan.hayward@arm.com>
14615 David Sherwood <david.sherwood@arm.com>
14617 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
14618 an inline wrapper that provides a location.
14619 (gimple_build_vector): Likewise.
14620 * gimple-fold.c (gimple_build_vector_from_val): New function.
14621 (gimple_build_vector): Likewise.
14622 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
14623 functions to build the initial value. Always return a gimple value.
14624 (get_initial_defs_for_reduction): Likewise. Only compute
14626 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
14627 vect_init_vector on the results from get_initial_def(s)_for_reduction.
14628 (vectorizable_induction): Use gimple_build_vector rather than
14631 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14632 Alan Hayward <alan.hayward@arm.com>
14633 David Sherwood <david.sherwood@arm.com>
14635 * target.h (vec_perm_indices): New typedef.
14636 (auto_vec_perm_indices): Likewise.
14637 * optabs-query.h: Include target.h
14638 (can_vec_perm_p): Take a vec_perm_indices *.
14639 * optabs-query.c (can_vec_perm_p): Likewise.
14640 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
14641 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14642 * tree-vect-generic.c (lower_vec_perm): Likewise.
14643 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
14644 (vect_grouped_load_supported): Likewise.
14645 (vect_shift_permute_load_chain): Likewise.
14646 (vect_permute_store_chain): Use auto_vec_perm_indices.
14647 (vect_permute_load_chain): Likewise.
14648 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
14649 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
14650 Update uses of can_vec_perm_p.
14651 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
14652 mode with a number of elements. Take a vec_perm_indices *.
14653 (vect_create_epilog_for_reduction): Update accordingly.
14654 Use auto_vec_perm_indices.
14655 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
14656 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
14657 (vect_transform_slp_perm_load): Likewise.
14658 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
14659 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
14660 (vect_gen_perm_mask_checked): Likewise.
14661 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
14662 (vect_gen_perm_mask_checked): Likewise.
14663 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
14664 (vectorizable_store): Likewise.
14665 (vectorizable_load): Likewise.
14666 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
14667 (vectorizable_bswap): Likewise.
14669 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14670 Alan Hayward <alan.hayward@arm.com>
14671 David Sherwood <david.sherwood@arm.com>
14673 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
14674 * tree.c (build_vector): Likewise.
14675 (build_vector_from_ctor): Update accordingly.
14676 (build_vector_from_val): Likewise.
14677 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14678 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14679 * tree-vect-generic.c (add_rshift): Likewise.
14680 (expand_vector_divmod): Likewise.
14681 (optimize_vector_constructor): Likewise.
14682 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
14683 (vect_transform_slp_perm_load): Likewise.
14684 (vect_schedule_slp_instance): Likewise.
14685 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
14686 (vectorizable_call): Likewise.
14687 (vect_gen_perm_mask_any): Likewise. Add elements in order.
14688 * expmed.c (make_tree): Likewise.
14689 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
14690 a vector passed to build_vector.
14691 (fold_convert_const): Likewise.
14692 (exact_inverse): Likewise.
14693 (fold_ternary_loc): Likewise.
14694 (fold_relational_const): Likewise.
14695 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
14696 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
14697 (const_unop): Likewise. Store the reduction accumulator in a
14698 variable rather than an array.
14699 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
14700 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
14701 the new vector, rather than constructing it after the input arrays.
14702 (native_interpret_vector): Use auto_vec<tree> when building
14703 a vector passed to build_vector. Add elements in order.
14704 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
14705 auto_vec<tree> when building a vector passed to build_vector.
14706 (vect_create_epilog_for_reduction): Likewise.
14707 (vectorizable_induction): Likewise.
14708 (get_initial_def_for_reduction): Likewise. Fix indentation of
14710 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
14712 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
14713 passed to build_vector.
14715 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14716 Alan Hayward <alan.hayward@arm.com>
14717 David Sherwood <david.sherwood@arm.com>
14719 * tree-core.h (tree_base::u): Add an "nelts" field.
14720 (tree_vector): Use VECTOR_CST_NELTS as the length.
14721 * tree.c (tree_size): Likewise.
14722 (make_vector): Initialize VECTOR_CST_NELTS.
14723 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
14724 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
14725 TYPE_VECTOR_SUBPARTS.
14726 * expr.c (const_vector_mask_from_tree): Consistently use "units"
14727 as the number of units, setting it from VECTOR_CST_NELTS.
14728 (const_vector_from_tree): Likewise.
14729 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
14730 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
14731 (fold_negate_expr_1): Likewise.
14732 (fold_convert_const): Likewise.
14733 (const_binop): Likewise. Differentiate the number of output and
14735 (const_unop): Likewise.
14736 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
14737 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
14738 in cases that did the opposite.
14740 2017-09-14 Richard Biener <rguenther@suse.de>
14742 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
14745 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
14747 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
14749 2017-09-14 Jakub Jelinek <jakub@redhat.com>
14752 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
14753 if and where to split a bb, except for splitting before debug insn
14754 sequences followed by non-label real insn. Delete debug insns
14755 in between basic blocks.
14757 * combine.c (make_compound_operation_int): Formatting fixes.
14759 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
14760 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14761 * config/netbsd.h (LINK_EH_SPEC): Likewise.
14762 * config/sol2.h (LINK_EH_SPEC): Likewise.
14763 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14764 * config/s390/linux.h (LINK_SPEC): Likewise.
14765 * config/freebsd.h (LINK_EH_SPEC): Likewise.
14766 * config/openbsd.h (LINK_EH_SPEC): Likewise.
14767 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14768 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
14769 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
14770 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14771 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
14772 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
14774 2017-09-13 Jakub Jelinek <jakub@redhat.com>
14776 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
14778 (ENDFILE_LINUX_SPEC): Likewise.
14779 (LINK_EH_SPEC): Likewise.
14780 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
14781 (LINK_OS_LINUX_SPEC32): Likewise.
14782 (LINK_OS_LINUX_SPEC64): Likewise.
14783 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
14784 (LINK_OS_LINUX_SPEC): Likewise.
14786 2017-09-13 Martin Liska <mliska@suse.cz>
14788 PR middle-end/82154
14789 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
14790 CASE_HIGH is NULL_TREE.
14792 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14793 Alan Hayward <alan.hayward@arm.com>
14794 David Sherwood <david.sherwood@arm.com>
14796 * target.def (secondary_memory_needed): New hook.
14797 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
14798 instead of SECONDARY_MEMORY_NEEDED.
14799 (secondary_memory_needed_mode): Likewise.
14800 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
14801 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
14802 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
14803 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
14804 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
14805 * doc/tm.texi: Regenerate.
14806 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
14807 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
14808 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14809 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
14810 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
14811 * config/i386/i386.c (inline_secondary_memory_needed): Put the
14812 mode argument first and change the reg_class arguments to reg_class_t.
14813 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
14814 Make static. Update the call to inline_secondary_memory_needed.
14815 (ix86_register_move_cost): Update the call to
14816 inline_secondary_memory_needed.
14817 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14818 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
14820 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
14821 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14823 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
14824 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
14825 * config/mips/mips.c (mips_secondary_memory_needed): Make static
14826 and match hook interface. Add comment from mips.h.
14827 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14828 * config/mmix/mmix.md (truncdfsf2): Refer to
14829 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14831 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
14832 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
14833 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14834 (pa_secondary_memory_needed): New function.
14835 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
14836 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
14837 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14838 (pdp11_secondary_memory_needed): Make static and match hook interface.
14839 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
14840 * config/powerpcspe/powerpcspe-protos.h
14841 (rs6000_secondary_memory_needed_ptr): Delete.
14842 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
14844 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14845 (rs6000_option_override_internal): Assign to
14846 targetm.secondary_memory_needed rather than
14847 rs6000_secondary_memory_needed_ptr.
14848 (rs6000_secondary_memory_needed): Match hook interface.
14849 (rs6000_debug_secondary_memory_needed): Likewise.
14850 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
14851 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
14852 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
14853 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14854 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
14855 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
14857 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
14858 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14859 (rs6000_option_override_internal): Assign to
14860 targetm.secondary_memory_needed rather than
14861 rs6000_secondary_memory_needed_ptr.
14862 (rs6000_secondary_memory_needed): Match hook interface.
14863 (rs6000_debug_secondary_memory_needed): Likewise.
14864 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
14865 * config/s390/s390.c (s390_secondary_memory_needed): New function.
14866 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14867 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
14868 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14869 (sparc_secondary_memory_needed): New function.
14870 * lra-constraints.c (check_and_process_move): Refer to
14871 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14873 (curr_insn_transform): Likewise.
14874 (process_alt_operands): Use targetm.secondary_memory_needed
14875 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14876 (check_secondary_memory_needed_p): Likewise.
14877 (choose_split_class): Likewise.
14878 * reload.c: Unconditionally include code that was previously
14879 conditional on SECONDARY_MEMORY_NEEDED.
14880 (push_secondary_reload): Use targetm.secondary_memory_needed
14881 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14882 (push_reload): Likewise.
14883 * reload1.c: Unconditionally include code that was previously
14884 conditional on SECONDARY_MEMORY_NEEDED.
14885 (choose_reload_regs): Use targetm.secondary_memory_needed
14886 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14887 (gen_reload): Likewise.
14888 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
14890 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14891 Alan Hayward <alan.hayward@arm.com>
14892 David Sherwood <david.sherwood@arm.com>
14894 * target.def (secondary_memory_needed_mode): New hook:
14895 * targhooks.c (default_secondary_memory_needed_mode): Declare.
14896 * targhooks.h (default_secondary_memory_needed_mode): New function.
14897 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
14898 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
14899 * doc/tm.texi: Regenerate.
14900 * lra-constraints.c (check_and_process_move): Use
14901 targetm.secondary_memory_needed_mode instead of
14902 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
14903 (curr_insn_transform): Likewise.
14904 * reload.c (get_secondary_mem): Likewise.
14905 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14906 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
14908 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14909 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14910 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
14911 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14912 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
14914 * config/powerpcspe/powerpcspe-protos.h
14915 (rs6000_secondary_memory_needed_mode): Delete.
14916 * config/powerpcspe/powerpcspe.c
14917 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14918 (rs6000_secondary_memory_needed_mode): Make static.
14919 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14920 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
14922 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14924 (rs6000_secondary_memory_needed_mode): Make static.
14925 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14926 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
14927 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14928 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14929 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14931 (sparc_secondary_memory_needed_mode): New function.
14932 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
14934 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
14936 * config/aarch64/constraints.md (Umq): New constraint.
14937 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
14939 (mov<mode>): Update condition.
14941 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14943 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
14944 when bitposition is the same.
14946 2017-09-13 Richard Biener <rguenther@suse.de>
14948 * dwarf2out.c (output_die_symbol): Remove.
14949 (output_die): Do not output a DIEs symbol.
14951 2017-09-13 Richard Biener <rguenther@suse.de>
14953 PR middle-end/82128
14954 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
14955 default-def to avoid breaking iterator update with the weird
14956 interaction with cgraph_update_edges_for_call_stmt_node.
14958 2017-09-13 Richard Biener <rguenther@suse.de>
14960 * tree-cfg.c (verify_gimple_assign_binary): Add verification
14961 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
14962 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
14963 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
14965 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
14967 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
14968 Disable pc relative literal load irrespective of
14969 TARGET_FIX_ERR_A53_84341 for default.
14971 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
14973 * config/sparc/sparc.c (output_return): Output the source location of
14974 the insn in the delay slot, if any.
14975 (output_sibcall): Likewise.
14977 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
14980 * common.opt (-static-pie): New alias.
14981 (shared): Negate static-pie.
14982 (-no-pie): Update help text.
14984 (static-pie): New option.
14985 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
14986 -static-pie support.
14987 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
14988 (LINK_EH_SPEC): Likewise.
14989 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14990 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14991 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14992 * gcc.c (LINK_COMMAND_SPEC): Likewise.
14993 (init_gcc_specs): Likewise.
14994 (init_spec): Likewise.
14995 (display_help): Update help message for -pie.
14996 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
14999 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
15001 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
15002 (movdi_aarch64): Likewise.
15003 (movti_aarch64): Likewise.
15005 2017-09-12 Simon Wright <simon@pushface.org>
15008 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
15009 calculation of the minor version, always output as 0.
15011 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15014 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15015 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
15016 on it early, rather than manual conversion late. For
15017 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
15018 instead of performing manual conversion.
15020 2017-09-12 Carl Love <cel@us.ibm.com>
15022 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
15023 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
15025 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15026 VMULOSW): Add definitions.
15027 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15028 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15029 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15030 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15031 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15033 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15035 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
15037 (movti_aarch64): Likewise.
15038 (movdf_aarch64): Likewise.
15039 (movtf_aarch64): Likewise.
15040 (load_pairdi): Likewise.
15041 (store_pairdi): Likewise.
15042 (load_pairdf): Likewise.
15043 (store_pairdf): Likewise.
15044 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
15045 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
15046 (ldr_got_small_<mode>): Likewise.
15047 (ldr_got_small_28k_<mode>): Likewise.
15048 (ldr_got_tiny): Likewise.
15049 * config/aarch64/iterators.md (ldst_sz): New.
15050 (ldpstp_sz): Likewise.
15051 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
15053 (thunderx_load): Split load_8 to load_16.
15054 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
15056 (thunderx2t99_storepair_basic): Split store_8 to store_16.
15057 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
15058 (xgene1_store_pair): Split store_8 to store_16.
15059 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
15060 (falkor_st_0_st_sd): Split store_8 to store_16.
15062 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15064 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
15065 and store1/2/3/4 to store_4/8/12/16.
15066 * config/aarch64/aarch64.md: Update for rename.
15067 * config/arm/arm.md: Likewise.: Likewise.
15068 * config/arm/arm.c: Likewise.
15069 * config/arm/thumb1.md: Likewise.
15070 * config/arm/thumb2.md: Likewise.
15071 * config/arm/vfp.md: Likewise.
15072 * config/arm/arm-generic.md: Likewise.
15073 * config/arm/arm1020e.md: Likewise.
15074 * config/arm/arm1026ejs.md: Likewise.
15075 * config/arm/arm1136jfs.md: Likewise.
15076 * config/arm/arm926ejs.md: Likewise.
15077 * config/arm/cortex-a15.md: Likewise.
15078 * config/arm/cortex-a17.md: Likewise.
15079 * config/arm/cortex-a5.md: Likewise.
15080 * config/arm/cortex-a53.md: Likewise.
15081 * config/arm/cortex-a57.md: Likewise.
15082 * config/arm/cortex-a7.md: Likewise.
15083 * config/arm/cortex-a8.md: Likewise.
15084 * config/arm/cortex-a9.md: Likewise.
15085 * config/arm/cortex-m4.md: Likewise.
15086 * config/arm/cortex-m7.md: Likewise.
15087 * config/arm/cortex-r4.md: Likewise.
15088 * config/arm/exynos-m1.md: Likewise.
15089 * config/arm/fa526.md: Likewise.
15090 * config/arm/fa606te.md: Likewise.
15091 * config/arm/fa626te.md: Likewise.
15092 * config/arm/fa726te.md: Likewise.
15093 * config/arm/fmp626.md: Likewise.
15094 * config/arm/iwmmxt.md: Likewise.
15095 * config/arm/ldmstm.md: Likewise.
15096 * config/arm/marvell-pj4.md: Likewise.
15097 * config/arm/xgene1.md: Likewise.
15098 * config/aarch64/thunderx.md: Likewise.
15099 * config/aarch64/thunderx2t99.md: Likewise.
15100 * config/aarch64/falkor.md: Likewise.
15102 2017-09-12 Martin Liska <mliska@suse.cz>
15104 * attribs.c (private_lookup_attribute): New function.
15105 * attribs.h (private_lookup_attribute): Declared here.
15106 (lookup_attribute): Called from this place.
15108 2017-09-12 Richard Biener <rguenther@suse.de>
15110 PR tree-optimization/82157
15111 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
15112 stmts with side-effects.
15114 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15115 Alan Hayward <alan.hayward@arm.com>
15116 David Sherwood <david.sherwood@arm.com>
15118 * target.def (hard_regno_nregs): New hook.
15119 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
15120 * targhooks.h (default_hard_regno_nregs): Declare.
15121 * targhooks.c (default_hard_regno_nregs): New function.
15122 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
15123 (TARGET_HARD_REGNO_NREGS): ...this hook.
15124 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
15125 (CLASS_MAX_NREGS): Likewise.
15126 * doc/tm.texi: Regenerate.
15127 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
15128 instead of HARD_REGNO_NREGS.
15129 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
15130 HARD_REGNO_NREGS in the comment.
15131 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
15132 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
15133 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
15134 Return an unsigned int.
15135 (TARGET_HARD_REGNO_NREGS): Redefine.
15136 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
15137 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
15138 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15139 (arc_hard_regno_nregs): New function.
15140 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
15141 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
15142 (arm_hard_regno_nregs): New function.
15143 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
15144 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
15145 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
15146 (TARGET_HARD_REGNO_NREGS): Redefine.
15147 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
15148 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
15149 (HARD_REGNO_NREGS): Delete.
15150 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
15151 (cr16_hard_regno_nregs): New function.
15152 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
15153 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
15154 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
15155 (cris_hard_regno_nregs): New function.
15156 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
15157 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
15158 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15159 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
15160 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
15161 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
15162 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
15163 (frv_hard_regno_nregs): Make static. Take and return an
15165 (frv_class_max_nregs): Remove outdated copy of documentation.
15166 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
15167 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
15168 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
15169 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
15170 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
15171 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
15172 (TARGET_HARD_REGNO_NREGS): Redefine.
15173 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
15174 (CLASS_MAX_NREGS): Update comment.
15175 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
15176 (ia64_hard_regno_nregs): New function.
15177 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
15178 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
15179 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
15180 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
15181 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
15183 (m32c_hard_regno_nregs): Likewise. Make static.
15184 (TARGET_HARD_REGNO_NREGS): Redefine.
15185 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
15186 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
15187 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
15188 (m68k_hard_regno_nregs): New function.
15189 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
15190 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
15191 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
15192 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
15193 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
15194 Take and return an unsigned int.
15195 (TARGET_HARD_REGNO_NREGS): Redefine.
15196 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
15197 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15198 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
15199 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
15200 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
15201 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
15202 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
15203 (msp430_hard_regno_nregs): Make static. Take and return an
15205 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
15206 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
15207 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
15208 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
15209 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
15210 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
15211 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
15212 (TARGET_HARD_REGNO_NREGS): Redefine.
15213 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
15214 (PA_HARD_REGNO_NREGS): ...this.
15215 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
15216 (PA_HARD_REGNO_NREGS): ...this.
15217 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15218 (pa_hard_regno_nregs): New function.
15219 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
15220 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
15221 (pdp11_hard_regno_nregs): New function.
15222 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
15223 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
15224 (rs6000_hard_regno_nregs_hook): New function.
15225 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
15226 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
15227 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
15228 Take and return an unsigned int. Move earlier in file.
15229 (TARGET_HARD_REGNO_NREGS): Redefine.
15230 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
15231 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
15232 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
15233 (rl78_hard_regno_nregs): Make static. Take and return an
15235 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
15236 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
15237 (rs6000_hard_regno_nregs_hook): New function.
15238 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
15239 * config/rx/rx.c (rx_hard_regno_nregs): New function.
15240 (TARGET_HARD_REGNO_NREGS): Redefine.
15241 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
15242 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
15243 instead of HARD_REGNO_NREGS.
15244 (s390_hard_regno_nregs): New function.
15245 (s390_hard_regno_mode_ok): Add comment from s390.h.
15246 (TARGET_HARD_REGNO_NREGS): Redefine.
15247 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
15248 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
15249 (sh_hard_regno_nregs): New function.
15250 (sh_pass_in_reg_p): Use it.
15251 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
15252 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15253 (sparc_hard_regno_nregs): New function.
15254 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
15255 * config/spu/spu.c (spu_hard_regno_nregs): New function.
15256 (spu_function_arg_advance): Use it, supplying a valid register number.
15257 (TARGET_HARD_REGNO_NREGS): Redefine.
15258 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
15259 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
15260 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
15261 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
15262 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
15263 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
15264 (CLASS_MAX_NREGS): Remove copy of old documentation.
15265 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
15266 (visium_hard_regno_nregs): New function.
15267 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
15268 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
15269 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15270 xtensa_hard_regno_nregs): New function.
15271 * system.h (HARD_REGNO_NREGS): Poison.
15273 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15275 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
15276 hard_regno_nregs instead of HARD_REGNO_NREGS.
15277 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
15278 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
15279 (c6x_expand_epilogue): Likewise.
15280 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
15281 (frv_read_iacc_argument): Likewise.
15282 * config/sh/sh.c: Include regs.h.
15283 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15284 (regs_used): Likewise.
15285 (output_stack_adjust): Likewise.
15286 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
15287 * expmed.c: Include regs.h.
15288 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15289 * ree.c: Include regs.h.
15290 (combine_reaching_defs): Use hard_regno_nregs instead of
15292 (add_removable_extension): Likewise.
15294 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15296 * regs.h (hard_regno_nregs): Turn into a function.
15297 (end_hard_regno): Update accordingly.
15298 * caller-save.c (setup_save_areas): Likewise.
15299 (save_call_clobbered_regs): Likewise.
15300 (replace_reg_with_saved_mem): Likewise.
15301 (insert_restore): Likewise.
15302 (insert_save): Likewise.
15303 * combine.c (can_change_dest_mode): Likewise.
15304 (move_deaths): Likewise.
15305 (distribute_notes): Likewise.
15306 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
15307 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
15308 (rs6000_split_multireg_move): Likewise.
15309 (rs6000_register_move_cost): Likewise.
15310 (rs6000_memory_move_cost): Likewise.
15311 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
15312 (rs6000_split_multireg_move): Likewise.
15313 (rs6000_register_move_cost): Likewise.
15314 (rs6000_memory_move_cost): Likewise.
15315 * cselib.c (cselib_reset_table): Likewise.
15316 (cselib_lookup_1): Likewise.
15317 * emit-rtl.c (set_mode_and_regno): Likewise.
15318 * function.c (aggregate_value_p): Likewise.
15319 * ira-color.c (setup_profitable_hard_regs): Likewise.
15320 (check_hard_reg_p): Likewise.
15321 (calculate_saved_nregs): Likewise.
15322 (assign_hard_reg): Likewise.
15323 (improve_allocation): Likewise.
15324 (calculate_spill_cost): Likewise.
15325 * ira-emit.c (modify_move_list): Likewise.
15326 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
15327 (ira_hard_reg_in_set_p): Likewise.
15328 * ira.c (setup_reg_mode_hard_regset): Likewise.
15329 (clarify_prohibited_class_mode_regs): Likewise.
15330 (check_allocation): Likewise.
15331 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15332 (lra_setup_reg_renumber): Likewise.
15333 (setup_try_hard_regno_pseudos): Likewise.
15334 (spill_for): Likewise.
15335 (assign_hard_regno): Likewise.
15336 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
15337 * lra-constraints.c (in_class_p): Likewise.
15338 (lra_constraint_offset): Likewise.
15339 (simplify_operand_subreg): Likewise.
15340 (lra_constraints): Likewise.
15341 (split_reg): Likewise.
15342 (split_if_necessary): Likewise.
15343 (invariant_p): Likewise.
15344 (inherit_in_ebb): Likewise.
15345 * lra-lives.c (process_bb_lives): Likewise.
15346 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
15347 (get_hard_regs): Likewise.
15348 (do_remat): Likewise.
15349 * lra-spills.c (assign_spill_hard_regs): Likewise.
15350 * mode-switching.c (create_pre_exit): Likewise.
15351 * postreload.c (reload_combine_recognize_pattern): Likewise.
15352 * recog.c (peep2_find_free_register): Likewise.
15353 * regcprop.c (kill_value_regno): Likewise.
15354 (set_value_regno): Likewise.
15355 (copy_value): Likewise.
15356 (maybe_mode_change): Likewise.
15357 (find_oldest_value_reg): Likewise.
15358 (copyprop_hardreg_forward_1): Likewise.
15359 * regrename.c (check_new_reg_p): Likewise.
15360 (regrename_do_replace): Likewise.
15361 * reload.c (push_reload): Likewise.
15362 (combine_reloads): Likewise.
15363 (find_dummy_reload): Likewise.
15364 (operands_match_p): Likewise.
15365 (find_reloads): Likewise.
15366 (find_equiv_reg): Likewise.
15367 (reload_adjust_reg_for_mode): Likewise.
15368 * reload1.c (count_pseudo): Likewise.
15369 (count_spilled_pseudo): Likewise.
15370 (find_reg): Likewise.
15371 (clear_reload_reg_in_use): Likewise.
15372 (free_for_value_p): Likewise.
15373 (allocate_reload_reg): Likewise.
15374 (choose_reload_regs): Likewise.
15375 (reload_adjust_reg_for_temp): Likewise.
15376 (emit_reload_insns): Likewise.
15377 (delete_output_reload): Likewise.
15378 * rtlanal.c (subreg_get_info): Likewise.
15379 * sched-deps.c (sched_analyze_reg): Likewise.
15380 * sel-sched.c (init_regs_for_mode): Likewise.
15381 (mark_unavailable_hard_regs): Likewise.
15382 (choose_best_reg_1): Likewise.
15383 (verify_target_availability): Likewise.
15384 * valtrack.c (dead_debug_insert_temp): Likewise.
15385 * var-tracking.c (track_loc_p): Likewise.
15386 (emit_note_insn_var_location): Likewise.
15387 * varasm.c (make_decl_rtl): Likewise.
15388 * reginfo.c (choose_hard_reg_mode): Likewise.
15389 (init_reg_modes_target): Refer directly to
15390 this_target_regs->x_hard_regno_nregs.
15392 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15394 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
15395 instead of hard_regno_nregs.
15397 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15399 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
15400 end_hard_regno instead of hard_regno_nregs.
15401 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
15402 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
15403 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
15404 * ira-color.c (improve_allocation): Likewise.
15405 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15406 * lra-lives.c (mark_regno_live): Likewise.
15407 (mark_regno_dead): Likewise.
15408 * lra-remat.c (operand_to_remat): Likewise.
15409 * lra.c (collect_non_operand_hard_regs): Likewise.
15410 * postreload.c (reload_combine_note_store): Likewise.
15411 (move2add_valid_value_p): Likewise.
15412 * reload.c (regno_clobbered_p): Likewise.
15414 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15416 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
15418 * config/v850/v850.c (v850_reorg): Likewise.
15419 * reload.c (refers_to_regno_for_reload_p): Likewise.
15420 (find_equiv_reg): Likewise.
15421 * reload1.c (reload_reg_reaches_end_p): Likewise.
15423 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15425 * caller-save.c (add_used_regs): Use REG_NREGS instead of
15427 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
15428 * config/arm/arm.c (output_move_neon): Likewise.
15429 (arm_attr_length_move_neon): Likewise.
15430 (neon_split_vcombine): Likewise.
15431 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
15432 (c6x_mark_reg_written): Likewise.
15433 (c6x_dwarf_register_span): Likewise.
15434 * config/i386/i386.c (ix86_save_reg): Likewise.
15435 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
15436 (rws_access_reg): Likewise.
15437 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
15438 * mode-switching.c (create_pre_exit): Likewise.
15439 * ree.c (combine_reaching_defs): Likewise.
15440 (add_removable_extension): Likewise.
15441 * regcprop.c (find_oldest_value_reg): Likewise.
15442 (copyprop_hardreg_forward_1): Likewise.
15443 * reload.c (reload_inner_reg_of_subreg): Likewise.
15444 (push_reload): Likewise.
15445 (combine_reloads): Likewise.
15446 (find_dummy_reload): Likewise.
15447 (reload_adjust_reg_for_mode): Likewise.
15448 * reload1.c (find_reload_regs): Likewise.
15449 (forget_old_reloads_1): Likewise.
15450 (reload_reg_free_for_value_p): Likewise.
15451 (reload_adjust_reg_for_temp): Likewise.
15452 (emit_reload_insns): Likewise.
15453 (delete_output_reload): Likewise.
15454 * sel-sched.c (choose_best_reg_1): Likewise.
15455 (choose_best_pseudo_reg): Likewise.
15457 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15458 Alan Hayward <alan.hayward@arm.com>
15459 David Sherwood <david.sherwood@arm.com>
15461 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
15462 * target.def (slow_unaligned_access): New hook.
15463 * targhooks.h (default_slow_unaligned_access): Declare.
15464 * targhooks.c (default_slow_unaligned_access): New function.
15465 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
15466 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
15467 * doc/tm.texi: Regenerate.
15468 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
15469 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
15470 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
15472 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
15473 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
15475 (rs6000_slow_unaligned_access): New function.
15476 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15477 (expand_block_compare): Likewise.
15478 (expand_strn_compare): Likewise.
15479 (rs6000_rtx_costs): Likewise.
15480 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
15481 (riscv_slow_unaligned_access): Likewise.
15482 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
15483 (riscv_slow_unaligned_access_p): ...this and make static.
15484 (riscv_option_override): Update accordingly.
15485 (riscv_slow_unaligned_access): New function.
15486 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15487 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
15488 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15489 (rs6000_slow_unaligned_access): New function.
15490 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15491 (rs6000_rtx_costs): Likewise.
15492 * config/rs6000/rs6000-string.c (expand_block_compare)
15493 (expand_strn_compare): Use targetm.slow_unaligned_access instead
15494 of SLOW_UNALIGNED_ACCESS.
15495 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
15496 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
15497 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
15498 of SLOW_UNALIGNED_ACCESS.
15499 * expmed.c (simple_mem_bitfield_p): Likewise.
15500 * expr.c (alignment_for_piecewise_move): Likewise.
15501 (emit_group_load_1): Likewise.
15502 (emit_group_store): Likewise.
15503 (copy_blkmode_from_reg): Likewise.
15504 (emit_push_insn): Likewise.
15505 (expand_assignment): Likewise.
15506 (store_field): Likewise.
15507 (expand_expr_real_1): Likewise.
15508 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
15509 * lra-constraints.c (simplify_operand_subreg): Likewise.
15510 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15511 * gimple-ssa-store-merging.c: Likewise in block comment at start
15513 * tree-ssa-strlen.c: Include target.h.
15514 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
15515 of SLOW_UNALIGNED_ACCESS.
15516 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
15518 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15520 PR rtl-optimization/82185
15521 * expmed.c (emit_store_flag_int): Only test tem if it has been
15524 2017-09-12 Richard Biener <rguenther@suse.de>
15526 PR middle-end/82149
15527 * match.pd ((FTYPE) N CMP CST): Fix typo.
15529 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
15531 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
15533 (mips_near_type_p): Add 'short_call' attribute as a synonym
15535 * doc/extend.texi (short_call): Document new function attribute.
15537 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15540 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
15541 assertion check that in the condition.
15542 (get_atomic_generic_size): Likewise. Before testing if parameter
15543 has pointer type, if it has array type, call for C++
15544 default_conversion to perform array-to-pointer conversion.
15546 2017-09-12 Richard Biener <rguenther@suse.de>
15548 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
15549 for operations we cannot scalarize.
15551 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
15553 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
15554 vectors heap vectors. Clean up comments.
15555 Make visited_bbs a reference.
15556 (profitable_jump_thread_path): Make GC
15557 vectors heap vectors. Clean up comments.
15559 (convert_and_register_jump_thread_path): Make GC vectors heap
15561 (check_subpath_and_update_thread_path): Same. Clean up comments.
15562 Make visited_bbs a reference.
15563 (handle_phi): Abstract common code to to
15564 register_jump_thread_path_if_profitable.
15565 Rename VAR_BB to DEF_BB.
15567 Make GC vectors heap vectors.
15568 Make visited_bbs a reference.
15569 (handle_assignment): Same.
15570 (register_jump_thread_path_if_profitable): New.
15571 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
15573 Make GC vectors heap vectors. Clean up comments.
15574 Make visited_bbs a reference.
15575 (find_jump_threads_backwards): Make visited_bbs live in the stack.
15576 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
15579 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
15582 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
15583 words of E_DImode object are reachable by xtensa_uimm8x4 access.
15585 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
15587 Revert r251800 and r251799.
15589 2017-09-11 Martin Jambor <mjambor@suse.cz>
15592 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
15593 arguments in advance.
15594 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
15595 use it to find predecessor edges.
15596 (naive_outof_ssa): Collect vector of predecessors.
15598 2017-09-08 Jason Merrill <jason@redhat.com>
15600 PR c++/70029 - ICE with ref-qualifier and -flto
15601 * langhooks.h (struct lang_hooks_for_types): Add
15602 copy_lang_qualifiers.
15603 * attribs.c (build_type_attribute_qual_variant): Use it.
15604 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
15606 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
15607 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
15609 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
15612 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
15613 (*mulsi3_sp64): New instruction.
15614 (mulsi3): New expander.
15616 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
15618 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
15620 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15622 * sancov.c: Include memmodel.h.
15624 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
15627 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
15630 2017-09-07 Carl Love <cel@us.ibm.com>
15632 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
15633 the sldi instruction.
15635 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
15637 * sancov.c: Include tm_p.h.
15639 2017-09-07 Jakub Jelinek <jakub@redhat.com>
15642 * output.h (switch_to_other_text_partition): New declaration.
15643 * varasm.c (switch_to_other_text_partition): New function.
15644 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
15645 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
15646 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
15647 to the other text partition before emitting LCL label and switch back
15648 after emitting the word after it.
15650 2017-09-07 Richard Biener <rguenther@suse.de>
15652 * passes.def (pass_split_crit_edges): Remove instance before PRE.
15653 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
15654 critical edges here, after loop init.
15655 (pass_data_pre): Remove PROP_no_crit_edges flags.
15656 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
15657 for valueization of call args to avoid leaking VN_TOP.
15658 (visit_use): Assert we do not visit default defs.
15659 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
15660 Use error_mark_node to more easily detect leaking VN_TOP.
15661 All default-defs are varying, not VN_TOP. Mark them visited.
15662 (run_scc_vn): Make code match comment.
15664 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15666 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
15667 OPTION_MASK_FLOAT128_KEYWORD.
15668 (POWERPC_MASKS): Likewise.
15669 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
15670 support for the -mfloat128-type option, and make -mfloat128
15671 default on PowerPC Linux systems. Define or undefine
15672 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
15673 Define __float128 to be __ieee128 if IEEE 128-bit support is
15674 enabled, or undefine it.
15675 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
15676 Delete defining __FLOAT128_TYPE__.
15677 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
15678 -mfloat128-type option and make -mfloat128 default on PowerPC
15680 (TARGET_FLOAT128_TYPE): Likewise.
15681 (-mfloat128-type): Likewise.
15682 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15683 Delete the -mfloat128-type option and make -mfloat128 default on
15684 PowerPC Linux systems. Always use __ieee128 to be the keyword for
15685 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
15686 128-bit floating point is enabled. Change tests from using
15687 -mfloat128-type to -mfloat128.
15688 (rs6000_mangle_type): Use the correct mangling for the __float128
15689 type even if normal long double is restricted to 64-bits.
15690 (floatn_mode): Enable the _Float128 type by default on VSX Linux
15692 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
15693 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
15694 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
15696 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
15697 documentation for -mfloat128.
15699 2017-09-06 Olivier Hainque <hainque@adacore.com>
15701 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
15703 2017-09-06 Wish Wu <wishwu007@gmail.com>
15704 Jakub Jelinek <jakub@redhat.com>
15706 * asan.c (initialize_sanitizer_builtins): Add
15707 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
15708 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
15709 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
15710 BT_FN_VOID_UINT64_PTR variables.
15711 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
15712 (BT_FN_VOID_UINT16_UINT16): Likewise.
15713 (BT_FN_VOID_UINT32_UINT32): Likewise.
15714 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
15715 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
15716 (BT_FN_VOID_UINT64_PTR): Likewise.
15717 * common.opt (flag_sanitize_coverage): New variable.
15718 (fsanitize-coverage=trace-pc): Remove.
15719 (fsanitize-coverage=): Add.
15720 * flag-types.h (enum sanitize_coverage_code): New enum.
15721 * fold-const.c (fold_range_test): Disable non-short-circuit
15722 optimization if flag_sanitize_coverage.
15723 (fold_truth_andor): Likewise.
15724 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15725 * opts.c (COVERAGE_SANITIZER_OPT): Define.
15726 (coverage_sanitizer_opts): New array.
15727 (get_closest_sanitizer_option): Add OPTS argument, handle also
15728 OPT_fsanitize_coverage_.
15729 (parse_sanitizer_options): Adjusted to also handle
15730 OPT_fsanitize_coverage_.
15731 (common_handle_option): Add OPT_fsanitize_coverage_.
15732 * sancov.c (instrument_comparison, instrument_switch): New function.
15733 (sancov_pass): Add trace-cmp support.
15734 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
15735 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
15736 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
15737 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
15738 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
15739 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
15740 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
15741 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
15742 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
15743 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
15745 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15747 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
15748 error. Only quit immediately if parsing is complete.
15749 (BEGIN): Initialize fatal_err and parse_done.
15750 (begin fpu, end fpu): Check number of arguments.
15751 (begin arch, end arch): Likewise.
15752 (begin cpu, end cpu): Likewise.
15753 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
15754 (optalias): Likewise.
15756 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15758 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
15759 * config/arm/arm-isa.h: Delete. Move definitions to ...
15760 * arm-cpus.in: ... here. Use new feature and fgroup values.
15761 * config/arm/arm.c (arm_option_override): Use lower case for feature
15763 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
15764 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
15765 * config/arm/parsecpu.awk (END): Add new command 'isa'.
15767 (print_isa_bits_for): New function.
15768 (gen_isa): New function.
15769 (gen_comm_data): Use print_isa_bits_for.
15770 (define feature): New keyword.
15771 (define fgroup): New keyword.
15772 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
15773 (arm-isa.h): Add rule to generate file.
15774 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
15775 case for feature bit names.
15777 2017-09-06 Richard Biener <rguenther@suse.de>
15779 * tree-ssa-pre.c (NECESSARY): Remove.
15780 (create_expression_by_pieces): Do not touch pass-local flags.
15781 (insert_into_preds_of_block): Likewise.
15782 (do_pre_regular_insertion): Likewise.
15783 (eliminate_insert): Likewise.
15784 (eliminate_dom_walker::before_dom_children): Likewise.
15785 (fini_eliminate): Do not look at inserted_exprs.
15786 (mark_operand_necessary): Remove.
15787 (remove_dead_inserted_code): Replace with simple work-list
15788 algorithm based on inserted_exprs and SSA uses.
15789 (pass_pre::execute): Re-order fini_eliminate and
15790 remove_dead_inserted_code.
15792 2017-09-06 Olivier Hainque <hainque@adacore.com>
15794 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
15795 for VxWorks 7. Adjust surrounding comments.
15797 2017-09-06 Richard Biener <rguenther@suse.de>
15799 * gimple-ssa-strength-reduction.c
15800 (find_candidates_dom_walker::before_dom_children): Also allow
15803 2017-09-06 Richard Biener <rguenther@suse.de>
15805 PR tree-optimization/82108
15806 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
15807 for gap in the non-permutation SLP case.
15809 2017-09-06 Martin Jambor <mjambor@suse.cz>
15811 PR tree-optimization/82078
15812 * tree-sra.c (sort_and_splice_var_accesses): Move call to
15813 add_access_to_work_queue...
15814 (build_accesses_from_assign): ...here.
15815 (propagate_all_subaccesses): Make sure racc is the group
15816 representative, if there is one.
15818 2017-09-06 Jakub Jelinek <jakub@redhat.com>
15820 PR middle-end/82095
15821 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
15824 2017-09-06 Richard Biener <rguenther@suse.de>
15826 * gimple-ssa-strength-reduction.c
15827 (find_candidates_dom_walker::before_doom_children): Use a
15828 type and not a mode check.
15830 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
15833 * config/arm/predicates.md (arm_general_adddi_operand): Create new
15835 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
15837 2017-09-05 Jeff Law <law@redhat.com>
15839 PR tree-optimization/64910
15840 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
15841 cases where we have 3 or more operands.
15843 2017-09-05 Jakub Jelinek <jakub@redhat.com>
15845 PR middle-end/81768
15846 * omp-low.c (lower_omp_for): Recompute tree invariant if
15847 gimple_omp_for_initial/final is ADDR_EXPR.
15849 PR middle-end/81768
15850 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
15851 into gimple val before gimplification fo the COND_EXPR.
15853 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
15855 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
15856 REGION_COPY argument.
15857 (thread_through_all_blocks): Remove unused argument to
15858 duplicate_thread_path.
15860 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15861 Alan Hayward <alan.hayward@arm.com>
15862 David Sherwood <david.sherwood@arm.com>
15864 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
15865 Take a scalar_mode rather than a machine_mode.
15866 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15867 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
15868 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15869 (aarch64_gen_adjusted_ldpstp): Likewise.
15870 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
15872 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15873 Alan Hayward <alan.hayward@arm.com>
15874 David Sherwood <david.sherwood@arm.com>
15876 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
15877 Take a scalar_int_mode instead of a machine_mode.
15878 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15879 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15880 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15881 (aarch64_simd_attr_length_rglist): Delete.
15882 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
15883 a scalar_int_mode instead of a machine_mode.
15884 (aarch64_add_offset): Likewise.
15885 (aarch64_internal_mov_immediate): Likewise
15886 (aarch64_add_constant_internal): Likewise.
15887 (aarch64_add_constant): Likewise.
15888 (aarch64_movw_imm): Likewise.
15889 (aarch64_rtx_arith_op_extract_p): Likewise.
15890 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15891 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15892 Remove assert that the mode isn't a vector.
15893 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15894 (aarch64_expand_mov_immediate): Update calls after above changes.
15895 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
15896 (aarch64_and_bitmask_imm): Check for scalar integer modes.
15897 (aarch64_move_imm): Likewise.
15898 (aarch64_can_const_movi_rtx_p): Likewise.
15899 (aarch64_strip_extend): Likewise.
15900 (aarch64_extr_rtx_p): Likewise.
15901 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
15902 a CONST_INT when the mode parameter is VOIDmode.
15903 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
15905 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15907 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
15908 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15909 (bitwise_type_for_mode): Update accordingly.
15911 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15913 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
15914 * stor-layout.c (mode_for_size_tree): Likewise.
15915 (mode_for_array): Update accordingly.
15916 (layout_decl): Likewise.
15917 (compute_record_mode): Likewise. Only set the mode once.
15919 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15921 * target.def (get_mask_mode): Change return type to opt_mode.
15923 * doc/tm.texi: Regenerate.
15924 * targhooks.h (default_get_mask_mode): Return an opt_mode.
15925 * targhooks.c (default_get_mask_mode): Likewise.
15926 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15927 * optabs-query.c (can_vec_mask_load_store_p): Update use of
15928 targetm.get_mask_mode.
15929 * tree.c (build_truth_vector_type): Likewise.
15931 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15933 * machmode.h (mode_for_vector): Return an opt_mode.
15934 * stor-layout.c (mode_for_vector): Likewise.
15935 (mode_for_int_vector): Update accordingly.
15936 (layout_type): Likewise.
15937 * config/i386/i386.c (emit_memmov): Likewise.
15938 (ix86_expand_set_or_movmem): Likewise.
15939 (ix86_expand_vector_init): Likewise.
15940 (ix86_get_mask_mode): Likewise.
15941 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
15943 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
15944 * expmed.c (extract_bit_field_1): Likewise.
15945 * expr.c (expand_expr_real_2): Likewise.
15946 * optabs-query.c (can_vec_perm_p): Likewise.
15947 (can_vec_mask_load_store_p): Likewise.
15948 * optabs.c (expand_vec_perm): Likewise.
15949 * targhooks.c (default_get_mask_mode): Likewise.
15950 * tree-vect-stmts.c (vectorizable_store): Likewise.
15951 (vectorizable_load): Likewise.
15952 (get_vectype_for_scalar_type_and_size): Likewise.
15954 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15956 * machmode.h (mode_for_int_vector): New function.
15957 * stor-layout.c (mode_for_int_vector): Likewise.
15958 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
15959 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
15960 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15961 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15962 (s390_expand_vcond): Likewise.
15964 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15966 * machmode.h (opt_machine_mode): New type.
15967 (opt_mode<T>): Allow construction from anything that can be
15969 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
15970 (mode_for_size): Return an opt_machine_mode.
15971 * stor-layout.c (mode_for_size): Likewise.
15972 (mode_for_size_tree): Update call accordingly.
15973 (bitwise_mode_for_mode): Likewise.
15974 (make_fract_type): Likewise.
15975 (make_accum_type): Likewise.
15976 * caller-save.c (replace_reg_with_saved_mem): Update call
15978 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15979 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15980 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15981 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15982 * expmed.c (extract_bit_field_1): Likewise.
15983 * reload.c (get_secondary_mem): Likewise.
15984 * varasm.c (assemble_integer): Likewise.
15985 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
15988 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15990 * machmode.h (decimal_float_mode_for_size): New function.
15991 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
15992 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
15993 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
15994 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
15995 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
15996 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
15998 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16000 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
16001 (get_builtin_sync_mode): Likewise.
16002 (expand_ifn_atomic_compare_exchange): Likewise.
16003 (expand_builtin_atomic_clear): Likewise.
16004 (expand_builtin_atomic_test_and_set): Likewise.
16005 (fold_builtin_atomic_always_lock_free): Likewise.
16006 * calls.c (compute_argument_addresses): Likewise.
16007 (emit_library_call_value_1): Likewise.
16008 (store_one_arg): Likewise.
16009 * combine.c (combine_instructions): Likewise.
16010 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
16011 * config/arm/arm.c (arm_function_value): Likewise.
16012 (aapcs_allocate_return_reg): Likewise.
16013 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
16014 * config/i386/i386.c (construct_container): Likewise.
16015 (ix86_gimplify_va_arg): Likewise.
16016 (ix86_expand_sse_cmp): Likewise.
16017 (emit_memmov): Likewise.
16018 (emit_memset): Likewise.
16019 (expand_small_movmem_or_setmem): Likewise.
16020 (ix86_expand_pextr): Likewise.
16021 (ix86_expand_pinsr): Likewise.
16022 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
16023 * config/microblaze/microblaze.c (microblaze_block_move_straight):
16025 * config/mips/mips.c (mips_function_value_1) Likewise.
16026 (mips_block_move_straight): Likewise.
16027 (mips_expand_ins_as_unaligned_store): Likewise.
16028 * config/powerpcspe/powerpcspe.c
16029 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16030 (rs6000_darwin64_record_arg_flush): Likewise.
16031 * config/rs6000/rs6000.c
16032 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16033 (rs6000_darwin64_record_arg_flush): Likewise.
16034 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
16035 (sparc_function_value_1): Likewise.
16036 * config/spu/spu.c (adjust_operand): Likewise.
16037 (spu_emit_branch_or_set): Likewise.
16038 (arith_immediate_p): Likewise.
16039 * emit-rtl.c (gen_lowpart_common): Likewise.
16040 * expr.c (expand_expr_real_1): Likewise.
16041 * function.c (assign_parm_setup_block): Likewise.
16042 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
16043 * reload1.c (alter_reg): Likewise.
16044 * stor-layout.c (mode_for_vector): Likewise.
16045 (layout_type): Likewise.
16047 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16049 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
16050 (spu_convert_move): Likewise.
16051 * lower-subreg.c (resolve_simple_move): Likewise.
16053 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16056 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
16057 define_insn to a define_expand.
16058 (altivec_vsum2sws_direct): New define_insn.
16059 (altivec_vsumsws): Convert from a define_insn to a define_expand.
16061 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16063 * config/arm/arm.c (arm_option_params_internal): Improve setting of
16066 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
16071 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
16072 realign stack if stack alignment needed is less than incoming
16075 2017-09-05 Marek Polacek <polacek@redhat.com>
16078 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
16081 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16083 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
16085 2017-09-05 Richard Biener <rguenther@suse.de>
16087 PR tree-optimization/82084
16088 * fold-const.c (can_native_encode_string_p): Handle wide characters.
16090 2017-09-05 Richard Biener <rguenther@suse.de>
16092 PR tree-optimization/82102
16093 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
16095 2017-09-05 Martin Liska <mliska@suse.cz>
16097 PR tree-optimization/82032
16098 * tree-cfg.c (generate_range_test): New function.
16099 * tree-cfg.h (generate_range_test): Declared here.
16100 * tree-cfgcleanup.c (convert_single_case_switch): New function.
16101 (cleanup_control_expr_graph): Use it.
16102 * tree-switch-conversion.c (try_switch_expansion): Remove
16104 (emit_case_nodes): Use generate_range_test.
16106 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16109 * config/i386/i386.md (*<btsc><mode>_mask): Add
16110 TARGET_USE_BT to insn constraint.
16111 (*btr<mode>_mask): Ditto.
16113 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
16115 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
16116 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
16118 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
16121 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
16122 TARGET_NEON and TARGET_IWMMXT.
16123 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
16124 TARGET_NEON and TARGET_IWMMXT.
16125 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
16127 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16129 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
16130 (ix86_rewrite_tls_address): Ditto.
16131 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
16132 (ix86_rewrite_tls_address_1): Ditto.
16133 (ix86_rewrite_tls_address): Ditto.
16134 * config/i386/predicates.md (tls_address_pattern): New predicate.
16135 * config/i386/i386.md (TLS address splitter): New splitter.
16137 2017-09-04 Richard Biener <rguenther@suse.de>
16139 PR tree-optimization/82084
16140 * fold-const.h (can_native_encode_string_p): Declare.
16141 * fold-const.c (can_native_encode_string_p): Factor out from ...
16142 (native_encode_string): ... here.
16143 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
16144 vectorizing stores from constants we later cannot handle.
16146 2017-09-04 Marek Polacek <polacek@redhat.com>
16149 * doc/invoke.texi: Update -Wtautological-compare documentation.
16151 2017-09-04 Jeff Law <law@redhat.com>
16153 PR tree-optimization/64910
16154 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
16155 swap the first and last operand if the last is a constant.
16157 2017-09-04 Marek Polacek <polacek@redhat.com>
16160 * convert.c (do_narrow): When sanitizing signed integer overflows,
16161 bail out for signed types.
16162 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
16164 2017-09-04 Richard Biener <rguenther@suse.de>
16166 PR tree-optimization/82060
16167 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16168 Move devirtualization after stmt folding and before EH/AB/noreturn
16169 cleanup to get the stmt refs canonicalized. Use a bool instead
16170 of gimple_modified_p since that doesn't work for NOPs. Schedule
16171 NOPs generated by folding for removal.
16173 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16174 Alan Hayward <alan.hayward@arm.com>
16175 David Sherwood <david.sherwood@arm.com>
16177 * coretypes.h (pad_direction): New enum.
16178 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
16179 (FUNCTION_ARG_PADDING): Likewise.
16180 * target.def (function_arg_padding): New hook.
16181 * targhooks.h (default_function_arg_padding): Declare.
16182 * targhooks.c (default_function_arg_padding): New function.
16183 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
16184 (TARGET_FUNCTION_ARG_PADDING): ...this.
16185 * doc/tm.texi: Regenerate.
16186 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
16187 instead of direction.
16188 (compute_argument_addresses): Likewise.
16189 (load_register_parameters): Likewise.
16190 (emit_library_call_value_1): Likewise.
16191 (store_one_arg): Use targetm.calls.function_arg_padding instead
16192 of FUNCTION_ARG_PADDING.
16193 (must_pass_in_stack_var_size_or_pad): Likewise.
16194 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
16195 (emit_group_store): Likewise.
16196 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
16197 instead of FUNCTION_ARG_PADDING.
16198 (emit_push_insn): Likewise, and propagate enum change throughout
16200 * function.h (direction): Delete.
16201 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
16203 * function.c (assign_parm_find_stack_rtl): Likewise.
16204 (assign_parm_setup_block_p): Likewise.
16205 (assign_parm_setup_block): Likewise.
16206 (gimplify_parameters): Likewise.
16207 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
16208 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
16210 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
16211 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16212 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
16213 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
16214 (aarch64_function_arg_padding): ...this new function.
16215 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
16216 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16217 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
16218 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16219 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
16220 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16221 (arm_pad_arg_upward): Replace with...
16222 (arm_function_arg_padding): ...this new function.
16223 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
16225 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
16226 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
16227 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16228 (ia64_hpux_function_arg_padding): Replace with...
16229 (ia64_function_arg_padding): ...this new function. Use pad_direction
16230 instead of direction. Check for TARGET_HPUX.
16231 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
16232 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16233 (iq2000_function_arg_padding): New function.
16234 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
16235 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
16236 (mips_function_arg_padding): ...this new function.
16237 (mips_pad_reg_upward): Update accordingly.
16238 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16239 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
16240 targetm.calls.function_arg_padding.
16241 (FUNCTION_ARG_PADDING): Delete.
16242 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16243 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
16244 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16245 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
16246 (nios2_block_reg_padding): Return pad_direction instead of direction.
16247 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
16248 instead of direction.
16249 (nios2_function_arg_padding): Likewise. Make static.
16250 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16251 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
16252 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
16253 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
16254 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16255 (pa_function_arg_padding): Make static. Return pad_direction instead
16257 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
16258 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16259 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
16260 instead of direction. Use targetm.calls.function_arg_padding.
16261 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
16262 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16263 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
16264 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
16265 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
16267 (function_arg_padding): Rename to...
16268 (rs6000_function_arg_padding): ...this. Make static. Return
16269 pad_direction instead of direction.
16270 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16271 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
16272 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16273 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
16274 instead of direction. Use targetm.calls.function_arg_padding.
16275 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
16276 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16277 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
16278 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
16279 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16280 (function_arg_padding): Rename to...
16281 (rs6000_function_arg_padding): ...this. Make static. Return
16282 pad_direction instead of direction.
16283 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16284 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
16285 * config/s390/s390.c (s390_function_arg_padding): New function.
16286 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16287 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
16288 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
16289 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16290 (function_arg_padding): Rename to...
16291 (sparc_function_arg_padding): ...this. Make static. Return
16292 pad_direction instead of direction.
16293 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
16294 * config/spu/spu.c (spu_function_arg_padding): New function.
16295 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16296 * system.h (FUNCTION_ARG_PADDING): Poison.
16298 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16299 Alan Hayward <alan.hayward@arm.com>
16300 David Sherwood <david.sherwood@arm.com>
16302 * target.def (modes_tieable_p): New hook.
16303 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
16304 (TARGET_MODES_TIEABLE_P): ...this.
16305 * doc/tm.texi.in: Regenerate.
16306 * hooks.h (hook_bool_mode_mode_true): Declare.
16307 * hooks.c (hook_bool_mode_mode_true): New function.
16308 * combine.c (subst): Use targetm.modes_tieable_p instead of
16310 * dse.c (find_shift_sequence): Likewise.
16311 * expmed.c (extract_low_bits): Likewise.
16312 * lower-subreg.c: Include target.h.
16313 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
16315 * rtlanal.c (rtx_cost): Likewise.
16316 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
16317 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
16318 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
16319 (TARGET_MODES_TIEABLE_P): Redefine.
16320 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
16321 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
16322 (TARGET_MODES_TIEABLE_P): Redefine.
16323 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
16324 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
16325 (arc_modes_tieable_p): New function.
16326 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
16327 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
16328 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
16329 (arm_modes_tieable_p): Make static.
16330 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
16331 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
16332 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
16333 (TARGET_MODES_TIEABLE_P): Redefine.
16334 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
16335 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
16336 (TARGET_MODES_TIEABLE_P): Redefine.
16337 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
16338 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
16339 (cr16_modes_tieable_p): New function.
16340 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
16341 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
16342 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
16343 (TRULY_NOOP_TRUNCATION): Update comment.
16344 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
16345 (TRULY_NOOP_TRUNCATION): Update comment.
16346 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
16347 (frv_modes_tieable_p): New function.
16348 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
16349 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
16350 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
16351 (TARGET_MODES_TIEABLE_P): Redefine.
16352 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
16353 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
16354 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
16355 (TARGET_MODES_TIEABLE_P): Redefine.
16356 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
16357 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
16358 (ia64_modes_tieable_p): New function.
16359 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
16360 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
16361 (iq2000_modes_tieable_p): New function.
16362 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
16363 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
16364 (lm32_modes_tieable_p): New function.
16365 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
16366 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
16367 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
16368 (TARGET_MODES_TIEABLE_P): Redefine.
16369 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
16370 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
16371 (m32r_modes_tieable_p): New function.
16372 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
16373 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
16374 (m68k_modes_tieable_p): New function.
16375 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
16376 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
16377 (mcore_modes_tieable_p): New function.
16378 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
16379 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
16381 (TARGET_MODES_TIEABLE_P): Redefine.
16382 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
16383 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
16384 * config/mips/mips.c (mips_modes_tieable_p): Make static.
16385 (TARGET_MODES_TIEABLE_P): Redefine.
16386 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
16387 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
16388 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
16389 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
16390 (mn10300_modes_tieable_p): ...this and make static.
16391 (TARGET_MODES_TIEABLE_P): Redefine.
16392 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
16393 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
16394 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
16395 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
16396 (msp430_modes_tieable_p): Make static.
16397 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
16398 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
16399 (TARGET_MODES_TIEABLE_P): Redefine.
16400 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
16401 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
16402 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
16403 (TARGET_MODES_TIEABLE_P): Redefine.
16404 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
16405 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
16406 * config/pa/pa.c (pa_modes_tieable_p): Make static.
16407 (TARGET_MODES_TIEABLE_P): Redefine.
16408 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
16409 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
16410 (pdp11_modes_tieable_p): New function.
16411 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
16412 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
16413 (rs6000_modes_tieable_p): New function.
16414 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16415 * config/powerpcspe/powerpcspe.md: Update comment.
16416 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
16417 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
16418 (TARGET_MODES_TIEABLE_P): Redefine.
16419 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
16420 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
16421 (rl78_modes_tieable_p): New function.
16422 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
16423 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
16424 (rs6000_modes_tieable_p): New function.
16425 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16426 * config/rs6000/rs6000.md: Update comment.
16427 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
16428 * config/rx/rx.c (rx_modes_tieable_p): New function.
16429 (TARGET_MODES_TIEABLE_P): Redefine.
16430 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
16431 * config/s390/s390.c (s390_modes_tieable_p): New function.
16432 (TARGET_MODES_TIEABLE_P): Redefine.
16433 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
16434 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
16435 (sh_modes_tieable_p): New function.
16436 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
16437 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
16438 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
16439 (sparc_modes_tieable_p): Make static.
16440 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
16441 * config/spu/spu.c (spu_modes_tieable_p): New function.
16442 (TARGET_MODES_TIEABLE_P): Redefine.
16443 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
16444 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
16445 (TARGET_MODES_TIEABLE_P): Redefine.
16446 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
16447 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
16448 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
16449 * config/v850/v850.c (v850_modes_tieable_p): New function.
16450 (TARGET_MODES_TIEABLE_P): Redefine.
16451 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
16452 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
16453 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
16454 (visium_modes_tieable_p): New function.
16455 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
16456 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
16457 (xtensa_modes_tieable_p): New function.
16458 * system.h (MODES_TIEABLE_P): Poison.
16460 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16461 Alan Hayward <alan.hayward@arm.com>
16462 David Sherwood <david.sherwood@arm.com>
16464 * target.def (hard_regno_mode_ok): New hook.
16465 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
16466 (TARGET_HARD_REGNO_MODE_OK): ...this.
16467 * doc/tm.texi.in: Regenerate.
16468 * hooks.h (hook_bool_uint_mode_true): Declare.
16469 * hooks.c (hook_bool_uint_mode_true): New function.
16470 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
16471 HARD_REGNO_MODE_OK.
16472 * genpreds.c (write_insn_preds_c): Add an include of target.h.
16473 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
16474 instead of HARD_REGNO_MODE_OK.
16475 * caller-save.c: Include target.h.
16476 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
16477 HARD_REGNO_MODE_OK.
16478 * combine.c (can_combine_p): Likewise.
16479 (combinable_i3pat): Likewise.
16480 (can_change_dest_mode): Likewise.
16481 * expr.c (init_expr_target): Likewise.
16482 (convert_move): Likewise.
16483 (convert_modes): Likewise.
16484 * ira.c (setup_prohibited_class_mode_regs): Likewise.
16485 (setup_prohibited_mode_move_regs): Likewise.
16486 * ira.h (target_ira): Likewise.
16487 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16488 * lra-constraints.c (process_alt_operands): Likewise.
16489 (split_reg): Likewise.
16490 * recog.c (peep2_find_free_register): Likewise.
16491 * ree.c (combine_reaching_defs): Likewise.
16492 * regcprop.c (maybe_mode_change): Likewise.
16493 * reginfo.c (init_reg_sets_1): Likewise.
16494 (choose_hard_reg_mode): Likewise.
16495 (simplifiable_subregs): Likewise.
16496 * regrename.c (check_new_reg_p): Likewise.
16497 * reload.c (find_valid_class): Likewise.
16498 (find_valid_class_1): Likewise.
16499 (reload_inner_reg_of_subreg): Likewise.
16500 (push_reload): Likewise.
16501 (combine_reloads): Likewise.
16502 (find_dummy_reload): Likewise.
16503 (find_reloads): Likewise.
16504 * reload1.c (find_reg): Likewise.
16505 (set_reload_reg): Likewise.
16506 (allocate_reload_reg): Likewise.
16507 (choose_reload_regs): Likewise.
16508 (reload_adjust_reg_for_temp): Likewise.
16509 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
16510 (simplify_subreg_regno): Likewise.
16511 * sel-sched.c (init_regs_for_mode): Likewise.
16512 * varasm.c (make_decl_rtl): Likewise.
16513 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
16514 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
16515 HARD_REGNO_MODE_OK.
16516 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
16517 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
16518 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16519 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
16520 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
16521 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16522 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
16523 (arc_mode_class): Delete.
16524 (HARD_REGNO_MODE_OK): Delete.
16525 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16526 (arc_hard_regno_mode_ok): Rename old array to...
16527 (arc_hard_regno_mode_ok_modes): ...this.
16528 (arc_conditional_register_usage): Update accordingly.
16529 (arc_mode_class): Make static.
16530 (arc_hard_regno_mode_ok): New function.
16531 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
16532 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
16533 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16534 (arm_hard_regno_mode_ok): Make static.
16535 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
16536 HARD_REGNO_MODE_OK.
16537 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
16538 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
16539 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
16541 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16542 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
16543 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
16544 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
16545 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
16546 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16547 * config/bfin/predicates.md (valid_reg_operand): Use
16548 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
16549 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
16550 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
16551 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16552 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
16553 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
16554 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16555 (cr16_hard_regno_mode_ok): Make static and return a bool.
16556 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
16557 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16558 (cris_hard_regno_mode_ok): New function.
16559 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
16560 (epiphany_mode_class): Delete.
16561 (HARD_REGNO_MODE_OK): Delete.
16562 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
16563 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16564 (hard_regno_mode_ok): Rename to...
16565 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
16566 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
16567 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
16568 HARD_REGNO_MODE_OK.
16569 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
16570 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
16571 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16572 (frv_hard_regno_mode_ok): Make static and return a bool.
16573 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
16574 HARD_REGNO_MODE_OK.
16575 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
16576 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
16577 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
16578 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
16580 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16581 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
16582 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
16583 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
16585 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16586 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
16587 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16588 (ia64_hard_regno_mode_ok): New function.
16589 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
16590 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16591 (iq2000_hard_regno_mode_ok): New function.
16592 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
16593 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16594 (lm32_hard_regno_mode_ok): New function.
16595 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
16596 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
16597 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
16598 instead of HARD_REGNO_MODE_OK.
16599 (m32c_hard_regno_ok): Rename to...
16600 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
16601 (m32c_cannot_change_mode_class): Update accordingly.
16602 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16603 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
16604 (m32r_mode_class): Delete.
16605 (HARD_REGNO_MODE_OK): Delete.
16606 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16607 (m32r_hard_regno_mode_ok): Rename to...
16608 (m32r_hard_regno_modes): ...this.
16609 (m32r_mode_class): Make static.
16610 (m32r_hard_regno_mode_ok): New function.
16611 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
16612 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
16613 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16614 (m68k_hard_regno_mode_ok): Make static.
16615 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
16616 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16617 (mcore_hard_regno_mode_ok): New function.
16618 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
16619 (HARD_REGNO_MODE_OK): Delete.
16620 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
16622 (microblaze_hard_regno_mode_ok_p): ...this and make static.
16623 (microblaze_hard_regno_mode_ok): New function.
16624 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16625 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
16626 (mips_hard_regno_mode_ok): Delete.
16627 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
16628 (mips_hard_regno_mode_ok_p): ...this and make static.
16629 (mips_hard_regno_mode_ok_p): Rename to...
16630 (mips_hard_regno_mode_ok_uncached): ...this.
16631 (mips_hard_regno_mode_ok): New function.
16632 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
16633 of HARD_REGNO_MODE_OK.
16634 (mips_option_override): Update after above name changes.
16635 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16636 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
16637 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
16638 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
16639 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
16640 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16641 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
16642 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
16643 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
16644 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16645 (msp430_hard_regno_mode_ok): Make static and return a bool.
16646 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
16647 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
16648 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
16650 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16651 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
16652 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
16653 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
16654 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
16655 (PA_HARD_REGNO_MODE_OK): ...this
16656 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
16657 (PA_HARD_REGNO_MODE_OK): ...this.
16658 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16659 (pa_hard_regno_mode_ok): New function.
16660 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
16661 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16662 (pdp11_hard_regno_mode_ok): New function.
16663 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
16664 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
16666 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
16668 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16669 (rs6000_hard_regno_mode_ok): Rename to...
16670 (rs6000_hard_regno_mode_ok_uncached): ...this.
16671 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16672 (rs6000_hard_regno_mode_ok): New function.
16673 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
16674 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
16675 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
16676 (riscv_hard_regno_mode_ok): ...this and make static.
16677 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16678 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
16679 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
16680 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16681 (rl78_hard_regno_mode_ok): Make static and return bool.
16682 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
16683 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
16685 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
16686 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16687 (rs6000_hard_regno_mode_ok): Rename to...
16688 (rs6000_hard_regno_mode_ok_uncached): ...this.
16689 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16690 (rs6000_hard_regno_mode_ok): New function.
16691 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
16692 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
16693 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16694 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
16695 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
16696 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
16697 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16698 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
16699 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
16700 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16701 (sh_hard_regno_mode_ok): Make static.
16702 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
16703 instead of HARD_REGNO_MODE_OK.
16704 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
16705 (sparc_mode_class): Delete.
16706 (HARD_REGNO_MODE_OK): Delete.
16707 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16708 (hard_regno_mode_classes): Make static.
16709 (sparc_mode_class): Likewise.
16710 (sparc_hard_regno_mode_ok): New function.
16711 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
16712 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
16713 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
16715 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16716 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
16717 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
16718 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
16719 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
16720 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16721 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
16722 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
16723 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16724 (visium_hard_regno_mode_ok): New function.
16725 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
16726 instead of HARD_REGNO_MODE_OK.
16727 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
16728 (HARD_REGNO_MODE_OK): Delete.
16729 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
16730 (xtensa_hard_regno_mode_ok_p): ...this and make static.
16731 (xtensa_option_override): Update accordingly.
16732 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16733 (xtensa_hard_regno_mode_ok): New function.
16734 * system.h (HARD_REGNO_MODE_OK): Poison.
16736 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16737 Alan Hayward <alan.hayward@arm.com>
16738 David Sherwood <david.sherwood@arm.com>
16740 * target.def (hard_regno_call_part_clobbered): New hook.
16741 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
16742 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
16743 * doc/tm.texi: Regenerate.
16744 * hooks.h (hook_bool_uint_mode_false): Declare.
16745 * hooks.c (hook_bool_uint_mode_false): New function.
16746 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16747 * cselib.c (cselib_process_insn): Use
16748 targetm.hard_regno_call_part_clobbered instead of
16749 HARD_REGNO_CALL_PART_CLOBBERED.
16750 * ira-conflicts.c (ira_build_conflicts): Likewise.
16751 * ira-costs.c (ira_tune_allocno_costs): Likewise.
16752 * lra-constraints.c (need_for_call_save_p): Likewise.
16753 * lra-lives.c: Include target.h.
16754 (check_pseudos_live_through_calls): Use
16755 targetm.hard_regno_call_part_clobbered instead of
16756 HARD_REGNO_CALL_PART_CLOBBERED.
16757 * regcprop.c: Include target.h.
16758 (copyprop_hardreg_forward_1): Use
16759 targetm.hard_regno_call_part_clobbered instead of
16760 HARD_REGNO_CALL_PART_CLOBBERED.
16761 * reginfo.c (choose_hard_reg_mode): Likewise.
16762 * regrename.c (check_new_reg_p): Likewise.
16763 * reload.c (find_equiv_reg): Likewise.
16764 * reload1.c (emit_reload_insns): Likewise.
16765 * sched-deps.c (deps_analyze_insn): Likewise.
16766 * sel-sched.c (init_regs_for_mode): Likewise.
16767 (mark_unavailable_hard_regs): Likewise.
16768 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
16769 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16770 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
16772 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16773 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16774 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
16776 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
16778 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16779 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16780 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
16782 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16783 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16784 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
16786 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16787 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
16789 * config/powerpcspe/powerpcspe.c
16790 (rs6000_hard_regno_call_part_clobbered): New function.
16791 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16792 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16793 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
16795 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16796 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16797 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
16799 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16800 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16801 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
16803 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16804 Alan Hayward <alan.hayward@arm.com>
16805 David Sherwood <david.sherwood@arm.com>
16807 * rtl.h (subreg_memory_offset): Declare.
16808 * emit-rtl.c (subreg_memory_offset): New function.
16809 * expmed.c (store_bit_field_1): Use it.
16810 * expr.c (undefined_operand_subword_p): Likewise.
16811 * simplify-rtx.c (simplify_subreg): Likewise.
16813 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
16815 PR rtl-optimization/57448
16818 * optabs.c (expand_atomic_load): Place compiler memory barriers if
16819 using atomic_load pattern.
16820 (expand_atomic_store): Likewise.
16822 2017-09-04 Jakub Jelinek <jakub@redhat.com>
16825 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
16826 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
16827 handling. Use replace_call_with_value with NULL instead of
16828 gsi_replace, unlink_stmt_vdef and release_defs.
16830 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
16833 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
16835 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16838 * rtl.h (emit_library_call_value_1): Declare.
16839 (emit_library_call): Replace declaration with a series of overloads.
16840 Remove the parameter count argument.
16841 (emit_library_call_value): Likewise.
16842 * calls.c (emit_library_call_value_1): Make global. Replace varargs
16843 with an "rtx_mode_t *".
16844 (emit_library_call_value): Delete.
16845 (emit_library_call): Likewise.
16846 * asan.c (asan_emit_stack_protection): Update calls accordingly.
16847 (asan_emit_allocas_unpoison): Likewise.
16848 * builtins.c (expand_builtin_powi): Likewise.
16849 (expand_asan_emit_allocas_unpoison): Likewise.
16850 * cfgexpand.c (expand_main_function): Likewise.
16851 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
16852 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
16853 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
16854 * config/arm/arm.c (arm_trampoline_init): Likewise.
16855 (arm_call_tls_get_addr): Likewise.
16856 (arm_expand_divmod_libfunc): Likewise.
16857 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
16858 (smulsi3_highpart): Likewise.
16859 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
16860 (c6x_expand_compare): Likewise.
16861 (c6x_expand_movmem): Likewise.
16862 * config/frv/frv.c (frv_trampoline_init): Likewise.
16863 * config/i386/i386.c (ix86_trampoline_init): Likewise.
16864 (ix86_expand_divmod_libfunc): Likewise.
16865 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
16866 (ia64_expand_compare): Likewise.
16867 (ia64_profile_hook): Likewise.
16868 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
16869 (nonlocal_goto): Likewise.
16870 (restore_stack_nonlocal): Likewise.
16871 * config/m32r/m32r.c (block_move_call): Likewise.
16872 (m32r_trampoline_init): Likewise.
16873 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
16874 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
16875 (m68k_call_m68k_read_tp): Likewise.
16876 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
16877 (microblaze_expand_divide): Likewise.
16878 * config/mips/mips.h (mips_args): Likewise.
16879 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
16880 (MIPS_ICACHE_SYNC): Likewise.
16881 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
16882 (nios2_trampoline_init): Likewise.
16883 * config/pa/pa.c (hppa_tls_call): Likewise.
16884 (pa_trampoline_init): Likewise.
16885 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
16886 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
16887 (expand_strn_compare): Likewise.
16888 (rs6000_generate_compare): Likewise.
16889 (rs6000_expand_float128_convert): Likewise.
16890 (output_profile_hook): Likewise.
16891 (rs6000_trampoline_init): Likewise.
16892 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
16893 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
16894 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
16895 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
16896 (rs6000_generate_compare): Likewise.
16897 (rs6000_expand_float128_convert): Likewise.
16898 (output_profile_hook): Likewise.
16899 (rs6000_trampoline_init): Likewise.
16900 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
16901 * config/sh/sh.c (sh_trampoline_init): Likewise.
16902 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
16903 (sparc_emit_float_lib_cmp): Likewise.
16904 (sparc32_initialize_trampoline): Likewise.
16905 (sparc64_initialize_trampoline): Likewise.
16906 (sparc_profile_hook): Likewise.
16907 * config/spu/spu.c (ea_load_store): Likewise.
16908 * config/spu/spu.md (floatunssidf2): Likewise.
16909 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
16910 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
16911 * config/visium/visium.c (expand_block_move_4): Likewise.
16912 (expand_block_move_2): Likewise.
16913 (expand_block_move_1): Likewise.
16914 (expand_block_set_4): Likewise.
16915 (expand_block_set_2): Likewise.
16916 (expand_block_set_1): Likewise.
16917 (visium_trampoline_init): Likewise.
16918 (visium_profile_hook): Likewise.
16919 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
16920 (xtensa_setup_frame_addresses): Likewise.
16921 (xtensa_trampoline_init): Likewise.
16922 * except.c (sjlj_emit_function_enter): Likewise.
16923 (sjlj_emit_function_exit): Likewise.
16924 * explow.c (allocate_dynamic_stack_space): Likewise.
16925 (probe_stack_range): Likewise.
16926 * expr.c (convert_mode_scalar): Likewise.
16927 * optabs.c (expand_binop): Likewise.
16928 (expand_twoval_binop_libfunc): Likewise.
16929 (expand_unop): Likewise.
16930 (prepare_cmp_insn): Likewise.
16931 (prepare_float_lib_cmp): Likewise.
16932 (expand_float): Likewise.
16933 (expand_fix): Likewise.
16934 (expand_fixed_convert): Likewise.
16935 (maybe_emit_sync_lock_test_and_set): Likewise.
16936 (expand_atomic_compare_and_swap): Likewise.
16937 (expand_mem_thread_fence): Likewise.
16938 (expand_atomic_fetch_op): Likewise.
16940 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
16942 * doc/generic.texi (OpenACC): Adjust URL.
16943 * doc/invoke.texi (C Dialect Options): Ditto.
16945 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
16947 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
16948 predicate for operand 1. Add (m,<S>) constraint.
16949 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
16950 Prevent memory operand 1 with register operand 2.
16952 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
16954 PR rtl-optimization/82024
16955 * combine.c (try_combine): If the combination result is a PARALLEL,
16956 and we only need to retain the SET in there that would be placed
16957 at I2, check that we can place that at I3 instead, before doing so.
16959 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16962 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
16964 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
16965 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
16968 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
16969 Jeff Law <law@redhat.com>
16971 * varasm.c (bss_initializer_p): Do not put constants into .bss
16972 (categorize_decl_for_section): Handle bss_initializer_p returning
16973 false when DECL_INITIAL is NULL.
16975 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16978 * config/s390/s390.c (s390_can_inline_p): New function.
16980 2017-09-01 Jeff Law <law@redhat.com>
16982 PR tree-optimization/82052
16983 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16984 Always initialize the returned slot after a hash table miss
16985 when INSERT is true.
16987 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
16989 * config/s390/s390.md (mem_signal_fence): Remove.
16990 * doc/md.texi (mem_signal_fence): Remove.
16991 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
16993 * target-insns.def (mem_signal_fence): Remove.
16995 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16998 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
17001 * asan.c (create_odr_indicator): Strip name encoding from assembler
17002 name before appending it after __odr_asan_.
17004 2017-09-01 Martin Liska <mliska@suse.cz>
17006 PR tree-optimization/82059
17007 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
17008 frequency only when an edge is redirected.
17010 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17012 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
17013 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
17014 (arc_conditional_register_usage): Remove ARC600 lp_count
17016 (arc_file_start): Emit Tag_ARC_CPU_variation.
17017 (arc_can_use_doloop_p): New conditions to use ZOLs.
17018 (hwloop_fail): New function.
17019 (hwloop_optimize): Likewise.
17020 (hwloop_pattern_reg): Likewise.
17021 (arc_doloop_hooks): New struct, to be used with reorg_loops.
17022 (arc_reorg_loops): New function, calls reorg_loops.
17023 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
17024 (arc600_corereg_hazard): Remove ZOL checking, case handled by
17026 (arc_loop_hazard): Remove function, functionality moved into
17028 (arc_hazard): Remove arc_loop_hazard call.
17029 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
17030 into hwloop_optimize.
17031 (arc_label_align): Remove ZOL handling.
17032 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
17033 * config/arc/arc.md (doloop_begin): Remove pattern.
17034 (doloop_begin_i): Likewise.
17035 (doloop_end_i): Likewise.
17036 (doloop_fallback): Likewise.
17037 (doloop_fallback_m): Likewise.
17038 (doloop_end): Reimplement expand.
17039 (arc_lp): New pattern for LP instruction.
17040 (loop_end): New pattern.
17041 (loop_fail): Likewise.
17042 (decrement_and_branch_until_zero): Likewise.
17043 * config/arc/arc.opt (mlpc-width): New option.
17044 * doc/invoke.texi (mlpc-width): Document option.
17046 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17048 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
17049 (arc_ccfsm_advance): Fix checking for delay slots.
17050 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
17052 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17054 * config/arc/arc.md (movqi_insn): Add stores to save constant long
17056 (movhi_insn): Update store instruction constraint which are saving
17057 6-bit short immediates.
17058 (movsi_insn): Consider also short scaled load operations.
17059 (zero_extendhisi2_i): Use Usd constraint instead of T.
17060 (extendhisi2_i): Add q constraint.
17061 (arc_clzsi2): Add type and length attributes.
17062 (arc_ctzsi2): Likewise.
17063 * config/arc/constraints.md (Usc): Update constraint, the
17064 assembler can parse two relocations for a single instruction.
17066 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17068 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
17069 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
17071 2017-08-31 Olivier Hainque <hainque@adacore.com>
17073 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
17074 match as powerpc-wrs-vxworks*.
17076 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
17078 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
17079 register constraint for by-element operand.
17080 (aarch64_mls_elt_merge<mode>): Likewise.
17082 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17084 * config/arc/arc.c (arc_can_follow_jump): Check for short
17087 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17089 * config.gcc: Use g.opt for arc.
17090 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
17091 functionality moved to ...
17092 (legitimate_scaled_address_p): New function, ...here.
17093 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
17094 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
17095 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
17097 (arc_override_options): Handle G option.
17098 (arc_output_pic_addr_const): Correct function definition.
17099 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
17100 (arc_decl_anon_ns_mem_p): Delete.
17101 (arc_in_small_data_p): Overhaul this function to take into
17102 consideration the value given via G option.
17103 (arc_rewrite_small_data_1): Renamed and corrected old
17104 arc_rewrite_small_data function.
17105 (arc_rewrite_small_data): New function.
17106 (small_data_pattern): Don't use pic_offset_table_rtx.
17107 * config/arc/arc.h (CC1_SPEC): Recognize G option.
17108 * config/arc/simdext.md (movmisalignv2hi): Use
17109 prepare_move_operands function.
17111 (movmisalign*): Likewise.
17112 * doc/invoke.texi (ARC options): Document -G option.
17114 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17116 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
17118 * config/arc/arc.c (arc_print_operand): Output scalled address for
17119 sdata whenever is possible.
17120 (arc_in_small_data_p): Allow sdata for 64bit datum when double
17121 load/stores are available.
17122 (compact_sda_memory_operand): Check for the alignment required by
17123 code density instructions.
17124 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
17126 * config/arc/constraints.md (Usd): Update constraint.
17127 (Us0): New constraint.
17128 (Usc): Update constraint.
17130 2017-08-31 Richard Biener <rguenther@suse.de>
17132 PR middle-end/82054
17133 * dwarf2out.c (dwarf2out_early_global_decl): Process each
17134 function only once.
17136 2017-08-31 Tamar Christina <tamar.christina@arm.com>
17138 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
17139 Resize type_signature.
17141 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17142 Alan Hayward <alan.hayward@arm.com>
17143 David Sherwood <david.sherwood@arm.com>
17145 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
17146 subregs whose inner modes can be stored in GPRs.
17147 (aarch64_classify_index): Likewise.
17149 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17150 Alan Hayward <alan.hayward@arm.com>
17151 David Sherwood <david.sherwood@arm.com>
17153 * config/aarch64/iterators.md (V_cmp_result): Rename to...
17154 (V_INT_EQUIV): ...this.
17155 (v_cmp_result): Rename to...
17156 (v_int_equiv): ...this.
17157 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
17158 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
17159 (copysign<mode>3): Likewise.
17160 (aarch64_simd_bsl<mode>_internal): Likewise.
17161 (aarch64_simd_bsl<mode>): Likewise.
17162 (vec_cmp<mode><mode>): Likewise.
17163 (vcond<mode><mode>): Likewise.
17164 (vcond<v_cmp_mixed><mode>): Likewise.
17165 (vcondu<mode><v_cmp_mixed>): Likewise.
17166 (aarch64_cm<optab><mode>): Likewise.
17167 (aarch64_cmtst<mode>): Likewise.
17168 (aarch64_fac<optab><mode>): Likewise.
17169 (vec_perm_const<mode>): Likewise.
17170 (vcond_mask_<mode><v_cmp_result>): Rename to...
17171 (vcond_mask_<mode><v_int_equiv>): ...this.
17172 (vec_cmp<mode><v_cmp_result>): Rename to...
17173 (vec_cmp<mode><v_int_equiv>): ...this.
17175 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17176 Alan Hayward <alan.hayward@arm.com>
17177 David Sherwood <david.sherwood@arm.com>
17179 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
17181 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
17182 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
17183 (UNSPEC_LD4_DREG): New unspecs.
17184 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
17185 (aarch64_ld2<mode>_dreg_be): Replace with...
17186 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
17188 (aarch64_ld3<mode>_dreg_le)
17189 (aarch64_ld3<mode>_dreg_be): Replace with...
17190 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
17192 (aarch64_ld4<mode>_dreg_le)
17193 (aarch64_ld4<mode>_dreg_be): Replace with...
17194 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
17197 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17199 PR tree-optimization/81987
17200 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
17201 insert an initializer in a location not dominated by the stride
17204 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17206 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
17207 on the entire header of the finally block in the fallthru case.
17209 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17211 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
17213 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
17215 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
17216 rs6000_emit_move_from_cr and call renamed function.
17217 (rs6000_emit_prologue): Call renamed functions.
17218 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
17219 movesi_from_cr, remove volatile CRs.
17221 2017-08-30 Jon Beniston <jon@beniston.com>
17222 Richard Biener <rguenther@suse.de>
17224 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
17225 of VECTOR_MODE_P check.
17226 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
17227 element vector types.
17229 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17231 * df.h (df_read_modify_subreg_p): Remove in favor of...
17232 * rtl.h (read_modify_subreg_p): ...this new function. Take a
17233 const_rtx instead of an rtx.
17234 * cprop.c (local_cprop_find_used_regs): Update accordingly.
17235 * df-problems.c (df_word_lr_mark_ref): Likewise.
17236 * ira-lives.c (mark_pseudo_reg_live): Likewise.
17237 (mark_pseudo_reg_dead): Likewise.
17238 (mark_ref_dead): Likewise.
17239 * reginfo.c (init_subregs_of_mode): Likewise.
17240 * sched-deps.c (sched_analyze_1): Likewise.
17241 * df-scan.c (df_def_record_1): Likewise.
17242 (df_uses_record): Likewise.
17243 (df_read_modify_subreg_p): Remove in favor of...
17244 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
17245 const_rtx instead of an rtx.
17247 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17248 Alan Hayward <alan.hayward@arm.com>
17249 David Sherwood <david.sherwood@arm.com>
17251 * rtl.h (partial_subreg_p): New function.
17252 * caller-save.c (save_call_clobbered_regs): Use it.
17253 * calls.c (expand_call): Likewise.
17254 * combine.c (combinable_i3pat): Likewise.
17255 (simplify_set): Likewise.
17256 (make_extraction): Likewise.
17257 (make_compound_operation_int): Likewise.
17258 (gen_lowpart_or_truncate): Likewise.
17259 (force_to_mode): Likewise.
17260 (make_field_assignment): Likewise.
17261 (reg_truncated_to_mode): Likewise.
17262 (record_truncated_value): Likewise.
17263 (move_deaths): Likewise.
17264 * cse.c (record_jump_cond): Likewise.
17265 (cse_insn): Likewise.
17266 * cselib.c (cselib_lookup_1): Likewise.
17267 * expmed.c (extract_bit_field_using_extv): Likewise.
17268 * function.c (assign_parm_setup_reg): Likewise.
17269 * ifcvt.c (noce_convert_multiple_sets): Likewise.
17270 * ira-build.c (create_insn_allocnos): Likewise.
17271 * lra-coalesce.c (merge_pseudos): Likewise.
17272 * lra-constraints.c (match_reload): Likewise.
17273 (simplify_operand_subreg): Likewise.
17274 (curr_insn_transform): Likewise.
17275 * lra-lives.c (process_bb_lives): Likewise.
17276 * lra.c (new_insn_reg): Likewise.
17277 (lra_substitute_pseudo): Likewise.
17278 * regcprop.c (mode_change_ok): Likewise.
17279 (maybe_mode_change): Likewise.
17280 (copyprop_hardreg_forward_1): Likewise.
17281 * reload.c (push_reload): Likewise.
17282 (find_reloads): Likewise.
17283 (find_reloads_subreg_address): Likewise.
17284 * reload1.c (alter_reg): Likewise.
17285 (eliminate_regs_1): Likewise.
17286 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17288 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
17290 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
17291 back to if statements, including unpack.
17293 2017-08-30 Martin Liska <mliska@suse.cz>
17295 PR inline-asm/82001
17296 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
17298 (func_checker::compare_asm_inputs_outputs): ... this function.
17299 (func_checker::compare_gimple_asm): Use the function to compare
17300 also ASM constrains.
17301 * ipa-icf-gimple.h: Rename the function.
17303 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17304 Alan Hayward <alan.hayward@arm.com>
17305 David Sherwood <david.sherwood@arm.com>
17307 * coretypes.h (complex_mode): New type.
17308 * gdbhooks.py (build_pretty_printer): Handle it.
17309 * machmode.h (complex_mode): New class.
17310 (complex_mode::includes_p): New function.
17311 (is_complex_int_mode): Likewise.
17312 (is_complex_float_mode): Likewise.
17313 * genmodes.c (get_mode_class): Handle complex mode classes.
17314 * function.c (expand_function_end): Use is_complex_int_mode.
17316 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17317 Alan Hayward <alan.hayward@arm.com>
17318 David Sherwood <david.sherwood@arm.com>
17320 * coretypes.h (scalar_mode_pod): New typedef.
17321 * gdbhooks.py (build_pretty_printer): Handle it.
17322 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
17323 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
17324 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
17325 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
17326 as_a <scalar_mode>.
17328 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17329 Alan Hayward <alan.hayward@arm.com>
17330 David Sherwood <david.sherwood@arm.com>
17332 * machmode.h (mode_for_vector): Take a scalar_mode instead
17334 * stor-layout.c (mode_for_vector): Likewise.
17335 * explow.c (promote_mode): Use as_a <scalar_mode>.
17336 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
17338 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17339 Alan Hayward <alan.hayward@arm.com>
17340 David Sherwood <david.sherwood@arm.com>
17342 * target.def (preferred_simd_mode): Take a scalar_mode
17343 instead of a machine_mode.
17344 * targhooks.h (default_preferred_simd_mode): Likewise.
17345 * targhooks.c (default_preferred_simd_mode): Likewise.
17346 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
17347 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
17348 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
17349 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
17350 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
17351 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
17352 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
17353 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
17355 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
17356 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
17357 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
17358 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
17359 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
17360 * doc/tm.texi: Regenerate.
17361 * optabs-query.c (can_vec_mask_load_store_p): Return false for
17364 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17365 Alan Hayward <alan.hayward@arm.com>
17366 David Sherwood <david.sherwood@arm.com>
17368 * target.def (scalar_mode_supported_p): Take a scalar_mode
17369 instead of a machine_mode.
17370 * targhooks.h (default_scalar_mode_supported_p): Likewise.
17371 * targhooks.c (default_scalar_mode_supported_p): Likewise.
17372 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
17373 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17374 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
17375 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
17376 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
17377 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
17378 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
17379 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
17380 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
17381 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
17382 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
17383 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
17385 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
17386 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
17387 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17388 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17389 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
17391 * doc/tm.texi: Regenerate.
17393 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17394 Alan Hayward <alan.hayward@arm.com>
17395 David Sherwood <david.sherwood@arm.com>
17397 * coretypes.h (opt_scalar_mode): New typedef.
17398 * gdbhooks.py (build_pretty_printers): Handle it.
17399 * machmode.h (mode_iterator::get_2xwider): Add overload for
17401 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
17403 * expr.c (convert_mode_scalar): Likewise.
17404 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17405 * optabs.c (expand_float): Likewise.
17406 (expand_fix): Likewise.
17407 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17409 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17410 Alan Hayward <alan.hayward@arm.com>
17411 David Sherwood <david.sherwood@arm.com>
17413 * optabs.c (expand_float): Explicitly check for scalars before
17414 using a branching expansion.
17415 (expand_fix): Likewise.
17417 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17418 Alan Hayward <alan.hayward@arm.com>
17419 David Sherwood <david.sherwood@arm.com>
17421 * expr.c (convert_mode): Split scalar handling out into...
17422 (convert_mode_scalar): ...this new function. Treat the modes
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 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
17431 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
17433 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17434 Alan Hayward <alan.hayward@arm.com>
17435 David Sherwood <david.sherwood@arm.com>
17437 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
17438 rather than a machine_mode.
17439 (fixed_from_string): Likewise.
17440 (fixed_convert): Likewise.
17441 (fixed_convert_from_int): Likewise.
17442 (fixed_convert_from_real): Likewise.
17443 (real_convert_from_fixed): Likewise.
17444 * fixed-value.c (fixed_from_double_int): Likewise.
17445 (fixed_from_string): Likewise.
17446 (fixed_convert): Likewise.
17447 (fixed_convert_from_int): Likewise.
17448 (fixed_convert_from_real): Likewise.
17449 (real_convert_from_fixed): Likewise.
17450 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
17452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17453 Alan Hayward <alan.hayward@arm.com>
17454 David Sherwood <david.sherwood@arm.com>
17456 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
17457 of separate mode class checks. Do not allow vector modes here.
17458 (immed_wide_int_const): Use as_a <scalar_mode>.
17459 * explow.c (trunc_int_for_mode): Likewise.
17460 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
17461 (wi::shwi): Likewise.
17462 (wi::min_value): Likewise.
17463 (wi::max_value): Likewise.
17464 * dwarf2out.c (loc_descriptor): Likewise.
17465 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
17466 for CONST_WIDE_INT.
17468 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17469 Alan Hayward <alan.hayward@arm.com>
17470 David Sherwood <david.sherwood@arm.com>
17472 * tree.h (SCALAR_TYPE_MODE): New macro.
17473 * expr.c (expand_expr_addr_expr_1): Use it.
17474 (expand_expr_real_2): Likewise.
17475 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
17476 (fold_convert_const_fixed_from_int): Likewise.
17477 (fold_convert_const_fixed_from_real): Likewise.
17478 (native_encode_fixed): Likewise
17479 (native_encode_complex): Likewise
17480 (native_encode_vector): Likewise.
17481 (native_interpret_fixed): Likewise.
17482 (native_interpret_real): Likewise.
17483 (native_interpret_complex): Likewise.
17484 (native_interpret_vector): Likewise.
17485 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
17486 (simd_clone_adjust_argument_types): Likewise.
17487 (simd_clone_init_simd_arrays): Likewise.
17488 (simd_clone_adjust): Likewise.
17489 * stor-layout.c (layout_type): Likewise.
17490 * tree.c (build_minus_one_cst): Likewise.
17491 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17492 * tree-inline.c (estimate_move_cost): Likewise.
17493 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
17494 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
17495 (vectorizable_reduction): Likewise.
17496 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
17497 (vect_recog_mixed_size_cond_pattern): Likewise.
17498 (check_bool_pattern): Likewise.
17499 (adjust_bool_pattern): Likewise.
17500 (search_type_for_mask_1): Likewise.
17501 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
17502 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17503 (vectorizable_load): Likewise.
17504 (vectorizable_store): Likewise.
17505 * ubsan.c (ubsan_encode_value): Likewise.
17506 * varasm.c (output_constant): Likewise.
17508 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17509 Alan Hayward <alan.hayward@arm.com>
17510 David Sherwood <david.sherwood@arm.com>
17512 * coretypes.h (scalar_mode): New class.
17513 * machmode.h (scalar_mode): Likewise.
17514 (scalar_mode::includes_p): New function.
17515 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
17516 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
17517 * genmodes.c (get_mode_class): Handle remaining scalar modes.
17518 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
17519 * expmed.c (store_bit_field_1): Likewise.
17520 (extract_bit_field_1): Likewise.
17521 * expr.c (write_complex_part): Likewise.
17522 (read_complex_part): Likewise.
17523 (emit_move_complex_push): Likewise.
17524 (expand_expr_real_2): Likewise.
17525 * function.c (assign_parm_setup_reg): Likewise.
17526 (assign_parms_unsplit_complex): Likewise.
17527 * optabs.c (expand_binop): Likewise.
17528 * rtlanal.c (subreg_get_info): Likewise.
17529 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17530 * varasm.c (output_constant_pool_2): Likewise.
17532 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17533 Alan Hayward <alan.hayward@arm.com>
17534 David Sherwood <david.sherwood@arm.com>
17536 * expmed.c (extract_high_half): Use scalar_int_mode and remove
17538 (expmed_mult_highpart_optab): Likewise.
17539 (expmed_mult_highpart): Likewise.
17541 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17542 Alan Hayward <alan.hayward@arm.com>
17543 David Sherwood <david.sherwood@arm.com>
17545 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
17546 instead of a machine_mode.
17547 (builtin_memset_read_str): Likewise.
17548 * builtins.c (c_readstr): Likewise.
17549 (builtin_memcpy_read_str): Likewise.
17550 (builtin_strncpy_read_str): Likewise.
17551 (builtin_memset_read_str): Likewise.
17552 (builtin_memset_gen_str): Likewise.
17553 (expand_builtin_signbit): Use scalar_int_mode for local variables.
17554 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
17555 instead of a machine_mode.
17556 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
17558 (make_extraction): Likewise.
17559 (try_widen_shift_mode): Take and return scalar_int_modes instead
17561 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
17562 a scalar_int_mode instead of a machine_mode.
17563 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
17564 (avr_addr_space_pointer_mode): Likewise.
17565 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
17566 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
17567 (msp430_unwind_word_mode): Likewise.
17568 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
17569 (spu_addr_space_pointer_mode): Likewise.
17570 (spu_addr_space_address_mode): Likewise.
17571 (spu_libgcc_cmp_return_mode): Likewise.
17572 (spu_libgcc_shift_count_mode): Likewise.
17573 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
17574 (rl78_addr_space_pointer_mode): Likewise.
17575 (fl78_unwind_word_mode): Likewise.
17576 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
17578 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
17579 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
17580 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
17581 (mips_valid_pointer_mode): Likewise.
17582 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
17583 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
17584 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
17586 (ft32_addr_space_address_mode): Likewise.
17587 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
17588 scalar_int_mode instead of a machine_mode.
17589 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
17591 (m32c_addr_space_address_mode): Likewise.
17592 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
17593 (rs6000_eh_return_filter_mode): Likewise.
17594 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
17595 (rs6000_eh_return_filter_mode): Likewise.
17596 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
17597 (s390_libgcc_shift_count_mode): Likewise.
17598 (s390_unwind_word_mode): Likewise.
17599 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
17601 * target.def (mode_rep_extended): Likewise.
17602 (valid_pointer_mode): Likewise.
17603 (addr_space.valid_pointer_mode): Likewise.
17604 (eh_return_filter_mode): Return a scalar_int_mode rather than
17606 (libgcc_cmp_return_mode): Likewise.
17607 (libgcc_shift_count_mode): Likewise.
17608 (unwind_word_mode): Likewise.
17609 (addr_space.pointer_mode): Likewise.
17610 (addr_space.address_mode): Likewise.
17611 * doc/tm.texi: Regenerate.
17612 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
17614 (do_jump): Use scalar_int_mode for local variables.
17615 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
17616 rather than a machine_mode.
17617 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
17618 (scompare_loc_descriptor_wide): Likewise.
17619 (scompare_loc_descriptor_narrow): Likewise.
17620 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
17622 * except.c (sjlj_emit_dispatch_table): Likewise.
17623 (expand_builtin_eh_copy_values): Likewise.
17624 * explow.c (convert_memory_address_addr_space_1): Likewise.
17625 Take a scalar_int_mode rather than a machine_mode.
17626 (convert_memory_address_addr_space): Take a scalar_int_mode rather
17627 than a machine_mode.
17628 (memory_address_addr_space): Use scalar_int_mode for local variables.
17629 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
17630 rather than a machine_mode.
17631 * expmed.c (mask_rtx): Likewise.
17632 (init_expmed_one_conv): Likewise.
17633 (expand_mult_highpart_adjust): Likewise.
17634 (extract_high_half): Likewise.
17635 (expmed_mult_highpart_optab): Likewise.
17636 (expmed_mult_highpart): Likewise.
17637 (expand_smod_pow2): Likewise.
17638 (expand_sdiv_pow2): Likewise.
17639 (emit_store_flag_int): Likewise.
17640 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
17642 (extract_low_bits): Likewise.
17643 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
17645 * expr.c (pieces_addr::adjust): Likewise.
17646 (can_store_by_pieces): Likewise.
17647 (store_by_pieces): Likewise.
17648 (clear_by_pieces_1): Likewise.
17649 (expand_expr_addr_expr_1): Likewise.
17650 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
17651 (expand_expr_real_1): Likewise.
17652 (try_casesi): Likewise.
17653 * final.c (shorten_branches): Likewise.
17654 * fold-const.c (fold_convert_const_int_from_fixed): Change the
17655 type of "mode" to machine_mode.
17656 * internal-fn.c (expand_arith_overflow_result_store): Take a
17657 scalar_int_mode rather than a machine_mode.
17658 (expand_mul_overflow): Use scalar_int_mode for local variables.
17659 * loop-doloop.c (doloop_modify): Likewise.
17660 (doloop_optimize): Likewise.
17661 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
17662 than a machine_mode.
17663 (expand_doubleword_shift_condmove): Likewise.
17664 (expand_doubleword_shift): Likewise.
17665 (expand_doubleword_clz): Likewise.
17666 (expand_doubleword_popcount): Likewise.
17667 (expand_doubleword_parity): Likewise.
17668 (expand_absneg_bit): Use scalar_int_mode for local variables.
17669 (prepare_float_lib_cmp): Likewise.
17670 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
17671 rather than a machine_mode.
17672 (convert_memory_address_addr_space): Likewise.
17673 (get_mode_bounds): Likewise.
17674 (get_address_mode): Return a scalar_int_mode rather than a
17676 * rtlanal.c (get_address_mode): Likewise.
17677 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
17678 than a machine_mode.
17679 * targhooks.c (default_mode_rep_extended): Likewise.
17680 (default_valid_pointer_mode): Likewise.
17681 (default_addr_space_valid_pointer_mode): Likewise.
17682 (default_eh_return_filter_mode): Return a scalar_int_mode rather
17683 than a machine_mode.
17684 (default_libgcc_cmp_return_mode): Likewise.
17685 (default_libgcc_shift_count_mode): Likewise.
17686 (default_unwind_word_mode): Likewise.
17687 (default_addr_space_pointer_mode): Likewise.
17688 (default_addr_space_address_mode): Likewise.
17689 * targhooks.h (default_eh_return_filter_mode): Likewise.
17690 (default_libgcc_cmp_return_mode): Likewise.
17691 (default_libgcc_shift_count_mode): Likewise.
17692 (default_unwind_word_mode): Likewise.
17693 (default_addr_space_pointer_mode): Likewise.
17694 (default_addr_space_address_mode): Likewise.
17695 (default_mode_rep_extended): Take a scalar_int_mode rather than
17697 (default_valid_pointer_mode): Likewise.
17698 (default_addr_space_valid_pointer_mode): Likewise.
17699 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
17701 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
17702 rather than a machine_mode.
17703 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
17704 for local variables.
17705 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17706 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
17707 than a machine_mode.
17709 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17710 Alan Hayward <alan.hayward@arm.com>
17711 David Sherwood <david.sherwood@arm.com>
17713 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
17714 the mode argument to scalar_int_mode.
17715 (do_jump_by_parts_zero_rtx): Likewise.
17716 (do_jump_by_parts_equality_rtx): Likewise.
17717 (do_jump_by_parts_greater): Take a mode argument.
17718 (do_jump_by_parts_equality): Likewise.
17719 (do_jump_1): Update calls accordingly.
17721 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17722 Alan Hayward <alan.hayward@arm.com>
17723 David Sherwood <david.sherwood@arm.com>
17725 * is-a.h (safe_dyn_cast): New function.
17726 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
17727 (jump_table_for_label): Likewise.
17728 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
17729 instead of an rtx_insn *.
17730 (shorten_branches): Use dyn_cast instead of LABEL_P and
17731 JUMP_TABLE_DATA_P. Use jump_table_for_label and
17732 rtx_jump_table_data::get_data_mode.
17733 (final_scan_insn): Likewise.
17735 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17736 Alan Hayward <alan.hayward@arm.com>
17737 David Sherwood <david.sherwood@arm.com>
17739 * combine.c (try_combine): Use is_a <scalar_int_mode> when
17740 trying to combine a full-register integer set with a subreg
17743 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17744 Alan Hayward <alan.hayward@arm.com>
17745 David Sherwood <david.sherwood@arm.com>
17747 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
17748 that is always either address_mode or pointer_mode.
17750 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17751 Alan Hayward <alan.hayward@arm.com>
17752 David Sherwood <david.sherwood@arm.com>
17754 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
17755 when the two are known to be equal.
17757 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17758 Alan Hayward <alan.hayward@arm.com>
17759 David Sherwood <david.sherwood@arm.com>
17761 * simplify-rtx.c (simplify_const_unary_operation): Use
17762 is_a <scalar_int_mode> instead of checking for a nonzero
17763 precision. Forcibly convert op_mode to a scalar_int_mode
17764 in that case. More clearly differentiate the operand and
17765 result modes and use the former when deciding what the value
17766 of a count-bits operation should be. Use is_int_mode instead
17767 of checking for a MODE_INT. Remove redundant check for whether
17768 this mode has a zero precision.
17770 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17771 Alan Hayward <alan.hayward@arm.com>
17772 David Sherwood <david.sherwood@arm.com>
17774 * optabs.c (widen_leading): Change the type of the mode argument
17775 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
17776 (widen_bswap): Likewise.
17777 (expand_parity): Likewise.
17778 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
17779 (expand_ffs): Likewise.
17780 (epand_unop): Check for scalar integer modes before calling the
17783 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17784 Alan Hayward <alan.hayward@arm.com>
17785 David Sherwood <david.sherwood@arm.com>
17787 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
17789 (expand_expr_real_1): Update call accordingly.
17791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17792 Alan Hayward <alan.hayward@arm.com>
17793 David Sherwood <david.sherwood@arm.com>
17795 * expmed.c (store_bit_field_using_insv): Add op0_mode and
17796 value_mode arguments. Use scalar_int_mode internally.
17797 (store_bit_field_1): Rename the new integer mode from imode
17798 to op0_mode and use it instead of GET_MODE (op0). Update calls
17799 to store_split_bit_field, store_bit_field_using_insv and
17800 store_fixed_bit_field.
17801 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
17802 Use scalar_int_mode internally. Use a bit count rather than a mode
17803 when calculating the largest bit size for get_best_mode.
17804 Update calls to store_split_bit_field and store_fixed_bit_field_1.
17805 (store_fixed_bit_field_1): Add mode and value_mode arguments.
17806 Remove assertion that OP0 has a scalar integer mode.
17807 (store_split_bit_field): Add op0_mode and value_mode arguments.
17808 Update calls to extract_fixed_bit_field.
17809 (extract_bit_field_using_extv): Add an op0_mode argument.
17810 Use scalar_int_mode internally.
17811 (extract_bit_field_1): Rename the new integer mode from imode to
17812 op0_mode and use it instead of GET_MODE (op0). Update calls to
17813 extract_split_bit_field, extract_bit_field_using_extv and
17814 extract_fixed_bit_field.
17815 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
17816 to extract_split_bit_field and extract_fixed_bit_field_1.
17817 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
17818 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
17819 on the target mode.
17820 (extract_split_bit_field): Add an op0_mode argument. Update call
17821 to extract_fixed_bit_field.
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 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
17828 * explow.c (hard_function_value): Likewise.
17829 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
17830 convert_to_mode call outside the loop.
17831 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
17832 for the mode iterator. Require the mode specified by max_pieces
17834 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
17836 (copy_blkmode_to_reg): Likewise.
17837 (set_storage_via_setmem): Likewise.
17838 * optabs.c (prepare_cmp_insn): Likewise.
17839 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17840 * stor-layout.c (finish_bitfield_representative): Likewise.
17842 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17843 Alan Hayward <alan.hayward@arm.com>
17844 David Sherwood <david.sherwood@arm.com>
17846 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
17847 * expr.c (convert_move): Use them.
17848 (convert_modes): Likewise.
17849 (store_expr_with_bounds): Likewise.
17851 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17852 Alan Hayward <alan.hayward@arm.com>
17853 David Sherwood <david.sherwood@arm.com>
17855 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
17856 parameter for the mode of "x". Remove the "known_x", "known_mode"
17857 and "known_ret" arguments. Change the type of the mode argument
17858 to scalar_int_mode.
17859 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
17860 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
17861 (reg_num_sign_bit_copies_for_combine): Likewise.
17862 * rtlanal.c (nonzero_bits1): Likewise.
17863 (num_sign_bit_copies1): Likewise.
17864 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
17865 (reg_num_sign_bit_copies_general): Likewise.
17866 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
17867 (reg_nonzero_bits_general): Likewise.
17869 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17870 Alan Hayward <alan.hayward@arm.com>
17871 David Sherwood <david.sherwood@arm.com>
17873 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
17874 than in subroutines. Return 1 for non-integer modes.
17875 (cached_num_sign_bit_copies): Change the type of the mode parameter
17876 to scalar_int_mode.
17877 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
17878 classes. Handle CONST_INT_P first and then check whether X also
17879 has a scalar integer mode. Check the same thing for inner registers
17880 of a SUBREG and for values that are being extended or truncated.
17882 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17883 Alan Hayward <alan.hayward@arm.com>
17884 David Sherwood <david.sherwood@arm.com>
17886 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
17887 in subroutines. Return the mode mask for non-integer modes.
17888 (cached_nonzero_bits): Change the type of the mode parameter
17889 to scalar_int_mode.
17890 (nonzero_bits1): Likewise. Remove early exit for other mode
17891 classes. Handle CONST_INT_P first and then check whether X
17892 also has a scalar integer mode.
17894 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17895 Alan Hayward <alan.hayward@arm.com>
17896 David Sherwood <david.sherwood@arm.com>
17898 * expr.c (widest_int_mode_for_size): Make the comment match the code.
17899 Return a scalar_int_mode and assert that the size is greater than
17901 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
17902 (op_by_pieces_d::op_by_pieces_d): Likewise.
17903 (op_by_pieces_d::run): Likewise.
17904 (can_store_by_pieces): Likewise.
17906 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17907 Alan Hayward <alan.hayward@arm.com>
17908 David Sherwood <david.sherwood@arm.com>
17910 * combine.c (extract_left_shift): Add a mode argument and update
17912 (make_compound_operation_int): Change the type of the mode parameter
17913 to scalar_int_mode and update the call to extract_left_shift.
17915 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17916 Alan Hayward <alan.hayward@arm.com>
17917 David Sherwood <david.sherwood@arm.com>
17919 * combine.c (simplify_and_const_int): Change the type of the mode
17920 parameter to scalar_int_mode.
17921 (simplify_and_const_int_1): Likewise. Update recursive call.
17923 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17924 Alan Hayward <alan.hayward@arm.com>
17925 David Sherwood <david.sherwood@arm.com>
17927 * combine.c (simplify_compare_const): Check that the mode is a
17928 scalar_int_mode (rather than VOIDmode) before testing its
17930 (simplify_comparison): Move COMPARISON_P handling out of the
17931 loop and restrict the latter part of the loop to scalar_int_modes.
17932 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
17933 and when considering SUBREG_REGs. Use is_int_mode instead of
17934 checking GET_MODE_CLASS against MODE_INT.
17936 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17937 Alan Hayward <alan.hayward@arm.com>
17938 David Sherwood <david.sherwood@arm.com>
17940 * combine.c (try_widen_shift_mode): Move check for equal modes to...
17941 (simplify_shift_const_1): ...here. Use scalar_int_mode for
17942 shift_unit_mode and for modes involved in scalar shifts.
17944 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17945 Alan Hayward <alan.hayward@arm.com>
17946 David Sherwood <david.sherwood@arm.com>
17948 * combine.c (force_int_to_mode): New function, split out from...
17949 (force_to_mode): ...here. Keep xmode up-to-date and use it
17950 instead of GET_MODE (x).
17952 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17953 Alan Hayward <alan.hayward@arm.com>
17954 David Sherwood <david.sherwood@arm.com>
17956 * optabs-query.h (extraction_insn::struct_mode): Change type to
17957 opt_scalar_int_mode and update comment.
17958 (extraction_insn::field_mode): Change type to scalar_int_mode.
17959 (extraction_insn::pos_mode): Likewise.
17960 * combine.c (make_extraction): Update accordingly.
17961 * optabs-query.c (get_traditional_extraction_insn): Likewise.
17962 (get_optab_extraction_insn): Likewise.
17963 * recog.c (simplify_while_replacing): Likewise.
17964 * expmed.c (narrow_bit_field_mem): Change the type of the mode
17965 parameter to opt_scalar_int_mode.
17967 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17968 Alan Hayward <alan.hayward@arm.com>
17969 David Sherwood <david.sherwood@arm.com>
17971 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
17972 to a scalar_int_mode instead of a machine_mode.
17973 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
17974 (get_best_mode): Return a boolean and use a pointer argument to store
17975 the selected mode. Replace the limit mode parameter with a bit limit.
17976 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
17977 for the values returned by bit_field_mode_iterator::next_mode.
17978 (store_bit_field): Update call to get_best_mode.
17979 (store_fixed_bit_field): Likewise.
17980 (extract_fixed_bit_field): Likewise.
17981 * expr.c (optimize_bitfield_assignment_op): Likewise.
17982 * fold-const.c (optimize_bit_field_compare): Likewise.
17983 (fold_truth_andor_1): Likewise.
17984 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
17985 Update for new type of m_mode.
17986 (get_best_mode): As above.
17988 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17989 Alan Hayward <alan.hayward@arm.com>
17990 David Sherwood <david.sherwood@arm.com>
17992 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
17993 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
17994 (store_bit_field): Check is_a <scalar_int_mode> before calling
17995 strict_volatile_bitfield_p.
17996 (extract_bit_field): Likewise.
17998 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17999 Alan Hayward <alan.hayward@arm.com>
18000 David Sherwood <david.sherwood@arm.com>
18002 * target.def (cstore_mode): Return a scalar_int_mode.
18003 * doc/tm.texi: Regenerate.
18004 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
18005 * targhooks.h (default_cstore_mode): Likewise.
18006 * targhooks.c (default_cstore_mode): Likewise, using a forced
18008 * expmed.c (emit_cstore): Expect the target of the cstore to be
18011 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18012 Alan Hayward <alan.hayward@arm.com>
18013 David Sherwood <david.sherwood@arm.com>
18015 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
18017 (niter_desc): Likewise mode.
18018 (iv_analyze): Add a mode parameter.
18020 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
18021 and change its type to scalar_int_mode.
18022 * loop-iv.c: Update commentary at head of file.
18023 (iv_constant): Pass the mode paraeter before the rtx it describes
18024 and change its type to scalar_int_mode. Remove VOIDmode handling.
18025 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
18026 (iv_extend): Likewise.
18027 (shorten_into_mode): Likewise.
18028 (iv_add): Use scalar_int_mode.
18029 (iv_mult): Likewise.
18030 (iv_shift): Likewise.
18031 (canonicalize_iv_subregs): Likewise.
18032 (get_biv_step_1): Pass the outer_mode parameter before the rtx
18033 it describes and change its mode to scalar_int_mode. Also change
18034 the type of the returned inner_mode to scalar_int_mode.
18035 (get_biv_step): Likewise, turning outer_mode from a pointer
18036 into a direct parameter. Update call to get_biv_step_1.
18037 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
18038 iv_constant and get_biv_step.
18039 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
18040 and change its type to scalar_int_mode. Don't initialise iv->mode
18041 to VOIDmode and remove later checks for its still being VOIDmode.
18042 Update calls to iv_analyze_op and iv_analyze_expr. Check
18043 is_a <scalar_int_mode> when changing the mode under consideration.
18044 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
18045 Update call to iv_analyze_expr.
18046 (iv_analyze_op): Add a mode parameter. Reject subregs whose
18047 inner register is not also a scalar_int_mode. Update call to
18049 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
18050 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
18051 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
18052 separate mode class checks. Update calls to iv_analyze. Remove
18053 fix-up of VOIDmodes after iv_analyze_biv.
18054 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
18055 don't have a scalar_int_mode. Update call to biv_p.
18057 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18058 Alan Hayward <alan.hayward@arm.com>
18059 David Sherwood <david.sherwood@arm.com>
18061 * cfgexpand.c (convert_debug_memory_address): Use
18062 as_a <scalar_int_mode>.
18063 * combine.c (expand_compound_operation): Likewise.
18064 (make_extraction): Likewise.
18065 (change_zero_ext): Likewise.
18066 (simplify_comparison): Likewise.
18067 * cse.c (cse_insn): Likewise.
18068 * dwarf2out.c (minmax_loc_descriptor): Likewise.
18069 (mem_loc_descriptor): Likewise.
18070 (loc_descriptor): Likewise.
18071 * expmed.c (init_expmed_one_mode): Likewise.
18072 (synth_mult): Likewise.
18073 (emit_store_flag_1): Likewise.
18074 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
18075 of a comparison with size.
18076 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
18077 (reduce_to_bit_field_precision): Likewise.
18078 * function.c (expand_function_end): Likewise.
18079 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
18080 * loop-doloop.c (doloop_modify): Likewise.
18081 * optabs.c (expand_binop): Likewise.
18082 (expand_unop): Likewise.
18083 (expand_copysign_absneg): Likewise.
18084 (prepare_cmp_insn): Likewise.
18085 (maybe_legitimize_operand): Likewise.
18086 * recog.c (const_scalar_int_operand): Likewise.
18087 * rtlanal.c (get_address_mode): Likewise.
18088 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18089 (simplify_cond_clz_ctz): Likewise.
18090 * tree-nested.c (get_nl_goto_field): Likewise.
18091 * tree.c (build_vector_type_for_mode): Likewise.
18092 * var-tracking.c (use_narrower_mode): Likewise.
18094 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18095 Alan Hayward <alan.hayward@arm.com>
18096 David Sherwood <david.sherwood@arm.com>
18098 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
18099 * builtins.c (expand_builtin_signbit): Use it.
18100 * cfgexpand.c (expand_debug_expr): Likewise.
18101 * dojump.c (do_jump): Likewise.
18102 (do_compare_and_jump): Likewise.
18103 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
18104 * expmed.c (make_tree): Likewise.
18105 * expr.c (expand_expr_real_2): Likewise.
18106 (expand_expr_real_1): Likewise.
18107 (try_casesi): Likewise.
18108 * fold-const-call.c (fold_const_call_ss): Likewise.
18109 * fold-const.c (unextend): Likewise.
18110 (extract_muldiv_1): Likewise.
18111 (fold_single_bit_test): Likewise.
18112 (native_encode_int): Likewise.
18113 (native_encode_string): Likewise.
18114 (native_interpret_int): Likewise.
18115 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
18116 * internal-fn.c (expand_addsub_overflow): Likewise.
18117 (expand_neg_overflow): Likewise.
18118 (expand_mul_overflow): Likewise.
18119 (expand_arith_overflow): Likewise.
18120 * match.pd: Likewise.
18121 * stor-layout.c (layout_type): Likewise.
18122 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18123 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
18124 * tree-ssanames.c (get_range_info): Likewise.
18125 * tree-switch-conversion.c (array_value_type) Likewise.
18126 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
18127 (vect_recog_divmod_pattern): Likewise.
18128 (vect_recog_mixed_size_cond_pattern): Likewise.
18129 * tree-vrp.c (extract_range_basic): Likewise.
18130 (simplify_float_conversion_using_ranges): Likewise.
18131 * tree.c (int_fits_type_p): Likewise.
18132 * ubsan.c (instrument_bool_enum_load): Likewise.
18133 * varasm.c (mergeable_string_section): Likewise.
18134 (narrowing_initializer_constant_valid_p): Likewise.
18135 (output_constant): Likewise.
18137 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18138 Alan Hayward <alan.hayward@arm.com>
18139 David Sherwood <david.sherwood@arm.com>
18141 * machmode.h (NARROWEST_INT_MODE): New macro.
18142 * expr.c (alignment_for_piecewise_move): Use it instead of
18143 GET_CLASS_NARROWEST_MODE (MODE_INT).
18144 (push_block): Likewise.
18145 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
18147 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18149 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18150 Alan Hayward <alan.hayward@arm.com>
18151 David Sherwood <david.sherwood@arm.com>
18153 * postreload.c (move2add_valid_value_p): Change the type of the
18154 mode parameter to scalar_int_mode.
18155 (move2add_use_add2_insn): Add a mode parameter and use it instead
18157 (move2add_use_add3_insn): Likewise.
18158 (reload_cse_move2add): Update accordingly.
18160 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18161 Alan Hayward <alan.hayward@arm.com>
18162 David Sherwood <david.sherwood@arm.com>
18164 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
18166 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
18167 * optabs.c (expand_unop): 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 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
18174 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
18175 (popcount_loc_descriptor): Likewise.
18176 (bswap_loc_descriptor): Likewise.
18177 (rotate_loc_descriptor): Likewise.
18178 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
18179 calling the functions above.
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 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
18187 (try_combine): Likewise.
18188 (simplify_if_then_else): Likewise.
18189 * cse.c (cse_insn): Likewise.
18190 * dwarf2out.c (mem_loc_descriptor): Likewise.
18191 * emit-rtl.c (gen_lowpart_common): Likewise.
18192 * simplify-rtx.c (simplify_truncation): Likewise.
18193 (simplify_binary_operation_1): Likewise.
18194 (simplify_const_relational_operation): Likewise.
18195 (simplify_ternary_operation): Likewise.
18196 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
18198 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18199 Alan Hayward <alan.hayward@arm.com>
18200 David Sherwood <david.sherwood@arm.com>
18202 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
18203 * reload.c (push_reload): Likewise.
18204 (find_reloads): Likewise.
18206 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18207 Alan Hayward <alan.hayward@arm.com>
18208 David Sherwood <david.sherwood@arm.com>
18210 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
18211 (make_compound_operation_int): Likewise.
18212 (change_zero_ext): Likewise.
18213 * expr.c (convert_move): Likewise.
18214 (convert_modes): Likewise.
18215 * fwprop.c (forward_propagate_subreg): Likewise.
18216 * loop-iv.c (get_biv_step_1): Likewise.
18217 * optabs.c (widen_operand): Likewise.
18218 * postreload.c (move2add_valid_value_p): Likewise.
18219 * recog.c (simplify_while_replacing): Likewise.
18220 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18221 (simplify_binary_operation_1): Likewise. Remove redundant
18222 mode equality check.
18224 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18225 Alan Hayward <alan.hayward@arm.com>
18226 David Sherwood <david.sherwood@arm.com>
18228 * combine.c (combine_simplify_rtx): Add checks for
18229 is_a <scalar_int_mode>.
18230 (simplify_if_then_else): Likewise.
18231 (make_field_assignment): Likewise.
18232 (simplify_comparison): Likewise.
18233 * ifcvt.c (noce_try_bitop): Likewise.
18234 * loop-invariant.c (canonicalize_address_mult): Likewise.
18235 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18238 Alan Hayward <alan.hayward@arm.com>
18239 David Sherwood <david.sherwood@arm.com>
18241 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
18242 is_a <scalar_int_mode> instead of != BLKmode.
18244 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18245 Alan Hayward <alan.hayward@arm.com>
18246 David Sherwood <david.sherwood@arm.com>
18248 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
18249 instead of != VOIDmode.
18250 * combine.c (if_then_else_cond): Likewise.
18251 (change_zero_ext): Likewise.
18252 * dwarf2out.c (mem_loc_descriptor): Likewise.
18253 (loc_descriptor): Likewise.
18254 * rtlanal.c (canonicalize_condition): Likewise.
18255 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
18257 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18258 Alan Hayward <alan.hayward@arm.com>
18259 David Sherwood <david.sherwood@arm.com>
18261 * simplify-rtx.c (simplify_binary_operation_1): Use
18262 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
18264 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18265 Alan Hayward <alan.hayward@arm.com>
18266 David Sherwood <david.sherwood@arm.com>
18268 * wide-int.h (int_traits<unsigned char>) New class.
18269 (int_traits<unsigned short>) Likewise.
18270 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
18271 Use GET_MODE_UNIT_PRECISION and remove redundant test for
18273 * combine.c (set_nonzero_bits_and_sign_copies): Use
18274 is_a <scalar_int_mode>.
18275 (find_split_point): Likewise.
18276 (combine_simplify_rtx): Likewise.
18277 (simplify_logical): Likewise.
18278 (expand_compound_operation): Likewise.
18279 (expand_field_assignment): Likewise.
18280 (make_compound_operation): Likewise.
18281 (extended_count): Likewise.
18282 (change_zero_ext): Likewise.
18283 (simplify_comparison): Likewise.
18284 * dwarf2out.c (scompare_loc_descriptor): Likewise.
18285 (ucompare_loc_descriptor): Likewise.
18286 (minmax_loc_descriptor): Likewise.
18287 (mem_loc_descriptor): Likewise.
18288 (loc_descriptor): Likewise.
18289 * expmed.c (init_expmed_one_mode): Likewise.
18290 * lra-constraints.c (lra_constraint_offset): Likewise.
18291 * optabs.c (prepare_libcall_arg): Likewise.
18292 * postreload.c (move2add_note_store): Likewise.
18293 * reload.c (operands_match_p): Likewise.
18294 * rtl.h (load_extend_op): Likewise.
18295 * rtlhooks.c (gen_lowpart_general): Likewise.
18296 * simplify-rtx.c (simplify_truncation): Likewise.
18297 (simplify_unary_operation_1): Likewise.
18298 (simplify_binary_operation_1): Likewise.
18299 (simplify_const_binary_operation): Likewise.
18300 (simplify_const_relational_operation): Likewise.
18301 (simplify_subreg): Likewise.
18302 * stor-layout.c (bitwise_mode_for_mode): Likewise.
18303 * var-tracking.c (adjust_mems): Likewise.
18304 (prepare_call_arguments): Likewise.
18306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18307 Alan Hayward <alan.hayward@arm.com>
18308 David Sherwood <david.sherwood@arm.com>
18310 * machmode.h (is_int_mode): New fuction.
18311 * combine.c (find_split_point): Use it.
18312 (combine_simplify_rtx): Likewise.
18313 (simplify_if_then_else): Likewise.
18314 (simplify_set): Likewise.
18315 (simplify_shift_const_1): Likewise.
18316 (simplify_comparison): Likewise.
18317 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
18318 * cse.c (notreg_cost): Likewise.
18319 (cse_insn): Likewise.
18320 * cselib.c (cselib_lookup_1): Likewise.
18321 * dojump.c (do_jump_1): Likewise.
18322 (do_compare_rtx_and_jump): Likewise.
18323 * dse.c (get_call_args): Likewise.
18324 * dwarf2out.c (rtl_for_decl_init): Likewise.
18325 (native_encode_initializer): Likewise.
18326 * expmed.c (emit_store_flag_1): Likewise.
18327 (emit_store_flag): Likewise.
18328 * expr.c (convert_modes): Likewise.
18329 (store_field): Likewise.
18330 (expand_expr_real_1): Likewise.
18331 * fold-const.c (fold_read_from_constant_string): Likewise.
18332 * gimple-ssa-sprintf.c (get_format_string): Likewise.
18333 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
18334 * optabs.c (expand_binop): Likewise.
18335 (expand_unop): Likewise.
18336 (expand_abs_nojump): Likewise.
18337 (expand_one_cmpl_abs_nojump): Likewise.
18338 * simplify-rtx.c (mode_signbit_p): Likewise.
18339 (val_signbit_p): Likewise.
18340 (val_signbit_known_set_p): Likewise.
18341 (val_signbit_known_clear_p): Likewise.
18342 (simplify_relational_operation_1): Likewise.
18343 * tree.c (vector_type_mode): Likewise.
18345 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18346 Alan Hayward <alan.hayward@arm.com>
18347 David Sherwood <david.sherwood@arm.com>
18349 * machmode.h (smallest_mode_for_size): Fix formatting.
18350 (smallest_int_mode_for_size): New function.
18351 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
18352 instead of smallest_mode_for_size.
18353 * combine.c (make_extraction): Likewise.
18354 * config/arc/arc.c (arc_expand_movmem): Likewise.
18355 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
18356 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
18357 * config/s390/s390.c (s390_expand_insv): Likewise.
18358 * config/sparc/sparc.c (assign_int_registers): Likewise.
18359 * config/spu/spu.c (spu_function_value): Likewise.
18360 (spu_function_arg): Likewise.
18361 * coverage.c (get_gcov_type): Likewise.
18362 (get_gcov_unsigned_t): Likewise.
18363 * dse.c (find_shift_sequence): Likewise.
18364 * expmed.c (store_bit_field_1): Likewise.
18365 * expr.c (convert_move): Likewise.
18366 (store_field): Likewise.
18367 * internal-fn.c (expand_arith_overflow): Likewise.
18368 * optabs-query.c (get_best_extraction_insn): Likewise.
18369 * optabs.c (expand_twoval_binop_libfunc): Likewise.
18370 * stor-layout.c (layout_type): Likewise.
18371 (initialize_sizetypes): Likewise.
18372 * targhooks.c (default_get_mask_mode): Likewise.
18373 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
18375 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18376 Alan Hayward <alan.hayward@arm.com>
18377 David Sherwood <david.sherwood@arm.com>
18379 * machmode.h (opt_mode::else_blk): New function.
18380 (int_mode_for_mode): Declare.
18381 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
18382 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
18384 * cfgexpand.c (expand_debug_expr): Likewise.
18385 * combine.c (gen_lowpart_or_truncate): Likewise.
18386 (gen_lowpart_for_combine): Likewise.
18387 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
18388 * config/avr/avr.c (avr_to_int_mode): Likewise.
18389 (avr_out_plus_1): Likewise.
18390 (avr_out_plus): Likewise.
18391 (avr_out_round): Likewise.
18392 * config/i386/i386.c (ix86_split_to_parts): Likewise.
18393 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
18394 (s390_expand_vcond): Likewise.
18395 * config/spu/spu.c (spu_split_immediate): Likewise.
18396 (spu_expand_mov): Likewise.
18397 * dse.c (get_stored_val): Likewise.
18398 * expmed.c (store_bit_field_1): Likewise.
18399 (convert_extracted_bit_field): Use int_mode_for_mode instead of
18401 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
18402 (extract_low_bits): Likewise.
18403 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
18404 handling rather than repeating the check.
18405 (emit_group_store): Likewise.
18406 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
18407 * optabs.c (expand_absneg_bit): Likewise.
18408 (expand_copysign_absneg): Likewise.
18409 (expand_copysign_bit): Likewise.
18410 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
18411 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18412 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18413 * var-tracking.c (prepare_call_arguments): Likewise.
18414 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
18415 int_mode_for_mode instead of mode_for_size.
18416 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
18418 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18419 Alan Hayward <alan.hayward@arm.com>
18420 David Sherwood <david.sherwood@arm.com>
18422 * machmode.h (int_mode_for_size): New function.
18423 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
18424 instead of mode_for_size.
18425 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
18427 * combine.c (expand_field_assignment): Use int_mode_for_size
18428 instead of mode_for_size.
18429 (make_extraction): Likewise.
18430 (simplify_shift_const_1): Likewise.
18431 (simplify_comparison): Likewise.
18432 * dojump.c (do_jump): Likewise.
18433 * dwarf2out.c (mem_loc_descriptor): Likewise.
18434 * emit-rtl.c (init_derived_machine_modes): Likewise.
18435 * expmed.c (flip_storage_order): Likewise.
18436 (convert_extracted_bit_field): Likewise.
18437 * expr.c (copy_blkmode_from_reg): Likewise.
18438 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
18439 * internal-fn.c (expand_mul_overflow): Likewise.
18440 * lower-subreg.c (simple_move): Likewise.
18441 * optabs-libfuncs.c (init_optabs): Likewise.
18442 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18443 * tree.c (vector_type_mode): Likewise.
18444 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18445 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
18446 * tree-vect-generic.c (expand_vector_parallel): Likewise.
18447 * tree-vect-stmts.c (vectorizable_load): Likewise.
18448 (vectorizable_store): 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 * coretypes.h (pod_mode): New type.
18455 (scalar_int_mode_pod): New typedef.
18456 * machmode.h (pod_mode): New class.
18457 (int_n_data_t::m): Change type to scalar_int_mode_pod.
18458 * genmodes.c (emit_mode_int_n): Update accordingly.
18459 * lower-subreg.h (target_lower_subreg): Change type to
18460 scalar_int_mode_pod.
18461 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
18462 scalar_int_mode_pod.
18464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18465 Alan Hayward <alan.hayward@arm.com>
18466 David Sherwood <david.sherwood@arm.com>
18468 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
18469 machine_mode to scalar_int_mode.
18470 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
18471 (rs6000_option_override_internal): Remove cast to int.
18472 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
18473 machine_mode to scalar_int_mode.
18474 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
18475 (rs6000_option_override_internal): Remove cast to int.
18476 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
18477 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
18479 * config/s390/s390.c (s390_expand_builtin): Likewise.
18480 * coretypes.h (scalar_int_mode): New type.
18481 (opt_scalar_int_mode): New typedef.
18482 * machmode.h (scalar_int_mode): New class.
18483 (scalar_int_mode::includes_p): New function.
18484 (byte_mode): Change type to scalar_int_mode.
18485 (word_mode): Likewise.
18486 (ptr_mode): Likewise.
18487 * emit-rtl.c (byte_mode): Likewise.
18488 (word_mode): Likewise.
18489 (ptr_mode): Likewise.
18490 (init_derived_machine_modes): Update accordingly.
18491 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
18492 and MODE_PARTIAL_INT.
18493 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
18494 opt_scalar_int_mode.
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 * target.def (libgcc_floating_mode_supported_p): Take a
18502 * doc/tm.texi: Regenerate.
18503 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
18505 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18506 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
18509 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18510 Alan Hayward <alan.hayward@arm.com>
18511 David Sherwood <david.sherwood@arm.com>
18513 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
18514 * doc/tm.texi: Regenerate.
18515 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
18516 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
18517 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
18518 * targhooks.h (default_floatn_mode): Likewise.
18519 * targhooks.c (default_floatn_mode): Likewise.
18520 * tree.c (build_common_tree_nodes): Update accordingly.
18522 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18523 Alan Hayward <alan.hayward@arm.com>
18524 David Sherwood <david.sherwood@arm.com>
18526 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
18527 (mode_iterator::iterate_p): Likewise.
18528 (mode_iterator::get_wider): Likewise.
18529 * expr.c (init_expr_target): Use opt_scalar_float_mode.
18531 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18532 Alan Hayward <alan.hayward@arm.com>
18533 David Sherwood <david.sherwood@arm.com>
18535 * coretypes.h (opt_scalar_float_mode): New typedef.
18536 * machmode.h (float_mode_for_size): New function.
18537 * emit-rtl.c (double_mode): Delete.
18538 (init_emit_once): Use float_mode_for_size.
18539 * stor-layout.c (layout_type): Likewise.
18540 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
18542 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18543 Alan Hayward <alan.hayward@arm.com>
18544 David Sherwood <david.sherwood@arm.com>
18546 * output.h (assemble_real): Take a scalar_float_mode.
18547 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
18548 * config/arm/arm.md (consttable_4): Likewise.
18549 (consttable_8): Likewise.
18550 (consttable_16): Likewise.
18551 * config/mips/mips.md (consttable_float): Likewise.
18552 * config/s390/s390.c (s390_output_pool_entry): Likewise.
18553 * varasm.c (assemble_real): Take a scalar_float_mode.
18554 (output_constant_pool_2): Update accordingly.
18555 (output_constant): Likewise.
18557 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18558 Alan Hayward <alan.hayward@arm.com>
18559 David Sherwood <david.sherwood@arm.com>
18561 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
18562 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
18563 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
18564 (native_encode_real): Likewise.
18565 (native_interpret_real): Likewise.
18566 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
18567 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18569 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18570 Alan Hayward <alan.hayward@arm.com>
18571 David Sherwood <david.sherwood@arm.com>
18573 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
18574 <scalar_float_mode>. Simplify.
18575 (gen_extend_conv_libfunc): Likewise.
18577 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18578 Alan Hayward <alan.hayward@arm.com>
18579 David Sherwood <david.sherwood@arm.com>
18581 * coretypes.h (scalar_float_mode): New type.
18582 * machmode.h (mode_traits::from_int): Use machine_mode if
18583 USE_ENUM_MODES is defined.
18584 (is_a): New function.
18586 (dyn_cast): Likewise.
18587 (scalar_float_mode): New class.
18588 (scalar_float_mode::includes_p): New function.
18589 (is_float_mode): Likewise.
18590 * gdbhooks.py (MachineModePrinter): New class.
18591 (build_pretty_printer): Use it for scalar_float_mode.
18592 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
18593 (format_helper::format_helper): Turn into a template.
18594 * genmodes.c (get_mode_class): New function.
18595 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
18596 or machine_mode if none.
18597 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
18598 as_a <scalar_float_mode>.
18599 * dwarf2out.c (mem_loc_descriptor): Likewise.
18600 (insert_float): Likewise.
18601 (add_const_value_attribute): Likewise.
18602 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18603 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
18604 (expand_unop): Update accordingly.
18605 (expand_abs_nojump): Likewise.
18606 (expand_copysign_absneg): Take a scalar_float_mode.
18607 (expand_copysign_bit): Likewise.
18608 (expand_copysign): Update accordingly.
18610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18611 Alan Hayward <alan.hayward@arm.com>
18612 David Sherwood <david.sherwood@arm.com>
18614 * coretypes.h (opt_mode): New class.
18615 * machmode.h (opt_mode): Likewise.
18616 (opt_mode::else_void): New function.
18617 (opt_mode::require): Likewise.
18618 (opt_mode::exists): Likewise.
18619 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
18620 (GET_MODE_2XWIDER_MODE): Likewise.
18621 (mode_iterator::get_wider): Update accordingly.
18622 (mode_iterator::get_2xwider): Likewise.
18623 (mode_iterator::get_known_wider): Likewise, turning into a template.
18624 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
18625 forcing a wider mode to exist.
18626 * config/cr16/cr16.h (LONG_REG_P): Likewise.
18627 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18628 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
18629 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
18630 * lower-subreg.c (init_lower_subreg): Likewise.
18631 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
18632 on the final iteration.
18633 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
18634 a wider mode exists before asking for a move pattern.
18635 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
18636 forcing a wider mode to exist.
18637 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
18638 returning false if no such mode exists.
18639 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
18640 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
18641 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
18642 Avoid checking for a MODE_INT if we already know the mode is not a
18644 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
18645 forcing a wider mode to exist.
18646 (expmed_mult_highpart_optab): Likewise.
18647 (expmed_mult_highpart): Likewise.
18648 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
18650 * lto-streamer-in.c (lto_input_mode_table): Likewise.
18651 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
18652 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
18653 * internal-fn.c (expand_mul_overflow): Update use of
18654 GET_MODE_2XWIDER_MODE.
18655 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18656 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
18657 GET_MODE_WIDER_MODE.
18658 (convert_plusminus_to_widen): Likewise.
18659 * tree-switch-conversion.c (array_value_type): Likewise.
18660 * var-tracking.c (emit_note_insn_var_location): Likewise.
18661 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18662 Return false inside rather than outside the loop if no wider mode
18664 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
18665 and GET_MODE_2XWIDER_MODE
18666 (can_compare_p): Use else_void.
18667 * gdbhooks.py (OptMachineModePrinter): New class.
18668 (build_pretty_printer): Use it for opt_mode.
18670 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18671 Alan Hayward <alan.hayward@arm.com>
18672 David Sherwood <david.sherwood@arm.com>
18674 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
18675 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
18677 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18678 Alan Hayward <alan.hayward@arm.com>
18679 David Sherwood <david.sherwood@arm.com>
18681 * machmode.h (mode_traits): New structure.
18682 (get_narrowest_mode): New function.
18683 (mode_iterator::start): Likewise.
18684 (mode_iterator::iterate_p): Likewise.
18685 (mode_iterator::get_wider): Likewise.
18686 (mode_iterator::get_known_wider): Likewise.
18687 (mode_iterator::get_2xwider): Likewise.
18688 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
18689 (FOR_EACH_MODE): Likewise.
18690 (FOR_EACH_MODE_FROM): Likewise.
18691 (FOR_EACH_MODE_UNTIL): Likewise.
18692 (FOR_EACH_WIDER_MODE): Likewise.
18693 (FOR_EACH_2XWIDER_MODE): Likewise.
18694 * builtins.c (expand_builtin_strlen): Use new mode iterators.
18695 * combine.c (simplify_comparison): Likewise
18696 * config/i386/i386.c (type_natural_mode): Likewise.
18697 * cse.c (cse_insn): Likewise.
18698 * dse.c (find_shift_sequence): Likewise.
18699 * emit-rtl.c (init_derived_machine_modes): Likewise.
18700 (init_emit_once): Likewise.
18701 * explow.c (hard_function_value): Likewise.
18702 * expmed.c (extract_fixed_bit_field_1): Likewise.
18703 (extract_bit_field_1): Likewise.
18704 (expand_divmod): Likewise.
18705 (emit_store_flag_1): Likewise.
18706 * expr.c (init_expr_target): Likewise.
18707 (convert_move): Likewise.
18708 (alignment_for_piecewise_move): Likewise.
18709 (widest_int_mode_for_size): Likewise.
18710 (emit_block_move_via_movmem): Likewise.
18711 (copy_blkmode_to_reg): Likewise.
18712 (set_storage_via_setmem): Likewise.
18713 (compress_float_constant): Likewise.
18714 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18715 * optabs-query.c (get_best_extraction_insn): Likewise.
18716 * optabs.c (expand_binop): Likewise.
18717 (expand_twoval_unop): Likewise.
18718 (expand_twoval_binop): Likewise.
18719 (widen_leading): Likewise.
18720 (widen_bswap): Likewise.
18721 (expand_parity): Likewise.
18722 (expand_unop): Likewise.
18723 (prepare_cmp_insn): Likewise.
18724 (prepare_float_lib_cmp): Likewise.
18725 (expand_float): Likewise.
18726 (expand_fix): Likewise.
18727 (expand_sfix_optab): Likewise.
18728 * postreload.c (move2add_use_add2_insn): Likewise.
18729 * reg-stack.c (reg_to_stack): Likewise.
18730 * reginfo.c (choose_hard_reg_mode): Likewise.
18731 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18732 * stor-layout.c (mode_for_size): Likewise.
18733 (smallest_mode_for_size): Likewise.
18734 (mode_for_vector): Likewise.
18735 (finish_bitfield_representative): Likewise.
18736 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
18737 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
18738 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18739 * var-tracking.c (prepare_call_arguments): Likewise.
18741 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18742 Alan Hayward <alan.hayward@arm.com>
18743 David Sherwood <david.sherwood@arm.com>
18745 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
18746 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
18747 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
18748 * machmode.h (mode_size): Move earlier in file.
18749 (mode_precision): Likewise.
18750 (mode_inner): Likewise.
18751 (mode_nunits): Likewise.
18752 (mode_unit_size): Likewise.
18753 (unit_unit_precision): Likewise.
18754 (mode_wider): Likewise.
18755 (mode_2xwider): Likewise.
18756 (machine_mode): New class.
18757 (mode_to_bytes): New function.
18758 (mode_to_bits): Likewise.
18759 (mode_to_precision): Likewise.
18760 (mode_to_inner): Likewise.
18761 (mode_to_unit_size): Likewise.
18762 (mode_to_unit_precision): Likewise.
18763 (mode_to_nunits): Likewise.
18764 (GET_MODE_SIZE): Use mode_to_bytes.
18765 (GET_MODE_BITSIZE): Use mode_to_bits.
18766 (GET_MODE_PRECISION): Use mode_to_precision.
18767 (GET_MODE_INNER): Use mode_to_inner.
18768 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
18769 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
18770 (GET_MODE_NUNITS): Use mode_to_nunits.
18771 * system.h (ALWAYS_INLINE): New macro.
18772 * config/powerpcspe/powerpcspe-c.c
18773 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
18774 int for arg1_mode and arg2_mode.
18776 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18777 Alan Hayward <alan.hayward@arm.com>
18778 David Sherwood <david.sherwood@arm.com>
18780 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18781 Prefix mode names with E_ in case statements.
18782 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18783 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
18784 (aarch64_split_simd_move): Likewise.
18785 (aarch64_gen_storewb_pair): Likewise.
18786 (aarch64_gen_loadwb_pair): Likewise.
18787 (aarch64_gen_store_pair): Likewise.
18788 (aarch64_gen_load_pair): Likewise.
18789 (aarch64_get_condition_code_1): Likewise.
18790 (aarch64_constant_pool_reload_icode): Likewise.
18791 (get_rsqrte_type): Likewise.
18792 (get_rsqrts_type): Likewise.
18793 (get_recpe_type): Likewise.
18794 (get_recps_type): Likewise.
18795 (aarch64_gimplify_va_arg_expr): Likewise.
18796 (aarch64_simd_container_mode): Likewise.
18797 (aarch64_emit_load_exclusive): Likewise.
18798 (aarch64_emit_store_exclusive): Likewise.
18799 (aarch64_expand_compare_and_swap): Likewise.
18800 (aarch64_gen_atomic_cas): Likewise.
18801 (aarch64_emit_bic): Likewise.
18802 (aarch64_emit_atomic_swap): Likewise.
18803 (aarch64_emit_atomic_load_op): Likewise.
18804 (aarch64_evpc_trn): Likewise.
18805 (aarch64_evpc_uzp): Likewise.
18806 (aarch64_evpc_zip): Likewise.
18807 (aarch64_evpc_ext): Likewise.
18808 (aarch64_evpc_rev): Likewise.
18809 (aarch64_evpc_dup): Likewise.
18810 (aarch64_gen_ccmp_first): Likewise.
18811 (aarch64_gen_ccmp_next): Likewise.
18812 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18813 (alpha_emit_xfloating_libcall): Likewise.
18814 (emit_insxl): Likewise.
18815 (alpha_arg_type): Likewise.
18816 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
18817 (arc_preferred_simd_mode): Likewise.
18818 (arc_secondary_reload): Likewise.
18819 (get_arc_condition_code): Likewise.
18820 (arc_print_operand): Likewise.
18821 (arc_legitimate_constant_p): Likewise.
18822 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18823 * config/arc/arc.md (casesi_load): Likewise.
18824 (casesi_compact_jump): Likewise.
18825 * config/arc/predicates.md (proper_comparison_operator): Likewise.
18826 (cc_use_register): Likewise.
18827 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18828 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18829 (arm_init_iwmmxt_builtins): Likewise.
18830 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
18831 (neon_expand_vector_init): Likewise.
18832 (arm_attr_length_move_neon): Likewise.
18833 (maybe_get_arm_condition_code): Likewise.
18834 (arm_emit_vector_const): Likewise.
18835 (arm_preferred_simd_mode): Likewise.
18836 (arm_output_iwmmxt_tinsr): Likewise.
18837 (thumb1_output_casesi): Likewise.
18838 (thumb2_output_casesi): Likewise.
18839 (arm_emit_load_exclusive): Likewise.
18840 (arm_emit_store_exclusive): Likewise.
18841 (arm_expand_compare_and_swap): Likewise.
18842 (arm_evpc_neon_vuzp): Likewise.
18843 (arm_evpc_neon_vzip): Likewise.
18844 (arm_evpc_neon_vrev): Likewise.
18845 (arm_evpc_neon_vtrn): Likewise.
18846 (arm_evpc_neon_vext): Likewise.
18847 (arm_validize_comparison): Likewise.
18848 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
18849 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
18850 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
18851 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
18852 (c6x_preferred_simd_mode): Likewise.
18853 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
18854 (epiphany_rtx_costs): Likewise.
18855 * config/epiphany/predicates.md (proper_comparison_operator):
18857 * config/frv/frv.c (condexec_memory_operand): Likewise.
18858 (frv_emit_move): Likewise.
18859 (output_move_single): Likewise.
18860 (output_condmove_single): Likewise.
18861 (frv_hard_regno_mode_ok): Likewise.
18862 (frv_matching_accg_mode): Likewise.
18863 * config/h8300/h8300.c (split_adds_subs): Likewise.
18864 (h8300_rtx_costs): Likewise.
18865 (h8300_print_operand): Likewise.
18866 (compute_mov_length): Likewise.
18867 (output_logical_op): Likewise.
18868 (compute_logical_op_length): Likewise.
18869 (compute_logical_op_cc): Likewise.
18870 (h8300_shift_needs_scratch_p): Likewise.
18871 (output_a_shift): Likewise.
18872 (compute_a_shift_length): Likewise.
18873 (compute_a_shift_cc): Likewise.
18874 (expand_a_rotate): Likewise.
18875 (output_a_rotate): Likewise.
18876 * config/i386/i386.c (classify_argument): Likewise.
18877 (function_arg_advance_32): Likewise.
18878 (function_arg_32): Likewise.
18879 (function_arg_64): Likewise.
18880 (function_value_64): Likewise.
18881 (ix86_gimplify_va_arg): Likewise.
18882 (ix86_legitimate_constant_p): Likewise.
18883 (put_condition_code): Likewise.
18884 (split_double_mode): Likewise.
18885 (ix86_avx256_split_vector_move_misalign): Likewise.
18886 (ix86_expand_vector_logical_operator): Likewise.
18887 (ix86_split_idivmod): Likewise.
18888 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
18889 (ix86_build_const_vector): Likewise.
18890 (ix86_build_signbit_mask): Likewise.
18891 (ix86_match_ccmode): Likewise.
18892 (ix86_cc_modes_compatible): Likewise.
18893 (ix86_expand_branch): Likewise.
18894 (ix86_expand_sse_cmp): Likewise.
18895 (ix86_expand_sse_movcc): Likewise.
18896 (ix86_expand_int_sse_cmp): Likewise.
18897 (ix86_expand_vec_perm_vpermi2): Likewise.
18898 (ix86_expand_vec_perm): Likewise.
18899 (ix86_expand_sse_unpack): Likewise.
18900 (ix86_expand_int_addcc): Likewise.
18901 (ix86_split_to_parts): Likewise.
18902 (ix86_vectorize_builtin_gather): Likewise.
18903 (ix86_vectorize_builtin_scatter): Likewise.
18904 (avx_vpermilp_parallel): Likewise.
18905 (inline_memory_move_cost): Likewise.
18906 (ix86_tieable_integer_mode_p): Likewise.
18907 (x86_maybe_negate_const_int): Likewise.
18908 (ix86_expand_vector_init_duplicate): Likewise.
18909 (ix86_expand_vector_init_one_nonzero): Likewise.
18910 (ix86_expand_vector_init_one_var): Likewise.
18911 (ix86_expand_vector_init_concat): Likewise.
18912 (ix86_expand_vector_init_interleave): Likewise.
18913 (ix86_expand_vector_init_general): Likewise.
18914 (ix86_expand_vector_set): Likewise.
18915 (ix86_expand_vector_extract): Likewise.
18916 (emit_reduc_half): Likewise.
18917 (ix86_emit_i387_round): Likewise.
18918 (ix86_mangle_type): Likewise.
18919 (ix86_expand_round_sse4): Likewise.
18920 (expand_vec_perm_blend): Likewise.
18921 (canonicalize_vector_int_perm): Likewise.
18922 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
18923 (expand_vec_perm_1): Likewise.
18924 (expand_vec_perm_interleave3): Likewise.
18925 (expand_vec_perm_even_odd_pack): Likewise.
18926 (expand_vec_perm_even_odd_1): Likewise.
18927 (expand_vec_perm_broadcast_1): Likewise.
18928 (ix86_vectorize_vec_perm_const_ok): Likewise.
18929 (ix86_expand_vecop_qihi): Likewise.
18930 (ix86_expand_mul_widen_hilo): Likewise.
18931 (ix86_expand_sse2_abs): Likewise.
18932 (ix86_expand_pextr): Likewise.
18933 (ix86_expand_pinsr): Likewise.
18934 (ix86_preferred_simd_mode): Likewise.
18935 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
18936 * config/i386/sse.md (*andnot<mode>3): Likewise.
18937 (<mask_codefor><code><mode>3<mask_name>): Likewise.
18938 (*<code><mode>3): Likewise.
18939 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
18940 (ia64_expand_atomic_op): Likewise.
18941 (ia64_arg_type): Likewise.
18942 (ia64_mode_to_int): Likewise.
18943 (ia64_scalar_mode_supported_p): Likewise.
18944 (ia64_vector_mode_supported_p): Likewise.
18945 (expand_vec_perm_broadcast): Likewise.
18946 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
18947 (iq2000_function_arg_advance): Likewise.
18948 (iq2000_function_arg): Likewise.
18949 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
18950 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
18951 (m68k_libcall_value): Likewise.
18952 (m68k_function_value): Likewise.
18953 (sched_attr_op_type): Likewise.
18954 * config/mcore/mcore.c (mcore_output_move): Likewise.
18955 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
18957 (microblaze_function_arg): Likewise.
18958 * config/mips/mips.c (mips16_build_call_stub): Likewise.
18959 (mips_print_operand): Likewise.
18960 (mips_mode_ok_for_mov_fmt_p): Likewise.
18961 (mips_vector_mode_supported_p): Likewise.
18962 (mips_preferred_simd_mode): Likewise.
18963 (mips_expand_vpc_loongson_even_odd): Likewise.
18964 (mips_expand_vec_unpack): Likewise.
18965 (mips_expand_vi_broadcast): Likewise.
18966 (mips_expand_vector_init): Likewise.
18967 (mips_expand_vec_reduc): Likewise.
18968 (mips_expand_msa_cmp): Likewise.
18969 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
18970 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
18971 (cc_flags_for_mode): Likewise.
18972 * config/msp430/msp430.c (msp430_print_operand): Likewise.
18973 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
18974 (nds32_output_casesi_pc_relative): Likewise.
18975 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18976 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
18977 (nvptx_gen_unpack): Likewise.
18978 (nvptx_gen_pack): Likewise.
18979 (nvptx_gen_shuffle): Likewise.
18980 (nvptx_gen_wcast): Likewise.
18981 (nvptx_preferred_simd_mode): Likewise.
18982 * config/pa/pa.c (pa_secondary_reload): Likewise.
18983 * config/pa/predicates.md (base14_operand): Likewise.
18984 * config/powerpcspe/powerpcspe-c.c
18985 (altivec_resolve_overloaded_builtin): Likewise.
18986 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
18988 (rs6000_preferred_simd_mode): Likewise.
18989 (output_vec_const_move): Likewise.
18990 (rs6000_expand_vector_extract): Likewise.
18991 (rs6000_split_vec_extract_var): Likewise.
18992 (reg_offset_addressing_ok_p): Likewise.
18993 (rs6000_legitimate_offset_address_p): Likewise.
18994 (rs6000_legitimize_address): Likewise.
18995 (rs6000_emit_set_const): Likewise.
18996 (rs6000_const_vec): Likewise.
18997 (rs6000_emit_move): Likewise.
18998 (spe_build_register_parallel): Likewise.
18999 (rs6000_darwin64_record_arg_recurse): Likewise.
19000 (swap_selector_for_mode): Likewise.
19001 (spe_init_builtins): Likewise.
19002 (paired_init_builtins): Likewise.
19003 (altivec_init_builtins): Likewise.
19004 (do_load_for_compare): Likewise.
19005 (rs6000_generate_compare): Likewise.
19006 (rs6000_expand_float128_convert): Likewise.
19007 (emit_load_locked): Likewise.
19008 (emit_store_conditional): Likewise.
19009 (rs6000_output_function_epilogue): Likewise.
19010 (rs6000_handle_altivec_attribute): Likewise.
19011 (rs6000_function_value): Likewise.
19012 (emit_fusion_gpr_load): Likewise.
19013 (emit_fusion_p9_load): Likewise.
19014 (emit_fusion_p9_store): Likewise.
19015 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
19016 (fusion_gpr_mem_load): Likewise.
19017 (fusion_addis_mem_combo_load): Likewise.
19018 (fusion_addis_mem_combo_store): Likewise.
19019 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
19020 (fusion_gpr_mem_load): Likewise.
19021 (fusion_addis_mem_combo_load): Likewise.
19022 (fusion_addis_mem_combo_store): Likewise.
19023 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19025 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
19026 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19027 (rs6000_preferred_simd_mode): Likewise.
19028 (output_vec_const_move): Likewise.
19029 (rs6000_expand_vector_extract): Likewise.
19030 (rs6000_split_vec_extract_var): Likewise.
19031 (reg_offset_addressing_ok_p): Likewise.
19032 (rs6000_legitimate_offset_address_p): Likewise.
19033 (rs6000_legitimize_address): Likewise.
19034 (rs6000_emit_set_const): Likewise.
19035 (rs6000_const_vec): Likewise.
19036 (rs6000_emit_move): Likewise.
19037 (rs6000_darwin64_record_arg_recurse): Likewise.
19038 (swap_selector_for_mode): Likewise.
19039 (paired_init_builtins): Likewise.
19040 (altivec_init_builtins): Likewise.
19041 (rs6000_expand_float128_convert): Likewise.
19042 (emit_load_locked): Likewise.
19043 (emit_store_conditional): Likewise.
19044 (rs6000_output_function_epilogue): Likewise.
19045 (rs6000_handle_altivec_attribute): Likewise.
19046 (rs6000_function_value): Likewise.
19047 (emit_fusion_gpr_load): Likewise.
19048 (emit_fusion_p9_load): Likewise.
19049 (emit_fusion_p9_store): Likewise.
19050 * config/rx/rx.c (rx_gen_move_template): Likewise.
19051 (flags_from_mode): Likewise.
19052 * config/s390/predicates.md (s390_alc_comparison): Likewise.
19053 (s390_slb_comparison): Likewise.
19054 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
19055 (s390_vector_mode_supported_p): Likewise.
19056 (s390_cc_modes_compatible): Likewise.
19057 (s390_match_ccmode_set): Likewise.
19058 (s390_canonicalize_comparison): Likewise.
19059 (s390_emit_compare_and_swap): Likewise.
19060 (s390_branch_condition_mask): Likewise.
19061 (s390_rtx_costs): Likewise.
19062 (s390_secondary_reload): Likewise.
19063 (__SECONDARY_RELOAD_CASE): Likewise.
19064 (s390_expand_cs): Likewise.
19065 (s390_preferred_simd_mode): Likewise.
19066 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
19067 * config/sh/sh.c (sh_print_operand): Likewise.
19068 (dump_table): Likewise.
19069 (sh_secondary_reload): Likewise.
19070 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19071 * config/sh/sh.md (casesi_worker_1): Likewise.
19072 (casesi_worker_2): Likewise.
19073 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
19074 (fcc_comparison_operator): Likewise.
19075 * config/sparc/sparc.c (sparc_expand_move): Likewise.
19076 (emit_soft_tfmode_cvt): Likewise.
19077 (sparc_preferred_simd_mode): Likewise.
19078 (output_cbranch): Likewise.
19079 (sparc_print_operand): Likewise.
19080 (sparc_expand_vec_perm_bmask): Likewise.
19081 (vector_init_bshuffle): Likewise.
19082 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
19083 (spu_vector_mode_supported_p): Likewise.
19084 (spu_expand_insv): Likewise.
19085 (spu_emit_branch_or_set): Likewise.
19086 (spu_handle_vector_attribute): Likewise.
19087 (spu_builtin_splats): Likewise.
19088 (spu_builtin_extract): Likewise.
19089 (spu_builtin_promote): Likewise.
19090 (spu_expand_sign_extend): Likewise.
19091 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
19092 (tilegx_simd_int): Likewise.
19093 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
19094 (tilepro_simd_int): Likewise.
19095 * config/v850/v850.c (const_double_split): Likewise.
19096 (v850_print_operand): Likewise.
19097 (ep_memory_offset): Likewise.
19098 * config/vax/vax.c (vax_rtx_costs): Likewise.
19099 (vax_output_int_move): Likewise.
19100 (vax_output_int_add): Likewise.
19101 (vax_output_int_subtract): Likewise.
19102 * config/visium/predicates.md (visium_branch_operator): Likewise.
19103 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
19104 (visium_print_operand_address): Likewise.
19105 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19106 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
19107 (xtensa_expand_conditional_branch): Likewise.
19108 (xtensa_copy_incoming_a7): Likewise.
19109 (xtensa_output_literal): Likewise.
19110 * dfp.c (decimal_real_maxval): Likewise.
19111 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19113 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19114 Alan Hayward <alan.hayward@arm.com>
19115 David Sherwood <david.sherwood@arm.com>
19117 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
19118 (mode_nunits_inline): Likewise.
19119 (mode_inner_inline): Likewise.
19120 (mode_unit_size_inline): Likewise.
19121 (mode_unit_precision_inline): Likewise.
19122 (emit_insn_modes_h): Likewise. Also emit a #define of the
19124 (emit_mode_wider): Add an E_ prefix to mode names.
19125 (emit_mode_complex): Likewise.
19126 (emit_mode_inner): Likewise.
19127 (emit_mode_adjustments): Likewise.
19128 (emit_mode_int_n): Likewise.
19129 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
19130 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
19131 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
19132 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
19133 (CRC32_BUILTIN, ENTRY): Likewise.
19134 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
19135 (aarch64_pop_regs): Likewise.
19136 (aarch64_process_components): Likewise.
19137 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
19138 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
19139 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
19140 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
19141 * config/arm/arm.c (arm_init_libfuncs): Likewise.
19142 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
19144 * config/i386/i386-builtin.def (pcmpestr): Likewise.
19145 (pcmpistr): Likewise.
19146 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
19147 * config/mmix/mmix.c (mmix_output_condition): Likewise.
19148 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19150 * config/rl78/rl78.c (mduc_regs): Likewise.
19151 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19152 (htm_expand_builtin): Likewise.
19153 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
19154 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
19155 * config/xtensa/xtensa.c (print_operand): Likewise.
19156 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
19157 (NUM_MODE_VECTOR_INT): Likewise.
19158 * genoutput.c (null_operand): Likewise.
19159 (output_operand_data): Likewise.
19160 * genrecog.c (print_parameter_value): Likewise.
19161 * lra.c (debug_operand_data): Likewise.
19163 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19165 * dwarf2out.c (scompare_loc_descriptor_wide)
19166 (scompare_loc_descriptor_narrow): New functions, split out from...
19167 (scompare_loc_descriptor): ...here.
19168 * expmed.c (emit_store_flag_int): New function, split out from...
19169 (emit_store_flag): ...here.
19171 2017-08-30 Richard Biener <rguenther@suse.de>
19173 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
19174 (dwarf2out_early_finish): Move setting of AT_pubnames from
19175 early debug output to early finish.
19177 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19179 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
19180 and -mdata-region to the assembler.
19182 2017-08-30 Richard Biener <rguenther@suse.de>
19184 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
19186 (gen_subprogram_die): Add DW_AT_object_pointer only early.
19187 (dwarf2out_early_global_decl): Only generate a DIE for the
19188 abstract origin if it doesn't already exist or is a declaration DIE.
19189 (resolve_addr): Do not add the linkage name twice when
19190 generating a stub DIE for the DW_TAG_GNU_call_site target.
19192 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19194 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19195 Use machine_mode rather than int for arg1_mode.
19197 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19200 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
19201 that the second argument of the built-in functions to unpack
19202 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
19203 switch statement instead a lot of if statements.
19204 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
19205 Allow 64-bit values to be in Altivec registers as well as
19206 traditional floating point registers.
19207 (pack<mode>, FMOVE128_VSX iterator): Likewise.
19209 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
19211 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
19212 MAX_REGS_PER_ADDRESS == 1.
19214 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
19216 * config/i386/i386.opt (flag_fentry): Do not init to -1.
19217 * config/i386/i386.c (ix86_option_override_internal): Simplify
19218 setting of opts->x_flag_entry.
19220 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19221 Jakub Jelinek <jakub@redhat.com>
19222 Richard Biener <rguenther@suse.de>
19224 PR tree-optimization/81503
19225 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
19226 folded constant fits in the target type; reorder tests for clarity.
19228 2017-08-29 Martin Liska <mliska@suse.cz>
19230 * passes.def: Include pass_lower_switch.
19231 * stmt.c (dump_case_nodes): Remove and move to
19232 tree-switch-conversion.
19233 (case_values_threshold): Likewise.
19234 (expand_switch_as_decision_tree_p): Likewise.
19235 (emit_case_decision_tree): Likewise.
19236 (expand_case): Likewise.
19237 (balance_case_nodes): Likewise.
19238 (node_has_low_bound): Likewise.
19239 (node_has_high_bound): Likewise.
19240 (node_is_bounded): Likewise.
19241 (emit_case_nodes): Likewise.
19242 (struct simple_case_node): New struct.
19243 (add_case_node): Remove.
19244 (emit_case_dispatch_table): Use vector instead of case_list.
19245 (reset_out_edges_aux): Remove.
19246 (compute_cases_per_edge): Likewise.
19247 (expand_case): Build list of simple_case_node.
19248 (expand_sjlj_dispatch_table): Use it.
19249 * tree-switch-conversion.c (struct case_node): Moved from
19250 stmt.c and adjusted.
19251 (emit_case_nodes): Likewise.
19252 (node_has_low_bound): Likewise.
19253 (node_has_high_bound): Likewise.
19254 (node_is_bounded): Likewise.
19255 (case_values_threshold): Likewise.
19256 (reset_out_edges_aux): Likewise.
19257 (compute_cases_per_edge): Likewise.
19258 (add_case_node): Likewise.
19259 (dump_case_nodes): Likewise.
19260 (balance_case_nodes): Likewise.
19261 (expand_switch_as_decision_tree_p): Likewise.
19262 (emit_jump): Likewise.
19263 (emit_case_decision_tree): Likewise.
19264 (try_switch_expansion): Likewise.
19265 (do_jump_if_equal): Likewise.
19266 (emit_cmp_and_jump_insns): Likewise.
19267 (fix_phi_operands_for_edge): New function.
19268 (record_phi_operand_mapping): Likewise.
19269 (class pass_lower_switch): New pass.
19270 (pass_lower_switch::execute): New function.
19271 (make_pass_lower_switch): Likewise.
19272 (conditional_probability):
19273 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
19274 * tree-pass.h: Add make_pass_lower_switch.
19276 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19279 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
19282 2017-08-29 Richard Biener <rguenther@suse.de>
19284 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
19285 we do not add a DW_AT_inline attribute twice.
19286 (gen_subprogram_die): Remove code setting DW_AT_inline on
19287 DECL_ABSTRACT_P nodes.
19289 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
19291 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
19292 calls to internal functions.
19293 (gimplify_modify_expr): Likewise.
19294 * tree-call-cdce.c (use_internal_fn): Likewise.
19295 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
19296 (convert_to_divmod): Set the nothrow flag.
19297 * tree-if-conv.c (predicate_mem_writes): Likewise.
19298 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19299 (vectorizable_call): Likewise.
19300 (vectorizable_store): Likewise.
19301 (vectorizable_load): Likewise.
19302 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
19303 (vect_recog_mask_conversion_pattern): Likewise.
19305 2017-08-29 Martin Liska <mliska@suse.cz>
19308 * gcc.c (driver_handle_option): Add new argument.
19309 * opts-common.c (handle_option): Pass
19310 target_option_override_hook.
19311 * opts-global.c (lang_handle_option): Add new option.
19312 (set_default_handlers): Add new argument.
19313 (decode_options): Likewise.
19314 * opts.c (target_handle_option): Likewise.
19315 (common_handle_option): Call target_option_override_hook.
19316 * opts.h (struct cl_option_handler_func): Add hook for
19317 target option override.
19318 (struct cl_option_handlers): Likewise.
19319 (set_default_handlers): Add new argument.
19320 (decode_options): Likewise.
19321 (common_handle_option): Likewise.
19322 (target_handle_option): Likewise.
19323 * toplev.c (toplev::main): Pass targetm.target_option.override
19326 2017-08-29 Richard Biener <rguenther@suse.de>
19327 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
19329 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
19330 life to the active subtree.
19332 2017-08-28 Jeff Law <law@redhat.com>
19334 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
19335 derive_equivalences.
19336 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
19337 Code moved into....
19338 (edge_info::derive_equivalences): New private member function
19340 * tree-ssa-dom.c (class edge_info): Changed from a struct
19341 to a class. Add ctor/dtor, methods and data members.
19342 (edge_info::edge_info): Renamed from allocate_edge_info.
19343 Initialize additional members.
19344 (edge_info::~edge_info): New.
19345 (free_dom_edge_info): Delete the edge info.
19346 (record_edge_info): Use new class & associated member functions.
19347 Tighten forms for testing for edge equivalences.
19348 (record_temporary_equivalences): Iterate over the simple
19349 equivalences rather than assuming there's only one per edge.
19350 (cprop_into_successor_phis): Iterate over the simple
19351 equivalences rather than assuming there's only one per edge.
19352 (optimize_stmt): Use operand_equal_p rather than pointer
19353 equality for mini-DSE code.
19355 2017-08-28 Nathan Sidwell <nathan@acm.org>
19357 * gcc.c (execute): Fold SIGPIPE handling into switch
19358 statement. Adjust internal error message.
19360 2017-08-28 Richard Biener <rguenther@suse.de>
19363 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
19364 Do nothing for removed DIEs.
19366 2017-08-28 Richard Biener <rguenther@suse.de>
19368 PR tree-optimization/81977
19369 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
19372 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
19375 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
19376 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
19377 using targetm.gen_mem_thread_fence.
19379 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
19382 * config/i386/i386.md (*<btsc><mode>): Change operand 2
19383 predicate to register_operand. Reorder operands.
19384 (*btr<mode>): Ditto.
19385 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
19386 (*btr<mode>_mask): Ditto.
19388 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
19390 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
19391 * config/rs6000/xmmintrin.h: New file.
19392 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
19394 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19397 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
19398 parameter and_insn and return it.
19399 (recombine_lvx_pattern): Insert a copy to ensure availability of
19400 the base register of the copied masking operation at the point of
19401 the instruction replacement.
19402 (recombine_stvx_pattern): Likewise.
19404 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
19406 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
19407 undocumented switches.
19408 (-mpower9-dform-vector): Likewise.
19409 (-mpower9-dform): Likewise.
19410 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19411 comments to delete references to -mpower9-dform* switches.
19412 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
19413 Delete reference to -mpower9-dform* switches, test for
19414 -mpower9-vector instead.
19415 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
19416 (OTHER_P9_VECTOR_MASKS): Likewise.
19417 (POWERPC_MASKS): Likewise.
19418 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
19419 tests against -mpower9-dform* to -mpower9-vector. Delete code
19420 that checked for -mpower9-dform* consistancy with other options.
19421 Add test for -mpower9-misc to enable other power9 switches.
19422 (rs6000_init_hard_regno_mode_ok): Likewise.
19423 (rs6000_option_override_internal): Likewise.
19424 (rs6000_emit_prologue): Likewise.
19425 (rs6000_emit_epilogue): Likewise.
19426 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
19427 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
19428 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
19430 (emit_fusion_p9_store): Likewise.
19431 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
19432 resetting these macros if the assembler does not support ISA 3.0
19434 (TARGET_P9_DFORM_VECTOR): Likewise.
19435 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
19436 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
19438 2017-08-25 Alan Modra <amodra@gmail.com>
19440 PR rtl-optimization/81747
19441 * cse.c (cse_extended_basic_block): Don't attempt to record
19442 equivalences for degenerate conditional branches that branch
19443 to their fall-through.
19445 2017-08-24 Martin Sebor <msebor@redhat.com>
19447 PR middle-end/81908
19448 * gimple-fold.c (size_must_be_zero_p): New function.
19449 (gimple_fold_builtin_memory_op): Call it.
19451 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
19453 * config/rs6000/mm_malloc.h: New file.
19455 2017-08-24 Bin Cheng <bin.cheng@arm.com>
19457 PR tree-optimization/81913
19458 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
19459 analysis when either IVs in condition can wrap.
19461 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
19463 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
19464 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
19466 2017-08-24 Richard Biener <rguenther@suse.de>
19469 * targhooks.c (default_target_can_inline_p): Properly
19470 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
19471 is present and always compare.
19472 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
19473 infer -mfpmath=sse from TARGET_SSE_P.
19474 (ix86_can_inline_p): Properly use target_option_default_node when
19475 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
19477 2017-08-24 Richard Biener <rguenther@suse.de>
19480 * dwarf2out.c (output_die): Handle flag_generate_offload like
19482 (output_comp_unit): Likewise.
19483 (gen_array_type_die): Likewise.
19484 (dwarf2out_early_finish): Likewise.
19485 (note_variable_value_in_expr): Likewise.
19486 (dwarf2out_finish): Likewise. Adjust assert.
19487 * cgraphunit.c (symbol_table::compile): Move setting of
19488 flag_generate_offload earlier ...
19489 (symbol_table::finalize_compilation_unit): ... here, before
19490 early debug finalization.
19492 2017-08-24 Richard Biener <rguenther@suse.de>
19494 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
19495 and ipa-fnsummary.h.
19496 (ix86_can_inline_p): When ix86_fpmath flags do not match
19497 check whether the callee uses FP math at all.
19499 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
19501 PR middle-end/81931
19502 * tree-ssanames.c (get_nonzero_bits): Use element_precision
19503 instead of TYPE_PRECISION.
19505 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
19506 Alan Hayward <alan.hayward@arm.com>
19507 David Sherwood <david.sherwood@arm.com>
19509 * combine.c (make_extraction): Use subreg_offset_from_lsb.
19511 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
19513 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
19515 (ix86_frame::stack_realign_allocate): New field.
19516 (struct machine_frame_state): Modify comments.
19517 (machine_frame_state::sp_realigned_fp_end): New field.
19518 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
19519 layout calculation.
19520 (sp_valid_at): Add assertion to assure no attempt to access invalid
19521 offset of a realigned stack.
19522 (fp_valid_at): Likewise.
19523 (choose_baseaddr): Modify comments.
19524 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
19525 ix86_expand_prologue.
19526 (ix86_expand_prologue): Modify stack realignment and allocation.
19527 (ix86_expand_epilogue): Modify comments.
19528 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
19529 avx2_runtime, avx512f, and avx512f_runtime.
19531 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
19533 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
19534 (mstackrealign): Do not init to -1.
19535 * config/i386/i386.c (ix86_option_override_internal):
19536 Check opts_set, not opts when setting default value of
19537 opts->x_ix86_force_align_arg_pointer.
19539 2017-08-23 Richard Biener <rguenther@suse.de>
19541 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
19542 lang_hooks.decl_printable_name.
19543 * print-rtl-function.c (print_rtx_function): Likewise.
19544 * tree-pretty-print.c (dump_function_header): Likewise.
19546 2017-08-23 Richard Biener <rguenther@suse.de>
19549 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
19550 -g0 at compile-time.
19552 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19554 PR middle-end/19706
19555 * doc/sourcebuild.texi (Other hardware attributes):
19558 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19560 PR middle-end/19706
19561 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
19562 Fix single-use check.
19564 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
19566 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
19568 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19570 * doc/install.texi: Modify to add more details on running selected
19573 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19575 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
19576 is combined with -mabi=ms.
19577 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
19580 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19582 PR tree-optimization/81488
19583 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
19584 and cached_basis fields.
19585 (MAX_SPREAD): New constant.
19586 (alloc_cand_and_find_basis): Initialize new fields.
19587 (clear_visited): New function.
19588 (create_phi_basis_1): Rename from create_phi_basis, set visited
19589 and cached_basis fields.
19590 (create_phi_basis): New wrapper function.
19591 (phi_add_costs_1): Rename from phi_add_costs, add spread
19592 parameter, set visited field, short-circuit when limits reached.
19593 (phi_add_costs): New wrapper function.
19594 (record_phi_increments_1): Rename from record_phi_increments, set
19596 (record_phi_increments): New wrapper function.
19597 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
19598 (phi_incr_cost): New wrapper function.
19599 (all_phi_incrs_profitable_1): Rename from
19600 all_phi_incrs_profitable, set visited field.
19601 (all_phi_incrs_profitable): New wrapper function.
19603 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19604 Alan Hayward <alan.hayward@arm.com>
19605 David Sherwood <david.sherwood@arm.com>
19607 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
19608 that takes the outer and inner modes.
19609 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
19610 comparison as the canonical test for a paradoxical subreg.
19611 * combine.c (simplify_set): Use paradoxical_subreg_p.
19612 (make_extraction): Likewise.
19613 (force_to_mode): Likewise.
19614 (rtx_equal_for_field_assignment_p): Likewise.
19615 (gen_lowpart_for_combine): Likewise.
19616 (simplify_comparison): Likewise.
19617 * cse.c (equiv_constant): Likewise.
19618 * expmed.c (store_bit_field_1): Likewise.
19619 * final.c (alter_subreg): Likewise.
19620 * fwprop.c (propagate_rtx): Likewise.
19621 (forward_propagate_subreg): Likewise.
19622 * ira-conflicts.c (ira_build_conflicts): Likewise.
19623 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
19624 * lra-constraints.c (curr_insn_transform): Likewise.
19625 (split_reg): Likewise.
19626 * lra-eliminations.c (move_plus_up): Likewise.
19627 (lra_eliminate_regs_1): Likewise.
19628 * recog.c (general_operand): Likewise.
19629 * ree.c (combine_reaching_defs): Likewise.
19630 * reload.c (push_reload): Likewise.
19631 (find_reloads): Likewise.
19632 * reload1.c (elimination_effects): Likewise.
19633 (compute_reload_subreg_offset): Likewise.
19634 (choose_reload_regs): Likewise.
19635 * rtlanal.c (subreg_lsb_1): Likewise.
19636 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19637 (simplify_subreg): Likewise.
19638 * var-tracking.c (track_loc_p): Likewise.
19639 * emit-rtl.c (byte_lowpart_offset): Likewise.
19640 (paradoxical_subreg_p): Delete out-of-line definition.
19642 2017-08-22 Jeff Law <law@redhat.com>
19644 PR tree-optimization/81741
19645 PR tree-optimization/71947
19646 * tree-ssa-dom.c: Include tree-inline.h.
19647 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
19648 equivalences if one is more expensive to compute than the other.
19649 * tree-ssa-scopedtables.h (class const_or_copies): Make
19650 record_const_or_copy_raw method private.
19651 (class avail_exprs_stack): New method simplify_binary_operation.
19652 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
19653 avail_exprs_stack::simplify_binary_operation as needed.
19654 (avail_exprs_stack::simplify_binary_operation): New function.
19656 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
19658 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
19659 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
19660 (DOT_SYMBOLS): Likewise.
19661 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
19662 (RELOCATABLE_NEEDS_FIXUP): Likewise.
19663 (RS6000_ABI_NAME): Likewise.
19664 (TARGET_CMODEL): Likewise.
19665 (TOC_SECTION_ASM_OP): Likewise.
19666 (SET_CMODEL): New macro.
19667 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
19669 2017-08-22 Richard Biener <rguenther@suse.de>
19671 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
19672 to allow for free-lang-data replacements similar to verify_type_variant.
19674 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19675 Alan Hayward <alan.hayward@arm.com>
19676 David Sherwood <david.sherwood@arm.com>
19678 * config/aarch64/aarch64.md (casesi): Use DImode rather than
19679 VOIDmode for the LABEL_REF.
19681 2017-08-22 Richard Biener <rguenther@suse.de>
19683 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
19685 2017-08-22 Richard Biener <rguenther@suse.de>
19687 * common.opt (feliminate-dwarf2-dups): Ignore.
19688 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
19689 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
19690 same_die_p_wrap, compute_section_prefix,
19691 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
19692 (comdat_symbol_id, comdat_symbol_number): Likewise.
19693 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
19695 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
19696 (output_die): Mark unreachable path unreachable.
19697 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
19698 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
19699 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
19700 (dwarf2out_early_finish): Likewise.
19702 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
19704 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
19706 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
19709 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
19710 not VAR_P. Filter attribute warnings with OPT_Wattributes.
19711 (avr_attribute_table) <io, io_low, address>: Initialize
19712 .decl_required with true.
19714 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
19716 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
19717 undocumented debugging options.
19718 (-mvsx-scalar-double): Likewise.
19719 (-mallow-df-permute): Likewise.
19720 (-mvectorize-builtins): Likewise.
19721 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19722 (rs6000_builtin_vectorized_function): Likewise.
19723 (rs6000_builtin_md_vectorized_function): Likewise.
19724 (rs6000_opt_vars): Likewise.
19726 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
19729 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
19730 (*btrq_imm): Rename from *btrq.
19731 (*btcq_imm): Rename from *btcq.
19732 (btsc): New code attribute.
19733 (*<btsc><mode>): New insn pattern.
19734 (*btr<mode>): Ditto.
19735 (*<btsc><mode>_mask): New insn_and_split pattern.
19736 (*btr<mode>_mask): Ditto.
19738 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19739 Alan Hayward <alan.hayward@arm.com>
19740 David Sherwood <david.sherwood@arm.com>
19742 * function.c (pad_below): Simplify padding calculation.
19744 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19745 Alan Hayward <alan.hayward@arm.com>
19746 David Sherwood <david.sherwood@arm.com>
19748 * target.def (function_prologue): Remove frame size argument.
19749 (function_epilogue): Likewise.
19750 * doc/tm.texi: Regenerate.
19751 * final.c (final_start_function): Update call to function_prologue.
19752 (final_end_function): Update call to function_epilogue.
19753 (default_function_pro_epilogue): Remove frame size argument.
19754 * output.h (default_function_pro_epilogue): Likewise.
19755 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
19756 (arm_output_function_prologue): Likewise.
19757 * config/frv/frv.c (frv_function_prologue): Likewise.
19758 (frv_function_epilogue): Likewise.
19759 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
19760 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
19761 (ia64_output_function_epilogue): Likewise.
19762 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
19763 (m32r_output_function_epilogue): Likewise.
19764 * config/microblaze/microblaze.c (microblaze_function_prologue)
19765 (microblaze_function_epilogue): Likewise.
19766 * config/mips/mips.c (mips_output_function_prologue): Likewise.
19767 (mips_output_function_epilogue): Likewise.
19768 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
19769 (mmix_target_asm_function_epilogue): Likewise.
19770 * config/msp430/msp430.c (msp430_start_function): Likewise.
19771 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
19772 (nds32_asm_function_epilogue): Likewise.
19773 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
19774 * config/pa/pa.c (pa_output_function_prologue): Likewise.
19775 (pa_output_function_epilogue): Likewise.
19776 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
19777 (rs6000_output_function_epilogue): Likewise.
19778 * config/rl78/rl78.c (rl78_start_function): Likewise.
19779 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
19780 (rs6000_output_function_epilogue): Likewise.
19781 * config/rx/rx.c (rx_output_function_prologue): Likewise.
19782 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
19783 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
19784 (sparc_asm_function_epilogue): Likewise.
19786 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19788 * tree.h (type_has_mode_precision_p): New function.
19789 * convert.c (convert_to_integer_1): Use it.
19790 * expr.c (expand_expr_real_2): Likewise.
19791 (expand_expr_real_1): Likewise.
19792 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
19793 * match.pd: Likewise.
19794 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
19795 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
19796 * tree-tailcall.c (process_assignment): Likewise.
19797 * tree-vect-loop.c (vectorizable_reduction): Likewise.
19798 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
19799 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
19800 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19801 (vectorizable_assignment): Likewise.
19802 (vectorizable_shift): Likewise.
19803 (vectorizable_operation): Likewise.
19804 * tree-vrp.c (register_edge_assert_for_2): Likewise.
19806 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
19808 * match.pd: Add pow (C, x) simplification.
19810 2017-08-21 Richard Biener <rguenther@suse.de>
19812 PR tree-optimization/81900
19813 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
19814 for blocks with abnormal predecessors.
19815 (compute_antic): Do not set visited flag prematurely.
19817 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
19820 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
19822 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19824 * stor-layout.h (vector_type_mode): Move to...
19825 * tree.h (vector_type_mode): ...here.
19826 * stor-layout.c (vector_type_mode): Move to...
19827 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
19829 2017-08-21 Richard Biener <rguenther@suse.de>
19831 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
19832 register_external_die hooks.
19833 (debug_false_tree_charstarstar_uhwistar): Declare.
19834 (debug_nothing_tree_charstar_uhwi): Likewise.
19835 * debug.c (do_nothing_debug_hooks): Adjust.
19836 (debug_false_tree_charstarstar_uhwistar): New do nothing.
19837 (debug_nothing_tree_charstar_uhwi): Likewise.
19838 * dbxout.c (dbx_debug_hooks): Adjust.
19839 (xcoff_debug_hooks): Likewise.
19840 * sdbout.c (sdb_debug_hooks): Likewise.
19841 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19842 * dwarf2out.c (macinfo_label_base): New global.
19843 (dwarf2out_register_external_die): New function for the
19844 register_external_die hook.
19845 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
19846 (dwarf2_debug_hooks): Use them.
19847 (dwarf2_lineno_debug_hooks): Adjust.
19848 (struct die_struct): Add with_offset flag.
19849 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
19850 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
19851 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
19852 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
19853 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
19854 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
19855 defining section names for the early LTO debug variants.
19856 (reset_indirect_string): New helper.
19857 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
19858 (print_dw_val): Add support for offsetted symbol references.
19859 (get_ultimate_context): Split out from is_cxx.
19860 (is_cxx): Use get_ultimate_context.
19861 (is_fortran): Add decl overload.
19862 (compute_comp_unit_symbol): Split out worker from
19863 compute_section_prefix.
19864 (compute_section_prefix): Call compute_comp_unit_symbol and
19865 set comdat_type_p here.
19866 (output_die): Skip DIE symbol output for the LTO added one.
19867 Handle DIE symbol references with offset.
19868 (output_comp_unit): Guard section name mangling properly.
19869 For LTO debug sections emit a symbol at the section beginning
19870 which we use to refer to its DIEs.
19871 (add_abstract_origin_attribute): For DIEs registered via
19872 dwarf2out_register_external_die directly refer to the early
19873 DIE rather than indirectly through the shadow one we created.
19874 Remove obsolete call to dwarf2out_abstract_function for
19875 non-function/block origins.
19876 (gen_array_type_die): When generating early LTO debug do
19877 not emit DW_AT_string_length.
19878 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
19879 late when in LTO. As suggested place a gcc_unreachable for
19880 the DECL_ABSTRACT_P case.
19881 (gen_subprogram_die): Avoid another specification DIE
19882 for early built declarations/definitions for the late LTO case.
19883 (gen_variable_die): Add type references for late duplicated VLA dies
19885 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
19886 we have the abstract instance already.
19887 (process_scope_var): Adjust decl DIE contexts in LTO which
19888 first puts them in limbo.
19889 (gen_decl_die): Do not generate type DIEs late apart from
19890 types for VLAs or for decls we do not yet have a DIE. Do not
19891 call dwarf2out_abstract_function late.
19892 (dwarf2out_early_global_decl): Make sure to create DIEs
19893 for abstract instances of a decl first.
19894 (dwarf2out_late_global_decl): Adjust comment.
19895 (output_macinfo_op): With multiple macro sections use
19896 macinfo_label_base to distinguish labels.
19897 (output_macinfo): Likewise. Update macinfo_label_base.
19898 Pass in the line info label.
19899 (note_variable_value_in_expr): When generating LTO resolve
19900 all variable values here by generating DIEs as needed.
19901 (init_sections_and_labels): Add early LTO debug flag parameter
19902 and generate different sections and names if set. Add generation
19903 counter for the labels so we can have multiple of them.
19904 (reset_dies): Helper to allow DIEs to be output multiple times.
19905 (dwarf2out_finish): When outputting DIEs to the fat part of an
19906 LTO object first reset DIEs.
19907 (dwarf2out_early_finish): Output early DIEs when generating LTO.
19908 (modified_type_die): Check for decl_ultimate_origin being self
19910 (gen_type_die_with_usage): Likewise.
19911 (gen_typedef_die): Allow decl_ultimate_origin being self.
19912 (set_decl_abstract_flags): Remove.
19913 (set_block_abstract_flags): Likewise.
19914 (dwarf2out_abstract_function): Treat the early generated DIEs
19915 as the abstract copy and only add DW_AT_inline and
19916 DW_AT_artificial here and call set_decl_origin_self.
19917 If the DIE has an abstract origin don't do anything.
19918 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
19919 if we have none yet (Go fails to build one, PR78628).
19920 (variably_modified_type_p): Prevent endless recursion for Ada
19921 cyclic pointer types.
19922 * lto-streamer-in.c: Include debug.h.
19923 (dref_queue): New global.
19924 (lto_read_tree_1): Stream in DIE references.
19925 (lto_input_tree): Register DIE references.
19926 (input_function): Stream DECL_DEBUG_ARGS.
19927 * lto-streamer-out.c: Include debug.h.
19928 (lto_write_tree_1): Output DIE references.
19929 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
19930 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
19931 (output_function): Stream DECL_DEBUG_ARGS.
19932 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
19933 Stream DECL_ABSTRACT_ORIGIN.
19934 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
19935 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
19936 DECL_CONTEXT for file-scope decls.
19937 * lto-streamer.h (struct dref_entry): Declare.
19938 (dref_queue): Likewise.
19939 * cfgexpand.c (pass_expand::execute): Do not call the
19940 outlining_inline_function hook here.
19941 * lto-wrapper.c (debug_obj): New global.
19942 (tool_cleanup): Unlink it if required.
19943 (debug_objcopy): New function.
19944 (run_gcc): Handle early debug sections in the IL files by
19945 extracting them to separate files, partially linkin them and
19946 feeding the result back as result to the linker.
19947 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
19948 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
19949 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
19950 sections into a separate segment.
19951 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
19953 (darwin_asm_dwarf_section): Likewise.
19954 (darwin_asm_output_dwarf_offset): Likewise.
19955 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
19957 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19958 Alan Hayward <alan.hayward@arm.com>
19959 David Sherwood <david.sherwood@arm.com>
19961 * read-md.h (md_reader::record_potential_iterator_use): Replace
19962 pointer argument with an rtx and an index.
19963 * read-rtl.c (iterator_group::apply_iterator): Likewise.
19964 (apply_mode_iterator): Likewise.
19965 (apply_code_iterator): Likewise.
19966 (apply_int_iterator): Likewise.
19967 (apply_subst_iterator): Likewise.
19968 (record_iterator_use): Likewise.
19969 (record_attribute_use): Likewise.
19970 (md_reader::record_potential_iterator_use): Likewise. Update calls
19971 to record_iterator_use and apply_iterator.
19972 (iterator_use): Replace ptr with x and index.
19973 (attribute_use): Likewise.
19974 (apply_attribute_uses): Update calls to apply_iterator.
19975 (apply_iterators): Likewise. Update initialization of iterator_use.
19976 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
19977 and record_potential_iterator_use.
19978 (rtx_reader::read_rtx_operand): Likewise.
19980 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19981 Alan Hayward <alan.hayward@arm.com>
19982 David Sherwood <david.sherwood@arm.com>
19984 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
19987 2017-08-21 Richard Biener <rguenther@suse.de>
19989 PR middle-end/81884
19990 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
19991 at struct end conservatively when comparing common bases.
19993 2017-08-21 Richard Biener <rguenther@suse.de>
19995 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
19996 (mem_ref_in_stmt): Remove.
19997 (determine_max_movement): Use ref index to get at the reference.
19998 (invariantness_dom_walker::before_dom_children): Deal with
19999 lim data already initialized.
20000 (gather_mem_refs_stmt): Initialize lim data and record ref index.
20002 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20004 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
20005 (TARGET_ISA_ROUND): Ditto.
20006 (TARGET_ROUND): Ditto.
20007 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
20008 * config/i386/i386.md: Ditto.
20009 * config/i386/sse.md: Ditto.
20010 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
20011 with OPTION_MASK_ISA_SSE4_1.
20013 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20016 * doc/extend.texi (x86 Built-in Functions): Correct the name of
20017 __builtin_ia32_lzcnt_u16.
20019 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
20022 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
20023 (rs6000_set_current_function): Rewrite function to use it.
20025 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
20028 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
20029 and TYPE_WARN_IF_NOT_ALIGN.
20030 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
20031 (handle_warn_if_not_align): New.
20032 (place_union_field): Call handle_warn_if_not_align.
20033 (place_field): Call handle_warn_if_not_align.
20034 Copy TYPE_WARN_IF_NOT_ALIGN.
20035 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
20036 (layout_type): Likewise.
20037 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
20039 (tree_decl_common): Add warn_if_not_align.
20040 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
20041 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
20042 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
20043 (DECL_WARN_IF_NOT_ALIGN): Likewise.
20044 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
20045 * doc/extend.texi: Document warn_if_not_aligned attribute.
20046 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
20048 2017-08-17 Martin Liska <mliska@suse.cz>
20051 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
20052 (get_data_dependence): Use it as pointer type.
20053 (distribute_loop): Likewise.
20055 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20057 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
20058 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
20059 (p8_vmrgow_<mode>_direct): New define_insn.
20060 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
20061 handle endianness for vmrgew and vmrgow permute patterns.
20063 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20065 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
20066 * config/rs6000/rs6000-cpus.def: Remove comment.
20067 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
20068 (POWERPC_MASKS): Likewise.
20069 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
20070 use of TARGET_VSX_TIMODE.
20071 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
20072 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
20073 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
20074 (rs6000_option_override_internal): Remove dead code.
20075 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
20076 (rs6000_legitimize_reload_address): Likewise.
20077 (rs6000_legitimate_address_p): Likewise.
20078 (rs6000_opt_masks): Delete "vsx-timode".
20079 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
20080 from function comment.
20081 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
20082 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
20083 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
20085 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
20086 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
20087 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
20088 (VSX_TI): Likewise.
20090 (define_peephole2): Likewise.
20092 2017-08-17 Martin Sebor <msebor@redhat.com>
20095 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
20096 past the end of an array.
20097 (test_pp_format): Add test cases.
20099 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
20101 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
20102 missing ECF_NOTHROW flags.
20104 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20107 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
20108 operands residing in integer registers.
20109 (*vsx_le_perm_load_<mode>): Likewise.
20110 (*vsx_le_perm_store_<mode>): Likewise.
20111 (define_peephole2): Add peepholes to optimize the above.
20113 2017-08-17 Marek Polacek <polacek@redhat.com>
20115 PR middle-end/81814
20116 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
20117 to mimic what shorten_compare did. Change the return type to bool.
20118 (fold_cond_expr_with_comparison): Update call to
20119 operand_equal_for_comparison_p.
20120 (fold_ternary_loc): Likewise.
20122 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
20124 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
20126 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
20128 2017-08-17 Richard Biener <rguenther@suse.de>
20130 * tree-ssa-structalias.c (solve_graph): When propagating
20131 to successors update the graphs succ edges and avoid duplicate work.
20133 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
20136 * config/i386/i386.c (ix86_option_override_internal): Save target
20137 specific options after ix86_stack_protector_guard_reg was changed.
20139 2017-08-17 Richard Biener <rguenther@suse.de>
20141 PR tree-optimization/81827
20142 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
20143 (new_var_info): Initialize it conservatively.
20144 (get_call_vi): Mark register vars.
20145 (new_scalar_tmp_constraint_exp): Likewise.
20146 (handle_rhs_call): Likewise.
20147 (handle_const_call): Likewise.
20148 (create_function_info_for): Likewise.
20149 (solve_constraints): Sort varinfos to separate register from
20150 non-register vars to pack points-to solution bitmaps during
20153 2017-08-17 Marek Polacek <polacek@redhat.com>
20155 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
20157 2017-08-17 Richard Biener <rguenther@suse.de>
20159 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
20160 to true when overflow is undefined and we saturated the result.
20162 2017-08-17 Alan Modra <amodra@gmail.com>
20165 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
20166 Don't use store multiple if only one reg needs saving.
20167 (interesting_frame_related_regno): New function.
20168 (rs6000_frame_related): Don't emit frame info for regs that
20170 (rs6000_emit_epilogue): Likewise.
20172 2017-08-16 Nathan Sidwell <nathan@acm.org>
20174 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
20175 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
20176 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
20177 * tree.c (free_lang_data_in_type): Use else-if chain. Always
20178 clear TYPE_LANG_1. Remove obsolete member-function stripping.
20179 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
20180 (verify_type): Adjust for TYPE_BINFO move.
20181 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
20182 process TYPE_BINFO directly.
20183 (hash_tree): Likewise.
20184 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20186 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20189 2017-08-16 David Malcolm <dmalcolm@redhat.com>
20191 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
20193 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
20196 * config/i386/i386.md (*anddi_1_btr): Change predicates of
20197 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
20198 Add ix86_binary_operator_ok to insn constraint.
20199 (*iordi_1_bts): Ditto.
20200 (*xordi_1_btc): Ditto.
20201 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
20202 Update corresponding peephole2 pattern.
20206 2017-08-16 Bin Cheng <bin.cheng@arm.com>
20208 PR tree-optimization/81832
20209 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
20210 copy loop header which has IFN_LOOP_DIST_ALIAS call.
20212 2017-08-16 Marek Polacek <polacek@redhat.com>
20215 * fold-const.c (fold_indirect_ref_1): Restore original behavior
20216 regarding size_zero_node.
20218 2017-08-16 Martin Liska <mliska@suse.cz>
20221 * config.gcc: Respect previously set extra_objs in case
20224 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
20226 PR tree-optimization/81835
20227 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
20228 the phi SSA_NAME. Check that the condition in a COND_EXPR does
20229 not depend on the phi.
20231 2017-08-16 Alan Modra <amodra@gmail.com>
20233 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
20236 2017-08-16 Alan Modra <amodra@gmail.com>
20238 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
20239 (save_reg_p): ..into this. Update all callers.
20240 (first_reg_to_save): Simplify.
20242 2017-08-16 Alan Modra <amodra@gmail.com>
20244 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
20247 2017-08-15 Joseph Myers <joseph@codesourcery.com>
20251 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
20252 constant count if that count is less than 32.
20254 2017-08-15 Nathan Sidwell <nathan@acm.org>
20256 * gcc.c (execute): Emit friendlier message if inferior is killed
20257 by an external cause.
20259 2017-08-15 Richard Biener <rguenther@suse.de>
20261 PR tree-optimization/81790
20262 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
20263 CONSTRUCTORs from simplifying and VN.
20265 2017-08-14 Martin Sebor <msebor@redhat.com>
20267 * builtin-attrs.def: Add comments.
20269 2017-08-14 Martin Sebor <msebor@redhat.com>
20272 * doc/extend.texi (attribute nonstring): Document new attribute.
20274 2017-08-14 Martin Sebor <msebor@redhat.com>
20277 * tree-diagnostic.c (default_tree_printer): Handle %G.
20278 * gimple-pretty-print.h (percent_G_format): Declare new function.
20279 * gimple-pretty-print.c (percent_G_format): Define.
20280 * tree-pretty-print.c (percent_K_format): Add argument.
20282 2017-08-14 Martin Sebor <msebor@redhat.com>
20284 PR translation/79998
20285 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
20286 Remove a stray space.
20288 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
20291 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
20292 (*iordi_1_bts): Ditto.
20293 (*xordi_1_btc): Ditto.
20295 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20298 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
20299 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20301 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
20302 quoted strings, and make more translator-friendly.
20303 (darwin_rs6000_override_options): Likewise.
20304 (rs6000_option_override_internal): Likewise.
20305 (rs6000_return_in_memory): Fix overlong line.
20306 (init_cmulative_args): Use quoted strings, and make more
20307 translator-friendly.
20308 (rs6000_pass_by_reference): Fix overlong line.
20309 (def_builtin): Use quoted strings.
20310 (altivec_expand_predicate_builtin): Use quoted strings, and make
20311 more translator-friendly.
20312 (htm_expand_builtin): Use quoted strings.
20313 (cpu_expand_builtin): Use quoted strings, and make more
20314 translator-friendly.
20315 (altivec_expand_builtin): Likewise.
20316 (paired_expand_predicate_builtin): Likewise.
20317 (rs6000_invalid_builtin): Likewise.
20318 (builtin_function_type): Use quoted strings.
20319 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
20320 more translator-friendly.
20321 (rs6000_trampoline_init): Likewise.
20322 (rs6000_handle_altivec_attribute): Likewise.
20323 (rs6000_inner_target_options): Use quoted strings.
20324 (rs6000_disable_incompatible_switches): Likewise.
20325 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
20326 strings, and make more translator-friendly.
20327 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
20329 2017-08-14 Bin Cheng <bin.cheng@arm.com>
20331 PR tree-optimization/81799
20332 * tree-loop-distribution.c (version_loop_by_alias_check): Force
20333 cond_expr to simple gimple operand.
20335 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
20337 PR middle-end/46932
20338 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
20340 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
20344 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
20345 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
20346 TARGET_GASISR_PROLOGUES.
20347 * config/avr/avr.c (avr_option_override): Same.
20348 (avr_pass_pre_proep::execute): Same.
20350 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
20353 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
20354 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
20356 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
20358 * config/i386/i386.md (*load_tp_<mode>): Redefine as
20359 define_insn_and_split. Split to a memory load from 0 in
20360 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
20361 using PTR mode iterator.
20362 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
20363 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
20364 (*add_tp_<mode>): Redefine as define_insn_and_split.
20365 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
20366 address space. Merge with *add_tp_x32 using PTR mode iterator.
20367 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
20368 Split to an add with a memory load from 0 in
20369 DEFAULT_TLS_SEG_REG address space.
20371 2017-08-12 Andrew Pinski <apinski@cavium.com>
20373 * config/aarch64/aarch64-option-extensions.def (rdma):
20374 Fix feature string to what Linux prints out in /proc/cpuinfo.
20376 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
20379 * dwarf2out.c (modified_type_die): For C typedef types that have
20380 an ultimate origin, process the ultimate origin instead of the
20382 (gen_typedef_die): Assert that input DECLs have no ultimate
20384 (gen_type_die_with_usage): For typedef variants that have an
20385 ultimate origin, just call gen_decl_die on the original DECL.
20386 (process_scope_var): Avoid creating DIEs for local typedefs and
20387 concrete static variables.
20389 2017-08-12 Alan Modra <amodra@gmail.com>
20393 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
20394 match gnu-user.h startfile.
20395 (ENDFILE_LINUX_SPEC): Similarly.
20397 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
20400 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
20402 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
20404 2017-08-11 Tamar Christina <tamar.christina@arm.com>
20405 * config/aarch64/aarch64.md (mov<mode>): Change.
20406 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20407 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
20408 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
20410 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
20412 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
20413 for storage order barriers.
20415 2017-08-11 Martin Liska <mliska@suse.cz>
20418 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
20419 variables of void type.
20421 2017-08-11 Martin Liska <mliska@suse.cz>
20423 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
20424 TARGET_SUPPORTS_ALIASES.
20425 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
20426 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
20427 (optimize_weakref): Likewise.
20428 * symtab.c (symtab_node::noninterposable_alias): Likewise.
20429 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
20430 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
20432 2017-08-11 Martin Liska <mliska@suse.cz>
20435 * config/i386/i386.c (make_resolver_func): Do complete
20436 refactoring of the function.
20438 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20441 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
20442 * config/i386/i386.c (ix86_stack_protect_guard): Use
20443 ix86_stack_protect_guard_symbol_str to generate varible declaration.
20444 * doc/invoke.texi (x86 Options): Document
20445 -mstack-protector-guard-symbol= option.
20447 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20449 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
20450 * config/i386/i386.c (ix86_split_stack_guard): New function.
20451 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
20452 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
20453 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
20454 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
20455 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
20456 (split_stack_space_check): Call ix86_split_stack_guard.
20458 2017-08-10 Martin Sebor <msebor@redhat.com>
20460 * print-tree.c (print_node): Print location using the established
20462 Replace spaces with colons.
20463 (debug_raw, debug): Ditto.
20465 2017-08-10 Martin Sebor <msebor@redhat.com>
20468 * pretty-print.c (pp_format): Correct the handling of %s precision.
20470 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
20473 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20475 (ix86_finalize_stack_frame_flags): This. Also clear
20476 frame_pointer_needed if -fno-omit-frame-pointer is used without
20478 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20479 with ix86_finalize_stack_frame_flags.
20480 (ix86_expand_epilogue): Likewise.
20481 (ix86_expand_split_stack_prologue): Likewise.
20482 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
20484 2017-08-10 Martin Liska <mliska@suse.cz>
20487 * c-attribs.c (handle_target_attribute):
20488 Report warning for an empty string argument of target attribute.
20490 2017-08-09 Jakub Jelinek <jakub@redhat.com>
20493 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
20495 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
20496 or DECL_NONLOCAL labels.
20497 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
20498 or DECL_NONLOCAL labels here.
20500 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
20502 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
20503 to indicate when early gimple folding has been disabled.
20504 (rs6000_gimple_fold_builtin): Add debug content.
20505 (rs6000_invalid_builtin): Fix whitespace.
20506 (rs6000_expand_builtin): Fix whitespace.
20507 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
20509 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
20512 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
20513 SAVE_MULTIPLE if not all the registers that saves, should be saved.
20515 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
20517 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
20518 (qdf24xx): Likewise.
20519 * config/aarch64/aarch64.md: Include falkor.md.
20520 * config/aarch64/falkor.md: New.
20522 2017-08-09 Marek Polacek <polacek@redhat.com>
20525 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
20526 * diagnostic.c (emit_diagnostic): Add a comment.
20527 (emit_diagnostic_valist): New function.
20529 2017-08-09 Marek Polacek <polacek@redhat.com>
20532 * input.c (make_location): New overload.
20533 * input.h (make_location): Declare.
20535 2017-08-08 Alan Modra <amodra@gmail.com>
20536 H.J. Lu <hongjiu.lu@intel.com>
20539 * gcc.c (NO_PIE_SPEC): Delete.
20540 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
20542 (LINK_PIE_SPEC): ..to here.
20543 (LINK_COMMAND_SPEC): Support -no-pie.
20544 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
20545 chain of crtbegin*.o selection, update for PIE_SPEC changes and
20547 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
20548 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
20549 (ENDFILE_CRTEND_SPEC): Similarly.
20551 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
20554 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
20555 (mstack-protector-guard-offset=): Ditto.
20556 * config/i386/i386.c (ix86_option_override): Handle
20557 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
20559 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
20560 ix86_stack_protect_guard_offset variables.
20561 (TARGET_STACK_PROTECT_GUARD): Always define.
20562 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
20563 and -mstack-protector-guard-offset= options.
20565 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20567 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
20568 boundary case for the last candidate.
20570 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20572 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
20573 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
20575 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20577 PR middle-end/19706
20578 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
20579 * config/aarch64/aarch64-builtins.c
20580 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
20581 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
20582 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
20584 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20585 Andrew Pinski <pinskia@gmail.com>
20587 PR middle-end/19706
20588 * internal-fn.def (XORSIGN): New.
20589 * optabs.def (xorsign_optab): New.
20590 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
20591 (convert_expand_mult_copysign): New.
20592 (pass_optimize_widening_mul::execute): Call
20593 convert_expand_mult_copysign.
20595 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20597 PR tree-optimization/81354
20598 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
20599 Insert on edges rather than explicitly creating landing pads.
20600 (analyze_candidates_and_replace): Commit edge inserts.
20602 2017-08-08 Richard Biener <rguenther@suse.de>
20604 PR middle-end/81719
20605 * tree-ssa-loop-niter.c: Include tree-dfa.h.
20606 (expand_simple_operations): Also look through ADDR_EXPRs with
20607 MEM_REF bases treating them as POINTER_PLUS_EXPR.
20609 2017-08-08 Richard Biener <rguenther@suse.de>
20611 PR tree-optimization/81723
20612 * tree-vect-slp.c (struct bst_traits): New hash traits.
20613 (bst_fail): New global.
20614 (vect_build_slp_tree_2): New worker, split out from ...
20615 (vect_build_slp_tree): ... this now wrapping it with using
20616 bst_fail set to cache SLP tree build fails. Properly handle
20618 (vect_analyze_slp_instance): Allocate and free bst_fail.
20620 2017-08-08 Martin Liska <mliska@suse.cz>
20623 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
20624 LABEL_DECLs that can be from a different function.
20626 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20628 PR tree-optimization/81744
20629 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
20630 loop's number of iterations.
20632 2017-08-08 Martin Liska <mliska@suse.cz>
20634 * asan.c: Include header files.
20635 * attribs.c (build_decl_attribute_variant): New function moved
20637 (build_type_attribute_qual_variant): Likewise.
20638 (cmp_attrib_identifiers): Likewise.
20639 (simple_cst_list_equal): Likewise.
20640 (omp_declare_simd_clauses_equal): Likewise.
20641 (attribute_value_equal): Likewise.
20642 (comp_type_attributes): Likewise.
20643 (build_type_attribute_variant): Likewise.
20644 (lookup_ident_attribute): Likewise.
20645 (remove_attribute): Likewise.
20646 (merge_attributes): Likewise.
20647 (merge_type_attributes): Likewise.
20648 (merge_decl_attributes): Likewise.
20649 (merge_dllimport_decl_attributes): Likewise.
20650 (handle_dll_attribute): Likewise.
20651 (attribute_list_equal): Likewise.
20652 (attribute_list_contained): Likewise.
20653 * attribs.h (lookup_attribute): New function moved from tree.[ch].
20654 (lookup_attribute_by_prefix): Likewise.
20655 * bb-reorder.c: Include header files.
20656 * builtins.c: Likewise.
20657 * calls.c: Likewise.
20658 * cfgexpand.c: Likewise.
20659 * cgraph.c: Likewise.
20660 * cgraphunit.c: Likewise.
20661 * convert.c: Likewise.
20662 * dwarf2out.c: Likewise.
20663 * final.c: Likewise.
20664 * fold-const.c: Likewise.
20665 * function.c: Likewise.
20666 * gimple-expr.c: Likewise.
20667 * gimple-fold.c: Likewise.
20668 * gimple-pretty-print.c: Likewise.
20669 * gimple.c: Likewise.
20670 * gimplify.c: Likewise.
20671 * hsa-common.c: Likewise.
20672 * hsa-gen.c: Likewise.
20673 * internal-fn.c: Likewise.
20674 * ipa-chkp.c: Likewise.
20675 * ipa-cp.c: Likewise.
20676 * ipa-devirt.c: Likewise.
20677 * ipa-fnsummary.c: Likewise.
20678 * ipa-inline.c: Likewise.
20679 * ipa-visibility.c: Likewise.
20681 * lto-cgraph.c: Likewise.
20682 * omp-expand.c: Likewise.
20683 * omp-general.c: Likewise.
20684 * omp-low.c: Likewise.
20685 * omp-offload.c: Likewise.
20686 * omp-simd-clone.c: Likewise.
20687 * opts-global.c: Likewise.
20688 * passes.c: Likewise.
20689 * predict.c: Likewise.
20690 * sancov.c: Likewise.
20691 * sanopt.c: Likewise.
20692 * symtab.c: Likewise.
20693 * toplev.c: Likewise.
20694 * trans-mem.c: Likewise.
20695 * tree-chkp.c: Likewise.
20696 * tree-eh.c: Likewise.
20697 * tree-into-ssa.c: Likewise.
20698 * tree-object-size.c: Likewise.
20699 * tree-parloops.c: Likewise.
20700 * tree-profile.c: Likewise.
20701 * tree-ssa-ccp.c: Likewise.
20702 * tree-ssa-live.c: Likewise.
20703 * tree-ssa-loop.c: Likewise.
20704 * tree-ssa-sccvn.c: Likewise.
20705 * tree-ssa-structalias.c: Likewise.
20706 * tree-ssa.c: Likewise.
20707 * tree-streamer-in.c: Likewise.
20708 * tree-vectorizer.c: Likewise.
20709 * tree-vrp.c: Likewise.
20710 * tsan.c: Likewise.
20711 * ubsan.c: Likewise.
20712 * varasm.c: Likewise.
20713 * varpool.c: Likewise.
20714 * tree.c: Remove functions moved to attribs.[ch].
20715 * tree.h: Likewise.
20716 * config/aarch64/aarch64.c: Add attrs.h header file.
20717 * config/alpha/alpha.c: Likewise.
20718 * config/arc/arc.c: Likewise.
20719 * config/arm/arm.c: Likewise.
20720 * config/avr/avr.c: Likewise.
20721 * config/bfin/bfin.c: Likewise.
20722 * config/c6x/c6x.c: Likewise.
20723 * config/cr16/cr16.c: Likewise.
20724 * config/cris/cris.c: Likewise.
20725 * config/darwin.c: Likewise.
20726 * config/epiphany/epiphany.c: Likewise.
20727 * config/fr30/fr30.c: Likewise.
20728 * config/frv/frv.c: Likewise.
20729 * config/ft32/ft32.c: Likewise.
20730 * config/h8300/h8300.c: Likewise.
20731 * config/i386/winnt.c: Likewise.
20732 * config/ia64/ia64.c: Likewise.
20733 * config/iq2000/iq2000.c: Likewise.
20734 * config/lm32/lm32.c: Likewise.
20735 * config/m32c/m32c.c: Likewise.
20736 * config/m32r/m32r.c: Likewise.
20737 * config/m68k/m68k.c: Likewise.
20738 * config/mcore/mcore.c: Likewise.
20739 * config/microblaze/microblaze.c: Likewise.
20740 * config/mips/mips.c: Likewise.
20741 * config/mmix/mmix.c: Likewise.
20742 * config/mn10300/mn10300.c: Likewise.
20743 * config/moxie/moxie.c: Likewise.
20744 * config/msp430/msp430.c: Likewise.
20745 * config/nds32/nds32-isr.c: Likewise.
20746 * config/nds32/nds32.c: Likewise.
20747 * config/nios2/nios2.c: Likewise.
20748 * config/nvptx/nvptx.c: Likewise.
20749 * config/pa/pa.c: Likewise.
20750 * config/pdp11/pdp11.c: Likewise.
20751 * config/powerpcspe/powerpcspe.c: Likewise.
20752 * config/riscv/riscv.c: Likewise.
20753 * config/rl78/rl78.c: Likewise.
20754 * config/rx/rx.c: Likewise.
20755 * config/s390/s390.c: Likewise.
20756 * config/sh/sh.c: Likewise.
20757 * config/sol2.c: Likewise.
20758 * config/sparc/sparc.c: Likewise.
20759 * config/spu/spu.c: Likewise.
20760 * config/stormy16/stormy16.c: Likewise.
20761 * config/tilegx/tilegx.c: Likewise.
20762 * config/tilepro/tilepro.c: Likewise.
20763 * config/v850/v850.c: Likewise.
20764 * config/vax/vax.c: Likewise.
20765 * config/visium/visium.c: Likewise.
20766 * config/xtensa/xtensa.c: Likewise.
20768 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20771 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
20772 constraints since the -mupper-regs-* switches have been
20774 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
20775 into a vector from a double word element that was extracted from
20776 another vector, and eliminate extra XXPERMDI instructions.
20777 (vsx_concat_<mode>_2): Likewise.
20778 (vsx_concat_<mode>_3): Likewise.
20779 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
20780 concat to allow optimizing inserts from previous extracts.
20782 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
20784 * config/i386/i386.c (ix86_stack_protect_guard): Generate
20785 memory reference to a SSP offset in TLS address space.
20786 (ix86_print_operand) <case '@'>: Remove.
20787 (ix86_print_operand_punct_valid_p): Remove '@' code.
20788 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
20789 UNSPEC_SP_TLS_TEST.
20790 (stack_tls_protect_set_<mode>): Remove.
20791 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
20792 (stack_tls_protect_test_<mode>): Remove.
20793 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
20795 2017-08-07 Olivier Hainque <hainque@adacore.com>
20798 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
20800 2017-08-07 Douglas Rupp <rupp@adacore.com>
20802 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
20803 variable was referenced as multidir in command.
20805 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20808 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
20811 2017-08-07 Martin Liska <mliska@suse.cz>
20813 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
20814 * config/rl78/rl78.c: Add include of attribs.h.
20815 * config/sh/sh.c: Likewise.
20816 * config/v850/v850.c: Likewise.
20818 2017-08-07 Tom de Vries <tom@codesourcery.com>
20820 PR middle-end/78266
20821 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
20823 2017-08-07 Martin Liska <mliska@suse.cz>
20825 * config/mips/mips.c: Include attribs.h.
20827 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
20830 * doc/invoke.texi: Document change in behvaior for -Ofast for
20833 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
20835 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
20837 (aarch64_pop_regs): Likewise.
20838 (aarch64_gen_load_pair): Likewise.
20839 (aarch64_save_callee_saves): Likewise.
20840 (aarch64_restore_callee_saves): Likewise.
20842 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20844 * config/i386/i386.c: Revert the last change.
20846 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20849 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20851 (ix86_finalize_stack_frame_flags): This. Also clear
20852 frame_pointer_needed if -fno-omit-frame-pointer is used without
20854 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20855 with ix86_finalize_stack_frame_flags.
20856 (ix86_expand_epilogue): Likewise.
20857 (ix86_expand_split_stack_prologue): Likewise.
20859 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20862 * config/i386/i386.c (get_builtin_code_for_version): Set priority
20863 to P_AES for Westmere.
20865 2017-08-07 Jonathan Yong <10walls@gmail.com>
20867 * config/i386/mingw.opt (fset-stack-executable): Removed.
20868 * config/i386/cygming.opt (fset-stack-executable): Moved
20870 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
20872 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
20874 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
20876 2017-08-07 Marek Polacek <polacek@redhat.com>
20878 PR middle-end/81737
20879 * fold-const.c (fold_indirect_ref_1): Check type_domain.
20881 2017-08-07 Martin Liska <mliska@suse.cz>
20883 * attribs.h (canonicalize_attr_name): New function.
20884 (cmp_attribs): Move from c-format.c and adjusted.
20885 (is_attribute_p): Moved from tree.h.
20886 * tree-inline.c: Add new includes.
20887 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
20888 (private_is_attribute_p): Remove.
20889 (private_lookup_attribute): Likewise.
20890 (private_lookup_attribute_by_prefix): Simplify.
20891 (remove_attribute): Use is_attribute_p.
20892 * tree.h: Remove removed declarations.
20894 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20896 PR middle-end/81698
20897 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
20898 instead of computing it in the function. Formatting fix.
20899 (expand_case): Don't rely on default_edge being the first edge,
20900 clear it if removing it, pass default_edge to
20901 emit_case_dispatch_table.
20902 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
20905 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
20907 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
20908 insn in the function, emit NOP after the insn.
20910 2017-08-06 Tom de Vries <tom@codesourcery.com>
20912 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
20915 2017-08-06 Tom de Vries <tom@codesourcery.com>
20917 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
20920 2017-08-04 Yury Gribov <tetra2005@gmail.com>
20922 PR tree-optimization/57371
20923 * match.pd: New pattern.
20925 2017-08-04 Marek Polacek <polacek@redhat.com>
20927 PR middle-end/81695
20928 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
20929 perform the computation in offset_int.
20931 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20933 PR tree-optimization/81136
20934 * tree-vectorizer.h: Include tree-hash-traits.h.
20935 (vec_base_alignments): New typedef.
20936 (vec_info): Add a base_alignments field.
20937 (vect_record_base_alignments): Declare.
20938 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
20940 (DR_IS_CONDITIONAL_IN_STMT): New macro.
20941 (create_data_ref): Add an is_conditional_in_stmt argument.
20942 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
20943 the is_conditional_in_stmt field.
20944 (data_ref_loc): Add an is_conditional_in_stmt field.
20945 (get_references_in_stmt): Set the is_conditional_in_stmt field.
20946 (find_data_references_in_stmt): Update call to create_data_ref.
20947 (graphite_find_data_references_in_stmt): Likewise.
20948 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
20949 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
20950 (vect_record_base_alignment): New function.
20951 (vect_record_base_alignments): Likewise.
20952 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
20953 for nested statements even if we fail to compute a misalignment.
20954 Use pooled base alignments for unconditional references.
20955 (vect_find_same_alignment_drs): Compare base addresses instead
20957 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
20958 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
20960 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20962 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
20963 Add an explicit name for the enum. Use auto_vec for slp_instances
20964 and grouped_stores.
20965 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
20967 (_bb_vec_info): Add a constructor and destructor.
20968 (vinfo_for_stmt): Return NULL for uids of -1 as well.
20969 (destroy_loop_vec_info): Delete.
20970 (vect_destroy_datarefs): Likewise.
20971 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
20972 (vec_info::vec_info): New function.
20973 (vec_info::~vec_info): Likewise.
20974 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
20975 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
20976 destroy_loop_vec_info.
20977 * tree-vect-loop.c (new_loop_vec_info): Replace with...
20978 (_loop_vec_info::_loop_vec_info): ...this.
20979 (destroy_loop_vec_info): Replace with...
20980 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
20981 the stmt_vec_infos. Leave handling of vec_info information to its
20982 destructor. Remove explicit vector releases.
20983 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
20984 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
20985 * tree-vect-slp.c (new_bb_vec_info): Replace with...
20986 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
20987 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
20988 (destroy_bb_vec_info): Replace with...
20989 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
20990 information to its destructor.
20991 (vect_slp_analyze_bb_1): Use new and delete instead of
20992 new_bb_vec_info and destroy_bb_vec_info.
20993 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
20996 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20998 * tree-data-ref.h (subscript): Add access_fn field.
20999 (data_dependence_relation): Add could_be_independent_p.
21000 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
21001 (same_access_functions): Move to tree-data-ref.c.
21002 * tree-data-ref.c (ref_contains_union_access_p): New function.
21003 (access_fn_component_p): Likewise.
21004 (access_fn_components_comparable_p): Likewise.
21005 (dr_analyze_indices): Add a reference to access_fn_component_p.
21006 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
21008 (constant_access_functions): Likewise.
21009 (add_other_self_distances): Likewise.
21010 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
21011 (initialize_data_dependence_relation): Use XCNEW and remove
21012 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
21013 of access functions that have the same type. Allow the
21014 subsequence to end with different bases in some circumstances.
21015 Record the chosen access functions in SUB_ACCESS_FN.
21016 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
21017 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
21018 (subscript_dependence_tester_1): Likewise dra and drb.
21019 (build_classic_dist_vector): Update calls accordingly.
21020 (subscript_dependence_tester): Likewise.
21021 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
21022 DDR_COULD_BE_INDEPENDENT_P.
21023 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
21024 comp_alias_ddrs instead of may_alias_ddrs.
21025 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
21027 (vect_analyze_data_ref_dependence): Use it if
21028 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
21029 distance vectors if that fails.
21030 (dependence_distance_ge_vf): New function.
21031 (vect_prune_runtime_alias_test_list): Use it. Don't clear
21032 LOOP_VINFO_MAY_ALIAS_DDRS.
21034 2017-08-04 Richard Biener <rguenther@suse.de>
21036 PR middle-end/81705
21037 * fold-const.c (fold_binary_loc): Properly restrict
21038 minus_var0 && minus_var1 case when associating undefined overflow
21041 2017-08-04 Tom de Vries <tom@codesourcery.com>
21043 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
21045 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21047 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21048 Don't start diagnostic messages with a capital letter.
21049 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21051 (rs6000_invalid_builtin): Likewise.
21052 (rs6000_trampoline_init): Likewise.
21054 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21057 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
21058 after setting changeable df flags.
21060 2017-08-03 Richard Biener <rguenther@suse.de>
21062 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
21063 up if the use is in USE - X.
21065 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
21067 * toplev.c (dumpfile.h): New include.
21068 (internal_error_reentered): New static function. Use it...
21069 (internal_error_function): ...here to handle reentered internal_error.
21071 2017-08-03 Richard Biener <rguenther@suse.de>
21073 PR middle-end/81148
21074 * fold-const.c (split_tree): Add minus_var and minus_con
21075 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
21076 here but always use minus_*.
21077 (associate_trees): Assert we never associate with MINUS_EXPR
21078 and NULL first operand. Do not recurse for PLUS_EXPR operands
21079 when associating as MINUS_EXPR either.
21080 (fold_binary_loc): Track minus_var and minus_con.
21082 2017-08-03 Tom de Vries <tom@codesourcery.com>
21085 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
21086 ACCEL_COMPILER, apply finish_options on
21087 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
21089 2017-08-03 Tom de Vries <tom@codesourcery.com>
21092 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
21093 function_entry_patch_area_size > 0.
21095 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21098 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
21099 instead of 10??LU, perform unit multiplication in wide_int,
21100 don't change alloc_object_size_limit if the limit is larger
21103 PR tree-optimization/81655
21104 PR tree-optimization/81588
21105 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
21106 the case when ranges[i].low and high are 1 for unsigned type with
21109 PR middle-end/81052
21110 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
21111 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
21113 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21115 * tree-vrp.h: Add include guard.
21117 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
21120 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
21121 (ud2): New insn pattern.
21122 * config/i386/i386.c (ix86_expand_epilogue):
21123 For naked functions, generate ud2 instead of trap insn.
21125 2017-08-02 Marek Polacek <polacek@redhat.com>
21128 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
21130 2017-08-02 Tom de Vries <tom@codesourcery.com>
21131 Cesar Philippidis <cesar@codesourcery.com>
21133 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
21134 Add missing edge probabilities.
21136 2017-08-02 Tamar Christina <tamar.christina@arm.com>
21138 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
21139 Correct endianness.
21141 2017-08-02 Jakub Jelinek <jakub@redhat.com>
21143 PR middle-end/79499
21144 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
21145 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
21146 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
21147 prologue_seq sequences - if any.
21149 2017-08-02 Richard Biener <rguenther@suse.de>
21151 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
21152 via vectors if supported, integer extracts via punning if supported
21153 or otherwise vector extracts.
21155 2017-08-02 Richard Biener <rguenther@suse.de>
21157 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
21159 (bitmap_insert_into_set): ... this.
21161 2017-08-02 Richard Biener <rguenther@suse.de>
21163 PR tree-optimization/81633
21165 2015-08-17 Alan Hayward <alan.hayward@arm.com>
21167 PR tree-optimization/71752
21168 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
21170 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
21172 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
21173 (machine_function::call_ms2sysv_pad_out): Remove field.
21174 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
21175 (ix86_compute_frame_layout): Likewise.
21177 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
21180 * config/i386/i386.c (ix86_set_func_type): Disallow naked
21181 attribute with interrupt attribute.
21183 2017-08-01 Andrew Pinski <apinski@cavium.com>
21185 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
21186 BIT_INSERT_EXPR's operand 1
21187 to see if the types precision matches.
21189 2017-08-01 Martin Liska <mliska@suse.cz>
21191 PR middle-end/70140
21192 * builtins.c (expand_builtin_memcpy_args): Remove.
21193 (expand_builtin_memcpy): Call newly added function
21194 expand_builtin_memory_copy_args.
21195 (expand_builtin_memcpy_with_bounds): Likewise.
21196 (expand_builtin_mempcpy): Remove last argument.
21197 (expand_builtin_mempcpy_with_bounds): Likewise.
21198 (expand_builtin_memory_copy_args): New function created from
21199 expand_builtin_mempcpy_args with small modifications.
21200 (expand_builtin_mempcpy_args): Remove.
21201 (expand_builtin_stpcpy): Remove unused argument.
21202 (expand_builtin): Likewise.
21203 (expand_builtin_with_bounds): Likewise.
21205 2017-08-01 Martin Liska <mliska@suse.cz>
21208 Make mempcpy more optimal (PR middle-end/70140).
21210 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21213 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
21214 __builtin_vec_cmpne verify both arguments are compatible vectors
21215 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
21216 verify arg1_type is a pointer or array type. For __builtin_vec_st,
21217 move computation of aligned to after checking the argument types.
21221 * config/rs6000/vsx.md (vextract_fp_from_shorth,
21222 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
21225 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
21226 Doug Rupp <rupp@adacore.com>
21227 Olivier Hainque <hainque@adacore.com>
21229 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
21230 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
21232 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
21233 for TARGET_OS_CPP_BUILTIN.
21234 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
21235 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
21237 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
21238 passing required abi options to the assembler for EABI configurations.
21239 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
21240 of .text.hot and .text.unlikely sections for kernel modules when
21241 using ARM style exceptions.
21242 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
21243 options. Add EXTRA_CC1_SPEC.
21244 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
21246 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
21248 (ARM_TARGET2_DWARF_FORMAT): Define.
21249 * config/arm/t-vxworks: Adjust multilib control to removal of the
21250 Diab command line options.
21252 2017-08-01 Martin Liska <mliska@suse.cz>
21254 PR gcov-profile/81561
21255 * gcov.c (unblock): Make unblocking safe as we need to preserve
21256 index correspondence of blocks and block_lists.
21258 2017-08-01 Richard Biener <rguenther@suse.de>
21260 PR tree-optimization/81181
21261 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
21262 (compute_antic): ... end of iteration here.
21264 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
21266 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
21267 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
21268 (ftree-slp-vectorize): Likewise.
21269 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
21270 can no longer be set independent of flag_tree_loop_vectorize.
21271 * omp-general.c (emp_max_vf): Likewise.
21272 * opts.c (enable_fdo_optimizations): Remove references to
21273 flag_tree_vectorize, these are now implicit.
21274 (common_handle_option): Remove handling for OPT_ftree_vectorize,
21275 and leave it for the options machinery.
21277 2017-08-01 Martin Liska <mliska@suse.cz>
21279 PR middle-end/70140
21280 * builtins.c (expand_builtin_memcpy_args): Remove.
21281 (expand_builtin_memcpy): Call newly added function
21282 expand_builtin_memory_copy_args.
21283 (expand_builtin_memcpy_with_bounds): Likewise.
21284 (expand_builtin_mempcpy): Remove last argument.
21285 (expand_builtin_mempcpy_with_bounds): Likewise.
21286 (expand_builtin_memory_copy_args): New function created from
21287 expand_builtin_mempcpy_args with small modifications.
21288 (expand_builtin_mempcpy_args): Remove.
21289 (expand_builtin_stpcpy): Remove unused argument.
21290 (expand_builtin): Likewise.
21291 (expand_builtin_with_bounds): Likewise.
21293 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21296 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
21297 print "ds:" only for immediates in generic address space.
21299 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21302 * config/i386/i386.c (ix86_funciton_naked): New prototype.
21303 (ix86_function_ok_for_sibcall): Return false for naked functions.
21305 2017-08-01 Richard Biener <rguenther@suse.de>
21307 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
21308 (compute_antic): Seed worklist with exit block predecessors.
21309 * cfganal.c (dfs_find_deadend): For a cycle return the source
21310 of the edge closing it.
21312 2017-08-01 Tamar Christina <tamar.christina@arm.com>
21314 * config/aarch64/aarch64.c
21315 (aarch64_can_const_movi_rtx_p): Move 0 check.
21317 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21319 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
21320 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
21322 * match.pd: Ditto in address comparison pattern.
21324 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21326 PR tree-optimization/81627
21327 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
21328 closed ssa form for store-store chain.
21330 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21332 PR tree-optimization/81620
21333 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
21334 for store-store chain.
21336 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21338 PR tree-optimization/81588
21339 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
21340 ranges[i].in_p, invert comparison code ccode. For >/>=,
21341 swap rhs1 and rhs2 and comparison code unconditionally,
21342 for </<= don't do that. Don't swap rhs1/rhs2 again if
21343 ranges[i].in_p, instead invert comparison code ccode if
21344 opcode or oe->rank is BIT_IOR_EXPR.
21347 * optabs.def (vec_extract_optab, vec_init_optab): Change from
21348 a direct optab to conversion optab.
21349 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
21350 with GET_MODE_INNER as last argument instead of optab_handler.
21351 * expmed.c (extract_bit_field_1): Likewise. Use vector from
21352 vector extraction if possible and optab is available.
21353 * expr.c (store_constructor): Use convert_optab_handler instead
21354 of optab_handler. Use vector initialization from smaller
21355 vectors if possible and optab is available.
21356 * tree-vect-stmts.c (vectorizable_load): Likewise.
21357 * doc/md.texi (vec_extract, vec_init): Document that the optabs
21358 now have two modes.
21359 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
21360 of vec_init from half-sized vectors with the same element mode.
21361 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
21362 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
21363 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
21364 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
21365 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
21366 after mode in gen_vec_extract* calls.
21367 (vec_extract<mode>): Renamed to ...
21368 (vec_extract<mode><ssescalarmodelower>): ... this.
21369 (vec_extract<mode><ssehalfvecmodelower>): New expander.
21370 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
21371 element mode after mode in gen_vec_init* calls.
21372 (VEC_INIT_HALF_MODE): New mode iterator.
21373 (vec_init<mode>): Renamed to ...
21374 (vec_init<mode><ssescalarmodelower>): ... this.
21375 (vec_init<mode><ssehalfvecmodelower>): New expander.
21376 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
21377 (vec_extractv2sfsf): ... this.
21378 (vec_initv2sf): Renamed to ...
21379 (vec_initv2sfsf): ... this.
21380 (vec_extractv2si): Renamed to ...
21381 (vec_extractv2sisi): ... this.
21382 (vec_initv2si): Renamed to ...
21383 (vec_initv2sisi): ... this.
21384 (vec_extractv4hi): Renamed to ...
21385 (vec_extractv4hihi): ... this.
21386 (vec_initv4hi): Renamed to ...
21387 (vec_initv4hihi): ... this.
21388 (vec_extractv8qi): Renamed to ...
21389 (vec_extractv8qiqi): ... this.
21390 (vec_initv8qi): Renamed to ...
21391 (vec_initv8qiqi): ... this.
21392 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
21393 (vec_init<mode>): Renamed to ...
21394 (vec_init<mode><VEC_base_l>): ... this.
21395 (vec_extract<mode>): Renamed to ...
21396 (vec_extract<mode><VEC_base_l>): ... this.
21397 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
21398 (vec_initv2sfsf): ... this.
21399 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
21400 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21401 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
21402 element mode after mode in gen_vec_init* calls.
21403 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
21404 (vec_init<mode><Vel>): ... this.
21405 (vec_extract<mode>): Renamed to ...
21406 (vec_extract<mode><Vel>): ... this.
21407 * config/aarch64/iterators.md (Vel): New mode attribute.
21408 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
21409 Add element mode after mode in gen_vec_extract* calls.
21410 * config/s390/vector.md (non_vec_l): New mode attribute.
21411 (vec_extract<mode>): Renamed to ...
21412 (vec_extract<mode><non_vec_l>): ... this.
21413 (vec_init<mode>): Renamed to ...
21414 (vec_init<mode><non_vec_l>): ... this.
21415 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
21416 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
21418 * config/arm/iterators.md (V_elem_l): New mode attribute.
21419 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
21420 (vec_extract<mode><V_elem_l>): ... this.
21421 (vec_extractv2di): Renamed to ...
21422 (vec_extractv2didi): ... this.
21423 (vec_init<mode>): Renamed to ...
21424 (vec_init<mode><V_elem_l>): ... this.
21425 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
21426 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
21427 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
21428 Add element mode after gen_vec_extract* calls.
21429 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
21430 (vec_init<mode><unitmode>): ... this.
21431 (vec_extract<mode>): Renamed to ...
21432 (vec_extract<mode><unitmode>): ... this.
21433 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
21434 (vec_init<mode><unitmode>): ... this.
21435 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
21436 (vec_initv2sfsf): ... this.
21437 (vec_extractv2sf): Renamed to ...
21438 (vec_extractv2sfsf): ... this.
21439 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
21440 Add element mode after gen_vec_extract* calls.
21441 * config/mips/mips.md (unitmode): New mode iterator.
21442 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
21443 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
21444 * config/spu/spu.md (inner_l): New mode attribute.
21445 (vec_init<mode>): Renamed to ...
21446 (vec_init<mode><inner_l>): ... this.
21447 (vec_extract<mode>): Renamed to ...
21448 (vec_extract<mode><inner_l>): ... this.
21449 * config/sparc/sparc.md (veltmode): New mode iterator.
21450 (vec_init<VMALL:mode>): Renamed to ...
21451 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
21452 * config/ia64/vect.md (vec_initv2si): Renamed to ...
21453 (vec_initv2sisi): ... this.
21454 (vec_initv2sf): Renamed to ...
21455 (vec_initv2sfsf): ... this.
21456 (vec_extractv2sf): Renamed to ...
21457 (vec_extractv2sfsf): ... this.
21458 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
21459 (vec_init<mode>): Renamed to ...
21460 (vec_init<mode><VEC_base_l>): ... this.
21461 (vec_extract<mode>): Renamed to ...
21462 (vec_extract<mode><VEC_base_l>): ... this.
21463 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
21464 (vec_initv2sfsf): ... this.
21465 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
21466 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21467 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
21468 gen_vec_init* calls.
21470 2017-08-01 Richard Biener <rguenther@suse.de>
21472 PR tree-optimization/81297
21473 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
21474 TREE_OVERFLOW from INTEGER_CSTs.
21476 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
21478 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
21480 2017-07-31 Carl Love <cel@us.ibm.com>
21482 * config/rs6000/rs6000-c: Add support for built-in functions
21483 vector signed char vec_xl_be (signed long long, signed char *);
21484 vector unsigned char vec_xl_be (signed long long, unsigned char *);
21485 vector signed int vec_xl_be (signed long long, signed int *);
21486 vector unsigned int vec_xl_be (signed long long, unsigned int *);
21487 vector signed long long vec_xl_be (signed long long, signed long long *);
21488 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
21489 vector signed short vec_xl_be (signed long long, signed short *);
21490 vector unsigned short vec_xl_be (signed long long, unsigned short *);
21491 vector double vec_xl_be (signed long long, double *);
21492 vector float vec_xl_be (signed long long, float *);
21493 * config/rs6000/altivec.h (vec_xl_be): Add #define.
21494 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
21495 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
21497 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
21498 (altivec_expand_builtin): Add switch statement to call
21499 altivec_expand_xl_be for each builtin.
21500 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
21501 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
21502 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
21503 __builtin_vsx_le_be_v16qi.
21504 * doc/extend.texi: Update the built-in documentation file for the
21505 new built-in functions.
21507 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21510 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
21512 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21514 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21516 * config.gcc: Add z14.
21517 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
21518 CPU model numbers for z13s and z14.
21519 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
21521 * config/s390/s390-opts.h (enum processor_type): Rename
21522 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21523 * config/s390/s390.c (processor_table): Add field for CPU name to
21524 be passed to Binutils.
21525 (s390_asm_output_machine_for_arch): Use the new field in
21526 processor_table for Binutils.
21527 (s390_expand_builtin): Replace arch12 with z14.
21528 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21529 (s390_get_sched_attrmask): Likewise.
21530 (s390_get_unit_mask): Likewise.
21531 * config/s390/s390.opt: Add z14 to processor_type enum.
21533 2017-07-31 Martin Jambor <mjambor@suse.cz>
21536 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
21537 regardless of optimization level.
21539 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
21540 Martin Liska <mliska@suse.cz>
21542 * predict.def: Remove old comment and adjust probability.
21543 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
21544 PREDICT statements.
21546 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21549 * config/i386/i386.c (ix86_function_naked): New function.
21550 (ix86_can_use_return_insn_p): Return false for naked functions.
21551 (ix86_expand_prologue): Skip prologue for naked functions.
21552 (ix86_expand_epilogue): Skip epilogue for naked functions
21553 and emit trap instruction.
21554 (ix86_warn_func_return): New function.
21555 (ix86_attribute_table): Add "naked" attribute specification.
21556 (TARGET_WARN_FUNC_RETURN): Define.
21557 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
21559 2017-07-31 Martin Liska <mliska@suse.cz>
21561 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
21562 (dump_gimple_bb_header): Always dump BB info.
21563 (pp_cfg_jump): Do not append info about BB when dumping a jump.
21565 2017-07-31 Martin Liska <mliska@suse.cz>
21568 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
21569 also with current_function_decl non-null equality.
21571 2017-07-31 Jakub Jelinek <jakub@redhat.com>
21574 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
21575 change type to the element type, instead add eltype variable and
21576 use it where we are interested in the element type.
21578 PR tree-optimization/81603
21579 * ipa-polymorphic-call.c
21580 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
21581 offset arithmetic in offset_int, bail out if the resulting bit offset
21582 doesn't fit into shwi.
21584 2017-07-31 Martin Liska <mliska@suse.cz>
21586 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
21587 (gimplify_save_expr): Fix comment.
21589 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
21592 * config/i386/i386.c (ix86_function_arg): Update arguments for
21594 (ix86_compute_frame_layout): Set the initial stack offset to
21595 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
21596 INCOMING_FRAME_SP_OFFSET.
21597 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
21598 stack before exception handler returns.
21599 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
21600 the 'ERROR_CODE' for exception handler.
21602 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
21604 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
21605 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
21606 (ASM_OUTPUT_REG_POP): Ditto.
21607 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
21608 instead of asm_fprintf to output pure string.
21610 2017-07-29 Jakub Jelinek <jakub@redhat.com>
21612 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
21613 to imported_module_or_decl hook.
21614 (debug_nothing_tree_tree_tree_bool): Remove.
21615 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
21616 * debug.c (do_nothing_debug_hooks): Use
21617 debug_nothing_tree_tree_tree_bool_bool instead of
21618 debug_nothing_tree_tree_tree_bool.
21619 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21620 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
21621 * sdbout.c (sdb_debug_hooks): Likewise.
21622 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
21623 (gen_namespace_die): Add DW_AT_export_symbols attribute if
21625 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
21626 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
21627 attribute, don't add anything.
21629 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21631 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
21632 (fold_build2_stat_loc): Likewise.
21633 (fold_build3_stat_loc): Likewise.
21634 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
21635 (fold_build1_loc): Remove macro.
21636 (fold_build2_loc): Likewise.
21637 (fold_build3_loc): Likewise.
21639 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21641 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
21642 (gimple_build_debug_bind_source_stat): Likewise.
21643 * gimple.h (gimple_build_debug_bind): Remove macro.
21644 (gimple_build_debug_bind_source): Likewise.
21646 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21648 * bitmap.c (bitmap_alloc): Adjust.
21649 (bitmap_gc_alloc): Likewise.
21650 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
21652 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21654 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
21655 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
21656 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
21657 (bitmap_gc_alloc_stat): Likewise.
21658 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
21660 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21662 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
21663 * rtl.h (shallow_copy_rtx): Remove macro.
21665 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21667 * emit-rtl.c (gen_raw_REG): Adjust.
21668 * gengenrtl.c (gendef): Likewise.
21669 * rtl.c (rtx_alloc_stat): Remove _stat from name.
21670 * rtl.h (rtx_alloc): Remove macro.
21672 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21674 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
21675 (build_tree_list_stat): Likewise.
21676 * tree.h (build_tree_list): Remove macro.
21677 (build_tree_list_vec): Likewise.
21679 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21681 * tree.c (make_vector_stat): Remove _stat from name.
21682 (build_vector_stat): Likewise.
21683 * tree.h (make_vector_stat): Remove macro.
21684 (build_vector_stat): Likewise.
21686 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21688 * tree.h (build_var_debug_value): Remove prototype.
21690 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21692 * tree.c (tree_cons_stat): Remove _stat from name.
21693 * tree.h (tree_cons): Remove macro.
21695 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21697 * tree.c (build_vl_exp_stat): Remove _stat from name.
21698 * tree.h (build_vl_exp): Remove macro.
21700 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21702 * tree.c (build_decl_stat): Remove _stat from name.
21703 * tree.h (build_decl): Remove macro.
21705 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21707 * gimple.c (gimple_build_with_ops_stat): Adjust.
21708 (gimple_alloc_stat): Remove _stat from name.
21709 * gimple.h (gimple_alloc): Remove macro.
21711 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21713 * tree.c (make_tree_vec_stat): Remove _stat from name.
21714 (grow_tree_vec_stat): Likewise.
21715 * tree.h (make_tree_vec_stat): Adjust prototype.
21716 (grow_tree_vec_stat): Likewise.
21717 (make_tree_vec): Remove macro.
21718 (grow_tree_vec): Likewise.
21720 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21722 * fold-const.c (fold_build1_stat_loc): Adjust.
21723 (fold_build2_stat_loc): Likewise.
21724 (fold_build3_stat_loc): Likewise.
21725 * tree.c (build0_stat): Remove _stat from name.
21726 (build1_stat): Likewise.
21727 (build2_stat): Likewise.
21728 (build3_stat): Likewise.
21729 (build4_stat): Likewise.
21730 (build5_stat): Likewise.
21731 * tree.h (build1_loc): Remove macro, and rename _stat function
21733 (build2_loc): Likewise.
21734 (build3_loc): Likewise.
21735 (build4_loc): Likewise.
21736 (build5_loc): Likewise.
21738 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21740 * tree.c (make_int_cst_stat): Remove _stat from name.
21741 * tree.h (make_int_cst_stat): Adjust prototype.
21742 (make_int_cst): Remove macro.
21744 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21746 * tree.c (make_tre_binfo_stat): Remove _stat from name.
21747 * tree.h (make_tree_binfo_stat): Adjust prototype.
21748 (make_tree_binfo): Remove.
21750 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21752 * tree.c (copy_node_stat): Rename to copy_node.
21753 (build_distinct_type_copy): Adjust.
21754 * tree.h (copy_node_stat): Adjust prototype.
21755 (copy_node): Remove macro.
21757 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21759 * tree.c (make_node_stat): rename to make_node.
21760 (build_tree_list_stat): Adjust.
21761 (build0_stat): Likewise.
21762 (build2_stat): Likewise.
21763 (build3_stat): Likewise.
21764 (build4_stat): Likewise.
21765 (build5_stat): Likewise.
21766 (build_decl_stat): Likewise.
21767 * tree.h (make_node_stat): Adjust prototype.
21768 (make_node): remove macro.
21770 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
21772 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
21773 (PPC_FEATURE2_SCV): Likewise.
21774 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
21776 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21778 * config/aarch64/aarch64.c
21779 (aarch64_internal_mov_immediate): Add new special pattern.
21780 * config/aarch64/aarch64.md (*movdi_aarch64):
21781 Add reg/32bit const mov case.
21783 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21784 Richard Sandiford <richard.sandiford@linaro.org>
21786 * config/aarch64/aarch64.md (mov<mode>): Generalize.
21787 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21788 Add integer and movi cases.
21789 (movi-split-hf-df-sf split, fp16): New.
21790 (enabled): Added TARGET_FP_F16INST.
21791 * config/aarch64/iterators.md (GPF_HF): New.
21792 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
21794 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21796 * config/aarch64/aarch64.c
21797 (aarch64_simd_container_mode): Add prototype.
21798 (aarch64_expand_mov_immediate): Add HI support.
21799 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
21800 (aarch64_can_const_movi_rtx_p): New.
21801 (aarch64_preferred_reload_class):
21802 Remove restrictions of using FP registers for certain SIMD operations.
21803 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
21804 (aarch64_valid_floating_const): Add integer move validation.
21805 (aarch64_simd_imm_scalar_p): Remove.
21806 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
21807 (aarch64_legitimate_constant_p): Expand list of supported cases.
21808 * config/aarch64/aarch64-protos.h
21809 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
21810 (aarch64_reinterpret_float_as_int): New.
21811 (aarch64_simd_imm_scalar_p): Remove.
21812 * config/aarch64/constraints.md (Uvi): New.
21813 (Dd): Split into Ds and new Dd.
21814 * config/aarch64/aarch64.md (*movsi_aarch64):
21816 (*movdi_aarch64): Add SIMD mov case.
21818 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21820 * tree-predcom.c: (struct chain): Handle store-store chain in which
21821 stores for elimination only store loop invariant values.
21822 (execute_pred_commoning_chain): Ditto.
21823 (prepare_initializers_chain_store_elim): Ditto.
21824 (prepare_finalizers): Ditto.
21825 (is_inv_store_elimination_chain): New function.
21826 (initialize_root_vars_store_elim_1): New function.
21828 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21830 * tree-predcom.c: Revise general description of the pass.
21831 (enum chain_type): New enum type for store elimination.
21832 (struct chain): New field supporting store elimination.
21833 (struct component): Ditto.
21834 (dump_chain): Dump store-stores chain.
21835 (release_chain): Release resources.
21836 (split_data_refs_to_components): Compute and create component
21837 contains only stores for elimination.
21838 (get_chain_last_ref_at): New function.
21839 (make_invariant_chain): Initialization.
21840 (make_rooted_chain): Specify chain type in parameter and record it.
21841 (add_looparound_copies): Skip for store-stores chain.
21842 (determine_roots_comp): Compute type of chain and pass it to
21844 (initialize_root_vars_store_elim_2): New function.
21845 (finalize_eliminated_stores): New function.
21846 (remove_stmt): Handle store for elimination.
21847 (execute_pred_commoning_chain): Execute predictive commoning on
21848 store-store chains.
21849 (determine_unroll_factor): Skip unroll for store-stores chain.
21850 (prepare_initializers_chain_store_elim): New function.
21851 (prepare_initializers_chain): Hanlde store-store chain.
21852 (prepare_finalizers_chain, prepare_finalizers): New function.
21853 (tree_predictive_commoning_loop): Return integer value indicating
21854 if loop is unrolled or lcssa form is corrupted.
21855 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
21857 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21859 * tree-predcom.c (initialize_root): Delete.
21860 (execute_pred_commoning_chain): Initialize root vars and replace
21861 reference of non-combined chain directly, rather than call above
21864 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21866 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
21867 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
21869 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21871 * tree-predcom.c (struct chain): New field init_seq.
21872 (release_chain): Release init_seq.
21873 (prepare_initializers_chain): Record intialization stmts in above
21875 (insert_init_seqs): New function.
21876 (tree_predictive_commoning_loop): Call insert_init_seqs.
21878 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21880 * tree-predcom.c (determine_roots_comp): Skip trivial components.
21882 2017-07-28 Richard Biener <rguenther@suse.de>
21884 * match.pd: Remove superfluous :c.
21885 * genmatch.c (simplify::id): Add member.
21886 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
21888 (current_id): New global.
21889 (dt_node::parent): Move from ...
21890 (dt_operand::parent): ... here. Add for_id member.
21891 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
21892 (decision_tree::find_node): Relax order requirement when
21893 merging DT_TRUE nodes to ones inbetween the current simplify
21894 and the one we try to merge with. Add diagnostic whenever
21895 we need to enforce pattern order by not merging.
21896 (decision_tree::insert): Set current_id.
21897 (decision_tree::print_node): Dump parent node and for_id.
21898 (parser::last_id): Add member.
21899 (parser::push_simplify): Assign unique id.
21900 (parser::parser): Initialize last_id.
21902 2017-07-28 Martin Liska <mliska@suse.cz>
21905 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
21906 gimple_build_debug_bind.
21908 2017-07-28 Richard Biener <rguenther@suse.de>
21910 PR tree-optimization/81502
21911 * match.pd: Add pattern combining BIT_INSERT_EXPR with
21913 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
21915 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
21916 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
21917 for BIT_FIELD_REF args.
21918 * fold-const.c (make_bit_field_ref): Likewise.
21919 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
21921 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21924 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
21925 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
21926 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
21927 Or it into SANITIZER_UNDEFINED.
21928 * ubsan.c: Include gimple-fold.h and varasm.h.
21929 (ubsan_expand_ptr_ifn): New function.
21930 (instrument_pointer_overflow): New function.
21931 (maybe_instrument_pointer_overflow): New function.
21932 (instrument_object_size): Formatting fix.
21933 (pass_ubsan::execute): Call instrument_pointer_overflow
21934 and maybe_instrument_pointer_overflow.
21935 * internal-fn.c (expand_UBSAN_PTR): New function.
21936 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
21937 * sanitizer.def (__ubsan_handle_pointer_overflow,
21938 __ubsan_handle_pointer_overflow_abort): New builtins.
21939 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
21940 * internal-fn.def (UBSAN_PTR): New internal function.
21941 * opts.c (sanitizer_opts): Add pointer-overflow.
21942 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
21943 * fold-const.c (build_range_check): Compute pointer range check in
21944 integral type if pointer arithmetics would be needed. Formatting
21947 2017-07-28 Martin Liska <mliska@suse.cz>
21950 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
21951 parameters that are of a variable-length.
21953 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
21955 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
21957 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
21958 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
21959 (CRT_CALL_STATIC_FUNCTION): Likewise.
21960 (ASM_DEFAULT_SPEC): New define.
21961 (ASM_SPEC32): Likewise.
21962 (ASM_SPEC64): Likewise.
21963 (ASM_SPEC_COMMON): Likewise.
21964 (ASM_SPEC): Likewise.
21965 (INVALID_64BIT): Likewise.
21966 (LINK_OS_DEFAULT_SPEC): Likewise.
21967 (LINK_OS_SPEC32): Likewise.
21968 (LINK_OS_SPEC64): Likewise.
21969 (POWERPC_LINUX): Likewise.
21970 (PTRDIFF_TYPE): Likewise.
21971 (RESTORE_FP_PREFIX): Likewise.
21972 (RESTORE_FP_SUFFIX): Likewise.
21973 (SAVE_FP_PREFIX): Likewise.
21974 (SAVE_FP_SUFFIX): Likewise.
21975 (SIZE_TYPE): Likewise.
21976 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
21977 (TARGET_64BIT): Likewise.
21978 (TARGET_64BIT): Likewise.
21979 (TARGET_AIX): Likewise.
21980 (WCHAR_TYPE_SIZE): Likewise.
21981 (WCHAR_TYPE): Undefine.
21982 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
21983 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
21984 (CPP_OS_RTEMS_SPEC): Delete.
21985 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
21986 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
21988 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
21990 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21992 PR tree-optimization/81578
21993 * tree-parloops.c (build_new_reduction): Bail out if
21994 reduction_code isn't one of the standard OpenMP reductions.
21995 Move the details printing after that decision.
21997 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21999 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
22000 related to reload_in_progress.
22001 (splat_input_operand): Likewise.
22002 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
22004 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
22006 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
22007 (TARGET_INSTANTIATE_DECLS): Likewise.
22008 (legitimate_indexed_address_p): Delete reload_in_progress code.
22009 (rs6000_debug_legitimate_address_p): Likewise.
22010 (rs6000_eliminate_indexed_memrefs): Likewise.
22011 (rs6000_emit_le_vsx_store): Likewise.
22012 (rs6000_emit_move_si_sf_subreg): Likewise.
22013 (rs6000_emit_move): Likewise.
22014 (register_to_reg_type): Likewise.
22015 (rs6000_pre_atomic_barrier): Likewise.
22016 (rs6000_machopic_legitimize_pic_address): Likewise.
22017 (rs6000_allocate_stack_temp): Likewise.
22018 (rs6000_address_for_fpconvert): Likewise.
22019 (rs6000_address_for_altivec): Likewise.
22020 (rs6000_secondary_memory_needed_rtx): Delete function.
22021 (rs6000_check_sdmode): Likewise.
22022 (rs6000_alloc_sdmode_stack_slot): Likewise.
22023 (rs6000_instantiate_decls): Likewise.
22024 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
22025 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
22026 Delete reload_in_progress.
22027 (*vec_reload_and_plus_<mptrsize>): Likewise.
22028 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
22029 (vsx_div_v2di): Likewise.
22030 (vsx_udiv_v2di): Likewise.
22032 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22034 * config/rs6000/rs6000.opt (mlra): Replace with stub.
22035 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
22036 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
22037 (rs6000_debug_reg_global): Delete print of LRA status.
22038 (rs6000_option_override_internal): Delete dead LRA related code.
22039 (rs6000_lra_p): Delete function.
22040 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
22042 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
22044 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
22045 * config/riscv/rtems.h: New file.
22047 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22048 Sudakshina Das <sudi.das@arm.com>
22050 * config/aarch64/aarch64.md
22051 (define_split for and<mode>3nr_compare): Move
22052 non aarch64_logical_operand to a register.
22053 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
22054 register immediate operand to a register.
22055 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
22057 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22059 PR middle-end/81564
22060 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
22062 2017-07-27 Richard Biener <rguenther@suse.de>
22064 PR tree-optimization/81573
22065 PR tree-optimization/81494
22066 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
22067 multi defuse cycle case.
22069 2017-07-27 Richard Biener <rguenther@suse.de>
22071 PR tree-optimization/81571
22072 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
22075 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
22077 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
22078 earlier and only if MASK_FPU is set. Adjust formatting.
22080 2017-07-27 Martin Liska <mliska@suse.cz>
22082 * opt-functions.awk: Add validation of value of Init.
22083 * optc-gen.awk: Pass new argument.
22085 2017-07-27 Martin Liska <mliska@suse.cz>
22087 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
22088 Fix wrong condition.
22090 2017-07-27 Martin Liska <mliska@suse.cz>
22092 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
22093 BBs and edges seen by autoFDO.
22095 2017-07-27 Richard Biener <rguenther@suse.de>
22097 PR tree-optimization/81502
22098 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
22099 with incompatible but same sized type.
22100 (execute_update_addresses_taken): Likewise.
22102 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
22104 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
22105 flag_tree_loop_vectorize rather than flag_tree_vectorize.
22107 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22110 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
22111 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
22112 Change s_operand to memory_operand.
22114 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
22116 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
22117 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
22118 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
22119 Emit instructions rather than returning an expression. Handle TFmode
22120 and KFmode by casting to TImode.
22121 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
22122 (rs6000_emit_le_vsx_store): Likewise.
22123 * config/rs6000/vsx.md (VSX_TI): New iterator.
22124 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
22125 (*vsx_le_undo_permute_<mode>): Likewise.
22126 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
22127 emit the split sequence.
22128 (*vsx_le_perm_store_<mode>): Likewise.
22130 2017-07-27 Jakub Jelinek <jakub@redhat.com>
22132 PR tree-optimization/81555
22133 PR tree-optimization/81556
22134 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
22135 if true, force CHANGED for the recursive invocation.
22136 (reassociate_bb): Remember original length of ops array, pass
22137 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
22139 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
22140 attributes for noipa attribute. For naked attribute use
22141 lookup_attribute first before lookup_attribute_spec.
22142 * final.c (rest_of_handle_final): Disable IPA RA for functions with
22144 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
22145 for functions with noipa attribute.
22146 (cgraph_externally_visible_p): Return true for functions with noipa
22148 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
22149 for functions with noipa attribute.
22150 * doc/extend.texi: Document noipa function attribute.
22151 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
22152 also for functions with noipa attribute.
22153 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
22155 2017-07-26 Andrew Pinski <apinski@cavium.com>
22157 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
22158 vec_unalign_load_cost and vec_unalign_store_cost.
22160 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
22162 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
22163 -mvsx-small-integer option.
22164 (ISA_3_0_MASKS_IEEE): Likewise.
22165 (OTHER_VSX_VECTOR_MASKS): Likewise.
22166 (POWERPC_MASKS): Likewise.
22167 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
22168 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
22169 code, only testing for DImode being allowed in non-VSX floating
22171 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
22172 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
22174 (rs6000_option_override_internal): Delete -mvsx-small-integer.
22175 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
22176 TARGET_P8_VECTOR test.
22177 (rs6000_secondary_reload_simple_move): Likewise.
22178 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
22179 since TARGET_P9_VECTOR was already tested.
22180 (rs6000_opt_masks): Remove -mvsx-small-integer.
22181 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
22182 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22184 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
22185 test for TARGET_VEXTRACTUB was used, and that uses
22187 (p9 extract splitter): Likewise.
22188 (vsx_extract_<mode>_di_p9): Likewise.
22189 (vsx_extract_<mode>_store_p9): Likewise.
22190 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
22191 for TARGET_P9_VECTOR was used. Delete code that is now dead with
22192 the elimination of TARGET_VSX_SMALL_INTEGER.
22193 (vsx_extract_<mode>_p8): Likewise.
22194 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
22195 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
22196 (vsx_set_<mode>_p9): Likewise.
22197 (vsx_set_v4sf_p9): Likewise.
22198 (vsx_set_v4sf_p9_zero): Likewise.
22199 (vsx_insert_extract_v4sf_p9): Likewise.
22200 (vsx_insert_extract_v4sf_p9_2): Likewise.
22201 * config/rs6000/rs6000.md (sign extend splitter): Change
22202 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
22203 (floatsi<mode>2_lfiwax_mem): Likewise.
22204 (floatunssi<mode>2_lfiwzx_mem): Likewise.
22205 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
22206 since a test for TARGET_P9_VECTOR was used.
22207 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22208 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
22209 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22210 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22211 TARGET_P8_VECTOR test.
22212 (fix_trunc<mode>si2_stfiwx): Likewise.
22213 (fix_trunc<mode>si2_internal): Likewise.
22214 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
22215 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22217 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22218 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22219 TARGET_P8_VECTOR test.
22220 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22221 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
22222 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22224 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22225 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
22226 since a test for TARGET_P9_VECTOR was used.
22227 (splitter for loading small constants): Likewise.
22229 2017-07-26 Andrew Pinski <apinski@cavium.com>
22231 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
22234 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
22237 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
22238 (fp_valid_at): Likewise.
22240 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22242 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
22243 (qdf24xx_addrcost_table): Likewise.
22244 (cortexa57_tunings): Update to use generic_branch_cost.
22245 (cortexa72_tunings): Likewise.
22246 (cortexa73_tunings): Likewise.
22247 (qdf24xx_tunings): Likewise.
22249 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22251 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
22252 (thunderx2t99_branch_cost): Likewise.
22253 (cortexa35_tunings): Update to use generic_branch_cost.
22254 (cortexa53_tunings): Likewise.
22255 (cortexa57_tunings): Likewise.
22256 (cortexa72_tunings): Likewise.
22257 (cortexa73_tunings): Likewise.
22258 (thunderx2t99_tunings): Likewise.
22260 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22262 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
22263 (sparc_option_override): Honour MASK_FSMULD.
22264 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
22265 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
22266 * config/sparc/sparc.opt (mfsmuld): New option.
22267 * doc/invoke.texi (mfsmuld): Document option.
22269 2017-07-26 Marek Polacek <polacek@redhat.com>
22271 PR middle-end/70992
22272 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
22274 2017-07-26 Richard Biener <rguenther@suse.de>
22276 * gimple-match-head.c (do_valueize): Return OP if valueize
22278 (get_def): New helper to get at the def stmt of a SSA name
22279 if valueize allows.
22280 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
22281 do_valueize to get at the def stmt.
22282 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
22284 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
22286 PR middle-end/46932
22287 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
22289 2017-07-26 Martin Liska <mliska@suse.cz>
22292 * function.c (expand_function_start): Make expansion of
22293 nonlocal_goto_save_area after parm_birth_insn.
22295 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22297 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
22298 from all CPU target flags enable members.
22300 2017-07-26 Richard Biener <rguenther@suse.de>
22302 * genmatch.c (dt_simplify::gen): Make iterator vars const.
22303 (decision_tree::gen): Make 'type' const.
22304 (write_predicate): Likewise.
22306 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22308 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
22309 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
22310 (rs6000_option_override_internal): Likewise.
22311 (rs6000_expand_vector_set): Likewise.
22312 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
22313 (TARGET_UPPER_REGS_SF): Likewise.
22314 (TARGET_UPPER_REGS_DI): Likewise.
22315 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
22316 (TARGET_DIRECT_MOVE_64BIT): Likewise.
22317 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
22318 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22319 (Splitters for DI constants in Altivec registers): Likewise.
22320 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
22321 (vsx_set_v4sf_p9): Likewise.
22322 (vsx_set_v4sf_p9_zero): Likewise.
22323 (vsx_insert_extract_v4sf_p9): Likewise.
22324 (vsx_insert_extract_v4sf_p9_2): Likewise.
22326 2017-07-25 Carl Love <cel@us.ibm.com>
22328 * doc/extend.texi: Update the built-in documentation file for the
22329 existing built-in functions
22330 vector signed char vec_cnttz (vector signed char);
22331 vector unsigned char vec_cnttz (vector unsigned char);
22332 vector signed short vec_cnttz (vector signed short);
22333 vector unsigned short vec_cnttz (vector unsigned short);
22334 vector signed int vec_cnttz (vector signed int);
22335 vector unsigned int vec_cnttz (vector unsigned int);
22336 vector signed long long vec_cnttz (vector signed long long);
22337 vector unsigned long long vec_cnttz (vector unsigned long long);
22339 2017-07-25 Andrew Pinski <apinski@cavium.com>
22341 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
22342 accesses where the use is for the first operand of a BIT_INSERT.
22344 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22347 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22348 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22350 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22352 * config/i386/gstabs.h: Delete.
22353 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
22355 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
22357 * config/i386/i386.c (ix86_decompose_address): Do not check for
22358 register RTX when looking at index_reg or base_reg.
22359 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
22361 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
22363 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
22364 to update EH info here.
22366 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22368 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
22370 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22372 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
22374 2017-07-25 Torsten Duwe <duwe@suse.de>
22376 * common.opt: Introduce -fpatchable-function-entry
22377 command line option, and its variables function_entry_patch_area_size
22378 and function_entry_patch_area_start.
22379 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
22380 including a two-value parser.
22381 * target.def (print_patchable_function_entry): New target hook.
22382 * targhooks.h (default_print_patchable_function_entry): New function.
22383 * targhooks.c (default_print_patchable_function_entry): Likewise.
22384 * toplev.c (process_options): Switch off IPA-RA if
22385 patchable function entries are being generated.
22386 * varasm.c (assemble_start_function): Look at the
22387 patchable-function-entry command line switch and current
22388 function attributes and maybe generate NOP instructions by
22389 calling the print_patchable_function_entry hook.
22390 * doc/extend.texi: Document patchable_function_entry attribute.
22391 * doc/invoke.texi: Document -fpatchable_function_entry
22392 command line option.
22393 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
22395 * doc/tm.texi: Re-generate.
22397 2017-07-25 Jakub Jelinek <jakub@redhat.com>
22400 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
22401 TARGET_AVX512DQ rather than TARGET_AVX512BW.
22403 2017-07-25 Tamar Christina <tamar.christina@arm.com>
22405 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
22407 2017-07-25 Richard Biener <rguenther@suse.de>
22409 PR tree-optimization/81455
22410 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
22411 not walk in cycles when looking for guards.
22413 2017-07-25 Richard Biener <rguenther@suse.de>
22415 PR tree-optimization/81529
22416 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
22417 when optimizing backedge uses.
22419 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
22421 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
22423 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
22424 to dl_section_ref. On AIX, append an expression to subtract
22425 the size of the section length to dl_section_ref.
22427 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
22429 * configure.ac: If any of the config.* scripts fail, exit 1.
22430 * configure: Regenerate.
22432 2017-07-25 Richard Biener <rguenther@suse.de>
22434 PR middle-end/81546
22435 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
22436 of immediate uses, be more verbose on errors.
22438 2017-07-25 Richard Biener <rguenther@suse.de>
22440 PR tree-optimization/81510
22441 * tree-vect-loop.c (vect_is_simple_reduction): When the
22442 reduction stmt is not inside the loop bail out.
22444 2017-07-25 Richard Biener <rguenther@suse.de>
22446 PR tree-optimization/81303
22447 * tree-vect-loop-manip.c (vect_loop_versioning): Build
22448 profitability check against LOOP_VINFO_NITERSM1.
22450 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22452 * domwalk.c (cmp_bb_postorder): Simplify.
22453 (sort_bbs_postorder): New function. Use it...
22454 (dom_walker::walk): ...here to optimize common cases.
22456 2017-07-25 Martin Liska <mliska@suse.cz>
22459 * ipa-visibility.c (function_and_variable_visibility): Make the
22460 redirection just on target that supports aliasing.
22461 Fix GNU coding style.
22463 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
22466 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
22468 * config/arm/rtems.h: Likewise.
22469 * config/bfin/rtems.h: Likewise.
22470 * config/i386/rtemself.h: Likewise.
22471 * config/lm32/rtems.h: Likewise.
22472 * config/m32c/rtems.h: Likewise.
22473 * config/m68k/rtemself.h: Likewise.
22474 * config/microblaze/rtems.h: Likewise.
22475 * config/mips/rtems.h: Likewise.
22476 * config/moxie/rtems.h: Likewise.
22477 * config/nios2/rtems.h: Likewise.
22478 * config/powerpcspe/rtems.h: Likewise.
22479 * config/rs6000/rtems.h: Likewise.
22480 * config/rtems.h: Likewise.
22481 * config/sh/rtems.h: Likewise.
22482 * config/sh/rtemself.h: Likewise.
22483 * config/sparc/rtemself.h: Likewise.
22485 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22488 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
22489 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
22490 * tree-ssa-structalias.c (alias_get_name): Same.
22492 2017-07-25 Bin Cheng <bin.cheng@arm.com>
22495 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
22496 instructions if no du chain is found.
22498 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22500 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
22502 2017-07-25 Richard Biener <rguenther@suse.de>
22504 PR middle-end/81505
22505 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
22508 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22510 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
22511 upper-regs options.
22512 (ISA_2_7_MASKS_SERVER): Likewise.
22513 (ISA_3_0_MASKS_IEEE): Likewise.
22514 (OTHER_P8_VECTOR_MASKS): Likewise.
22515 (OTHER_VSX_VECTOR_MASKS): Likewise.
22516 (POWERPC_MASKS): Likewise.
22517 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
22518 duplicate list of options.
22519 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
22520 explicit -mupper-regs options.
22521 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
22522 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
22523 alias for -mupper-regs-df.
22524 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
22525 (rs6000_init_hard_regno_mode_ok): Likewise.
22526 (rs6000_option_override_internal): Likewise.
22527 (rs6000_opt_masks): Likewise.
22528 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
22529 options in terms of whether -mvsx or -mpower8-vector was used.
22530 (TARGET_UPPER_REGS_DI): Likewise.
22531 (TARGET_UPPER_REGS_SF): Likewise.
22532 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
22533 -mupper-regs-* options.
22535 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22537 * passes.c (emergency_dump_function): Print some empty lines and a
22538 header before the RTL dump.
22540 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22542 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
22544 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
22547 * config/aarch64/aarch64.c (aarch64_classify_symbol):
22548 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
22550 2017-07-24 Carl Love <cel@us.ibm.com>
22552 * config/rs6000/rs6000-c.c: Add support for built-in functions
22553 vector float vec_extract_fp32_from_shorth (vector unsigned short);
22554 vector float vec_extract_fp32_from_shortl (vector unsigned short);
22555 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
22556 vec_extract_fp_from_shortl): Add defines for the two builtins.
22557 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
22558 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
22560 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
22561 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
22562 * doc/extend.texi: Update the built-in documentation file for the
22563 new built-in function.
22565 2017-07-24 Jakub Jelinek <jakub@redhat.com>
22568 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
22570 * doc/generic.texi: Likewise.
22571 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22572 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22574 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
22576 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
22577 (aarch64_mls_elt_merge<mode>): Likewise.
22579 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
22581 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22582 having __cxa_atexit.
22584 2017-07-23 Michael Collison <michael.collison@arm.com>
22586 * config/arm/arm.c (arm_option_override): Deprecate
22587 use of -mstructure-size-boundary.
22588 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
22589 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
22591 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22594 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22595 Reduce cost estimate for direct moves.
22597 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
22600 * config/i386/i386.c (ix86_option_override_internal): Disable
22601 BMI, BMI2 and TBM instructions for -m16.
22603 2017-07-21 Carl Love <cel@us.ibm.com>
22605 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22606 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22607 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22608 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22609 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22610 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22611 VMULOSW): New enum "unspec" values.
22612 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22613 altivec_vmulosw): New patterns.
22614 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22615 VMULOSW): Add definitions.
22617 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
22619 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
22620 (qdf24xx): Likewise.
22621 * config/aarch64/aarch64-options-extensions.def (rdma); New.
22622 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
22623 (AARCH64_FL_V8_1): Renumber.
22624 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
22625 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
22626 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
22627 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
22628 rdma to feature modifiers list.
22630 2017-07-21 Yury Gribov <tetra2005@gmail.com>
22632 PR middle-end/56727
22633 * ipa-visibility (function_and_variable_visibility): Convert
22634 recursive PLT call to direct call if appropriate.
22636 2017-07-21 Andrew Pinski <apinski@cavium.com>
22638 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
22639 operand 1 to see if the types precision matches.
22640 * fold-const.c (operand_equal_p): Likewise.
22642 2017-07-21 Richard Biener <rguenther@suse.de>
22644 PR tree-optimization/81303
22645 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
22646 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
22647 (vect_peeling_hash_get_lowest_cost): Adjust.
22648 (vect_enhance_data_refs_alignment): Likewise. Use
22649 vect_get_peeling_costs_all_drs to compute the penalty for no
22650 peeling to match up costs.
22652 2017-07-21 Richard Biener <rguenther@suse.de>
22654 PR tree-optimization/81500
22655 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
22656 we didn't identify a reduction path.
22658 2017-07-21 Tom de Vries <tom@codesourcery.com>
22659 Cesar Philippidis <cesar@codesourcery.com>
22661 PR gcov-profile/81442
22662 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
22665 2017-07-21 Tom de Vries <tom@codesourcery.com>
22668 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
22670 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
22671 nvptx_override_options_after_change.
22673 2017-07-21 Ulrich Drepper <drepper@redhat.com>
22675 * dwarf2out.c (output_file_names): Avoid double testing for
22676 dwarf_version >= 5.
22678 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
22680 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
22682 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22684 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
22686 (try_crossjump_to_edge): Do not punt on partitioned functions.
22688 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22690 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
22691 Put all BBs reachable only via paths crossing cold region to cold
22693 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
22695 2016-07-21 Richard Biener <rguenther@suse.de>
22697 PR tree-optimization/81303
22698 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
22699 into account prologue and epilogue iterations when raising
22700 min_profitable_iters to sth at least covering one vector iteration.
22702 2017-07-21 Tamar Christina <tamar.christina@arm.com>
22704 * config/arm/arm.c (arm_test_cpu_arch_dat):
22707 2017-07-20 Nathan Sidwell <nathan@acm.org>
22709 Remove TYPE_METHODS.
22710 * tree.h (TYPE_METHODS): Delete.
22711 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
22712 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
22713 (dbxout_type_methods): Scan TYPE_FIELDS.
22714 (dbxout_type): Don't check TYPE_METHODS here.
22715 * function.c (use_register_for_decl): Always ignore register for
22716 class types when not optimizing.
22717 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
22718 * tree.c (free_lang_data_in_type): Stitch out member functions and
22719 templates from TYPE_FIELDS.
22720 (build_distinct_type_copy, verify_type_variant,
22721 verify_type): Member fns are on TYPE_FIELDS.
22722 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
22723 * tree-pretty-print.c (dump_generic_node): Likewise.
22725 2017-07-20 Jakub Jelinek <jakub@redhat.com>
22728 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
22729 V2TImode and V4TImode.
22730 (ix86_expand_vector_extract): Likewise.
22731 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
22732 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
22733 (ssescalarmode): Handle V4TImode and V2TImode.
22734 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
22735 (*vec_extractv2ti, *vec_extractv4ti): New insns.
22736 (VEXTRACTI128_MODE): New mode iterator.
22737 (splitter for *vec_extractv?ti first element): New.
22738 (VEC_INIT_MODE): New mode iterator.
22739 (vec_init<mode>): Consolidate 3 expanders into one using
22740 VEC_INIT_MODE mode iterator.
22742 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22744 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
22745 non_spilled_static_chain_regno_p.
22747 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22749 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
22751 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
22753 * bb-reorder.c (connect_traces): Allow copying of blocks within
22756 2017-07-20 Richard Biener <rguenther@suse.de>
22758 * gimple.h (gimple_phi_result): Add gphi * overload.
22759 (gimple_phi_result_ptr): Likewise.
22760 (gimple_phi_arg): Likewise. Adjust index assert to only
22761 allow actual argument accesses rather than all slots available
22763 (gimple_phi_arg_def): Add gphi * overload.
22764 * tree-phinodes.c (make_phi_node): Initialize only actual
22766 (resize_phi_node): Clear memory not covered by old node,
22767 do not initialize excess argument slots.
22768 (reserve_phi_args_for_new_edge): Initialize new argument slot
22771 2017-07-20 Bin Cheng <bin.cheng@arm.com>
22773 PR tree-optimization/81388
22775 2016-07-21 Bin Cheng <bin.cheng@arm.com>
22777 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
22778 by removing computation of may_be_zero.
22780 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22781 Tom de Vries <tom@codesourcery.com>
22783 PR middle-end/81030
22784 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
22785 when gimple level profile disagrees with what RTL expander did.
22787 2017-07-20 Richard Biener <rguenther@suse.de>
22789 PR tree-optimization/61171
22790 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
22791 (vect_analyze_stmt): Add slp instance parameter.
22792 (vectorizable_reduction): Likewise.
22793 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
22794 (vect_is_simple_reduction): Deal with chains not detected
22795 as SLP reduction chain, specifically not properly associated
22796 chains containing a mix of plus/minus.
22797 (get_reduction_op): Remove.
22798 (get_initial_defs_for_reduction): Simplify, pass in whether
22799 this is a reduction chain, pass in the SLP node for the PHIs.
22800 (vect_create_epilog_for_reduction): Get the SLP instance as
22802 (vectorizable_reduction): Get the SLP instance as arg.
22803 During analysis remember the SLP node with the PHIs in the
22804 instance. Simplify getting at the vectorized reduction PHIs.
22805 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
22806 through SLP instance.
22807 (vect_slp_analyze_operations): Likewise.
22808 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
22809 (vect_transform_stmt): Likewise.
22811 2017-07-20 Tom de Vries <tom@codesourcery.com>
22813 PR tree-optimization/81489
22814 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
22815 read of phi arg location to before loop that modifies phi.
22817 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22819 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
22822 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22824 PR middle-end/81331
22825 * except.c (execute): Fix ordering issue.
22827 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22829 PR rtl-optimization/81423
22830 * combine.c (make_compound_operation_int): Don't try to optimize
22831 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
22833 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22835 PR rtl-optimization/81423
22836 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
22837 with a constant that is -1 in the truncated to mode.
22839 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22841 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
22842 (determine_unlikely_bbs): ... here.
22843 * predict.h (propagate_unlikely_bbs_forward): Declare.
22844 * cfgexpand.c (pass_expand::execute): Use it.
22845 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
22847 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
22848 propagate_unlikely_bbs_forward.
22850 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22852 PR middle-end/81331
22853 * except.c (maybe_add_nop_after_section_switch): New function.
22856 2017-07-19 Tom de Vries <tom@codesourcery.com>
22858 * gimple.h (gimple_phi_set_arg): Make assert more strict.
22860 2017-07-19 Tom de Vries <tom@codesourcery.com>
22862 * gimple.h (gimple_phi_arg): Make assert more strict.
22864 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
22866 * config.gcc (powerpc*-*-*): Add mmintrin.h.
22867 * config/rs6000/mmintrin.h: New file.
22868 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
22870 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22872 PR tree-optimization/81346
22873 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
22875 2017-07-19 Tom de Vries <tom@codesourcery.com>
22877 * config/nvptx/nvptx.md (VECIM): Add V2DI.
22879 2017-07-19 Tom de Vries <tom@codesourcery.com>
22881 * config/nvptx/nvptx-modes.def: Add V2DImode.
22882 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
22883 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
22884 (nvptx_output_mov_insn): Handle lack of mov.b128.
22885 (nvptx_print_operand): Handle 'H' and 'L' codes.
22886 (nvptx_vector_mode_supported): Allow V2DImode.
22887 (nvptx_preferred_simd_mode): New function.
22888 (nvptx_data_alignment): New function.
22889 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
22890 nvptx_preferred_simd_mode.
22891 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
22893 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
22895 2017-07-19 Tom de Vries <tom@codesourcery.com>
22897 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
22898 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
22899 (nvptx_vector_mode_supported): New function. Allow V2SImode.
22900 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
22901 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
22902 (mov<VECIM>_insn): New define_insn.
22903 (define_expand "mov<VECIM>): New define_expand.
22905 2017-07-19 Tom de Vries <tom@codesourcery.com>
22907 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
22909 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22911 PR tree-optimization/81346
22912 * fold-const.h (fold_div_compare, range_check_type): Declare.
22913 * fold-const.c (range_check_type): New function.
22914 (build_range_check): Use range_check_type.
22915 (fold_div_compare): No longer static, rewritten into
22916 a match.pd helper function.
22917 (fold_comparison): Don't call fold_div_compare here.
22918 * match.pd (X / C1 op C2): New optimization using fold_div_compare
22919 as helper function.
22921 2017-07-19 Nathan Sidwell <nathan@acm.org>
22923 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
22924 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
22925 * tree.c (find_decls_types_r, verify_type): Use
22926 TYPE_{MIN,MAX}_VALUE_RAW.
22927 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
22928 (hash_tree): Likewise.
22929 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
22931 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
22934 2017-07-18 Tom de Vries <tom@codesourcery.com>
22936 PR middle-end/81464
22937 * omp-expand.c (expand_omp_for_static_chunk): Handle
22938 equal-argument loop exit phi.
22940 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
22943 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
22944 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
22945 operand 2 predicate.
22946 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
22947 operand 2 predicate.
22948 (ror,rol -> rorx splitters): Use const_int_operand as
22949 operand 2 predicate.
22951 2017-06-18 Richard Biener <rguenther@suse.de>
22953 PR tree-optimization/81410
22954 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
22955 the gap in the ! slp_perm SLP case after each group.
22957 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22959 PR middle-end/81463
22960 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
22963 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22965 PR middle-end/81462
22966 * predict.c (set_even_probabilities): Cleanup; do not affect
22967 probabilities that are already known.
22968 (combine_predictions_for_bb): Call even when count is set.
22970 2017-07-18 Nathan Sidwell <nathan@acm.org>
22972 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
22975 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22978 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
22979 optimization for loop niter analysis.
22981 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
22984 * config/avr/avr.c (avr_optimize_casesi): Don't use
22985 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
22987 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
22989 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
22990 body_cost_vec from _vect_peel_extended_info.
22991 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
22992 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
22995 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22997 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
22999 2017-07-18 Richard Biener <rguenther@suse.de>
23001 PR tree-optimization/80620
23002 PR tree-optimization/81403
23003 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
23004 info when re-using a VN table entry.
23006 2017-07-18 Richard Biener <rguenther@suse.de>
23008 PR tree-optimization/81418
23009 * tree-vect-loop.c (vectorizable_reduction): Properly compute
23010 vectype_in. Verify that with lane-reducing reduction operations
23011 we have a single def-use cycle.
23013 2017-07-17 Carl Love <cel@us.ibm.com>
23015 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
23017 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23018 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23019 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23020 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23021 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23022 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23023 VMULOSW): New enum "unspec" values.
23024 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23025 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23026 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23027 altivec_vmulosw): New patterns.
23028 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23029 VMULOSW): Add definitions.
23031 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
23033 * config/alpha/alpha.c: Include predict.h.
23035 2017-07-17 Yury Gribov <tetra2005@gmail.com>
23037 * tree-vrp.c (compare_assert_loc): Fix comparison function
23038 to return predictable results.
23040 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23042 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
23044 (subdi3): Likewise.
23045 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
23046 * doc/invoke.texi (mexpand-adddi): Update text.
23048 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23050 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
23051 that also clobbers the CC register. The old expand code is moved
23053 (*arc_clzsi2): ... here.
23054 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
23055 the CC register. The old expand code is moved to ...
23056 (arc_ctzsi2): ... here.
23058 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23060 * config/arc/arc.opt (mindexed-loads): Use initial value
23061 TARGET_INDEXED_LOADS_DEFAULT.
23062 (mauto-modify-reg): Use initial value
23063 TARGET_AUTO_MODIFY_REG_DEFAULT.
23064 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23065 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23066 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23067 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23069 2017-07-17 Martin Liska <mliska@suse.cz>
23072 * opts.c (finish_options): Do not allow -fgnu-tm
23073 w/ -fsanitize={kernel-,}address. Say sorry.
23075 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23078 * tree-loop-distribution.c (classify_partition): Only assert on
23079 numer of iterations.
23080 (merge_dep_scc_partitions): Delete prameter. Update function call.
23081 (distribute_loop): Remove code handling loop with unknown niters.
23082 (pass_loop_distribution::execute): Skip loop with unknown niters.
23084 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23087 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
23088 function sort_partitions_by_post_order.
23090 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23092 PR tree-optimization/81374
23093 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
23094 the max index of basic blocks, rather than number of basic blocks.
23096 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23098 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
23100 (arc_legitimate_pic_operand_p): Likewise.
23101 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
23103 (arc_needs_pcl_p): Likewise.
23104 (arc_legitimate_pc_offset_p): Likewise.
23105 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
23106 function is also used in constrains.md.
23107 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
23108 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
23109 PLUS. Only return true/false in known cases, otherwise assert.
23110 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
23111 is already called in arc_legitimate_constant_p.
23112 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
23114 (LEGITIMATE_PIC_OPERAND_P): Use
23115 arc_raw_symbolic_reference_mentioned_p function.
23116 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
23121 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23122 Andrew Burgess <andrew.burgess@embecosm.com>
23124 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
23125 (arc_return_address_register): New function.
23126 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
23127 (arc_handle_fndecl_attribute): Add naked attribute.
23128 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
23129 (TARGET_WARN_FUNC_RETURN): Likewise.
23130 (arc_allocate_stack_slots_for_args): New function.
23131 (arc_warn_func_return): Likewise.
23132 (machine_function): Change type fn_type.
23133 (arc_compute_function_type): Consider new naked function type,
23134 change function return type.
23135 (arc_must_save_register): Adapt to handle new
23136 arc_compute_function_type's return type.
23137 (arc_expand_prologue): Likewise.
23138 (arc_expand_epilogue): Likewise.
23139 (arc_return_address_regs): Delete.
23140 (arc_return_address_register): New function.
23141 (arc_epilogue_uses): Use above function.
23142 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
23143 (arc_function_type): Change encoding, add naked type.
23144 (ARC_INTERRUPT_P): Change to handle the new encoding.
23145 (ARC_FAST_INTERRUPT_P): Likewise.
23146 (ARC_NORMAL_P): Define.
23147 (ARC_NAKED_P): Likewise.
23148 (arc_compute_function_type): Delete prototype.
23149 * config/arc/arc.md (in_ret_delay_slot): Use
23150 arc_return_address_register function.
23151 (simple_return): Likewise.
23152 (p_return_i): Likewise.
23154 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23156 PR tree-optimization/81428
23157 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
23158 can't be built for those types.
23160 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23162 Remove stuff dead since r239246.
23164 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
23165 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
23166 (avr_inform_devices): Remove dead stuff.
23168 2017-07-17 Tamar Christina <tamar.christina@arm.com>
23170 * config/arm/arm_neon.h: Fix softp typo.
23172 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23174 PR tree-optimization/81365
23175 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
23176 aggregate moves onto bb predecessor edges, make sure there are no
23177 loads that could alias the lhs in between the start of bb and the
23180 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23183 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
23184 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
23185 [LSHIFTRT, outer_code = TRUNCATE]: Same.
23187 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23189 PR tree-optimization/81396
23190 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
23191 (init_symbolic_number): Initialize it to 1.
23192 (perform_symbolic_merge): Add n_ops from both operands into the new
23194 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
23195 without base_addr as useless if they need more than one operation.
23196 (bswap_replace): Handle !bswap case for NULL base_addr.
23198 2017-07-17 Tom de Vries <tom@codesourcery.com>
23201 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
23204 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
23206 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
23207 conditional builtin define __FIX_LEON3FT_B2BST.
23209 2017-07-17 Daniel Cederman <cederman@gaisler.com>
23211 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
23212 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
23215 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23217 PR rtl-optimization/81424
23218 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
23219 to remove potential trapping from operands if -fnon-call-exceptions.
23221 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23223 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
23224 profile_proability for scalling.
23225 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
23227 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23229 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
23231 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23233 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
23234 fixpoint arithmetics.
23236 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23238 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
23239 fixpoint arithmetics.
23241 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23243 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
23244 fixpoint arithmetics.
23246 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23248 * profile-count.h (profile_probability::from_reg_br_prob_note,
23249 profile_probability::to_reg_br_prob_note): New functions.
23250 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
23251 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
23252 * predict.c (probability_reliable_p): Update.
23253 (edge_probability_reliable_p): Update.
23254 (br_prob_note_reliable_p): Update.
23255 (invert_br_probabilities): Update.
23256 (add_reg_br_prob_note): New function.
23257 (combine_predictions_for_insn): Update.
23258 * asan.c (asan_clear_shadow): Update.
23259 * cfgbuild.c (compute_outgoing_frequencies): Update.
23260 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23261 (update_br_prob_note): Update.
23262 (rtl_verify_edges): Update.
23263 (purge_dead_edges): Update.
23264 (fixup_reorder_chain): Update.
23265 * emit-rtl.c (try_split): Update.
23266 * ifcvt.c (cond_exec_process_insns): Update.
23267 (cond_exec_process_if_block): Update.
23268 (dead_or_predicable): Update.
23269 * internal-fn.c (expand_addsub_overflow): Update.
23270 (expand_neg_overflow): Update.
23271 (expand_mul_overflow): Update.
23272 * loop-doloop.c (doloop_modify): Update.
23273 * loop-unroll.c (compare_and_jump_seq): Update.
23274 * optabs.c (emit_cmp_and_jump_insn_1): Update.
23275 * predict.h: Update.
23276 * reorg.c (mostly_true_jump): Update.
23278 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
23279 * config/alpha/alpha.c (emit_unlikely_jump): Update.
23280 * config/arc/arc.c: (emit_unlikely_jump): Update.
23281 * config/arm/arm.c: (emit_unlikely_jump): Update.
23282 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
23283 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
23284 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
23285 (ix86_print_operand): Update.
23286 (ix86_split_fp_branch): Update.
23287 (predict_jump): Update.
23288 * config/ia64/ia64.c (ia64_print_operand): Update.
23289 * config/mmix/mmix.c (mmix_print_operand): Update.
23290 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
23291 (rs6000_expand_split_stack_prologue): Update.
23292 * config/rs6000/rs6000.c: Update.
23293 * config/s390/s390.c (s390_expand_vec_strlen): Update.
23294 (s390_expand_vec_movstr): Update.
23295 (s390_expand_cs_tdsi): Update.
23296 (s390_expand_split_stack_prologue): Update.
23297 * config/sh/sh.c (sh_print_operand): Update.
23298 (expand_cbranchsi4): Update.
23299 (expand_cbranchdi4): Update.
23300 * config/sparc/sparc.c (output_v9branch): Update.
23301 * config/spu/spu.c (get_branch_target): Update.
23302 (ea_load_store_inline): Update.
23303 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
23304 * config/tilepro/tilepro.c: Update.
23306 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23308 * gimplify.c (mostly_copy_tree_r): Revert latest change.
23309 (gimplify_save_expr): Likewise.
23311 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23313 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
23315 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23317 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
23319 * timevar.def (TV_IPA_FNSUMMARY): Define.
23321 2017-07-16 Daniel Cederman <cederman@gaisler.com>
23323 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
23324 to back store errata sensitive sequence from being generated.
23325 (sqrtdf2_fix): Likewise.
23327 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23329 * tree-ssa-threadupdate.c (compute_path_counts,
23330 update_joiner_offpath_counts): Use profile_probability.
23332 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
23335 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23337 * config/arm/arm-c.c (arm_cpu_builtins): Define
23338 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23340 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23342 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23343 array entries to represent __ieee128 versions of the
23344 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
23345 scalar_extract_sig, and scalar_insert_exp built-in functions.
23346 (altivec_resolve_overloaded_builtin): Add special case handling
23347 for the __builtin_scalar_insert_exp function, as represented by
23348 the P9V_BUILTIN_VEC_VSIEDP constant.
23349 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
23350 exponent support for __ieee128 argument.
23351 (VSESQP): Add scalar extract signature support for __ieee128
23353 (VSTDCNQP): Add scalar test negative support for __ieee128
23355 (VSIEQP): Add scalar insert exponent support for __int128 argument
23356 with __ieee128 result.
23357 (VSIEQPF): Add scalar insert exponent support for __ieee128
23358 argument with __ieee128 result.
23359 (VSTDCQP): Add scalar test data class support for __ieee128
23361 (VSTDCNQP): Add overload support for scalar test negative with
23362 __ieee128 argument.
23363 (VSTDCQP): Add overload support for scalar test data class
23364 __ieee128 argument.
23365 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
23366 UNSPEC_VSX_SXSIGDP.
23367 (UNSPEC_VSX_SIEXPQP): New constant.
23368 (xsxexpqp): New insn for VSX scalar extract exponent quad
23370 (xsxsigqp): New insn for VSX scalar extract significand quad
23372 (xsiexpqpf): New insn for VSX scalar insert exponent quad
23373 precision with floating point argument.
23374 (xststdcqp): New expand for VSX scalar test data class quad
23376 (xststdcnegqp): New expand for VSX scalar test negative quad
23378 (xststdcqp): New insn to match expansions for VSX scalar test data
23379 class quad precision and VSX scalar test negative quad precision.
23380 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
23381 special case operand checking to enforce that second operand of
23382 VSX scalar test data class with quad precision argument is a 7-bit
23384 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23385 prototypes and descriptions of __ieee128 versions of
23386 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
23387 scalar_test_data_class, and scalar_test_neg built-in functions.
23389 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23391 PR tree-optimization/81162
23392 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
23393 replace a negate with an add.
23395 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
23397 * doc/invoke.texi (arm/-mcpu): Document +crypto.
23399 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23401 * config/arm/arm-c.c (arm_cpu_builtins): Define
23402 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23404 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23406 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
23407 (armv8-r): Set ARM Cortex-R52 as default CPU.
23408 * config/arm/arm-tables.opt: Regenerate.
23409 * config/arm/arm-tune.md: Regenerate.
23410 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23412 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
23413 extension for -mcpu=cortex-r52.
23415 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23417 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
23418 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
23419 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
23420 (fp-armv8): Define it as FP_ARMv8 only.
23421 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
23422 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
23424 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
23425 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
23426 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
23427 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
23428 than TARGET_FPU_ARMV8.
23429 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
23430 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
23431 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
23433 * config/arm/neon.md (neon_vrint): Likewise.
23434 (neon_vcvt): Likewise.
23435 (neon_<fmaxmin_op><mode>): Likewise.
23436 (<fmaxmin><mode>3): Likewise.
23437 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
23438 * config/arm/predicates.md (arm_cond_move_operator): Check against
23439 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
23441 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
23443 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
23444 to top of function.
23446 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23448 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
23449 loop in comment with memset.
23451 2017-07-14 Martin Liska <mliska@suse.cz>
23453 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
23454 * dwarf2out.c (is_java): Remove the function.
23455 (output_pubname): Remove usage of the function.
23456 (lower_bound_default): Remove usage of DW_LANG_Java.
23457 (gen_compile_unit_die): Likewise.
23458 * gcc.c: Remove compiler defaults for .java and .zip files.
23459 * gimple-expr.c (remove_suffix): Change as there's no longer
23460 extension than 4-letter one.
23461 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
23462 (gimplify_save_expr): Likewise.
23463 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
23464 as it's possible even for other languages than Java.
23465 * langhooks.h (struct lang_hooks): Remove Java from a comment.
23466 * lto-opts.c (lto_write_options): Remove reference to Java.
23467 * opts.c (strip_off_ending): Update file extension handling.
23468 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
23469 * tree-eh.c (lower_resx): Likewise.
23470 * tree.c (free_lang_data_in_type): Remove dead code.
23471 (find_decls_types_r): Likewise.
23472 (build_common_builtin_nodes): Remove Java from a comment.
23473 (verify_type): Remove dead code.
23474 * varasm.c (assemble_external): Remove Java from a comment.
23476 2017-07-14 Martin Liska <mliska@suse.cz>
23478 * opts.c (finish_options): Add quotes.
23479 (common_handle_option): Likewise.
23481 2017-07-14 Martin Liska <mliska@suse.cz>
23483 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
23484 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
23485 Remove N_SO_PASCAL.
23486 * dwarf2out.c (lower_bound_default): Do not handle
23488 (gen_compile_unit_die): Likewise.
23489 * gcc.c: Remove default extension binding for GNU Pascal.
23490 * stmt.c: Remove Pascal language from a comment.
23491 * xcoffout.c: Likewise.
23493 2017-07-13 David Malcolm <dmalcolm@redhat.com>
23496 * diagnostic-show-locus.c (fixit_cmp): New function.
23497 (layout::layout): Sort m_fixit_hints.
23498 (column_range::column_range): Assert that the values are valid.
23499 (struct char_span): New struct.
23500 (correction::overwrite): New method.
23501 (struct source_line): New struct.
23502 (line_corrections::add_hint): Add assertions. Reimplement memcpy
23503 calls in terms of classes source_line and char_span, and
23504 correction::overwrite.
23505 (selftest::test_overlapped_fixit_printing_2): New function.
23506 (selftest::diagnostic_show_locus_c_tests): Call it.
23508 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
23510 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
23511 early if there is no lhs.
23513 2017-07-13 Martin Liska <mliska@suse.cz>
23515 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
23516 (gen_reference_type_die): Likewise.
23517 * stor-layout.c: Remove Pascal-related comment.
23519 2017-07-13 Martin Liska <mliska@suse.cz>
23521 * opts.c (finish_options): Add quotes to error messages.
23522 (parse_sanitizer_options): Likewise.
23524 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
23526 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
23528 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
23530 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
23532 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23534 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
23536 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
23538 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
23541 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
23542 provides the hardware capability bits, define the macro
23543 __BUILTIN_CPU_SUPPORTS__.
23544 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
23545 if GLIBC does not provide the hardware capability bits. Add a
23546 gcc_unreachable call if the built-in cpu function is neither
23547 __builtin_cpu_is nor __builtin_cpu_supports.
23548 (rs6000_get_function_versions_dispatcher): Change the warning
23549 that an old GLIBC is used which does not export the capability
23550 bits to be an error.
23551 * doc/extend.texi (target_clones attribute): Document the
23552 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
23553 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
23554 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
23555 the macros defined by GCC if the newer GLIBC is available.
23557 2017-07-12 Jeff Law <law@redhat.com>
23559 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
23560 remaining includes slightly.
23561 * config/riscv/riscv-builtins.c: Include profile-count.h.
23563 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23566 * config/avr/avr.c (avr_set_current_function): In diagnostic
23567 messages: Quote keywords and (parts of) identifiers.
23568 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
23571 2017-07-12 Carl Love <cel@us.ibm.com>
23573 * config/rs6000/rs6000-c.c: Add support for built-in functions
23574 vector bool char vec_revb (vector bool char);
23575 vector bool short vec_revb (vector short char);
23576 vector bool int vec_revb (vector bool int);
23577 vector bool long long vec_revb (vector bool long long);
23578 * doc/extend.texi: Update the built-in documentation file for the
23579 new built-in functions.
23581 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23583 * config/s390/s390.md: Remove movcc splitter.
23585 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23587 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
23588 load/store on condition.
23590 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23593 * config/avr/avr.c (avr_encode_section_info)
23594 [progmem && !TREE_READONLY]: Error if progmem object needs
23597 2017-07-11 Michael Collison <michael.collison@arm.com>
23599 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
23602 2017-07-11 Carl Love <cel@us.ibm.com>
23604 * config/rs6000/rs6000-c.c: Add support for builtins
23605 vector unsigned int vec_parity_lsbb (vector signed int);
23606 vector unsigned int vec_parity_lsbb (vector unsigned int);
23607 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
23608 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
23609 vector unsigned long long vec_parity_lsbb (vector signed long long);
23610 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
23611 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
23612 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
23613 * doc/extend.texi: Update the built-in documentation file for the
23614 new built-in functions.
23616 2017-07-11 David Malcolm <dmalcolm@redhat.com>
23618 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
23619 (layout::m_primary_loc): New field.
23620 (layout::layout): Initialize new field. Move location filtering
23621 logic from here to...
23622 (layout::maybe_add_location_range): ...this new method. Add
23623 support for filtering to just the lines already specified by other
23625 (layout::will_show_line_p): New method.
23626 (gcc_rich_location::add_location_if_nearby): New method.
23627 (selftest::test_add_location_if_nearby): New test function.
23628 (selftest::diagnostic_show_locus_c_tests): Call it.
23629 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
23632 2017-07-11 Tom de Vries <tom@codesourcery.com>
23634 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
23635 (bb_first_real_insn): New function.
23636 (nvptx_single): Add extra initialization of broadcasted condition
23639 2017-07-11 Nathan Sidwell <nathan@acm.org>
23641 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
23643 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
23645 * doc/extend.texi (AVR Function Attributes): Remove weblink to
23646 Binutils doc as TEXI will mess them up.
23647 * doc/invoke.texi (AVR Options): Same here.
23649 2017-07-11 Daniel Cederman <cederman@gaisler.com>
23651 * config/sparc/sparc.opt (mfix-ut700): New option.
23652 (mfix-gr712rc): Likewise.
23653 (sparc_fix_b2bst): New variable.
23654 * doc/invoke.texi (SPARC options): Document them.
23655 (ARM options): Fix warnings.
23656 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
23657 instructions to prevent sequences that can trigger the store-store
23658 errata for certain LEON3FT processors.
23659 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
23660 (sparc_option_override): Set sparc_fix_b2bst appropriately.
23661 * config/sparc/sparc.md (fix_b2bst): New attribute.
23662 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
23664 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
23667 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
23669 (*rsqrtsf2_sse): Ditto.
23671 (div<mode>3): Macroize insn from divdf3 and divsf3
23672 using MODEF mode iterator.
23674 2017-07-10 Martin Sebor <msebor@redhat.com>
23676 PR tree-optimization/80397
23677 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
23678 instead of testing for equality to INTEGER_TYPE.
23680 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
23682 * config.gcc: Remove uclibc from arc target spec.
23684 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
23686 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
23688 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23691 * lto-wrapper.c (remove_option): New function.
23692 (merge_and_complain): Merge PIC/PIE options more realistically.
23694 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23696 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
23700 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
23701 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
23702 * config.in: Regenerate.
23703 * configure: Regenerate.
23704 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
23705 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
23706 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
23707 (TARGET_GASISR_PROLOGUES): ...target mask.
23708 * common/config/avr/avr-common.c
23709 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
23710 Set -mgas-isr-prologues.
23711 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
23712 INSERT_PASS_BEFORE for it.
23713 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
23714 * config/avr/avr.c (avr_option_override)
23715 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
23716 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
23717 (avr_attribute_table) <no_gccisr>: Add new function attribute.
23718 (avr_set_current_function) <is_no_gccisr>: Init machine field.
23719 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
23721 (make_avr_pass_pre_proep): New function.
23722 (emit_push_sfr) <treg>: Add argument to function and use it
23723 instead of TMP_REG.
23724 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
23725 and set machine->gasisr.yes.
23726 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
23727 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
23728 __gcc_isr.n_pushed to .L__stack_usage.
23729 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
23730 (avr_asm_final_postscan_insn): ...this new static function.
23731 * config/avr/avr.h (machine_function)
23732 <is_no_gccisr, use_L__stack_usage>: New fields.
23733 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
23734 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
23735 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
23736 (gasisr, *gasisr): New expander and insn.
23737 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23738 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
23739 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
23741 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
23743 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
23746 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23748 Move jump-tables out of .text again.
23751 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
23752 (ASM_OUTPUT_ADDR_VEC): New function.
23753 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
23754 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
23755 INSN_ADDRESSes as asm comment.
23756 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
23757 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
23758 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
23759 * config/avr/avr.md (*tablejump): Adjust comment.
23760 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
23761 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
23763 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
23764 (avr_output_addr_vec): New proto.
23765 (avr_log_t) <insn_addresses>: New field.
23767 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
23770 * config/i386/i386.c (ix86_function_arg_advance): Set
23771 outgoing_args_on_stack to true if there are outgoing arguments
23773 (ix86_function_arg): Likewise.
23774 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
23775 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
23776 * config/i386/i386.h (machine_function): Add
23777 outgoing_args_on_stack.
23779 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
23781 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23782 supporting pthreds.
23783 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
23785 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
23787 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
23788 (REAL_H): Remove $(MACHMODE_H).
23789 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
23791 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
23792 $(MACHMODE_H) and double-int.h.
23793 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
23795 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
23798 2017-07-07 Andrew Pinski <apinski@cavium.com>
23800 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
23801 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
23803 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23805 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
23806 Add warning if GCC was not configured to link against a GLIBC that
23807 exports the hardware capability bits.
23808 (make_resolver_func): Make resolver function private and not a
23809 COMDAT function. Create the name with clone_function_name instead
23810 of make_unique_name.
23813 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
23814 correct operand in doing the split.
23816 2017-07-07 Carl Love <cel@us.ibm.com>
23818 * config/rs6000/rs6000-c: Add support for built-in function
23819 vector unsigned short vec_pack_to_short_fp32 (vector float,
23821 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
23822 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
23823 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
23824 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
23825 (convert_4f32_8i16): Add define_expand.
23826 * doc/extend.texi: Update the built-in documentation file for the
23827 new built-in function.
23829 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23831 * config/sparc/m8.md: New file.
23832 * config/sparc/sparc.md: Include m8.md.
23834 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23836 * config/sparc/sparc.opt: New option -mvis4b.
23837 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
23838 (sparc_option_override): Handle VIS4B.
23839 (enum sparc_builtins): Define
23840 SPARC_BUILTIN_DICTUNPACK{8,16,32},
23841 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
23842 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
23843 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
23844 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
23845 (check_constant_argument): New function.
23846 (sparc_vis_init_builtins): Define builtins
23847 __builtin_vis_dictunpack{8,16,32},
23848 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
23849 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
23850 __builtin_vis_fpcmpde{8,16,32}shl and
23851 __builtin_vis_fpcmpur{8,16,32}shl.
23852 (sparc_expand_builtin): Check that the constant operands to
23853 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
23854 constant and in range.
23855 * config/sparc/sparc-c.c (sparc_target_macros): Handle
23857 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
23858 (SPARC_IMM5_P): Likewise.
23859 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
23860 (enabled): Handle vis4b.
23861 (UNSPEC_DICTUNPACK): New unspec.
23862 (UNSPEC_FPCMPSHL): Likewise.
23863 (UNSPEC_FPUCMPSHL): Likewise.
23864 (UNSPEC_FPCMPDESHL): Likewise.
23865 (UNSPEC_FPCMPURSHL): Likewise.
23866 (cpu_feature): New CPU feature `vis4b'.
23867 (dictunpack{8,16,32}): New insns.
23868 (FPCSMODE): New mode iterator.
23869 (fpcscond): New code iterator.
23870 (fpcsucond): Likewise.
23871 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
23872 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
23873 (fpcmpde{8,16,32}{si,di}shl): Likewise.
23874 (fpcmpur{8,16,32}{si,di}shl): Likewise.
23875 * config/sparc/constraints.md: Define constraints `q' for unsigned
23876 2-bit integer constants and `t' for unsigned 5-bit integer
23878 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
23880 (imm5_operand_dictunpack16): Likewise.
23881 (imm5_operand_dictunpack32): Likewise.
23882 (imm2_operand): Likewise.
23883 * doc/invoke.texi (SPARC Options): Document -mvis4b.
23884 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
23885 ditunpack* and fpcmp*shl builtins.
23887 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23889 * config.gcc: Handle m8 in --with-{cpu,tune} options.
23890 * config.in: Add HAVE_AS_SPARC6 define.
23891 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
23893 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
23895 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
23896 (CPP_CPU_SPEC): Handle m8.
23897 (ASM_CPU_SPEC): Likewise.
23898 * config/sparc/sparc-opts.h (enum processor_type): Add
23900 * config/sparc/sparc.c (m8_costs): New struct.
23901 (sparc_option_override): Handle TARGET_CPU_m8.
23902 (sparc32_initialize_trampoline): Likewise.
23903 (sparc64_initialize_trampoline): Likewise.
23904 (sparc_issue_rate): Likewise.
23905 (sparc_register_move_cost): Likewise.
23906 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
23907 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
23908 (ASM_CPU64_DEFAULT_SPEC): Likewise.
23909 (CPP_CPU_SPEC): Handle M8.
23910 (ASM_CPU_SPEC): Likewise.
23911 (AS_M8_FLAG): Define.
23912 * config/sparc/sparc.md: Add m8 to the cpu attribute.
23913 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
23914 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
23916 * configure: Regenerate.
23917 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
23920 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23922 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
23924 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
23925 ("*movdi_insn_sp32"): Do not set v3pipe.
23926 ("*movsi_insn"): Likewise.
23927 ("*movdi_insn_sp64"): Likewise.
23928 ("*movsf_insn"): Likewise.
23929 ("*movdf_insn_sp32"): Likewise.
23930 ("*movdf_insn_sp64"): Likewise.
23931 ("*zero_extendsidi2_insn_sp64"): Likewise.
23932 ("*sign_extendsidi2_insn"): Likewise.
23933 ("*mov<VM32:mode>_insn"): Likewise.
23934 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23935 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23936 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23937 ("<vlop:code><VL:mode>3"): Likewise.
23938 ("*not_<vlop:code><VL:mode>3"): Likewise.
23939 ("*nand<VL:mode>_vis"): Likewise.
23940 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
23941 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
23942 ("one_cmpl<VL:mode>2"): Likewise.
23943 ("faligndata<VM64:mode>_vis"): Likewise.
23944 ("alignaddrsi_vis"): Likewise.
23945 ("alignaddrdi_vis"): Likweise.
23946 ("alignaddrlsi_vis"): Likewise.
23947 ("alignaddrldi_vis"): Likewise.
23948 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23949 ("bmaskdi_vis"): Likewise.
23950 ("bmasksi_vis"): Likewise.
23951 ("bshuffle<VM64:mode>_vis"): Likewise.
23952 ("cmask8<P:mode>_vis"): Likewise.
23953 ("cmask16<P:mode>_vis"): Likewise.
23954 ("cmask32<P:mode>_vis"): Likewise.
23955 ("pdistn<P:mode>_vis"): Likewise.
23956 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23958 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23960 * config/sparc/sparc.md ("subtype"): New insn attribute.
23961 ("*wrgsr_sp64"): Set insn subtype.
23962 ("*rdgsr_sp64"): Likewise.
23963 ("alignaddrsi_vis"): Likewise.
23964 ("alignaddrdi_vis"): Likewise.
23965 ("alignaddrlsi_vis"): Likewise.
23966 ("alignaddrldi_vis"): Likewise.
23967 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23968 ("fexpand_vis"): Likewise.
23969 ("fpmerge_vis"): Likewise.
23970 ("faligndata<VM64:mode>_vis"): Likewise.
23971 ("bshuffle<VM64:mode>_vis"): Likewise.
23972 ("cmask8<P:mode>_vis"): Likewise.
23973 ("cmask16<P:mode>_vis"): Likewise.
23974 ("cmask32<P:mode>_vis"): Likewise.
23975 ("fchksm16_vis"): Likewise.
23976 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
23977 ("fmean16_vis"): Likewise.
23978 ("fp<plusminus_insn>64_vis"): Likewise.
23979 ("<plusminus_insn>v8qi3"): Likewise.
23980 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23981 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
23982 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
23983 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
23984 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
23985 ("*movqi_insn"): Likewise.
23986 ("*movhi_insn"): Likewise.
23987 ("*movsi_insn"): Likewise.
23988 ("movsi_pic_gotdata_op"): Likewise.
23989 ("*movdi_insn_sp32"): Likewise.
23990 ("*movdi_insn_sp64"): Likewise.
23991 ("movdi_pic_gotdata_op"): Likewise.
23992 ("*movsf_insn"): Likewise.
23993 ("*movdf_insn_sp32"): Likewise.
23994 ("*movdf_insn_sp64"): Likewise.
23995 ("*zero_extendhisi2_insn"): Likewise.
23996 ("*zero_extendqihi2_insn"): Likewise.
23997 ("*zero_extendqisi2_insn"): Likewise.
23998 ("*zero_extendqidi2_insn"): Likewise.
23999 ("*zero_extendhidi2_insn"): Likewise.
24000 ("*zero_extendsidi2_insn_sp64"): Likewise.
24001 ("ldfsr"): Likewise.
24002 ("prefetch_64"): Likewise.
24003 ("prefetch_32"): Likewise.
24004 ("tie_ld32"): Likewise.
24005 ("tie_ld64"): Likewise.
24006 ("*tldo_ldub_sp32"): Likewise.
24007 ("*tldo_ldub1_sp32"): Likewise.
24008 ("*tldo_ldub2_sp32"): Likewise.
24009 ("*tldo_ldub_sp64"): Likewise.
24010 ("*tldo_ldub1_sp64"): Likewise.
24011 ("*tldo_ldub2_sp64"): Likewise.
24012 ("*tldo_ldub3_sp64"): Likewise.
24013 ("*tldo_lduh_sp32"): Likewise.
24014 ("*tldo_lduh1_sp32"): Likewise.
24015 ("*tldo_lduh_sp64"): Likewise.
24016 ("*tldo_lduh1_sp64"): Likewise.
24017 ("*tldo_lduh2_sp64"): Likewise.
24018 ("*tldo_lduw_sp32"): Likewise.
24019 ("*tldo_lduw_sp64"): Likewise.
24020 ("*tldo_lduw1_sp64"): Likewise.
24021 ("*tldo_ldx_sp64"): Likewise.
24022 ("*mov<VM32:mode>_insn"): Likewise.
24023 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24024 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24026 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24028 * config/sparc/sparc.md ("type"): New insn type viscmp.
24029 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
24031 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
24032 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
24033 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24034 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
24036 ("n7_vis_logical_11cycle"): Likewise.
24037 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
24038 * config/sparc/niagara2.md ("niag3_vis": Likewise.
24039 * config/sparc/niagara.md ("niag_vis"): Likewise.
24040 * config/sparc/ultra3.md ("us3_fga"): Likewise.
24041 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
24043 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24045 * config/sparc/sparc.md: New instruction type `bmask'.
24046 (bmaskdi_vis): Use the `bmask' type.
24047 (bmasksi_vis): Likewise.
24048 * config/sparc/ultra3.md (us3_array): Likewise.
24049 * config/sparc/niagara7.md (n7_array): Likewise.
24050 * config/sparc/niagara4.md (n4_array): Likewise.
24051 * config/sparc/niagara2.md (niag2_vis): Likewise.
24052 (niag3_vis): Likewise.
24053 * config/sparc/niagara.md (niag_vis): Likewise.
24055 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24057 * ipa-comdats.c: Remove optimize check from gate.
24058 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
24059 for functions not optimized.
24060 (ipa_fn_summary_read): Skip optimize check.
24061 (ipa_fn_summary_write): Likewise.
24062 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
24064 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
24066 (can_inline_edge_p): Check flag_pcc_struct_return for match.
24067 (check_callers): Give up on caller which is not optimized.
24068 (inline_small_functions): Likewise.
24069 (ipa_inline): Do not give up when not optimizing.
24070 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
24071 away unoptimizes cdtors.
24072 (whole_program_function_and_variable_visibility): Do
24073 ipa_discover_readonly_nonaddressable_vars in LTO mode.
24074 * ipa.c (process_references): Do not check optimize.
24075 (symbol_table::remove_unreachable_nodes): Update optimize check.
24076 (set_writeonly_bit): Update optimize check.
24077 (pass_ipa_cdtor_merge::gate): Do not check optimize.
24078 (pass_ipa_single_use::gate): Remove.
24080 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24082 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
24083 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
24084 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
24085 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
24086 permute_load, permute_store, adjust_extract, adjust_splat,
24087 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
24088 replace_swap_with_copy, dump_swap_insn_table,
24089 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
24090 recombine_lvx_pattern, recombine_stvx_pattern,
24091 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
24092 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
24093 to file rs6000-p8swap.c.
24094 * config/rs6000/rs6000-p8swap.c: New file.
24095 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
24096 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
24097 and rs6000*-*-* targets.
24099 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24101 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
24103 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24105 * lto-wrapper.c (merge_and_complain): Do not merge
24106 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
24107 fsigned_zeros, ftrapping_math, fwrapv.
24108 (append_compiler_options): Do not track these options.
24109 (append_linker_options): Likewie
24111 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24113 * cgraphunit.c (cgraph_node::finalize_function): When
24114 !flag_toplevel_reorde set no_reorder flag.
24115 (varpool_node::finalize_decl): Likewise.
24116 (symbol_table::compile): Drop no toplevel reorder path.
24118 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24120 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
24121 edges; zero probability is not better than uninitialized.
24123 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24125 * asan.h (asan_sanitize_allocas_p): Declare.
24126 * asan.c (asan_sanitize_allocas_p): New function.
24127 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
24128 (handle_builtin_alloca): Likewise.
24129 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
24130 if !asan_sanitize_allocas_p.
24131 * params.def (asan-instrument-allocas): Add new option.
24132 * params.h (ASAN_PROTECT_ALLOCAS): Define.
24133 * opts.c (common_handle_option): Disable allocas sanitization for
24136 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24138 * asan.c: Include gimple-fold.h.
24139 (get_last_alloca_addr): New function.
24140 (handle_builtin_stackrestore): Likewise.
24141 (handle_builtin_alloca): Likewise.
24142 (asan_emit_allocas_unpoison): Likewise.
24143 (get_mem_refs_of_builtin_call): Add new parameter, remove const
24144 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
24145 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
24146 (instrument_builtin_call): Pass gimple iterator to
24147 get_mem_refs_of_builtin_call.
24148 (last_alloca_addr): New global.
24149 * asan.h (asan_emit_allocas_unpoison): Declare.
24150 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
24151 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
24152 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
24153 if function calls alloca.
24154 * gimple-fold.c (replace_call_with_value): Remove static keyword.
24155 * gimple-fold.h (replace_call_with_value): Declare.
24156 * internal-fn.c: Include asan.h.
24157 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
24158 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
24160 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24162 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
24163 (C_SELFTEST_FLAGS): New.
24164 (CPP_SELFTEST_FLAGS): New.
24165 (SELFTEST_DEPS): New, from deps of s-selftest.
24166 (C_SELFTEST_DEPS): New, from deps of s-selftest.
24167 (CPP_SELFTEST_DEPS): New.
24168 (selftest): Add dependency on s-selftest-c++.
24169 (s-selftest): Rename to...
24170 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
24171 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
24172 than SELFTEST_FLAGS.
24173 (selftest-gdb): Rename to...
24174 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
24176 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
24177 (selftest-valgrind): Rename to...
24178 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
24180 (selftest-valgrind): Reintroduce as an alias for
24181 selftest-c-valgrind.
24182 (s-selftest-c++): New.
24183 (selftest-c++-gdb): New.
24184 (selftest-c++-valgrind): New.
24186 2017-07-06 Olivier Hainque <hainque@adacore.com>
24188 * gcc.c (process_command): When deciding if undefined variables
24189 should be ignored when processing specs, accept "gcc -v" as well.
24191 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24193 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
24194 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
24196 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24198 * config/arm/arm-cpus.in (armv8-r): Add new entry.
24199 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
24200 * config/arm/arm-tables.opt: Regenerate.
24201 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
24203 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
24205 2017-07-06 Carl Love <cel@us.ibm.com>
24207 * ChangeLog: Clean up from mid air collision
24209 2017-07-06 Carl Love <cel@us.ibm.com>
24211 * config/rs6000/rs6000-c.c: Add support for built-in functions
24212 vector signed int vec_subc (vector signed int, vector signed int);
24213 vector signed __int128 vec_subc (vector signed __int128,
24214 vector signed __int128);
24215 vector unsigned __int128 vec_subc (vector unsigned __int128,
24216 vector unsigned __int128);
24217 vector signed int vec_sube (vector signed int, vector signed int,
24218 vector signed int);
24219 vector unsigned int vec_sube (vector unsigned int,
24220 vector unsigned int,
24221 vector unsigned int);
24222 vector signed __int128 vec_sube (vector signed __int128,
24223 vector signed __int128,
24224 vector signed__int128);
24225 vector unsigned __int128 vec_sube (vector unsigned __int128,
24226 vector unsigned __int128,
24227 vector unsigned __int128);
24228 vector signed int vec_subec (vector signed int, vector signed int,
24229 vector signed int);
24230 vector unsigned int vec_subec (vector unsigned int,
24231 vector unsigned int,
24232 vector unsigned int);
24233 vector signed __int128 vec_subec (vector signed __int128,
24234 vector signed __int128,
24235 vector signed__int128);
24236 vector unsigned __int128 vec_subec (vector unsigned __int128,
24237 vector unsigned __int128,
24238 vector unsigned __int128);
24239 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
24240 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
24241 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
24242 BU_ALTIVEC_OVERLOAD_X definitions.
24243 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
24244 * doc/extend.texi: Update the built-in documentation file for the new
24245 built-in functions.
24247 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24250 * diagnostic-show-locus.c (layout::layout): Use start and finish
24251 spelling location for the start and finish of each range.
24252 * genmatch.c (linemap_client_expand_location_to_spelling_point):
24253 Add unused aspect param.
24254 * input.c (expand_location_1): Add "aspect" param, and use it
24255 to access the correct part of the location.
24256 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
24258 (expand_location_to_spelling_point): Likewise.
24259 (linemap_client_expand_location_to_spelling_point): Add "aspect"
24260 param, and pass it to expand_location_1.
24262 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
24264 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
24265 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
24266 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
24267 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
24268 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
24269 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
24270 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
24271 _mm_maskz_getmant_ss): New intrinsics.
24272 (__builtin_ia32_getexpss128_mask): Changed to ...
24273 __builtin_ia32_getexpss128_round ... this.
24274 (__builtin_ia32_getexpsd128_mask): Changed to ...
24275 __builtin_ia32_getexpsd128_round ... this.
24276 * config/i386/i386-builtin-types.def
24277 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
24278 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
24279 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
24280 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
24281 __builtin_ia32_getmantss_mask_round): New builtins.
24282 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
24283 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
24284 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
24285 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
24286 * config/i386/sse.md
24287 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
24288 avx512f_sgetexp<mode><mask_scalar_name>
24289 <round_saeonly_scalar_name> ... this.
24290 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
24291 %0, %1, %2<round_saeonly_op3>}): Changed to ...
24292 vgetexp<ssescalarmodesuffix>
24293 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24294 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
24295 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
24296 avx512f_vgetmant<mode><mask_scalar_name>
24297 <round_saeonly_scalar_name> ... this.
24298 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24299 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
24300 vgetmant<ssescalarmodesuffix>
24301 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
24302 %0<mask_scalar_operand4>, %1, %2
24303 <round_saeonly_scalar_mask_op4>, %3} ... this.
24304 * config/i386/subst.md (mask_scalar_operand4,
24305 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
24306 round_saeonly_scalar_nimm_predicate): New subst attributes.
24308 2017-07-06 Julia Koval <julia.koval@intel.com>
24310 * config/i386/i386.c (ix86_erase_embedded_rounding):
24311 Remove code for old rounding pattern.
24313 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
24315 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
24317 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
24319 * doc/sourcebuild.texi (Test Directives, Variants of
24320 dg-require-support): Add documentation for dg-require-stack-check.
24322 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
24324 * config/i386/subst.md (mask_scalar, round_scalar,
24325 round_saeonly_scalar): New meta-templates.
24326 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
24327 round_scalar_mask_operand3, round_scalar_mask_op3,
24328 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
24329 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
24330 round_saeonly_scalar_constraint,
24331 round_saeonly_scalar_prefix): New subst attribute.
24332 * config/i386/sse.md
24333 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
24334 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
24335 <round_scalar_name> ... this.
24336 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
24337 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
24338 <round_scalar_name> ... this.
24339 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
24340 <sse>_vm<code><mode>3<mask_scalar_name>
24341 <round_saeonly_scalar_name> ... this.
24342 (v<plusminus_mnemonic><ssescalarmodesuffix>
24343 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24344 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24345 v<plusminus_mnemonic><ssescalarmodesuffix>
24346 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24347 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24348 (v<multdiv_mnemonic><ssescalarmodesuffix>
24349 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24350 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24351 v<multdiv_mnemonic><ssescalarmodesuffix>
24352 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24353 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24354 (v<maxmin_float><ssescalarmodesuffix>
24355 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
24356 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
24357 v<maxmin_float><ssescalarmodesuffix>
24358 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24359 %0<mask_scalar_operand3>, %1, %<iptr>2
24360 <round_saeonly_scalar_mask_op3>} ... this.
24362 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
24364 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
24365 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
24367 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
24368 Alan Hayward <alan.hayward@arm.com>
24369 David Sherwood <david.sherwood@arm.com>
24371 * combine.c (simplify_if_then_else): Remove "enum" before
24373 * compare-elim.c (can_eliminate_compare): Likewise.
24374 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
24376 (aarch64_lookup_simd_builtin_type): Likewise.
24377 (aarch64_simd_builtin_type): Likewise.
24378 (aarch64_init_simd_builtin_types): Likewise.
24379 (aarch64_simd_expand_args): Likewise.
24380 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
24382 (aarch64_reverse_mask): Likewise.
24383 (aarch64_simd_emit_reg_reg_move): Likewise.
24384 (aarch64_gen_adjusted_ldpstp): Likewise.
24385 (aarch64_ccmp_mode_to_code): Likewise.
24386 (aarch64_operands_ok_for_ldpstp): Likewise.
24387 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24388 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
24390 (aarch64_min_divisions_for_recip_mul): Likewise.
24391 (aarch64_reassociation_width): Likewise.
24392 (aarch64_get_condition_code_1): Likewise.
24393 (aarch64_simd_emit_reg_reg_move): Likewise.
24394 (aarch64_simd_attr_length_rglist): Likewise.
24395 (aarch64_reverse_mask): Likewise.
24396 (aarch64_operands_ok_for_ldpstp): Likewise.
24397 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24398 (aarch64_gen_adjusted_ldpstp): Likewise.
24399 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
24401 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
24402 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
24403 (arm_lookup_simd_builtin_type): Likewise.
24404 (arm_simd_builtin_type): Likewise.
24405 (arm_init_simd_builtin_types): Likewise.
24406 (arm_expand_builtin_args): Likewise.
24407 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
24408 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
24409 (ft32_setup_incoming_varargs): Likewise.
24410 (ft32_function_arg): Likewise.
24411 (ft32_function_arg_advance): Likewise.
24412 (ft32_pass_by_reference): Likewise.
24413 (ft32_arg_partial_bytes): Likewise.
24414 (ft32_valid_pointer_mode): Likewise.
24415 (ft32_addr_space_pointer_mode): Likewise.
24416 (ft32_addr_space_legitimate_address_p): Likewise.
24417 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
24419 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
24420 (ix86_emit_outlined_ms2sysv_restore): Likewise.
24421 (iamcu_alignment): Likewise.
24422 (canonicalize_vector_int_perm): Likewise.
24423 (ix86_noce_conversion_profitable_p): Likewise.
24424 (ix86_mpx_bound_mode): Likewise.
24425 (ix86_operands_ok_for_move_multiple): Likewise.
24426 * config/microblaze/microblaze-protos.h
24427 (microblaze_expand_conditional_branch_reg): Likewise.
24428 * config/microblaze/microblaze.c
24429 (microblaze_expand_conditional_branch_reg): Likewise.
24430 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
24432 (rs6000_reassociation_width): Likewise.
24433 (rs6000_invalid_binary_op): Likewise.
24434 (fusion_p9_p): Likewise.
24435 (emit_fusion_p9_load): Likewise.
24436 (emit_fusion_p9_store): Likewise.
24437 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
24439 (riscv_hard_regno_mode_ok_p): Likewise.
24440 (riscv_address_insns): Likewise.
24441 (riscv_split_symbol): Likewise.
24442 (riscv_legitimize_move): Likewise.
24443 (riscv_function_value): Likewise.
24444 (riscv_hard_regno_nregs): Likewise.
24445 (riscv_expand_builtin): Likewise.
24446 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
24447 (riscv_build_integer): Likewise.
24448 (riscv_split_integer): Likewise.
24449 (riscv_legitimate_constant_p): Likewise.
24450 (riscv_cannot_force_const_mem): Likewise.
24451 (riscv_regno_mode_ok_for_base_p): Likewise.
24452 (riscv_valid_base_register_p): Likewise.
24453 (riscv_valid_offset_p): Likewise.
24454 (riscv_valid_lo_sum_p): Likewise.
24455 (riscv_classify_address): Likewise.
24456 (riscv_legitimate_address_p): Likewise.
24457 (riscv_address_insns): Likewise.
24458 (riscv_load_store_insns): Likewise.
24459 (riscv_force_binary): Likewise.
24460 (riscv_split_symbol): Likewise.
24461 (riscv_force_address): Likewise.
24462 (riscv_legitimize_address): Likewise.
24463 (riscv_move_integer): Likewise.
24464 (riscv_legitimize_const_move): Likewise.
24465 (riscv_legitimize_move): Likewise.
24466 (riscv_address_cost): Likewise.
24467 (riscv_subword): Likewise.
24468 (riscv_output_move): Likewise.
24469 (riscv_canonicalize_int_order_test): Likewise.
24470 (riscv_emit_int_order_test): Likewise.
24471 (riscv_function_arg_boundary): Likewise.
24472 (riscv_pass_mode_in_fpr_p): Likewise.
24473 (riscv_pass_fpr_single): Likewise.
24474 (riscv_pass_fpr_pair): Likewise.
24475 (riscv_get_arg_info): Likewise.
24476 (riscv_function_arg): Likewise.
24477 (riscv_function_arg_advance): Likewise.
24478 (riscv_arg_partial_bytes): Likewise.
24479 (riscv_function_value): Likewise.
24480 (riscv_pass_by_reference): Likewise.
24481 (riscv_setup_incoming_varargs): Likewise.
24482 (riscv_print_operand): Likewise.
24483 (riscv_elf_select_rtx_section): Likewise.
24484 (riscv_save_restore_reg): Likewise.
24485 (riscv_for_each_saved_reg): Likewise.
24486 (riscv_register_move_cost): Likewise.
24487 (riscv_hard_regno_mode_ok_p): Likewise.
24488 (riscv_hard_regno_nregs): Likewise.
24489 (riscv_class_max_nregs): Likewise.
24490 (riscv_memory_move_cost): Likewise.
24491 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
24492 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
24493 (rl78_addr_space_address_mode): Likewise.
24494 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
24496 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
24497 (rs6000_reassociation_width): Likewise.
24498 (rs6000_invalid_binary_op): Likewise.
24499 (fusion_p9_p): Likewise.
24500 (emit_fusion_p9_load): Likewise.
24501 (emit_fusion_p9_store): Likewise.
24502 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
24503 (ok_for_simple_move_operands): Likewise.
24504 (ok_for_simple_move_strict_operands): Likewise.
24505 (ok_for_simple_arith_logic_operands): Likewise.
24506 (visium_legitimize_reload_address): Likewise.
24507 (visium_select_cc_mode): Likewise.
24508 (output_cbranch): Likewise.
24509 (visium_split_double_move): Likewise.
24510 (visium_expand_copysign): Likewise.
24511 (visium_expand_int_cstore): Likewise.
24512 (visium_expand_fp_cstore): Likewise.
24513 * config/visium/visium.c (visium_pass_by_reference): Likewise.
24514 (visium_function_arg): Likewise.
24515 (visium_function_arg_advance): Likewise.
24516 (visium_libcall_value): Likewise.
24517 (visium_setup_incoming_varargs): Likewise.
24518 (visium_legitimate_constant_p): Likewise.
24519 (visium_legitimate_address_p): Likewise.
24520 (visium_legitimize_address): Likewise.
24521 (visium_secondary_reload): Likewise.
24522 (visium_register_move_cost): Likewise.
24523 (visium_memory_move_cost): Likewise.
24524 (prepare_move_operands): Likewise.
24525 (ok_for_simple_move_operands): Likewise.
24526 (ok_for_simple_move_strict_operands): Likewise.
24527 (ok_for_simple_arith_logic_operands): Likewise.
24528 (visium_function_value_1): Likewise.
24529 (rtx_ok_for_offset_p): Likewise.
24530 (visium_legitimize_reload_address): Likewise.
24531 (visium_split_double_move): Likewise.
24532 (visium_expand_copysign): Likewise.
24533 (visium_expand_int_cstore): Likewise.
24534 (visium_expand_fp_cstore): Likewise.
24535 (visium_split_cstore): Likewise.
24536 (visium_select_cc_mode): Likewise.
24537 (visium_split_cbranch): Likewise.
24538 (output_cbranch): Likewise.
24539 (visium_print_operand_address): Likewise.
24540 * expmed.c (flip_storage_order): Likewise.
24541 * expmed.h (emit_cstore): Likewise.
24542 (flip_storage_order): Likewise.
24543 * genrecog.c (validate_pattern): Likewise.
24544 * hsa-gen.c (gen_hsa_addr): Likewise.
24545 * internal-fn.c (expand_arith_overflow): Likewise.
24546 * ira-color.c (allocno_copy_cost_saving): Likewise.
24547 * lra-assigns.c (find_hard_regno_for_1): Likewise.
24548 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
24549 (process_invariant_for_inheritance): Likewise.
24550 * lra-eliminations.c (move_plus_up): Likewise.
24551 * omp-low.c (lower_oacc_reductions): Likewise.
24552 * simplify-rtx.c (simplify_subreg): Likewise.
24553 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
24554 (TARGET_CHKP_BOUND_MODE): Likewise..
24555 * targhooks.c (default_chkp_bound_mode): Likewise.
24556 (default_setup_incoming_vararg_bounds): Likewise.
24557 * targhooks.h (default_chkp_bound_mode): Likewise.
24558 (default_setup_incoming_vararg_bounds): Likewise.
24559 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
24560 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
24561 (have_whole_vector_shift): Likewise.
24562 * tree-vect-stmts.c (vectorizable_load): Likewise.
24563 * doc/tm.texi: Regenerate.
24565 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24567 Graceful degrade if Binutils PR21472 is not available.
24570 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
24571 .rodata in flash test fails.
24572 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
24573 * confgure: Regenerate.
24574 * config.in: Regenerate.
24575 * config/avr/avr.c (avr_asm_named_section)
24576 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
24577 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
24578 (avr_asm_init_sections): Same.
24580 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24582 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
24583 (fma<VH:mode>4_intrinsic): Likewise.
24584 (*fmsub<VCVTF:mode>4): Likewise.
24585 (*fmsub<VH:mode>4_intrinsic): Likewise.
24587 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24590 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
24591 Don't depend on "optimize > 0".
24592 (out_movhi_r_mr, out_movqi_mr_r): Same.
24593 (out_movhi_mr_r, out_movqi_r_mr): Same.
24594 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
24595 io_address_operand on "optimize > 0".
24597 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24599 * tree-loop-distribution.c: Add general explanantion on the pass.
24600 (generate_loops_for_partition): Mark distributed loop.
24601 (pg_add_dependence_edges): New parameter. Handle alias data
24602 dependence specially and record it in the parameter if asked.
24603 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
24604 (init_partition_graph_vertices, add_partition_graph_edge): New.
24605 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
24606 (free_partition_graph_vdata, build_partition_graph): New.
24607 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
24608 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
24609 (data_ref_segment_size, latch_dominated_by_data_ref): New.
24610 (compute_alias_check_pairs, version_loop_by_alias_check): New.
24611 (version_for_distribution_p, finalize_partitions): New.
24612 (distribute_loop): Handle alias data dependence specially. Factor
24613 out loop fusion code as functions and call these functions.
24615 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24617 * tree-loop-distribution.c (classify_partition): New parameter and
24618 better handle reduction statement.
24619 (rdg_build_partitions): Revise comment.
24620 (distribute_loop): Compute statements in all partitions and pass it
24621 to classify_partition.
24623 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24625 * tree-loop-distribution.c (enum partition_type): New.
24626 (struct partition): New field type.
24627 (partition_merge_into): Add parameter. Update partition type.
24628 (data_dep_in_cycle_p, update_type_for_merge): New functions.
24629 (build_rdg_partition_for_vertex): Compute partition type.
24630 (rdg_build_partitions): Dump partition type.
24631 (distribute_loop): Update calls to partition_merge_into.
24633 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24635 * tree-loop-distribution.c (struct ddr_hasher): New.
24636 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
24638 (classify_partition): Call get_data_dependence.
24639 (pg_add_dependence_edges): Ditto.
24640 (distribute_loop): Release data dependence hash table.
24642 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24644 * tree-loop-distribution.c (ref_base_address): Delete.
24645 (similar_memory_accesses): Rename ...
24646 (share_memory_accesses): ... to this. Check if partitions access
24647 the same memory reference.
24648 (distribute_loop): Call share_memory_accesses.
24650 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24652 * tree-loop-distribution.c (struct partition): New field recording
24653 its data reference.
24654 (partition_alloc, partition_free): Init and release data refs.
24655 (partition_merge_into): Merge data refs.
24656 (build_rdg_partition_for_vertex): Collect data refs for partition.
24657 (pg_add_dependence_edges): Change parameters from vector to bitmap.
24659 (distribute_loop): Remve data refs from vertice data of partition
24662 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24664 * tree-loop-distribution.c (params.h): Include header file.
24665 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
24666 (datarefs_vec): New global var.
24667 (create_rdg_vertices): Use datarefs_vec directly.
24668 (free_rdg): Don't free data references.
24669 (build_rdg): Update use. Don't free data references.
24670 (distribute_loop): Compute global variable for data references.
24671 Bail out if there are too many data references.
24673 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24675 * tree-loop-distribution.c (loop_nest): New global var.
24676 (build_rdg): Use loop directly, rather than loop nest.
24677 (pg_add_dependence_edges): Remove loop nest parameter. Use global
24679 (distribute_loop): Compute global variable loop nest. Update use.
24681 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24683 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
24684 (partition_merge_into): New parameter. Dump reason for fusion.
24685 (distribute_loop): Update use of partition_merge_into.
24687 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24689 * tree-loop-distribution.c (bb_top_order_index): New.
24690 (bb_top_order_index_size, bb_top_order_cmp): New.
24691 (stmts_from_loop): Use topological order.
24692 (pass_loop_distribution::execute): Compute and release topological
24693 order for basic blocks.
24695 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24697 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
24700 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24702 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
24703 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
24704 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
24705 * internal-fn.def (LOOP_DIST_ALIAS): New.
24706 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
24707 (fold_loop_internal_call): ... this.
24708 (vect_loop_dist_alias_call): New function.
24709 (set_uid_loop_bbs): Call fold_loop_internal_call.
24710 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
24713 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24716 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
24717 Require dead FLAGS_REG at the beginning of a peephole.
24719 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24722 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
24723 arguments in the call to __builtin_ia32_sbb_u32.
24724 (_subborrow_u64): Swap _X and _Y arguments in the call to
24725 __builtin_ia32_sbb_u64.
24727 2017-07-04 Jakub Jelinek <jakub@redhat.com>
24730 * tree-vrp.c (compare_assert_loc): Turn into a function template
24731 with stable template parameter. Only test if a->e is NULL,
24732 !a->e == !b->e has been verified already. Use e == NULL or
24733 e != NULL instead of e or ! e tests. If stable is true, don't use
24734 iterative_hash_expr, on the other side allow a or b or both NULL
24735 and sort the NULLs last.
24736 (process_assert_insertions): Sort using compare_assert_loc<false>
24737 instead of compare_assert_loc, later sort using
24738 compare_assert_loc<true> before calling process_assert_insertions_for
24739 in a loop. Use break instead of continue once seen NULL pointer.
24741 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24743 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24744 Cortex-R7 and Cortex-R8 processors.
24746 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24748 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
24749 uninitialized while src is not.
24751 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
24753 * common/config/arm/arm-common.c: Adjust include path for
24755 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
24756 (arm-cpu.h): Create in build directory. Adjust dependency rules.
24757 (arm-cpu-data.h): Likewise.
24758 (arm-cpu-cdata.h): Likewise.
24759 * config/arm/arm-cpu.h: Delete.
24760 * config/arm/arm-cpu-cdata.h: Delete.
24761 * config/arm/arm-cpu-data.h: Delete.
24763 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
24765 * config/arm/arm-cpus.in (cortex-a55): New.
24766 (cortex-a75): Likewise.
24767 (cortex-a75.cortex-a55): Likewise.
24768 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
24770 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
24771 * config/arm/arm-cpu-cdata.h: Regenerate.
24772 * config/arm/arm-cpu-data.h: Regenerate.
24773 * config/arm/arm-cpu.h: Regenerate.
24774 * config/arm/arm-tables.opt: Regenerate.
24775 * config/arm/arm-tune.md: Regenerate.
24777 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24779 * haifa-sched.c (sched_create_recovery_edges): Update profile.
24781 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24783 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
24786 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
24788 PR tree-optimization/81292
24789 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
24790 full_string_p, also call adjust_related_strinfos if the adjustment
24791 is simple, otherwise invalidate related strinfos.
24793 2017-07-04 Martin Liska <mliska@suse.cz>
24796 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
24797 newly created variable as DECL_IGNORED_P.
24799 2017-07-04 Martin Liska <mliska@suse.cz>
24802 * ipa-inline.c (inline_small_functions):
24803 Use xstrdup_for_dump.
24805 2017-07-04 Tom de Vries <tom@codesourcery.com>
24807 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
24809 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24812 * config/darwin.c (darwin_function_switched_text_sections):
24815 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
24817 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
24819 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24821 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
24823 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24825 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
24826 min_profitable_iters, and th as inclusive lower bounds.
24827 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
24828 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
24829 for min_profitable_iters and min_profitable_estimate.
24830 (vect_transform_loop): Treat th as an inclusive lower bound.
24831 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24833 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24836 * config/darwin.c (darwin_function_switched_text_sections):
24837 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
24838 in two pieces, and suppress the use of buf.
24840 2017-07-03 Nathan Sidwell <nathan@acm.org>
24842 * hash-table.h (hash_table_mod1): Fix indentation.
24844 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24846 PR middle-end/81290
24847 * predict.c (force_edge_cold): Be more careful about propagation
24849 * profile-count.h (profile_probability::guessed,
24850 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
24852 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
24854 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
24856 * doc/invoke.texi (rcpc architecture extension): Document it.
24858 2017-07-03 Richard Biener <rguenther@suse.de>
24860 PR tree-optimization/60510
24861 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
24862 the scalar reduction PHI and use it.
24863 (vectorizable_reduction): Properly guard the single_defuse_cycle
24864 path for non-SLP reduction chains where we cannot use it.
24865 Rework reduc_def/index and vector type deduction. Rework
24866 vector operand gathering during reduction op code-gen.
24867 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
24868 chains dissolve the chain and leave it to non-SLP reduction
24871 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24873 * tree-data-ref.h (dr_alignment): Declare.
24874 * tree-data-ref.c (dr_alignment): New function.
24875 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
24876 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
24878 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
24880 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24882 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
24883 and base_misalignment fields.
24884 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
24885 * tree-data-ref.c: Include builtins.h.
24886 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
24887 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
24888 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
24889 * tree-vect-data-refs.c: Include tree-cfg.h.
24890 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
24891 fields instead of calculating an alignment here.
24892 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
24893 innermost_loop_behavior fields.
24895 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24897 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
24899 (DR_STEP_ALIGNMENT): New macro.
24900 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
24901 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
24902 (create_data_ref): Print it.
24903 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
24904 to tell whether the step preserves vector (mis)alignment.
24905 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24906 Move the check for an integer step and generalise to all INTEGER_CST.
24907 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
24908 Print the outer step alignment.
24910 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24912 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
24913 with offset_alignment.
24914 (DR_ALIGNED_TO): Delete.
24915 (DR_OFFSET_ALIGNMENT): New macro.
24916 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
24917 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
24918 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
24919 (create_data_ref): Likewise.
24920 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24921 (vect_analyze_data_refs): Likewise.
24922 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
24923 creating dummy innermost behavior.
24925 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24927 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
24928 with a "innermost_loop_behavior *" and refeence tree.
24929 * tree-data-ref.c (dr_analyze_innermost): Likewise.
24930 (create_data_ref): Update call accordingly.
24931 * tree-predcom.c (find_looparound_phi): Likewise.
24933 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24935 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
24936 fields with dr_wrt_vec_loop.
24937 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
24938 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
24939 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
24940 (vect_dr_behavior): New function.
24941 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24942 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
24943 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
24944 track whether the step preserves the misalignment.
24945 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24946 Use vect_dr_behavior.
24947 (vect_setup_realignment): Update call accordingly.
24948 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
24949 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
24950 call to vect_create_addr_base_for_vector_ref.
24951 (vect_create_cond_for_align_checks): Likewise.
24952 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
24953 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
24954 (vect_recog_mask_conversion_pattern): Likewise.
24955 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
24956 (new_stmt_vec_info): Remove redundant zeroing.
24958 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24960 * common/config/arm/arm-common.c (arm_be8_option): New function.
24961 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
24962 (ISA_ARMv6): Add isa_bit_be8.
24963 * config/arm/arm.h (arm_be8_option): Add prototype.
24964 (BE8_SPEC_FUNCTION): New define.
24965 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
24966 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
24967 (mlittle-endian): Similarly.
24968 (mbe8, mbe32): New options.
24969 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
24970 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
24972 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24974 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
24976 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24978 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
24979 (cleanup_tree_cfg_bb): Use it.
24980 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
24982 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
24984 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24987 * loop-doloop.c (add_test): Update profile.
24989 2017-07-03 Martin Liska <mliska@suse.cz>
24992 * sanopt.c (rewrite_usage_of_param): New function.
24993 (sanitize_rewrite_addressable_params): Likewise.
24994 (pass_sanopt::execute): Call rewrite_usage_of_param.
24996 2017-07-03 Richard Biener <rguenther@suse.de>
24998 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
24999 back to using VIEW_CONVERT_EXPR.
25001 2017-07-03 Martin Liska <mliska@suse.cz>
25004 * doc/extend.texi: Document when a resolver function is
25005 generated for target_clones.
25007 2017-07-03 Martin Liska <mliska@suse.cz>
25009 * asan.c (asan_emit_stack_protection): Unpoison just red zones
25010 and shadow memory of auto variables which are subject of
25011 use-after-scope sanitization.
25012 (asan_expand_mark_ifn): Add do set only when is_poison.
25014 2016-07-03 Richard Biener <rguenther@suse.de>
25016 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
25018 (vect_force_simple_reduction): Record reduction def -> phi mapping.
25019 (vectorizable_reduction): Perform reduction PHI creation when
25020 visiting a reduction PHI and adjust and simplify code generation
25021 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
25022 (vect_transform_loop): Visit reduction PHIs.
25023 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
25024 defs into the SLP tree.
25025 (vect_build_slp_tree): Reduction defs terminate the recursion.
25026 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
25028 (vect_get_vec_defs_for_stmt_copy): Export.
25029 (vect_get_vec_defs): Likewise.
25030 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
25032 (vect_get_vec_defs_for_stmt_copy): Declare.
25033 (vect_get_vec_defs): Likewise.
25035 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25037 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
25038 parameter with a "loop" parameter and use it instead of the
25039 loop containing DR_STMT. Don't check simple_iv when doing
25040 BB analysis. Describe the two analysis modes in the comment.
25042 2017-07-03 Tom de Vries <tom@codesourcery.com>
25044 PR tree-optimization/69468
25045 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
25046 (find_same_succ_bb): Handle ignore_edge_flags.
25048 2017-07-03 Tom de Vries <tom@codesourcery.com>
25050 PR tree-optimization/81192
25051 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
25053 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
25055 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
25057 2017-07-03 Tom de Vries <tom@codesourcery.com>
25059 PR tree-optimization/81192
25060 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
25061 BB_SAME_SUCC (bb) == NULL.
25063 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25065 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
25068 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25070 * dumpfile.c: Include profile-count.h
25071 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
25073 (insert_cond_bb): Update profile.
25074 * tree-cfg.h (insert_cond_bb): Update prototype.
25075 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
25076 * tree-dump.c: Do not include tree-cfg.
25078 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25080 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
25082 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25084 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
25087 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25089 * tree-complex.c (expand_complex_div_wide): update profile.
25091 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25092 Alan Hayward <alan.hayward@arm.com>
25093 David Sherwood <david.sherwood@arm.com>
25095 * Makefile.in (MACHMODE_H): Remove insn-modes.h
25096 (CORETYPES_H): New define.
25097 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
25098 (insn-modes-inline.h, s-modes-inline-h): New rules.
25099 (generated_files): Add insn-modes-inline.h.
25100 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
25101 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
25102 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
25103 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
25104 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
25105 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
25106 (build/gencodes.o, build/genconditions.o): Likewise.
25107 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
25108 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
25109 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
25110 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
25111 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
25112 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
25113 * coretypes.h: Include everything up to real.h for generators.
25114 Include insn-modes.h first. Include wide-int-print.h after
25115 wide-int.h. Include insn-modes-inline.h and then machmode.h.
25116 * machmode.h: Don't include insn-modes.h here.
25117 * function-tests.c: Remove includes of signop.h, machmode.h,
25118 double-int.h and wide-int.h.
25120 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
25122 * optc-save-gen.awk: Likewise.
25123 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
25124 * godump.c: Remove include of wide-int-print.h.
25125 * pretty-print.h: Likewise.
25126 * wide-int-print.cc: Likewise.
25127 * wide-int.cc: Likewise.
25128 * hash-map-tests.c: Remove include of signop.h.
25129 * hash-set-tests.c: Likewise.
25130 * rtl-tests.c: Likewise.
25131 * mkconfig.sh: Remove include of machmode.h.
25132 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
25134 (emit_insn_modes_inline_h): ...this new function. Emit the code
25135 into an insn-modes-inline.h header file, adding appropriate
25136 include guards and end comments.
25137 (emit_insn_modes_c_header): Remove include of machmode.h.
25138 (emit_min_insn_modes_c_header): Include coretypes.h rather than
25140 (main): Handle -i flag and call emit_insn_modes_inline_h when
25143 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25145 * tree-ssa-strlen.c (strinfo): Rename the length field to
25146 nonzero_chars. Add a full_string_p field.
25147 (compare_nonzero_chars, zero_length_string_p): New functions.
25148 (get_addr_stridx): Add an offset_out parameter.
25149 Use compare_nonzero_chars.
25150 (get_stridx): Update accordingly. Use compare_nonzero_chars.
25151 (new_strinfo): Update after above changes to strinfo.
25152 (set_endptr_and_length): Set full_string_p.
25153 (get_string_length): Update after above changes to strinfo.
25154 (unshare_strinfo): Update call to new_strinfo.
25155 (maybe_invalidate): Likewise.
25156 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
25157 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
25158 as a uhwi instead of an shwi. Update after above changes to
25159 strinfo and new_strinfo.
25160 (zero_length_string): Assert that chainsi contains full strings.
25161 Use zero_length_string_p. Update call to new_strinfo.
25162 (adjust_related_strinfos): Update after above changes to strinfo.
25163 Copy full_string_p from origsi.
25164 (adjust_last_stmt): Use zero_length_string_p.
25165 (handle_builtin_strlen): Update after above changes to strinfo and
25166 new_strinfo. Install the lhs as the string length if the previous
25167 entry didn't describe a full string.
25168 (handle_builtin_strchr): Update after above changes to strinfo
25170 (handle_builtin_strcpy): Likewise.
25171 (handle_builtin_strcat): Likewise.
25172 (handle_builtin_malloc): Likewise.
25173 (handle_pointer_plus): Likewise.
25174 (handle_builtin_memcpy): Likewise. Track nonzero characters
25175 that aren't necessarily followed by a nul terminator.
25176 (handle_char_store): Likewise.
25178 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25180 PR tree-optimization/80769
25181 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
25182 for malloc and calloc. Document the new invariant that all related
25183 strinfos have delayed lengths or none do.
25184 (verify_related_strinfos): Move earlier in file.
25185 (set_endptr_and_length): New function, split out from...
25186 (get_string_length): ...here. Also set the lengths of related
25188 (zero_length_string): Assert that chainsi has known (rather than
25190 (adjust_related_strinfos): Likewise.
25192 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25194 PR tree-optimization/81136
25195 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
25196 assert that two references with the same misalignment have the same
25197 compile-time misalignment if those compile-time misalignments
25200 2017-07-01 Andi Kleen <ak@linux.intel.com>
25202 * print-tree.c (print_node): Print all attributes.
25204 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25206 * cfg.c (scale_bbs_frequencies): New function.
25207 * cfg.h (scale_bbs_frequencies): Declare it.
25208 * cfgloopanal.c (single_likely_exit): Cleanup.
25209 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
25211 (scale_loop_profile): Likewise.
25212 (loop_version): Likewise.
25213 (create_empty_loop_on_edge): Update.
25214 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
25215 scale_loop_frequencies, scale_loop_profile, loopify,
25216 loop_version): Update prototypes.
25217 * modulo-sched.c (sms_schedule): Update.
25218 * predict.c (unlikely_executed_edge_p): Also check probability.
25219 (probably_never_executed_edge_p): Fix typo.
25220 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25221 * tree-parloops.c (gen_parallel_loop): Update.
25222 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
25223 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25224 * tree-ssa-loop-split.c (split_loop): Update.
25225 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25226 * tree-vect-loop-manip.c (vect_do_peeling): Update.
25227 (vect_loop_versioning): Update.
25228 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25230 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25232 * trans-mem.c (split_bb_make_tm_edge): Update profile.
25234 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25236 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
25237 to keep profile consistent.
25239 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25241 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
25242 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
25243 * profile-count.h (max_safe_multiplier): Make unsigned.
25244 (profile_count::guessed_zero): New.
25246 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25248 * bb-reorder.c (fix_up_crossing_landing_pad,
25249 fix_crossing_conditional_branches): Use make_single_succ_edge
25250 to keep profile consistent.
25252 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25254 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
25257 2017-07-01 Jakub Jelinek <jakub@redhat.com>
25260 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
25261 the right scopes, make sure cond_jump isn't preserved between multiple
25262 iterations. Search for fallthru edge whenever there are 3+ edges and
25263 use find_fallthru_edge for it.
25265 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25267 Patch by Alexander Monakov <amonakov@ispras.ru>
25268 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
25269 probabilities consistently.
25271 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25273 * pa.c (pa_expand_compare_and_swap_loop): Update call of
25274 emit_cmp_and_jump_insns.
25276 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25279 * tree-inline.c (expand_call_inline): Combine profile statuses.
25281 2017-06-30 Andrew Pinski <apinski@cavium.com>
25283 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
25284 fold_stmt returned true.
25286 2017-06-30 Nathan Sidwell <nathan@acm.org>
25288 * ggc.h (empty_string): Delete.
25289 * cfgexpand.c (expand_asm_stmt): Use plain "".
25290 * optabs.c (expand_asm_memory_barrier): Likewise.
25291 * stringpool.c (empty_string): Delete.
25292 (digit_vector, digit_string): Delete.
25293 (ggc_alloc_string): Use plain "", don't optimize single digit
25294 strings. Use ggc_alloc_atomic.
25296 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
25298 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
25299 comparison set and one other set, use the cost of the non-comparison
25302 2017-06-30 Nathan Sidwell <nathan@acm.org>
25304 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
25307 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
25309 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
25310 loops. Remove now unneeded calls to gimple_switch_set_label() that
25311 just set removed labels to NULL_TREE.
25313 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
25315 * tree-ssanames.c (set_range_info_raw): Abstract from ...
25316 (set_range_info): ...here. Only call set_range_info_raw if domain
25318 (set_nonzero_bits): Call set_range_info_raw.
25319 * tree-ssanames.h (set_range_info_raw): New.
25321 2017-06-30 Jakub Jelinek <jakub@redhat.com>
25324 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
25325 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
25326 of nonimmediate_operand and <store_mask_constraint> instead of m
25327 for the input operand. For V8FI iterator, always split if input
25328 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
25329 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
25330 <store_mask_predicate> instead of register_operand and
25331 <store_mask_constraint> instead of v for the input operand. Make
25332 sure both operands aren't MEMs for if not <mask_applied>.
25334 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
25336 * lto-wrapper.c (copy_file) Close both file descriptors before
25339 2017-06-30 Martin Liska <mliska@suse.cz>
25342 * multiple_target.c (create_dispatcher_calls): Make ifunc
25343 also for function that don't have calls or are not referenced.
25345 2017-06-30 Richard Biener <rguenther@suse.de>
25347 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
25348 analyze the first scalar stmt. Move vector type computation
25349 for the BB case here from ...
25350 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
25351 live operation processing in the SLP case properly.
25353 2017-06-30 Richard Biener <rguenther@suse.de>
25355 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
25357 2017-06-30 Martin Liska <mliska@suse.cz>
25360 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
25361 before BUILT_IN_UNWIND_RESUME when ASAN is used.
25363 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
25365 * doc/invoke.texi (AArch64): Add missing options and remove redundant
25368 2017-06-30 Richard Biener <rguenther@suse.de>
25370 PR tree-optimization/81249
25371 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
25372 condition reduction result to original scalar type.
25374 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25376 * profile-count.h (enum profile_quality): Fix typos and whitespace
25379 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25381 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
25382 type for branch probabilities.
25384 2017-06-29 Julian Brown <julian@codesourcery.com>
25385 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25387 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
25388 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
25389 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
25390 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
25392 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25394 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
25395 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
25396 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
25397 CC usage from generic code to here.
25398 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
25399 CC usage into the target macros.
25401 2017-06-29 Maya Rashish <coypu@sdf.org>
25403 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
25406 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25408 * arm/arm-builtins.c: Include profile-count.h
25409 * except.c (sjlj_emit_function_enter): Use
25410 profile_probability::unlikely.
25412 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25414 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
25415 and tocrel_offset be pointer args rather than implicitly using
25417 (legitimate_constant_pool_address_p, rs6000_emit_move,
25418 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
25419 tocrel_offset and use in toc_relative_expr_p call.
25420 (print_operand, print_operand_address): Use static tocrel_base_oac
25421 and tocrel_offset_oac.
25422 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
25425 2017-06-29 Maya Rashish <coypu@sdf.org>
25427 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
25429 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
25431 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
25432 objects, take into account only the alignment of 'op0' and 'mode1' if
25435 2017-06-29 Steve Ellcey <sellcey@cavium.com>
25437 * ccmp.c (ccmp_tree_comparison_p): New function.
25438 (ccmp_candidate_p): Update to use above function.
25439 (get_compare_parts): New function.
25440 (expand_ccmp_next): Update to use new functions.
25441 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
25443 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
25444 take mode as argument.
25445 * ccmp.h (expand_ccmp_expr): Add mode as argument.
25446 * expr.c (expand_expr_real_1): Pass mode as argument.
25448 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
25450 * combine.c (combine_instructions): Print insns to dump_file, together
25453 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25455 * asan.c (asan_emit_stack_protection): Update.
25456 (create_cond_insert_point): Update.
25457 * auto-profile.c (afdo_propagate_circuit): Update.
25458 * basic-block.h (struct edge_def): Turn probability to
25459 profile_probability.
25460 (EDGE_FREQUENCY): Update.
25461 * bb-reorder.c (find_traces_1_round): Update.
25462 (better_edge_p): Update.
25463 (sanitize_hot_paths): Update.
25464 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
25465 (make_single_succ_edge): Update.
25466 (check_bb_profile): Update.
25467 (dump_edge_info): Update.
25468 (update_bb_profile_for_threading): Update.
25469 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
25470 probabilitycount to 0.
25471 * cfgbuild.c (compute_outgoing_frequencies): Update.
25472 * cfgcleanup.c (try_forward_edges): Update.
25473 (outgoing_edges_match): Update.
25474 (try_crossjump_to_edge): Update.
25475 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
25476 (expand_gimple_tailcall): Update.
25477 (construct_init_block): Use make_single_succ_edge.
25478 (construct_exit_block): Use make_single_succ_edge.
25479 * cfghooks.c (verify_flow_info): Update.
25480 (redirect_edge_succ_nodup): Update.
25481 (split_edge): Update.
25482 (account_profile_record): Update.
25483 * cfgloopanal.c (single_likely_exit): Update.
25484 * cfgloopmanip.c (scale_loop_profile): Update.
25485 (set_zero_probability): Remove.
25486 (duplicate_loop_to_header_edge): Update.
25487 * cfgloopmanip.h (loop_version): Update prototype.
25488 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
25489 (force_nonfallthru_and_redirect): Update.
25490 (update_br_prob_note): Update.
25491 (rtl_verify_edges): Update.
25492 (purge_dead_edges): Update.
25493 (rtl_lv_add_condition_to_bb): Update.
25494 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
25495 * cgraphunit.c (init_lowered_empty_function): Update.
25496 (cgraph_node::expand_thunk): Update.
25497 * cilk-common.c: Include profile-count.h
25498 * dojump.c (inv): Remove.
25499 (jumpifnot): Update.
25500 (jumpifnot_1): Update.
25501 (do_jump_1): Update.
25503 (do_jump_by_parts_greater_rtx): Update.
25504 (do_compare_rtx_and_jump): Update.
25505 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
25506 do_jump_1. do_compare_rtx_and_jump): Update prototype.
25507 * dwarf2cfi.c: Include profile-count.h
25508 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
25509 (sjlj_emit_dispatch_table): Likewise.
25510 * explow.c: Include profile-count.h
25511 * expmed.c (emit_store_flag_force): Update.
25512 (do_cmp_and_jump): Update.
25513 * expr.c (compare_by_pieces_d::generate): Update.
25514 (compare_by_pieces_d::finish_mode): Update.
25515 (emit_block_move_via_loop): Update.
25516 (store_expr_with_bounds): Update.
25517 (store_constructor): Update.
25518 (expand_expr_real_2): Update.
25519 (expand_expr_real_1): Update.
25520 * expr.h (try_casesi, try_tablejump): Update prototypes.
25521 * gimple-pretty-print.c (dump_probability): Update.
25522 (dump_profile): New.
25523 (dump_gimple_label): Update.
25524 (dump_gimple_bb_header): Update.
25525 * graph.c (draw_cfg_node_succ_edges): Update.
25526 * hsa-gen.c (convert_switch_statements): Update.
25527 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
25528 (find_if_case_1): Update.
25529 (find_if_case_2): Update.
25530 * internal-fn.c (expand_arith_overflow_result_store): Update.
25531 (expand_addsub_overflow): Update.
25532 (expand_neg_overflow): Update.
25533 (expand_mul_overflow): Update.
25534 (expand_vector_ubsan_overflow): Update.
25535 * ipa-cp.c (good_cloning_opportunity_p): Update.
25536 * ipa-split.c (split_function): Use make_single_succ_edge.
25537 * ipa-utils.c (ipa_merge_profiles): Update.
25538 * loop-doloop.c (add_test): Update.
25539 (doloop_modify): Update.
25540 * loop-unroll.c (compare_and_jump_seq): Update.
25541 (unroll_loop_runtime_iterations): Update.
25542 * lra-constraints.c (lra_inheritance): Update.
25543 * lto-streamer-in.c (input_cfg): Update.
25544 * lto-streamer-out.c (output_cfg): Update.
25545 * mcf.c (adjust_cfg_counts): Update.
25546 * modulo-sched.c (sms_schedule): Update.
25547 * omp-expand.c (expand_omp_for_init_counts): Update.
25548 (extract_omp_for_update_vars): Update.
25549 (expand_omp_ordered_sink): Update.
25550 (expand_omp_for_ordered_loops): Update.
25551 (expand_omp_for_generic): Update.
25552 (expand_omp_for_static_nochunk): Update.
25553 (expand_omp_for_static_chunk): Update.
25554 (expand_cilk_for): Update.
25555 (expand_omp_simd): Update.
25556 (expand_omp_taskloop_for_outer): Update.
25557 (expand_omp_taskloop_for_inner): Update.
25558 * omp-simd-clone.c (simd_clone_adjust): Update.
25559 * optabs.c (expand_doubleword_shift): Update.
25560 (expand_abs): Update.
25561 (emit_cmp_and_jump_insn_1): Update.
25562 (expand_compare_and_swap_loop): Update.
25563 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
25564 * predict.c (predictable_edge_p): Update.
25565 (edge_probability_reliable_p): Update.
25566 (set_even_probabilities): Update.
25567 (combine_predictions_for_insn): Update.
25568 (combine_predictions_for_bb): Update.
25569 (propagate_freq): Update.
25570 (estimate_bb_frequencies): Update.
25571 (force_edge_cold): Update.
25572 * profile-count.c (profile_count::dump): Add missing space into dump.
25573 (profile_count::debug): Add newline.
25574 (profile_count::differs_from_p): Explicitly convert to unsigned.
25575 (profile_count::stream_in): Update.
25576 (profile_probability::dump): New member function.
25577 (profile_probability::debug): New member function.
25578 (profile_probability::differs_from_p): New member function.
25579 (profile_probability::differs_lot_from_p): New member function.
25580 (profile_probability::stream_in): New member function.
25581 (profile_probability::stream_out): New member function.
25582 * profile-count.h (profile_count_quality): Rename to ...
25583 (profile_quality): ... this one.
25584 (profile_probability): New.
25585 (profile_count): Update.
25586 * profile.c (compute_branch_probabilities): Update.
25587 * recog.c (peep2_attempt): Update.
25588 * sched-ebb.c (schedule_ebbs): Update.
25589 * sched-rgn.c (find_single_block_region): Update.
25590 (compute_dom_prob_ps): Update.
25591 (schedule_region): Update.
25592 * sel-sched-ir.c (compute_succs_info): Update.
25593 * stmt.c (struct case_node): Update.
25594 (do_jump_if_equal): Update.
25595 (get_outgoing_edge_probs): Update.
25596 (conditional_probability): Update.
25597 (emit_case_dispatch_table): Update.
25598 (expand_case): Update.
25599 (expand_sjlj_dispatch_table): Update.
25600 (emit_case_nodes): Update.
25601 * targhooks.c: Update.
25602 * tracer.c (better_p): Update.
25603 (find_best_successor): Update.
25604 * trans-mem.c (expand_transaction): Update.
25605 * tree-call-cdce.c: Update.
25606 * tree-cfg.c (gimple_split_edge): Upate.
25607 (move_sese_region_to_fn): Upate.
25608 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
25609 * tree-eh.c (lower_resx): Upate.
25610 (cleanup_empty_eh_move_lp): Upate.
25611 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25612 * tree-inline.c (copy_edges_for_bb): Update.
25613 (copy_cfg_body): Update.
25614 * tree-parloops.c (gen_parallel_loop): Update.
25615 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
25616 (gimple_gen_time_profiler): Update.
25617 * tree-ssa-dce.c (remove_dead_stmt): Update.
25618 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25619 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
25620 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
25621 (unloop_loops): Update.
25622 (try_peel_loop): Update.
25623 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25624 * tree-ssa-loop-split.c (connect_loops): Update.
25625 (split_loop): Update.
25626 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25627 (hoist_guard): Update.
25628 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
25629 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
25630 (value_replacement): Update.
25631 * tree-ssa-reassoc.c (branch_fixup): Update.
25632 * tree-ssa-tail-merge.c (replace_block_by): Update.
25633 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
25634 (create_edge_and_update_destination_phis): Update.
25635 (compute_path_counts): Update.
25636 (recompute_probabilities): Update.
25637 (update_joiner_offpath_counts): Update.
25638 (freqs_to_counts_path): Update.
25639 (duplicate_thread_path): Update.
25640 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
25641 (struct switch_conv_info): Update.
25642 (gen_inbound_check): Update.
25643 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25644 (vect_do_peeling): Update.
25645 (vect_loop_versioning): Update.
25646 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25647 (optimize_mask_stores): Update.
25648 * ubsan.c (ubsan_expand_null_ifn): Update.
25649 * value-prof.c (gimple_divmod_fixed_value): Update.
25650 (gimple_divmod_fixed_value_transform): Update.
25651 (gimple_mod_pow2): Update.
25652 (gimple_mod_pow2_value_transform): Update.
25653 (gimple_mod_subtract): Update.
25654 (gimple_mod_subtract_transform): Update.
25655 (gimple_ic): Update.
25656 (gimple_stringop_fixed_value): Update.
25657 (gimple_stringops_transform): Update.
25658 * value-prof.h: Update.
25660 2017-06-29 Carl Love <cel@us.ibm.com>
25662 * config/rs6000/rs6000-c.c: Add support for built-in functions
25663 vector signed int vec_signed (vector float);
25664 vector signed long long vec_signed (vector double);
25665 vector signed int vec_signed2 (vector double, vector double);
25666 vector signed int vec_signede (vector double);
25667 vector signed int vec_signedo (vector double);
25668 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
25669 instruction generator.
25670 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25671 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
25672 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
25674 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
25675 vunsignede_v2df): Add define_expands.
25676 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
25677 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
25678 VEC_UNSIGNEDO): Add definitions.
25679 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25680 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
25681 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
25682 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
25683 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
25684 * config/rs6000/altivec.h (vec_signed, vec_signed2,
25685 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
25686 vec_unsignede, vec_unsignedo): Add builtin defines.
25687 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
25689 * doc/extend.texi: Update the built-in documentation file for the
25690 new built-in functions.
25692 2017-06-29 Richard Biener <rguenther@suse.de>
25694 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
25695 reduction chains to LOOP_VINFO_REDUCTIONS.
25696 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
25697 SLP reductions after processing reduction chains.
25699 2017-06-29 Nathan Sidwell <nathan@acm.org>
25701 * builtins.c (fold_builtin_FUNCTION): Use
25702 lang_hooks.decl_printable_name.
25704 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
25706 PR middle-end/81194
25707 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
25708 with only one label.
25709 * stmt.c (expand_case): Assert NCASES is greater than one.
25711 2017-06-29 Richard Biener <rguenther@suse.de>
25713 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
25715 (group_case_labels): Likewise.
25716 (find_taken_edge): Push sanity checking on val to workers...
25717 (find_taken_edge_cond_expr): ... here
25718 (find_taken_edge_switch_expr): ... and here, handle cases
25719 with just a default label.
25720 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
25721 (group_case_labels): Likewise.
25722 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
25723 group_case_labels does anything cleanup the CFG again.
25725 2017-06-29 Bin Cheng <bin.cheng@arm.com>
25727 PR tree-optimization/81196
25728 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
25729 exit condition comparing two IVs.
25731 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
25733 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
25734 profile to the dummy entry at the end of the list of architectures.
25735 * config/arm/arm-cpu-cdata.h: Regenerated.
25737 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25738 Michael Collison <michael.collison@arm.com>
25741 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
25743 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
25744 (*aarch64_reg_<mode>3_minus_mask): New pattern.
25745 (*aarch64_<optab>_reg_di3_mask2): New pattern.
25746 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
25747 of shift when the shift amount is masked with constant equal to
25748 the size of the mode.
25749 * config/aarch64/predicates.md (subreg_lowpart_operator): New
25752 2017-06-29 Martin Liska <mliska@suse.cz>
25754 * config/i386/i386.opt: Change range from [1,5] to [0,5].
25756 2017-06-29 Yury Gribov <tetra2005@gmail.com>
25759 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
25762 (edge_growth_cache_entry::edge_growth_cache_entry): New
25764 (reset_edge_growth_cache): Update to use constructor.
25766 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25768 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
25769 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25770 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25772 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
25774 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
25775 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
25777 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
25779 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
25780 (*-linux-uclibc*): Add t-uclibc tmake_file.
25781 * config/t-musl: New.
25782 * config/t-uclibc: New.
25784 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
25786 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
25788 (gen_comm_data): Emit architectural setting of arch_prof.
25789 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
25791 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
25792 (armv8-m.base, armv8-m.main): Likewise.
25793 * arm-protos.h (arm_build_target): Add profile field.
25794 (arch_option): Likewise.
25795 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
25797 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
25798 arm_active_target.profile.
25800 2017-06-28 Richard Biener <rguenther@suse.de>
25802 PR middle-end/81227
25803 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
25804 TYPE_OVERFLOW_WRAPS.
25805 * match.pd (negate_expr_p): Likewise.
25806 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
25807 fold_build2, not fold_binary.
25809 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25811 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
25812 Convert memory address to Pmode.
25813 (aarch64_print_operand): Assert MEM operands are always Pmode.
25815 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25818 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
25819 Remove redundant if.
25820 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
25821 * config/arm/aarch-common-protos.h
25822 (aarch_forward_to_shift_is_not_shifted_re): Remove.
25823 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
25825 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
25828 * multiple_target.c (create_dispatcher_calls): Set the default
25829 clone to be static, not public.
25831 2017-06-28 Richard Biener <rguenther@suse.de>
25833 * tree-vect-loop.c (vectorizable_reduction): Move special
25834 cond reduction IV var creation ...
25835 (vect_create_epilog_for_reduction): ... here. Remove induction_index
25836 parameter. Use STMT_VINFO_VECTYPE.
25837 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
25840 2017-06-28 Martin Liska <mliska@suse.cz>
25843 * ipa-visibility.c (non_local_p): Handle visibility.
25845 2017-06-28 Martin Liska <mliska@suse.cz>
25848 * common.opt: Add IntegerRange to various options.
25849 * opt-functions.awk (integer_range_info): New function.
25850 * optc-gen.awk: Add integer_range_info to cl_options struct.
25851 * opts-common.c (decode_cmdline_option): Handle
25852 CL_ERR_INT_RANGE_ARG.
25853 (cmdline_handle_error): Likewise.
25854 * opts.c (print_filtered_help): Show valid interval in
25855 when --help is provided.
25856 * opts.h (struct cl_option): Add range_min and range_max fields.
25857 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
25859 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
25861 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
25862 (x * C EQ/NE y * C): New transformation.
25864 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
25866 * genmultilib (combination_space): Accept '+' in option names.
25868 2017-06-28 Martin Liska <mliska@suse.cz>
25871 * asan.c (instrument_derefs): Bail out inner references
25872 that are hard register variables.
25874 2017-06-28 Jakub Jelinek <jakub@redhat.com>
25877 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
25878 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
25880 2017-06-28 Richard Biener <rguenther@suse.de>
25882 * tree-vectorizer.h (vect_get_vec_defs): Remove.
25883 (vect_get_slp_defs): Adjust.
25884 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
25886 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
25888 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
25889 get_initial_defs_for_reduction instead of vect_get_vec_defs.
25890 (vectorizable_reduction): Adjust.
25891 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
25893 (vect_get_slp_defs): Likewise.
25894 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
25895 (vectorizable_bswap): Adjust.
25896 (vectorizable_call): Likewise.
25897 (vectorizable_conversion): Likewise.
25898 (vectorizable_assignment): Likewise.
25899 (vectorizable_shift): Likewise.
25900 (vectorizable_operation): Likewise.
25901 (vectorizable_store): Likewise.
25902 (vectorizable_condition): Likewise.
25903 (vectorizable_comparison): Likewise.
25905 2017-06-28 Michael Collison <michael.collison@arm.com>
25908 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
25910 (arm_gen_movmemqi): Removed unused variable 'i'.
25911 Convert 'for' loop into 'while' loop.
25912 (arm_expand_prologue): Remove last unnecessary set of insn.
25913 (thumb_pop): Remove unused variable 'pushed_words'.
25914 (thumb_exit): Remove last unnecessary set of regs_to_pop.
25916 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25918 * config/s390/predicates.md: Use s390_rel_address_ok_p.
25919 * config/s390/s390-protos.h: Add prototype of
25920 s390_rel_address_ok_p.
25921 * config/s390/s390.c (s390_got_symbol): New function.
25922 (s390_rel_address_ok_p): New function.
25923 (legitimize_pic_address): Use s390_rel_address_ok_p.
25924 (s390_load_got): Use s390_got_symbol.
25925 (s390_option_override): Issue error if
25926 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
25927 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
25929 * config/s390/s390.opt: New option mpic-data-is-text-relative.
25931 2017-06-27 Andrew Pinski <apinski@cavium.com>
25933 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
25934 (X * copysign (1.0, X)): New pattern.
25935 (X * copysign (1.0, -X)): New pattern.
25936 (copysign (-1.0, CST)): New pattern.
25938 2017-06-27 Joseph Myers <joseph@codesourcery.com>
25940 * genmultilib (combination_space): Remove variable.
25941 Validate reuse rules against regular expression for any sequence
25942 of multilib options in any order.
25944 2017-06-27 Michael Collison <michael.collison@arm.com>
25946 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
25947 call aarch64_split_simd_combine.
25948 * (aarch64_combine_internal<mode>): Delete pattern.
25949 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
25950 Allow register and subreg operands.
25952 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25954 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
25955 specific need, just fallback on defaults.
25956 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
25958 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25959 Olivier Hainque <hainque@adacore.com>
25961 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
25963 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
25964 targets. Pick a default if no particular attempt applied.
25965 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
25968 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25970 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
25971 (x86_64-wrs-vxworks7): Likewise.
25973 2017-06-27 Marek Polacek <polacek@redhat.com>
25976 * ubsan.c (instrument_null): Check get_base_address's result for null.
25978 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25980 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
25982 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25984 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
25985 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
25986 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
25987 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
25988 New function types.
25989 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
25990 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
25991 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
25992 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
25993 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
25994 BUILT_IN_FEUPDATEENV): New builtins.
25995 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
25996 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
25997 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
25998 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
26000 (builtin_structptr_types): Adjust size.
26001 * tree.c (builtin_structptr_types): Add four entries.
26003 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26004 Olivier Hainque <hainque@adacore.com>
26006 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
26007 (TLS_SYM): New local macro, forcing reference to __tls__ on
26008 link command lines for VxWorks 7 RTPs, triggering initialization
26010 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
26011 OS features TLS support, true for RTPs on VxWorks 7.
26012 * config/vxworks.c (vxworks_override_options): Setup emutls
26015 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26017 * predict.c (test_prediction_value_range): Use -1U instead of -1
26018 to avoid narrowing conversion warning.
26019 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
26020 to avoid narrowing conversion warning.
26021 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
26023 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
26025 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26027 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
26028 64bit configurations.
26029 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
26030 (SIZE_TYPE): Likewise.
26031 * config/vxworks.c (vxworks_emutls_var_fields): Use
26032 long_unsigned_type_node instead of unsigned_type_node as the offset
26033 field type, which is "pointer" mode in emutls.c.
26035 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26038 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
26040 PR middle-end/81207
26041 * gimple-fold.c (replace_call_with_call_and_fold): Handle
26042 gimple_vuse copying separately from gimple_vdef copying.
26044 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26046 * value-prof.c (free_hist): Remove call to memset and the enclosing if
26049 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26050 Olivier Hainque <hainque@adacore.com>
26052 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
26053 for all vxworks7 targets.
26054 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
26055 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
26056 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
26057 variations for VX6/VX7 and 32/64bits later on in ...
26058 (VXWORKS_LIB_SPEC): Leverage new macros.
26059 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
26060 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
26062 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26064 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
26065 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
26067 2017-06-26 Carl Love <cel@us.ibm.com>
26069 * config/rs6000/rs6000-c.c: Add support for built-in functions
26070 vector bool char vec_reve (vector bool char);
26071 vector signed char vec_reve (vector signed char);
26072 vector unsigned char vec_reve (vector unsigned char);
26073 vector bool int vec_reve (vector bool int);
26074 vector signed int vec_reve (vector signed int);
26075 vector unsigned int vec_reve (vector unsigned int);
26076 vector bool long long vec_reve (vector bool long long);
26077 vector signed long long vec_reve (vector signed long long);
26078 vector unsigned long long vec_reve (vector unsigned long long);
26079 vector bool short vec_reve (vector bool short);
26080 vector signed short vec_reve (vector signed short);
26081 vector double vec_reve (vector double);
26082 vector float vec_reve (vector float);
26083 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
26084 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
26085 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
26086 (altivec_vreve): New pattern.
26087 * config/rs6000/altivec.h (vec_reve): New define.
26088 * doc/extend.texi (vec_rev): Update the built-in documentation file
26089 for the new built-in functions.
26091 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26093 PR tree-optimization/71815
26094 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
26096 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
26098 (slsr_process_phi): Likewise.
26099 (replace_uncond_cands_and_profitable_phis): Don't replace a
26100 multiply candidate with a stride of 1 (copy or cast).
26101 (phi_incr_cost): Call uses_consumed_by_stmt rather than
26103 (lowest_cost_path): Likewise.
26104 (total_savings): Likewise.
26106 2017-06-26 Richard Biener <rguenther@suse.de>
26109 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
26110 Use def_builtin_pure for all gather builtins.
26112 2017-06-26 Richard Biener <rguenther@suse.de>
26114 PR tree-optimization/81203
26115 * tree-tailcall.c (find_tail_calls): Do not move stmts into
26116 non-dominating BBs.
26118 2017-06-26 Marek Polacek <polacek@redhat.com>
26121 * doc/invoke.texi: Document -Wmultistatement-macros.
26123 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
26125 * doc/sourcebuild.texi (ARM-specific attributes): Document new
26126 arm_neon_ok_no_float_abi effective target.
26128 2017-06-26 Richard Biener <rguenther@suse.de>
26130 PR tree-optimization/80928
26131 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
26132 (copy_bbs): Set BB_DUPLICATED flag early.
26133 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
26135 (execute_on_shrinking_pred): Likewise.
26136 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
26137 BB_DUPLICATED blocks.
26138 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
26139 iterate over all PHIs considering removal of *gsi.
26141 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
26143 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
26146 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26148 * config/rs6000/rs6000-string.c: (expand_block_clear,
26149 do_load_for_compare, select_block_compare_mode,
26150 compute_current_alignment, expand_block_compare,
26151 expand_strncmp_align_check, expand_strn_compare,
26152 expand_block_move, rs6000_output_load_multiple)
26153 Move functions related to string/block move/compare
26154 to a separate file.
26155 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
26156 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
26157 for this function which is now used in two files.
26158 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
26159 * config.gcc: Add rs6000-string.o to extra_objs for
26160 targets powerpc*-*-* and rs6000*-*-*.
26162 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
26165 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
26166 32-bit, since indexed is not valid for DImode.
26167 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
26168 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
26169 (define_peephole2 for Altivec d-form load): Add 32-bit support.
26170 (define_peephole2 for Altivec d-form store): Likewise.
26173 * multiple_target.c (create_dispatcher_calls): Only create the
26174 dispatcher call if the function is the default clone of a
26175 versioned function.
26177 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
26179 PR middle-end/80902
26180 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
26181 a call, force the call to not be a tail call.
26183 2017-06-23 Jeff Law <law@redhat.com>
26185 * doc/contrib.texi: Add entry for Steven Pemberton's work on
26188 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
26190 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
26191 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
26192 handling for early expansion of vector shifts (sl,sr,sra,rl).
26193 (builtin_function_type): Add vector shift right instructions
26194 to the unsigned argument list.
26196 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
26198 rtl-optimizatoin/79286
26199 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
26200 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
26201 trap. PIC register plus a const unspec without offset can never trap.
26203 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
26205 * tree.h (builtin_structptr_type): New type.
26206 (builtin_structptr_types): Declare new array.
26207 * tree.c (builtin_structptr_types): New array.
26208 (free_lang_data, build_common_tree_nodes): Use it.
26210 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
26213 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
26216 2017-06-22 Matt Turner <mattst88@gmail.com>
26218 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
26219 Lake models to skylake case. Assume skylake for unknown
26220 models with clflushopt.
26222 2017-06-22 Jeff Law <law@redhat.com>
26224 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
26225 frame sizes that do not satisfy aarch64_uimm12_shift.
26227 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
26229 * profile-count.h (apply_probability,
26230 apply_scale, probability_in): Fix checks for zero.
26232 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26234 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
26235 * doc/cppdiropts.texi (-I @var{dir}): Document it.
26237 2016-06-22 Richard Biener <rguenther@suse.de>
26239 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26240 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
26241 REDUC_MAX_EXPR support.
26242 (vectorizable_reduction): Likewise.
26243 (vect_create_epilog_for_reduction): Likewise.
26245 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
26247 * match.pd (A / (1 << B) -> A >> B): New.
26248 * generic-match-head.c: Include optabs-tree.h.
26249 * gimple-match-head.c: Likewise.
26250 * optabs-tree.h (target_supports_op_p): New.
26251 * optabs-tree.c (target_supports_op_p): New.
26253 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26255 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
26256 $gcc_cv_ld --help output.
26257 (gcc_cv_ld_demangle): Likewise.
26258 (gcc_cv_ld_eh_frame_hdr): Likewise.
26259 (gcc_cv_ld_pie): Likewise.
26260 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
26261 (gcc_cv_ld_buildid): Likewise.
26262 (gcc_cv_ld_sysroot): Likewise.
26263 (ld_bndplt_support): Likewise.
26264 (ld_pushpopstate_support): Likewise.
26265 * configure: Regenerate.
26266 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
26268 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26271 * config/i386/sse.md (round<mode>2): Renumber match_dup and
26272 operands indexes to avoid gap between operands and match_dups.
26274 2017-06-21 Andrew Pinski <apinski@cavium.com>
26276 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
26277 Increment Arith_shift and Arith_shift_reg by 1.
26278 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
26280 * config/aarch64/aarch64.c (thunderx_tunings): Enable
26281 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
26282 (aarch64_strip_extend): Add new argument and test for it.
26283 (aarch64_cheap_mult_shift_p): New function.
26284 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
26285 add a cost if it is true.
26286 Update calls to aarch64_strip_extend.
26287 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
26289 2017-06-21 Andrew Pinski <apinski@cavium.com>
26291 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
26293 (thunderxt88): Likewise.
26294 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
26295 (thunderx_prefetch_tune): New variable.
26296 (thunderx2t99_prefetch_tune): Update for the correct values.
26297 (thunderxt88_tunings): New variable.
26298 (thunderx_tunings): Use thunderx_prefetch_tune instead of
26299 generic_prefetch_tune.
26300 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
26302 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26304 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
26305 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
26306 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
26307 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
26308 (aarch64_atomic_cas<mode>, GPI): Likewise.
26310 2017-06-21 Martin Liska <mliska@suse.cz>
26312 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
26313 statements on cold and hot labels.
26314 * predict.c (tree_estimate_probability_bb): Remove the
26315 prediction from this place.
26317 2017-06-21 Martin Liska <mliska@suse.cz>
26319 PR tree-optimization/79489
26320 * gimplify.c (maybe_add_early_return_predict_stmt): New
26322 (gimplify_return_expr): Call the function.
26323 * predict.c (tree_estimate_probability_bb): Remove handling
26325 * predict.def: Update comment about early return predictor.
26326 * gimple-predict.h (is_gimple_predict): New function.
26327 * predict.def: Change default value of early return to 66.
26328 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
26330 * passes.def: Put pass_strip_predict_hints to the beginning of
26333 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
26335 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
26336 FUNCTION_DECL declarations.
26337 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
26339 (dwaf2out_decl): Likewise.
26340 * godump.c (go_early_global_decl): Skip call to the real debug hook
26341 for FUNCTION_DECL declarations.
26342 * passes.c (rest_of_decl_compilation): Skip call to the
26343 early_global_decl debug hook for FUNCTION_DECL declarations, unless
26344 -fdump-go-spec is passed.
26346 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26348 * config/i386/i386.c (struct builtin_isa): New field pure_p.
26349 Reorder for compactness.
26350 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
26351 (def_builtin_pure, def_builtin_pure2): New functions.
26352 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
26354 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26356 * match.pd (nop_convert): New predicate.
26357 ((A +- CST1) +- CST2): Allow some NOP conversions.
26359 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26362 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
26363 with ctors/dtors if GOVD_SHARED is set.
26365 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26367 * config/aarch64/aarch64.md (movti_aarch64):
26368 Emit mov rather than orr.
26369 (movtf_aarch64): Likewise.
26370 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
26371 Emit mov rather than orr.
26373 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26375 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
26376 Swap alternatives, make integer dup more expensive.
26378 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26380 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
26381 Return true for non-tls symbols.
26383 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
26385 * config/aarch64/aarch64-cores.def (cortex-a55): New.
26386 (cortex-a75): Likewise.
26387 (cortex-a75.cortex-a55): Likewise.
26388 * config/aarch64/aarch64-tune.md: Regenerate.
26389 * doc/invoke.texi (-mtune): Document new values for -mtune.
26391 2017-06-21 Tom de Vries <tom@codesourcery.com>
26393 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
26394 stack_size feature.
26395 (Effective-Target Keywords, Other attributes): Suggest using
26396 dg-add-options stack_size feature to get stack limit in stack_size
26397 effective target documentation.
26399 2017-06-21 Julian Brown <julian@codesourcery.com>
26400 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26402 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
26403 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
26404 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
26406 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
26407 attribute type list for neon_multiply.
26408 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
26409 attribute type list for neon_multiply.
26410 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
26411 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
26412 attribute type list for neon_multiply.
26413 * config/arm/types.md (crypto_pmull): Add.
26414 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
26415 attribute type list.
26417 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
26419 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
26422 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26424 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
26425 to make sure not to dereference a NULL cost_classes_ptr pointer.
26427 2017-06-20 Carl Love <cel@us.ibm.com>
26429 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26430 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
26431 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
26432 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
26433 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
26434 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
26435 VMULOSW): New enum "unspec" values.
26436 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
26437 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
26438 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
26439 altivec_vmulosw): New patterns.
26440 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
26441 VMULOSW): Add definitions.
26443 2017-06-20 Julia Koval <julia.koval@intel.com>
26445 * config/i386/i386.c: Fix rounding expand for new pattern.
26446 * config/i386/subst.md: Fix pattern (parallel -> unspec).
26448 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26450 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
26451 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
26453 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26455 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
26458 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26460 * config/aarch64/aarch64-cores.def: Rearrange to sort by
26461 architecture, then by implementer ID.
26462 * config/aarch64/aarch64-tune.md: Regenerate.
26464 2017-06-20 Richard Biener <rguenther@suse.de>
26466 PR middle-end/81097
26467 * fold-const.c (split_tree): Fold to type before negating.
26469 2017-06-20 David Malcolm <dmalcolm@redhat.com>
26471 * diagnostic-show-locus.c
26472 (selftest::test_fixit_deletion_affecting_newline): New function.
26473 (selftest::diagnostic_show_locus_c_tests): Call it.
26475 2017-06-20 Andreas Schwab <schwab@suse.de>
26478 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
26481 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26483 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
26484 __ARM_FEATURE_COPROC according to support.
26486 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26488 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26489 Rewritten to avoid overflow for > 32-bit pointers.
26492 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
26493 by removing enum keyword.
26494 (ubsan_type_descriptor): Likewise. Formatting fix.
26497 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
26498 splitter): Require TARGET_SSE2 in the condition.
26500 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
26503 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
26504 for doing vector set of SFmode on ISA 3.0.
26505 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
26506 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
26508 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
26509 SFmode value into a V4SF variable that was extracted from another
26510 V4SF variable without converting the element to double precision
26511 and back to single precision vector format.
26512 (vsx_insert_extract_v4sf_p9_2): Likewise.
26514 2017-06-19 Jakub Jelinek <jakub@redhat.com>
26516 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
26517 in UWHI to avoid undefined overflow.
26520 * ubsan.h (enum ubsan_encode_value_phase): New.
26521 (ubsan_encode_value): Change second argument to
26522 enum ubsan_encode_value_phase with default value of
26523 UBSAN_ENCODE_VALUE_GENERIC.
26524 * ubsan.c (ubsan_encode_value): Change second argument to
26525 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
26526 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
26527 create_tmp_var_raw instead of create_tmp_var and use a
26529 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
26530 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
26531 ubsan_encode_value callers.
26534 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
26535 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
26536 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
26538 2017-06-19 Richard Biener <rguenther@suse.de>
26540 PR middle-end/81118
26541 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
26542 estimates if we changed anything.
26544 2017-06-19 Richard Biener <rguenther@suse.de>
26546 PR tree-optimization/80887
26547 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
26548 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
26549 simplified lookups, then reset mprts_hook.
26550 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
26552 (try_to_simplify): Likewise.
26554 2017-06-19 Martin Liska <mliska@suse.cz>
26557 * gimplify.c (gimplify_switch_expr):
26558 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
26560 2017-06-19 Martin Liska <mliska@suse.cz>
26562 * doc/install.texi: Document that PGO runs in 4 stages.
26564 2017-06-19 Martin Liska <mliska@suse.cz>
26567 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
26568 to dispatcher function name.
26569 * multiple_target.c (replace_function_decl): New function.
26570 (create_dispatcher_calls): Redirect both edges and references.
26572 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
26574 * profile-count.c (profile_count::dump): Dump quality.
26575 (profile_count::differs_from_p): Update for unsigned val.
26576 * profile-count.h (profile_count_quality): New enum.
26577 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
26579 2017-06-19 Richard Biener <rguenther@suse.de>
26581 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
26582 struct function as arg.
26583 (estimate_numbers_of_iterations): Export overload with loop arg.
26584 (free_numbers_of_iterations_estimates_loop): Use an overload of
26585 free_numbers_of_iterations_estimates instead.
26586 * tree-cfg.c (remove_bb): Adjust.
26587 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
26588 * tree-parloops.c (gen_parallel_loop): Likewise.
26589 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
26591 (tree_unroll_loops_completely): Likewise.
26592 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
26593 Use an overload instead and export.
26594 (estimated_loop_iterations): Adjust.
26595 (max_loop_iterations): Likewise.
26596 (likely_max_loop_iterations): Likewise.
26597 (estimate_numbers_of_iterations): Take struct function as arg
26599 (loop_exits_before_overflow): Adjust.
26600 (free_numbers_of_iterations_estimates_loop): Use an overload.
26601 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
26602 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
26604 2017-06-19 Richard Biener <rguenther@suse.de>
26607 * ipa-prop.c (find_constructor_constant_at_offset): Handle
26608 RANGE_EXPR conservatively.
26610 2017-06-16 Carl Love <cel@us.ibm.com>
26612 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26613 definitions for vec_float, vec_float2, vec_floato,
26614 vec_floate built-ins.
26615 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
26616 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
26618 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
26619 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
26620 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
26621 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
26622 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
26623 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
26624 vec_floato): Add builtin defines.
26625 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
26626 Update the built-in documentation file for the new built-in
26629 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26631 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
26632 (mthumb): Mark as the negative of -marm.
26634 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26636 * doc/invoke.texi (ARM Options, -mcpu): Document supported
26638 (ARM Options, -mtune): Document that this accepts the same
26639 extension options as -mcpu.
26640 (ARM Options, -mfpu): Document addition of -mfpu=auto.
26642 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26644 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
26645 permitted extensions.
26647 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26649 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
26650 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
26651 (armv8-m.main): Add option +nodsp.
26652 * config/arm/arm-cpu-cdata.h: Regenerated.
26654 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26656 * config/arm/t-fuchsia: New file.
26657 * config.gcc (arm*-*-fuchsia*): Use it.
26659 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26661 * config/arm/t-symbian: Rewrite for new option infrastructure.
26663 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26665 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
26666 (MULTILIB_REQUIRED): Likewise.
26668 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26670 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
26671 (MULTILIB_RESUE): Likewise.
26672 (MULTILIB_MATCHES): Likewise.
26673 (MULTLIB_REQUIRED): Likewise.
26675 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26677 * config/arm/t-rtems: Rewrite for new option framework.
26679 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26681 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
26682 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
26683 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
26684 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
26685 * config/arm/t-multilib: ... here.
26686 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
26687 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
26688 armv7-a and armv8*-a when A-profile libraries have not been built.
26689 * config/arm/t-rmprofile: Rewrite.
26691 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26693 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
26694 with a backslash. Remove the backslash after substituting unescaped
26696 * doc/fragments.texi (MULTILIB_REUSE): Document it.
26698 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26700 * config.gcc: (arm*-*-*): When building a-profile libraries, force
26701 the driver to pass through the default setting of -mfloat-abi.
26702 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
26704 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
26705 (all_feat_combs): New rule.
26706 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
26708 * config/arm/t-aprofile: Rewrite.
26710 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26712 * config/arm/arm.h (FPUTYPE_AUTO): Define.
26713 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
26714 fpu is not specified by the user/command-line.
26715 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
26716 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
26717 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
26718 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
26719 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
26720 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
26722 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26724 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
26725 * config/arm/t-arm-elf: Rewritten.
26727 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26729 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
26730 have some floating-point instructions.
26731 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
26732 (TARGET_MAYBE_HARD_FLOAT): New macro.
26733 * config/arm/arm-builtins.c (arm_init_builtins): Use
26734 TARGET_MAYBE_HARD_FLOAT.
26735 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
26737 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26739 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
26740 (configargs.h): Include it.
26741 (arm_print_hint_for_fpu_option): New function.
26742 (arm_parse_fpu_option): New function.
26743 (candidate_extension): New class.
26744 (arm_canon_for_multilib): New function.
26745 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
26746 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
26747 (ARCH_CANONICAL_SPECS): New macro.
26748 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
26750 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26752 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
26753 are set after handling multilib fragments. Set target_cpu_default2
26756 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26758 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
26760 (arm*-*-*): Set target_cpu_default2 to a quoted string.
26761 * config/arm/parsecpu.awk (check_cpu): Validate any extension
26763 (check_arch): Likewise.
26764 * config/arm/arm.c (arm_configure_build_target): Handle
26765 TARGET_CPU_DEFAULT being a string constant. Scan any feature
26766 options in the default.
26768 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26770 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
26771 when an option is an alias of another.
26772 * config/arm/parsecpu.awk (optalias): New parser token.
26773 (gen_comm_data): Mark non-alias options as such. Emit entries
26774 for extension aliases.
26775 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
26776 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
26777 (armv6kz, armv6zk, armv6t2): Likewise.
26778 (armv7): Make vfpv3-d16 an alias.
26779 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
26781 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
26782 Sort in canonical order.
26783 (armv8-a): Sort in canonical order.
26784 (armv8.1-a, armv8.2-a): Likewise.
26785 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
26787 (cortex-a9): Sort in canonical order.
26788 * config/arm/arm.c (selftests.h): Include it.
26789 (arm_test_cpu_arch_data): New function.
26790 (arm_run_self_tests): New function.
26791 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
26792 (targetm): Move declaration to the end of the file.
26793 * arm-cpu-cdata.h: Regenerated.
26795 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26797 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
26798 call to target_mode_check describing the type of option passed.
26799 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
26800 (arm_target_thumb_only): Use arm_parse_arch_option_name or
26801 arm_parse_cpu_option_name to match parameters against list of
26803 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
26804 arm_arch_core_flags data structure.
26805 * config/arm/arm-cpu_cdata.h: Regenerated.
26807 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26809 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
26811 (arm_print_hint_for_cpu_option): Likewise.
26812 (arm_print_hint_for_arch_option): Likewise.
26813 (arm_parse_cpu_option_name): Likewise.
26814 (arm_parse_arch_option_name): Likewise.
26815 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
26816 of entries in the all_fpus list.
26817 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
26818 (arm_parse_cpu_option_name): Declare.
26819 (arm_parse_arch_option_name): Declare.
26820 (arm_parse_option_features): Declare.
26821 (arm_intialize_isa): Declare.
26822 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
26824 (gen_comm_data): ... here. Make definitions non-static.
26825 * config/arm/arm-cpu-data.h: Regenerated.
26826 * config/arm/arm-cpu-cdata.h: Regenerated.
26828 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26830 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
26831 (cpu_arch_extension): New structure.
26832 (cpu_arch_option, arch_option, cpu_option): New structures.
26833 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
26834 architecture types.
26835 (gen_data): Generate new format data tables.
26836 * config/arm/arm.c (cpu_tune): New structure.
26837 (cpu_option, processors): Delete.
26838 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
26839 (arm_print_hint_for_cpu_option): ... this and ...
26840 (arm_print_hint_for_arch_option): ... this.
26841 (arm_parse_arch_cpu_name): Delete. Replace with ...
26842 (arm_parse_cpu_option_name): ... this and ...
26843 (arm_parse_arch_option_name): ... this.
26844 (arm_unrecognized_feature): Change type of target parameter to
26846 (arm_parse_arch_cpu_features): Delete. Replace with ...
26847 (arm_parse_option_features): ... this.
26848 (arm_configure_build_target): Rework to use new configuration data
26850 (arm_print_tune_info): Rework for new configuration data tables.
26851 * config/arm/arm-cpu-data.h: Regenerated.
26852 * config/arm/arm-cpu.h: Regenerated.
26854 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26856 * Makefile.in (OBJS): Move sbitmap.o from here ...
26857 (OBJS-libcommon): ... to here.
26859 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26861 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
26862 (ISA_ALL_CRYPTO): New macro.
26863 (ISA_ALL_SIMD): New macro
26864 (ISA_ALL_FP): New macro.
26865 * config/arm/arm.c (fpu_bitlist): Update initializer.
26866 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
26868 (arm9e): Add fpu. Add option for nofp
26869 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
26870 (arm926ej-s, arm1026ej-s): Likewise.
26871 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
26872 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
26873 neon-fp16, neon-vfpv4, nofp and nosimd.
26874 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
26875 (cortex-a8): Add fpu. Add option for nofp.
26876 (cortex-a9): Add fpu. Add options for nosimd and nofp.
26877 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
26878 (cortex-r4f): Add fpu.
26879 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
26880 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
26882 (cortex-r8): Likewise.
26883 (cortex-m4): Add fpu. Add option for nofp.
26884 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
26885 (cortex-a17.cortex-a7): Likewise.
26886 (cortex-a32): Add fpu. Add options for crypto and nofp.
26887 (cortex-a35, cortex-a53): Likewise.
26888 (cortex-a57): Add fpu. Add option for crypto.
26889 (cortex-a72, cortex-a73): Likewise.
26890 (exynos-m1): Likewise.
26891 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
26892 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
26893 (cortex-m33): Add fpu. Add option for nofp.
26894 * config/arm/arm-cpu-cdata.h: Regenerated
26895 * config/arm/arm-cpu-data.h: Regenerated.
26897 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26899 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
26900 (armv5te, armv5tej): Likewise.
26901 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
26902 (armv7): Add options fp and vfpv3-d16.
26903 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
26904 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
26906 (armv7ve): Likewise.
26907 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
26908 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
26909 (armv8-a): Add nocrypto option.
26910 (armv8.1-a, armv8.2-a): Likewise.
26911 (armv8-m.main): add options fp, fp.dp and nofp.
26913 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26915 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
26917 (armv8-a+crc): Delete.
26918 (armv8.1-a): Add options simd, crypto and nofp.
26919 (armv8.2-a): Add options fp16, simd, crypto and nofp.
26920 (armv8.2-a+fp16): Delete.
26921 (armv8-m.main): Add option dsp.
26922 (armv8-m.main+dsp): Delete.
26923 (cortex-a8): Add fpu. Add nofp option.
26924 (cortex-a9): Add fpu. Add nofp and nosimd options.
26925 * config/arm/parsecpu.awk (gen_data): Generate option tables and
26926 link to main cpu and architecture data structures.
26927 (gen_comm_data): Only put isa attributes from the main architecture
26929 (option): New statement for architecture and CPU entries.
26930 * arm.c (struct cpu_option): New structure.
26931 (struct processors): Add entry for options.
26932 (arm_unrecognized_feature): New function.
26933 (arm_parse_arch_cpu_name): Ignore any characters after the first
26935 (arm_parse_arch_cpu_feature): New function.
26936 (arm_configure_build_target): Separate out any CPU and architecture
26937 features and parse separately. Don't error out if -mfpu=auto is
26938 used with only an architecture string.
26939 (arm_print_asm_arch_directives): New function.
26940 (arm_file_start): Call it.
26941 * config/arm/arm-cpu-cdata.h: Regenerated.
26942 * config/arm/arm-cpu-data.h: Likewise.
26943 * config/arm/arm-tables.opt: Likewise.
26945 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26947 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
26948 assembler when it is not -mfpu=auto.
26950 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26952 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
26953 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
26954 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
26955 (ASM_CPU_SPEC): Rewrite.
26956 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
26957 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
26958 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
26959 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
26960 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
26961 copied string is NUL-terminated. Also strip any characters prefixed
26963 (arm_rewrite_selected_arch): New function.
26964 (arm_rewrite_march): New function.
26966 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26968 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
26969 (x_arm_cpu_string, x_arm_tune_string): Likewise.
26970 (march, mcpu, mtune): Convert to string-based options.
26971 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
26972 (arm_parse_arch_cpu_name): New function.
26973 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
26974 identify selected architecture or CPU.
26975 (arm_option_save): New function.
26976 (TARGET_OPTION_SAVE): Redefine.
26977 (arm_option_restore): Restore string options.
26978 (arm_option_print): Print string options.
26980 2017-06-16 Martin Sebor <msebor@redhat.com>
26982 PR tree-optimization/80933
26983 PR tree-optimization/80934
26984 * builtins.c (fold_builtin_3): Do not handle bcmp here.
26985 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
26986 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
26987 (gimple_fold_builtin): Call them.
26989 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26991 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
26992 as unlikely; update profile.
26994 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26996 * predict.c (force_edge_cold): Handle declaring edges impossible
26999 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27001 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
27003 (try_unroll_loop_completely): Fix reporting.
27005 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27007 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
27009 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
27012 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
27013 if given a non-constant argument for an intrinsic which requires a
27016 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27018 * profile.c (compare_freqs): New function.
27019 (branch_prob): Sort edge list.
27020 (find_spanning_tree): Assume that the list is priority sorted.
27022 2017-06-16 Richard Biener <rguenther@suse.de>
27024 PR tree-optimization/81090
27025 * passes.def (pass_record_bounds): Remove.
27026 * tree-pass.h (make_pass_record_bounds): Likewise.
27027 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
27028 make_pass_record_bounds): Likewise.
27029 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
27030 not free niter estimates at the beginning but at the end.
27031 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
27033 2017-06-16 Richard Biener <rguenther@suse.de>
27035 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
27036 initializer to workaround ICE in host GCC 4.8.
27038 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27040 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
27042 (clone_inlined_nodes): Update.
27044 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27046 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
27047 prefetch settings, and enable prefetching by default at -O3.
27049 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27051 * config/aarch64/aarch64.c (aarch64_override_options_internal):
27052 Set flag_prefetch_loop_arrays according to tuning data.
27054 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27056 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
27057 New tune structure.
27058 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
27059 [Unrelated to main purpose of the patch] Place the pointer field last
27060 to enable type checking errors when tune structure are wrongly merged.
27061 * config/aarch64/aarch64.c (generic_prefetch_tune,)
27062 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
27063 (thunderx2t99_prefetch_tune): New tune constants.
27064 (tune_params *_tunings): Update all tunings (no functional change).
27065 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
27066 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
27067 from tunings structures.
27069 2017-06-16 Jakub Jelinek <jakub@redhat.com>
27072 * ubsan.c (instrument_null): Add T argument, use it instead
27073 of computing it based on IS_LHS.
27074 (instrument_object_size): Likewise.
27075 (pass_ubsan::execute): Adjust instrument_null and
27076 instrument_object_size callers to pass gimple_get_lhs or
27077 gimple_assign_rhs1 result to it. Use instrument_null instead of
27078 calling get_base_address and instrument_mem_ref. Handle
27079 aggregate call arguments for object-size sanitization.
27081 2017-06-16 Yury Gribov <tetra2005@gmail.com>
27083 PR tree-optimization/81089
27084 * tree-vrp.c (is_masked_range_test): Validate operands of
27087 2017-06-15 Martin Sebor <msebor@redhat.com>
27090 * dumpfile.c (dump_register): Avoid calling memset to initialize
27091 a class with a default ctor.
27092 * gcc.c (struct compiler): Remove const qualification.
27093 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
27094 * hash-table.h: Ditto.
27095 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
27097 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
27098 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
27100 * params.h (struct param_info): Make struct members non-const.
27101 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
27102 with default initialization.
27103 * vec.h (vec_copy_construct, vec_default_construct): New helper
27105 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
27106 with vec_copy_construct.
27107 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
27108 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
27109 * doc/invoke.texi (-Wclass-memaccess): Document.
27111 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27113 * emit-rtl.h (is_leaf): Update comment about local
27114 register allocator.
27116 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27119 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
27120 for a variable to have a section before checking if the section has a
27122 Set section to.persistent if persistent attribute is set.
27123 Warn if .persistent attribute is used on an automatic variable.
27125 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
27127 PR rtl-optimization/80474
27128 * reorg.c (update_block): Do not ignore instructions in a delay slot.
27130 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
27132 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
27135 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
27137 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
27138 (casesi): Emit bounds checking as RTL.
27139 (casesi_internal_mips16_<mode>): Remove bounds checking.
27141 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
27143 * config/xtensa/xtensa.c (xtensa_option_override): Append
27144 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
27145 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
27146 xtensa_doloop_hooks): Define unconditionally.
27147 (xtensa_reorg_loops): Only call reorg_loops in the presence of
27149 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
27150 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
27151 for it in xtensa_option_override.
27152 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
27153 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
27155 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
27157 * doc/cppopts.texi: Document '-' special value to -MF.
27159 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
27161 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
27162 (cortex_a53_fconst): Likewise.
27163 (cortex_a53_fpmul): Likewise.
27164 (cortex_a53_f_load_64): Likewise.
27165 (cortex_a53_f_load_many): Likewise.
27166 (cortex_a53_advsimd_alu): Likewise.
27167 (cortex_a53_advsimd_alu_q): Likewise.
27168 (cortex_a53_advsimd_mul): Likewise.
27169 (cortex_a53_advsimd_mul_q): Likewise.
27170 (fpmac bypass): Add new bypass for fpmac-fpmac case.
27171 Add missing fmul, r2f_cvt and fconst cases.
27173 2017-06-14 Richard Biener <rguenther@suse.de>
27175 PR middle-end/81088
27176 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
27178 (fold_binary_loc): When associating do not treat pre-existing
27179 TREE_OVERFLOW on literal constants as a reason to allow
27180 TREE_OVERFLOW on associated literal constants.
27182 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27184 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
27185 (MASK_FEATURES): New macro.
27186 * config/sparc/sparc.c (sparc_option_override): Remove the special
27187 handling of -mfpu and generalize it to all MASK_FEATURES switches.
27189 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27191 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
27192 a division of 0 if non-call exceptions are enabled.
27194 2017-06-14 Andrew Pinski <apinski@cavium.com>
27195 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27198 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
27199 Improve vector initialization code gen for only variable case.
27201 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27203 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
27205 2017-06-14 Richard Biener <rguenther@suse.de>
27207 PR tree-optimization/81083
27208 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
27211 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27213 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
27214 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
27215 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
27216 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
27217 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
27218 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
27220 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27222 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
27223 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
27225 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27227 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
27229 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27231 * config/rs6000/t-rtems: Don't handle SPE.
27233 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27235 * config/rs6000/t-linux: Don't handle SPE.
27237 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27239 * config/rs6000/eabispe.h: Delete file.
27241 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27243 * config/rs6000/t-spe: Delete file.
27245 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27247 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
27248 (rs6000_legitimate_offset_address_p): Return false for anything in
27249 V2SImode or V2SFmode.
27251 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27253 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
27254 except V2SF and V2SI. Rearrange the vector modes, and add comments.
27255 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
27257 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
27258 (rs6000_legitimate_offset_address_p): Ditto.
27259 (rs6000_emit_move): Ditto.
27260 (rs6000_init_builtins): Remove V4HI_type_node.
27262 2017-06-13 Martin Liska <mliska@suse.cz>
27265 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
27266 (gate_asan): Likewise.
27267 * asan.h (asan_no_sanitize_address_p): Remove the function.
27268 (sanitize_flags_p): New function.
27269 * builtins.def: Fix coding style.
27270 * common.opt: Use renamed enum value.
27271 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
27272 * doc/extend.texi: Document no_sanitize attribute.
27273 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
27274 to SANITIZE_UNDEFINED_NONDEFAULT.
27275 * gcc.c (sanitize_spec_function): Use the renamed enum value.
27276 * gimple-fold.c (optimize_atomic_compare_exchange_p):
27277 Use sanitize_flags_p.
27278 * gimplify.c (gimplify_function_tree): Likewise.
27279 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
27280 * opts.c (parse_no_sanitize_attribute): New function.
27281 (common_handle_option): Use renamed enum value.
27282 * opts.h (parse_no_sanitize_attribute): Declare.
27283 * tree.c (sanitize_flags_p): New function.
27284 * tree.h: Declared here.
27285 * tsan.c: Use sanitize_flags_p.
27286 * ubsan.c (ubsan_expand_null_ifn): Likewise.
27287 (instrument_mem_ref): Likewise.
27288 (instrument_bool_enum_load): Likewise.
27289 (do_ubsan_in_current_function): Remove the function.
27290 (pass_ubsan::execute): Use sanitize_flags_p.
27291 * ubsan.h: Remove do_ubsan_in_current_function
27292 * tree-cfg.c (print_no_sanitize_attr_value): New function.
27293 (dump_function_to_file): Use it here.
27295 2017-06-13 Martin Jambor <mjambor@suse.cz>
27297 PR tree-optimization/80803
27298 PR tree-optimization/81063
27299 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
27300 (propagate_subaccesses_across_link): Enqueue subtree whenever
27301 necessary instead of relying on the caller.
27303 2017-06-13 Martin Jambor <mjambor@suse.cz>
27305 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
27306 that have a first_link.
27307 (sort_and_splice_var_accesses): Do not check first_link before
27309 (subtree_mark_written_and_enqueue): Likewise.
27310 (propagate_all_subaccesses): Likewise and do not stop at first
27311 parent with a first_link.
27313 2017-06-13 Martin Jambor <mjambor@suse.cz>
27315 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
27318 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27320 * match.pd: New pattern.
27322 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27324 * tree-vrp.c (is_masked_range_test): New function.
27325 (register_edge_assert_for): Determine ranges for
27328 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27330 PR tree-optimization/67328
27331 * fold-const.c (maskable_range_p): New function.
27332 (build_range_check): Generate bittests if possible.
27334 2017-06-13 Martin Liska <mliska@suse.cz>
27336 * gimple-pretty-print.c (dump_probability): Add new argument.
27337 (dump_edge_probability): Dump both probability and count.
27338 (dump_gimple_label): Likewise.
27339 (dump_gimple_bb_header): Likewise.
27341 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
27344 * config/avr/avr-devices.c: Fix indentation.
27345 * config/avr/gen-avr-mmcu-specs.c: Dito.
27347 2017-06-13 Richard Biener <rguenther@suse.de>
27349 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
27350 instead get vector type from stmt_info.
27351 (vectorizable_reduction): Adjust. Remove dead code.
27353 2017-06-13 Richard Biener <rguenther@suse.de>
27355 PR middle-end/81065
27356 * fold-const.c (extract_muldiv_1): Remove bogus distribution
27357 case of C * (x * C2 + C3).
27358 (fold_addr_of_array_ref_difference): Properly fold index difference.
27360 2017-06-12 David S. Miller <davem@davemloft.net>
27363 * config/sparc/sparc.md (return expander): Emit frame blockage if
27364 function uses alloca.
27366 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
27368 * combine.c (make_field_assignment): Check len rather than the mode
27369 precision when calling force_to_mode.
27371 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
27373 Support multilibs and devices that see flash in RAM address range.
27376 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
27377 (avr_mcu_t) <flash_pm_offset>: New field.
27378 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
27379 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
27380 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
27381 (AVR_TINY_PM_OFFSET): Remove macro.
27382 * config/avr/avr.opt (-mshort-calls): New option.
27383 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
27384 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
27385 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
27386 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
27387 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
27388 instead of avr_arch->have_jmp_call.
27389 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
27390 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
27391 avr_arch->flash_pm_offset to define.
27392 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
27393 new field flash_pm_offset. Add entry for avrxmega3.
27394 (avr_texinfo): Add entry for avrxmega3.
27395 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
27396 attiny212, attiny214,
27397 attiny412, attiny414, attiny416, attiny417,
27398 attiny814, attiny816, attiny817,
27399 attiny1614, attiny1616, attiny1617,
27400 attiny3214, attiny3216, attiny3217.
27401 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
27402 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
27403 (avr_print_operand_address) [AVR_TINY]: Same.
27404 (avr_asm_init_sections) <readonly_data_section>: Only patch
27405 callback if avr_arch->flash_pm_offset = 0.
27406 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
27407 for rodata if avr_arch->flash_pm_offset != 0.
27408 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
27409 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
27410 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
27411 (m_options): Append opt_rcall.
27412 (m_dirnames): Append dir_rcall.
27413 * config/avr/t-multilib: Regenerate.
27415 * configure.ac [target=avr]: Check whether avrxmega3 default
27416 linker description file works as needed.
27417 * configure: Regenerate.
27418 * doc/avr-mmcu.texi: Regenerate.
27419 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
27420 <__AVR_ARCH__>: Document avrxmega3 and 103.
27421 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
27422 <__AVR_SHORT_CALLS__>: Document it.
27423 <__AVR_PM_BASE_ADDRESS__>: Document it.
27424 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
27425 (AVR Variable Attributes) <progmem>: Document this is
27426 not needed for avrxmega3.
27427 (AVR Named Address Spaces) <__flash>: Dito.
27429 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
27431 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
27433 2017-06-12 Doug Rupp <rupp@adacore.com>
27435 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
27436 Append vxworks-stdint.h to the tm_file list.
27437 * config/vxworks-stdint.h: New file.
27439 2017-06-12 Martin Liska <mliska@suse.cz>
27441 PR tree-optimization/81041
27442 * tree-profile.c (gimple_gen_ic_func_profiler):
27443 Create an extra BB in profile-generate
27444 (gimple_gen_time_profiler): Likewise.
27446 2017-06-12 Jakub Jelinek <jakub@redhat.com>
27448 PR tree-optimization/81003
27449 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
27450 (update_range_test): Use it instead of force_gimple_operand_gsi.
27452 2017-06-12 Richard Biener <rguenther@suse.de>
27454 PR tree-optimization/81053
27455 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
27456 with backedge value not defined in loop. Simplify def stmt
27459 2017-06-11 Tom de Vries <tom@codesourcery.com>
27462 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
27464 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
27465 nvptx_cannot_force_const_mem.
27467 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27469 * opts.c (finish_options): Move test for flag_split_stack after
27470 it has been initialized.
27472 2017-06-11 Jason Merrill <jason@redhat.com>
27474 * tree.h (id_equal): New.
27475 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
27476 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
27477 instead of strcmp of IDENTIFIER_POINTER.
27479 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27481 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
27482 (mark_all_inlined_calls_cdtor): Fix formating.
27483 (inline_transform): Rescale profile before inlining.
27485 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27487 * cgraph.h (cgraph_edge::clone): Update prototype.
27488 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
27489 (cgraph_node::create_clone): Update.
27490 (cgraph_node::create_version_clone): Update.
27491 * tree-inline.c (copy_bb): Update.
27492 (expand_call_inline): Update.
27494 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27496 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
27497 factored out from ...
27498 (rs6000_emit_prologue): ... here.
27500 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27502 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
27503 factored out from ...
27504 (rs6000_emit_prologue): ... here.
27506 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27508 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
27510 (handle_missing_profiles): Fix computation of tp_first_run.
27511 (counts_to_freqs): Do not touch freqs when count is 0.
27513 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27515 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
27518 2017-06-10 Tom de Vries <tom@codesourcery.com>
27520 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27521 attributes): Document signal effective target.
27523 2017-06-10 Tom de Vries <tom@codesourcery.com>
27525 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27526 Document effective target stack_size.
27528 2017-06-09 David Malcolm <dmalcolm@redhat.com>
27530 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
27531 to the edit_context if they can be auto-applied.
27533 2017-06-9 Ian Lance Taylor <iant@golang.org>
27535 * opts.c (finish_options): If -fsplit-stack, disable implicit
27536 -forder-blocks-and-partition.
27537 * doc/invoke.texi (Optimize Options): Document that when using
27538 -fsplit-stack -forder-blocks-and-partition is not implicitly
27541 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27543 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
27544 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
27545 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
27546 * builtins.def (abort, trap, unreachable): Declare cold.
27547 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
27548 * tree-core.h (ECF_COLD): New.
27549 * tree.c (set_call_expr_flags): Handle ECF_COLD.
27550 (build_common_builtin_nodes): Mark unreachable and abort as cold.
27552 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27554 * predict.c (unlikely_executed_stmt_p): Cleanup.
27556 2017-06-09 Richard Biener <rguenther@suse.de>
27558 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
27559 model if the ref is always written to.
27561 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27563 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
27565 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27567 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
27570 2017-06-09 Tom de Vries <tom@codesourcery.com>
27573 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
27574 "target cannot support label values" when encountering LABEL_REF.
27576 2017-06-09 Martin Liska <mliska@suse.cz>
27578 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
27579 (gimple_gen_ic_func_profiler): Emit direct comparison
27580 of __gcov_indirect_call_callee with NULL.
27581 (gimple_gen_time_profiler): Change probability from
27582 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
27584 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27586 * profile.c (edge_gcov_counts): Turn to pointer.
27587 (compute_branch_probabilities, compute_branch_probabilities): Update.
27588 (branch_prob): Do not clear edge_gcov_count.
27589 * profile.h (edge_gcov_counts): Turn to pointer.
27590 (edge_gcov_count): Update.
27592 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27594 * gimple.h (gimple_check_failed): Mark cold.
27596 2017-06-09 Richard Biener <rguenther@suse.de>
27598 PR tree-optimization/66623
27599 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
27600 refactor check_reduction into two parts, properly computing
27601 whether we have to check reduction validity for outer loop
27604 2017-06-09 Richard Biener <rguenther@suse.de>
27606 PR tree-optimization/79483
27607 * graphite-scop-detection.c (order): New global.
27608 (get_order): Compute bb to order mapping that satisfies code
27609 generation constraints.
27610 (cmp_pbbs): New helper.
27611 (build_scops): Start domwalk at entry block, sort generated
27614 2017-06-09 Richard Biener <rguenther@suse.de>
27616 PR middle-end/81007
27617 * ipa-polymorphic-call.c
27618 (ipa_polymorphic_call_context::restrict_to_inner_class):
27619 Skip FIELD_DECLs with error_mark_node type.
27620 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
27623 2017-06-09 Martin Liska <mliska@suse.cz>
27625 * predict.c (struct branch_predictor): New struct.
27626 (test_prediction_value_range): New test.
27627 (predict_c_tests): New function.
27628 * selftest-run-tests.c (selftest::run_tests): Run the function.
27629 * selftest.h: Declare new tests.
27631 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
27634 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
27635 gen_add3_insn did not fail.
27636 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
27637 r0, construct that number in a temporary reg and add that reg to r0.
27638 If asked to put the result in r0 as well, fail.
27640 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
27642 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27643 for early expansion of vec_eqv.
27645 2017-06-08 Jakub Jelinek <jakub@redhat.com>
27647 PR middle-end/81005
27648 * ubsan.c (instrument_null): Avoid pointless code temporary.
27649 (pass_ubsan::execute): Instrument aggregate arguments of calls.
27651 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
27655 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
27658 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
27661 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27663 * predict.c (unlikely_executed_edge_p): Move ahead.
27664 (probably_never_executed_edge_p): Use it.
27666 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27668 PR middle-end/79988
27669 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
27670 gimple_call_builtin_p call.
27672 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27674 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
27675 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
27676 rtl_check_failed_type2, rtl_check_failed_code1,
27677 rtl_check_failed_code2, rtl_check_failed_code_mode,
27678 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
27679 rtvec_check_failed_bounds, rtl_check_failed_flag,
27680 _fatal_insn_not_found, _fatal_insn): Likewise.
27681 * tree.h (tree_contains_struct_check_failed,
27682 tree_check_failed, tree_not_check_failed,
27683 tree_class_check_failed, tree_range_check_failed,
27684 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
27685 tree_vec_elt_check_failed, phi_node_elt_check_failed,
27686 tree_operand_check_failed, omp_clause_check_failed,
27687 omp_clause_operand_check_failed, omp_clause_range_check_failed):
27690 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27692 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
27693 flag_branch_probabilities.
27694 * ipa-inline.c (edge_badness): Likewise.
27695 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
27696 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
27697 * predict.c (maybe_hot_frequency_p): Likewise.
27698 (probably_never_executed): Likewise.
27699 * sched-ebb.c (schedule_ebbs): Likewise.
27700 * sched-rgn.c (find_single_block_region): Likewise.
27701 * tracer.c (tail_duplicate): Likewise.
27703 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27705 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
27706 longer requires x_flag_profile_use.
27708 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27710 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
27711 instead of flag_reorder_blocks_and_partition.
27712 * dbxout.c (dbxout_function_end): Likewise.
27713 * dwarf2out.c (gen_subprogram_die): Likewise.
27714 * haifa-sched.c (sched_create_recovery_edges): Likewise.
27715 * hw-doloop.c (reorg_loops): Likewise.
27716 * varasm.c (assemble_start_function,
27717 assemble_end_function): Likewise.
27718 (decide_function_section): Do not check for
27719 flag_reorder_blocks_and_partition.
27721 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27723 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27725 (chkp_get_hard_register_fake_addr_expr): Ditto.
27726 (chkp_build_addr_expr): Add check for hard reg case.
27727 (chkp_parse_array_and_component_ref): Ditto.
27728 (chkp_find_bounds_1): Ditto.
27729 (chkp_process_stmt): Don't generate bounds store for
27732 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27734 * predict.c (maybe_hot_bb_p): Do not check profile status.
27735 (maybe_hot_edge_p): Likewise.
27736 (probably_never_executed): Check for zero counts even if profile
27738 (unlikely_executed_edge_p): New function.
27739 (unlikely_executed_stmt_p): New function.
27740 (unlikely_executed_bb_p): New function.
27741 (set_even_probabilities): Use unlikely predicates.
27742 (combine_predictions_for_bb): Likewise.
27743 (predict_paths_for_bb): Likewise.
27744 (predict_paths_leading_to_edge): Likewise.
27745 (determine_unlikely_bbs): New function.
27746 (estimate_bb_frequencies): Use it.
27747 (compute_function_frequency): Use zero counts even if profile is
27749 * profile-count.h: Fix typo.
27751 2017-08-08 Julia Koval <julia.koval@intel.com>
27753 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
27754 _mm512_mask_cvtsepi16_storeu_epi8,
27755 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
27756 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
27757 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
27758 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
27759 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
27760 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
27761 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
27762 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
27763 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
27764 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
27765 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
27766 __builtin_ia32_pmovuswb256mem_mask,
27767 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
27768 __builtin_ia32_pmovwb512mem_mask): New builtins.
27770 2017-08-08 Julia Koval <julia.koval@intel.com>
27772 PR target/73350,80862
27773 * config/i386/subst.md (round): Fix round pattern.
27774 * config/i386/i386.c (ix86_erase_embedded_rounding):
27775 Fix erasing rounding for the fixed pattern.
27777 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27779 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
27781 2017-06-08 Martin Liska <mliska@suse.cz>
27783 PR gcov-profile/80911
27784 * gcov.c (block_info::block_info): New constructor.
27786 2017-06-07 Carl Love <cel@us.ibm.com>
27788 * config/rs6000/rs6000-c: The return type of the following
27789 built-in functions was implemented as int not long long. Fix sign
27790 of return value for the unsigned version of vec_mulo and vec_mule.
27791 vector unsigned long long vec_bperm (vector unsigned long long,
27792 vector unsigned char)
27793 vector signed long long vec_mule (vector signed int,
27795 vector unsigned long long vec_mule (vector unsigned int,
27796 vector unsigned int)
27797 vector signed long long vec_mulo (vector signed int,
27799 vector unsigned long long vec_mulo (vector unsigned int,
27800 vector unsigned int)
27801 * doc/extend.texi: Fix the documentation for the built-in
27804 2017-06-07 Carl Love <cel@us.ibm.com>
27807 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
27810 2017-06-07 Carl Love <cel@us.ibm.com>
27812 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
27813 support, Generate doublehv for signed int/float for BE case only.
27815 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
27817 * doc/invoke.texi (mcx16): Rewrite.
27819 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27821 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
27822 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
27823 *mov<mode>_softfloat, and an anonymous splitter): Use
27824 nonimmediate_operand instead of rs6000_nonimmediate_operand.
27826 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27828 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
27830 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27831 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
27832 (rs6000_debug_reg_global): Adjust.
27833 (rs6000_init_hard_regno_mode_ok): Adjust.
27834 (rs6000_dbx_register_number): Adjust.
27835 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
27836 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27837 Remove SPE_ACC and SPEFSCR.
27838 (REG_ALLOC_ORDER): Ditto.
27839 (FRAME_POINTER_REGNUM): Change to 111.
27840 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
27841 (REG_CLASS_NAMES): Ditto.
27842 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
27843 (REGISTER_NAMES): Ditto.
27844 (ADDITIONAL_REG_NAMES): Ditto.
27845 (rs6000_reg_names): Ditto.
27846 * config/rs6000/rs6000.md: Renumber some register number
27849 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27851 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
27853 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27854 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
27856 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
27857 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27858 Delete the SPE high registers.
27859 (REG_ALLOC_ORDER): Ditto.
27860 (enum reg_class): Remove SPE_HIGH_REGS.
27861 (REG_CLASS_NAMES): Ditto.
27862 (REG_CLASS_CONTENTS): Delete the SPE high registers.
27863 (REGISTER_NAMES): Ditto.
27864 (rs6000_reg_names): Ditto.
27865 * doc/tm.texi.in: Remove SPE as example.
27866 * doc/tm.texi: Regenerate.
27868 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27870 * config/rs6000/8540.md (ppc8540_brinc): Delete.
27871 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
27872 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
27873 * config/rs6000/rs6000.md (type): Remove "brinc".
27875 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27877 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
27878 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
27879 * config/rs6000/linuxspe.h: Delete file.
27880 * config/rs6000/rs6000.md: Don't include spe.md.
27881 * config/rs6000/spe.h: Delete file.
27882 * config/rs6000/spe.md: Delete file.
27883 * config/rs6000/t-rs6000: Remove spe.md.
27885 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27887 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
27888 (reg_or_none500mem_operand): Delete.
27889 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
27890 instead of reg_or_none500mem_operand.
27892 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27894 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27895 handling of SPE flags.
27896 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
27898 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27900 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
27902 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
27903 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
27904 paired_divv2sf3): Similar.
27905 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
27906 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
27907 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
27909 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
27910 Rename the paired_* instruction patterns.
27911 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
27913 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
27914 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
27915 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
27916 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
27917 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
27918 PAIRED_VECTOR_MODE.
27919 (struct machine_function): Delete field spe_insn_chain_scanned_p.
27920 (spe_func_has_64bit_regs_p): Delete.
27921 (spe_expand_predicate_builtin): Delete.
27922 (spe_expand_evsel_builtin): Delete.
27923 (TARGET_DWARF_REGISTER_SPAN): Do not define.
27924 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
27925 (invalid_e500_subreg): Delete.
27926 (rs6000_legitimize_address): Always force_reg op2 as well, for
27927 paired single memory accesses.
27928 (rs6000_member_type_forces_blk): Delete.
27929 (rs6000_spe_function_arg): Delete.
27930 (rs6000_expand_unop_builtin): Delete SPE handling.
27931 (rs6000_expand_binop_builtin): Ditto.
27932 (spe_expand_stv_builtin): Delete.
27933 (bdesc_2arg_spe): Delete.
27934 (spe_expand_builtin): Delete.
27935 (spe_expand_predicate_builtin): Delete.
27936 (spe_expand_evsel_builtin): Delete.
27937 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
27938 (spe_init_builtins): Delete.
27939 (spe_func_has_64bit_regs_p): Delete.
27940 (savres_routine_name): Delete "info" parameter. Adjust callers.
27941 (rs6000_emit_stack_reset): Ditto.
27942 (rs6000_dwarf_register_span): Delete.
27943 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
27944 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
27945 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
27947 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
27949 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
27950 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
27951 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
27952 mulv2sf3, divv2sf3): Delete expanders.
27954 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27956 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
27958 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27960 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
27961 * config/rs6000/rs6000.c: Ditto.
27963 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27965 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
27966 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
27967 comparison_operator.
27969 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27971 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
27972 * config/rs6000/rs6000.opt: Ditto.
27973 * config/rs6000/t-rtems: Ditto.
27975 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27977 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
27978 TARGET_E500_SINGLE by 0, simplify.
27979 * config/rs6000/rs6000.c: Ditto.
27980 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
27981 (spe_build_register_parallel): Delete.
27982 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
27983 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
27984 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
27985 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
27986 (E500_CONVERT): Delete.
27987 * config/rs6000/spe.md: Remove many patterns and all define_constants.
27989 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27991 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
27992 * config/rs6000/dfp.md: Ditto.
27993 (negdd2, *negdd2_fpr): Merge.
27994 (absdd2, *absdd2_fpr): Merge.
27995 (negtd2, *negtd2_fpr): Merge.
27996 (abstd2, *abstd2_fpr): Merge.
27997 * config/rs6000/e500.h: Delete file.
27998 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
27999 TARGET_FPRS by 1 and simplify.
28000 * config/rs6000/rs6000-c.c: Ditto.
28001 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
28002 TARGET_DF_SPE by 0.
28003 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
28005 * config/rs6000/rs6000.md: Ditto.
28006 (floatdidf2, *floatdidf2_fpr): Merge.
28007 (move_from_CR_gt_bit): Delete.
28008 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
28009 (E500_CR_IOR_COMPARE): Delete.
28010 (All patterns that require !TARGET_FPRS): Delete.
28011 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
28013 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28015 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
28017 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28019 * graphds.c (add_edge): Intitialize edge's attached data.
28020 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
28021 pointer parameter. Call pointed function on each edge during
28022 graph traversing. Skip traversing the edge when the function
28024 (graphds_dfs, graphds_scc): Ditto.
28025 (for_each_edge): New parameter. Pass the new parameter to callback
28027 * graphds.h (skip_edge_callback): New function pointer type.
28028 (graphds_dfs, graphds_scc): New function pointer parameter.
28029 (graphds_edge_callback, for_each_edge): New parameter.
28031 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28033 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
28034 out code checking if runtime alias check is possible to below ...
28035 Call the new function.
28036 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
28037 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
28039 2017-06-07 Marek Polacek <polacek@redhat.com>
28042 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
28043 TYPE_OVERFLOW_WRAPS check.
28045 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28047 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
28048 if versioning is required.
28049 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
28050 peeling with the check for versioning.
28052 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28054 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
28055 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
28056 Set true to new parameter if new ssa variable is defined.
28057 (vect_gen_vector_loop_niters): Refactor. Set range information
28058 for the new vector loop bound variable.
28059 (vect_do_peeling): Ditto.
28061 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28063 * tree-affine.c (ssa.h): Include header file.
28064 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
28065 has wrapping overflow behavior.
28067 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28069 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
28071 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28073 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
28074 (tree_to_aff_combination): ... here.
28076 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28078 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
28079 reg_pressure model function.
28080 (ivopts_global_cost_for_size): Delete.
28081 (determine_set_costs, iv_ca_recount_cost): Call new model function
28082 ivopts_estimate_reg_pressure.
28084 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28086 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
28087 expensive than udiv. Remove floating point cases from mod.
28089 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28091 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
28092 Increase idiv cost.
28094 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28096 * config/aarch64/aarch64.md
28097 (copysignsf3): Fix mask generation.
28099 2017-06-07 Jakub Jelinek <jakub@redhat.com>
28101 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
28103 (class dump_manager): Add register_dumps method.
28104 * dumpfile.c: Include langhooks.h.
28105 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
28106 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
28107 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
28108 (dump_manager::dump_register): Start with 512 entries instead of 32.
28109 (dump_manager::register_dumps): New method.
28110 * toplev.c (general_init): Instead of invoking register_dumps
28111 langhook, invoke register_dumps method on the dump manager.
28112 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
28115 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
28117 * doc/md.texi: Clarify the restrictions on a define_insn condition.
28118 Say that # requires an associated define_split to exist, and that
28119 the define_split must be suitable for use after register allocation.
28121 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28123 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
28124 (compute_outgoing_frequencies): Also initialize zero counts.
28125 (find_many_sub_basic_blocks): Do not produce uninitialized profile
28126 around loops; preserve more of profile when nothing changes.
28128 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
28130 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
28132 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
28133 * config/arm/arm-cpu-cdata.h: Regenerate.
28134 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
28135 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
28136 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
28137 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
28138 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
28140 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
28142 * config/arm/t-rmprofile: Likewise.
28143 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
28145 2017-06-06 David S. Miller <davem@davemloft.net>
28148 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
28149 blockage if function uses alloca.
28151 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28153 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
28154 New "uid" fields to hold pretty-print IDs of group and ref.
28155 Memory references are now identified as <group_id>:<ref_id>
28156 instead of using [random] addresses.
28157 (dump_mem_details): Simplify, no functional change.
28158 (dump_mem_ref): Simplify and make output more concise.
28159 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
28160 (find_or_create_group): Initialize group uid.
28161 (record_ref): Initialize ref uid. Improve debug output.
28162 (prune_group_by_reuse, should_issue_prefetch_p,)
28163 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
28164 (mark_nontemporal_store, determine_loop_nest_reuse):
28165 Improve debug output.
28167 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28169 * dbgcnt.def (prefetch): New debug counter.
28170 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
28171 (schedule_prefetches): Stop issueing prefetches if debug counter
28174 2017-06-06 Tom de Vries <tom@codesourcery.com>
28176 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
28177 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
28179 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28181 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
28182 Use aarch64_reg_or_zero predicate for operand 4.
28183 (aarch64_compare_and_swap<mode> define_insn_and_split):
28184 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
28185 (aarch64_store_exclusive<mode>): Likewise for operand 2.
28187 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
28189 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
28190 (arm_compute_save_core_reg_mask): This.
28191 (thumb1_compute_save_reg_mask): Rename into ...
28192 (thumb1_compute_save_core_reg_mask): This.
28193 (arm_compute_save_reg0_reg12_mask): Adapt comment.
28194 (arm_compute_frame_layout): Likewise.
28196 2017-06-06 Richard Biener <rguenther@suse.de>
28198 PR tree-optimization/80974
28199 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
28200 keep or clear leaders SSA info.
28202 2017-06-06 Tom de Vries <tom@codesourcery.com>
28204 * config/nvptx/nvptx.c (split_mode_p): New function.
28205 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
28207 2017-06-06 Tom de Vries <tom@codesourcery.com>
28209 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
28211 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28214 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
28217 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28219 * shrink-wrap.c (handle_simple_exit): Update profile.
28220 (try_shrink_wrapping): Upate profile.
28222 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28224 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
28225 (tree_guess_outgoing_edge_probabilities): New.
28226 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
28227 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
28229 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28231 * ipa-split.c (split_function): Initialize return bb profile.
28233 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28235 * profile.c (compute_branch_probabilities): Also initialize
28236 EXIT_BLOCK profile.
28238 2017-06-06 Richard Biener <rguenther@suse.de>
28240 PR tree-optimization/80928
28241 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
28242 (vect_analyze_loop_operations): Properly guard analysis for
28244 (vect_transform_loop): Likewise.
28245 (vect_analyze_loop_2): Also reset SLP type on PHIs.
28246 (vect_model_induction_cost): Do not cost for pure SLP.
28247 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
28248 of induction in inner loop vectorization.
28249 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
28250 (vect_get_and_check_slp_defs): Handle vect_induction_def.
28251 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
28253 (vect_analyze_slp_cost_1): Cost induction.
28254 (vect_detect_hybrid_slp_stmts): Handle PHIs.
28255 (vect_get_slp_vect_defs): Likewise.
28256 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
28257 (vect_transform_stmt): Handle SLP reductions.
28258 * tree-vectorizer.h (vectorizable_induction): Adjust.
28260 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28262 * config/rs6000/rs6000.c (make_resolver_func): Update
28263 init_lowered_empty_function call.
28265 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
28267 * doc/invoke.texi: Document the -fprofile-abs-path option.
28268 * common.opt (fprofile-abs-path): New option.
28269 * gcov-io.h (gcov_write_filename): Declare.
28270 * gcov-io.c (gcov_write_filename): New function.
28271 * coverage.c (coverage_begin_function): Use gcov_write_filename.
28272 * profile.c (output_location): Likewise.
28274 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28276 * shring-wrap.c: Revert accidental commit.
28278 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
28280 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
28282 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28284 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
28286 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
28287 profile in callgraph edge.
28288 * profile-count.h (apply_probability): If THIS is 0, then result is 0
28289 (apply_scale): Likewise.
28290 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
28291 Also scale profile when inlining function with zero profile.
28292 (initialize_cfun): Update exit block profile even when it is zero.
28293 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
28294 when profile is read.
28296 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28298 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
28299 (CLONE_*): New constants to define the processors we can generate
28300 code for with the target_clone attribute.
28301 (rs6000_clone_map): New array to identify which clone processors
28302 the current program is running on.
28303 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
28304 target_clone attribute.
28305 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
28306 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
28307 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
28308 (cpu_expand_builtin): Add support for target_clone attribute.
28309 (rs6000_valid_attribute_p): Allow "default" attribute.
28310 (get_decl_name): New debug function to simplify printing the
28311 current function name in debugging statements.
28312 (rs6000_clone_priority): New functions to support the target_clone
28313 attribute, and be able to generate code to switch between ISA 2.05
28314 through ISA 3.0 (power6 through power9).
28315 (rs6000_compare_version_priority): Likewise.
28316 (rs6000_get_function_versions_dispatcher): Likewise.
28317 (make_resolver_func): Likewise.
28318 (add_condition_to_bb): Likewise.
28319 (dispatch_function_versions): Likewise.
28320 (rs6000_generate_version_dispatcher_body): Likewise.
28321 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
28322 (fusion_gpr_load_p): Fix a spacing issue.
28323 * doc/extend.texi (Common Function Attributes): Document that the
28324 PowerPC supports the target_clone attribute.
28326 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28328 * config/arm/arm.h: explain F symbol found in description of ARM
28329 register allocation in its legend.
28331 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28333 * config/mips/frame-header-opt.c: Include profile-count.h.
28334 * config/riscv/riscv.c: Include profile-count.h
28336 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28338 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
28340 (sm_set_flag_if_changed): Add bbs field.
28341 (execute_sm_if_changed_flag_set): Pass BBS.
28342 (execute_sm): Update.
28344 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28346 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
28349 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28351 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
28352 (peephole2): New peephole2 to emit the above.
28353 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
28355 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28357 * config/aarch64/aarch64.c (define_peephole2 above
28358 *sub_<shift>_<mode>): New peephole.
28360 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28362 * config/i386/i386.c (make_resolver_func): Update.
28363 * Makefile.in: Add profile-count.h and profile-count.o
28364 * auto-profile.c (afdo_indirect_call): Update to new API.
28365 (afdo_set_bb_count): Update.
28366 (afdo_propagate_edge): Update.
28367 (afdo_propagate_circuit): Update.
28368 (afdo_calculate_branch_prob): Update.
28369 (afdo_annotate_cfg): Update.
28370 * basic-block.h: Include profile-count.h
28371 (struct edge_def): Turn count to profile_count.
28372 (struct basic_block_def): Likewie.
28373 (REG_BR_PROB_BASE): Move to profile-count.h
28374 (RDIV): Move to profile-count.h
28375 * bb-reorder.c (max_entry_count): Turn to profile_count.
28376 (find_traces): Update.
28377 (rotate_loop):Update.
28378 (connect_traces):Update.
28379 (sanitize_hot_paths):Update.
28380 * bt-load.c (migrate_btr_defs): Update.
28381 * cfg.c (RDIV): Remove.
28382 (init_flow): Use alloc_block.
28383 (alloc_block): Uninitialize count.
28384 (unchecked_make_edge): Uninitialize count.
28385 (check_bb_profile): Update.
28386 (dump_edge_info): Update.
28387 (dump_bb_info): Update.
28388 (update_bb_profile_for_threading): Update.
28389 (scale_bbs_frequencies_int): Update.
28390 (scale_bbs_frequencies_gcov_type): Update.
28391 (scale_bbs_frequencies_profile_count): New.
28392 * cfg.h (update_bb_profile_for_threading): Update.
28393 (scale_bbs_frequencies_profile_count): Declare.
28394 * cfgbuild.c (compute_outgoing_frequencies): Update.
28395 (find_many_sub_basic_blocks): Update.
28396 * cfgcleanup.c (try_forward_edges): Update.
28397 (try_crossjump_to_edge): Update.
28398 * cfgexpand.c (expand_gimple_tailcall): Update.
28399 (construct_exit_block): Update.
28400 * cfghooks.c (verify_flow_info): Update.
28401 (dump_bb_for_graph): Update.
28402 (split_edge): Update.
28403 (make_forwarder_block): Update.
28404 (duplicate_block): Update.
28405 (account_profile_record): Update.
28406 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
28407 (get_estimated_loop_iterations): Update.
28408 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
28409 (single_likely_exit): Update.
28410 * cfgloopmanip.c (scale_loop_profile): Update.
28412 (set_zero_probability): Update.
28413 (lv_adjust_loop_entry_edge): Update.
28414 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
28415 (purge_dead_edges): Update.
28416 (rtl_account_profile_record): Update.
28417 * cgraph.c (cgraph_node::create): Uninitialize count.
28418 (symbol_table::create_edge): Uninitialize count.
28419 (cgraph_update_edges_for_call_stmt_node): Update.
28420 (cgraph_edge::dump_edge_flags): Update.
28421 (cgraph_node::dump): Update.
28422 (cgraph_edge::maybe_hot_p): Update.
28423 * cgraph.h: Include profile-count.h
28424 (create_clone), create_edge, create_indirect_edge): Update.
28425 (cgraph_node): Turn count to profile_count.
28426 (cgraph_edge0: Likewise.
28427 (make_speculative, clone): Update.
28428 (create_edge): Update.
28429 (init_lowered_empty_function): Update.
28430 * cgraphclones.c (cgraph_edge::clone): Update.
28431 (duplicate_thunk_for_node): Update.
28432 (cgraph_node::create_clone): Update.
28433 * cgraphunit.c (cgraph_node::analyze): Update.
28434 (cgraph_node::expand_thunk): Update.
28435 * final.c (dump_basic_block_info): Update.
28436 * gimple-streamer-in.c (input_bb): Update.
28437 * gimple-streamer-out.c (output_bb): Update.
28438 * graphite.c (print_global_statistics): Update.
28439 (print_graphite_scop_statistics): Update.
28440 * hsa-brig.c: Include basic-block.h.
28441 * hsa-dump.c: Include basic-block.h.
28442 * hsa-gen.c (T sum_slice): Update.
28443 (convert_switch_statements):Update.
28444 * hsa-regalloc.c: Include basic-block.h.
28445 * ipa-chkp.c (chkp_produce_thunks): Update.
28446 * ipa-cp.c (struct caller_statistics): Update.
28447 (init_caller_stats): Update.
28448 (gather_caller_stats): Update.
28449 (ipcp_cloning_candidate_p): Update.
28450 (good_cloning_opportunity_p): Update.
28451 (get_info_about_necessary_edges): Update.
28452 (dump_profile_updates): Update.
28453 (update_profiling_info): Update.
28454 (update_specialized_profile): Update.
28455 (perhaps_add_new_callers): Update.
28456 (decide_about_value): Update.
28457 (ipa_cp_c_finalize): Update.
28458 * ipa-devirt.c (struct odr_type_warn_count): Update.
28459 (struct decl_warn_count): Update.
28460 (struct final_warning_record): Update.
28461 (possible_polymorphic_call_targets): Update.
28462 (ipa_devirt): Update.
28463 * ipa-fnsummary.c (redirect_to_unreachable): Update.
28464 * ipa-icf.c (sem_function::merge): Update.
28465 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
28466 * ipa-inline.c (compute_uninlined_call_time): Update.
28467 (compute_inlined_call_time): Update.
28468 (want_inline_small_function_p): Update.
28469 (want_inline_self_recursive_call_p): Update.
28470 (edge_badness): Update.
28471 (lookup_recursive_calls): Update.
28472 (recursive_inlining): Update.
28473 (inline_small_functions): Update.
28474 (dump_overall_stats): Update.
28475 (dump_inline_stats): Update.
28476 * ipa-profile.c (ipa_profile_generate_summary): Update.
28477 (ipa_propagate_frequency): Update.
28478 (ipa_profile): Update.
28479 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28480 * ipa-utils.c (ipa_merge_profiles): Update.
28481 * loop-doloop.c (doloop_modify): Update.
28482 * loop-unroll.c (report_unroll): Update.
28483 (unroll_loop_runtime_iterations): Update.
28484 * lto-cgraph.c (lto_output_edge): Update.
28485 (lto_output_node): Update.
28486 (input_node): Update.
28487 (input_edge): Update.
28488 (merge_profile_summaries): Update.
28489 * lto-streamer-in.c (input_cfg): Update.
28490 * lto-streamer-out.c (output_cfg): Update.
28491 * mcf.c (create_fixup_graph): Update.
28492 (adjust_cfg_counts): Update.
28493 (sum_edge_counts): Update.
28494 * modulo-sched.c (sms_schedule): Update.
28495 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
28496 * predict.c (maybe_hot_count_p): Update.
28497 (probably_never_executed): Update.
28498 (dump_prediction): Update.
28499 (combine_predictions_for_bb): Update.
28500 (propagate_freq): Update.
28501 (handle_missing_profiles): Update.
28502 (counts_to_freqs): Update.
28503 (rebuild_frequencies): Update.
28504 (force_edge_cold): Update.
28505 * predict.h: Include profile-count.h
28506 (maybe_hot_count_p, counts_to_freqs): UPdate.
28507 * print-rtl-function.c: Do not include cfg.h
28508 * print-rtl.c: Include basic-block.h
28509 * profile-count.c: New file.
28510 * profile-count.h: New file.
28511 * profile.c (is_edge_inconsistent): Update.
28512 (correct_negative_edge_counts): Update.
28513 (is_inconsistent): Update.
28514 (set_bb_counts): Update.
28515 (read_profile_edge_counts): Update.
28516 (compute_frequency_overlap): Update.
28517 (compute_branch_probabilities): Update; Initialize and deinitialize
28519 (branch_prob): Update.
28520 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
28521 (edge_gcov_count): New.
28522 (bb_gcov_count): New.
28523 * shrink-wrap.c (try_shrink_wrapping): Update.
28524 * tracer.c (better_p): Update.
28525 * trans-mem.c (expand_transaction): Update.
28526 (ipa_tm_insert_irr_call): Update.
28527 (ipa_tm_insert_gettmclone_call): Update.
28528 * tree-call-cdce.c: Update.
28529 * tree-cfg.c (gimple_duplicate_sese_region): Update.
28530 (gimple_duplicate_sese_tail): Update.
28531 (gimple_account_profile_record): Update.
28532 (execute_fixup_cfg): Update.
28533 * tree-inline.c (copy_bb): Update.
28534 (copy_edges_for_bb): Update.
28535 (initialize_cfun): Update.
28536 (freqs_to_counts): Update.
28537 (copy_cfg_body): Update.
28538 (expand_call_inline): Update.
28539 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
28540 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
28541 (try_unroll_loop_completely): Update.
28542 (try_peel_loop): Update.
28543 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
28544 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
28545 * tree-ssa-loop-split.c (connect_loops): Update.
28546 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
28547 * tree-ssa-reassoc.c (branch_fixup): Update.
28548 * tree-ssa-tail-merge.c (replace_block_by): Update.
28549 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
28550 (compute_path_counts): Update.
28551 (update_profile): Update.
28552 (recompute_probabilities): Update.
28553 (update_joiner_offpath_counts): Update.
28554 (estimated_freqs_path): Update.
28555 (freqs_to_counts_path): Update.
28556 (clear_counts_path): Update.
28557 (ssa_fix_duplicate_block_edges): Update.
28558 (duplicate_thread_path): Update.
28559 * tree-switch-conversion.c (case_bit_test_cmp): Update.
28560 (struct switch_conv_info): Update.
28561 * tree-tailcall.c (decrease_profile): Update.
28562 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
28563 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
28564 * value-prof.c (check_counter): Update.
28565 (gimple_divmod_fixed_value): Update.
28566 (gimple_mod_pow2): Update.
28567 (gimple_mod_subtract): Update.
28568 (gimple_ic_transform): Update.
28569 (gimple_stringop_fixed_value): Update.
28570 * value-prof.h (gimple_ic): Update.
28572 2017-06-02 Carl Love <cel@us.ibm.com>
28574 * config/rs6000/rs6000-c: Add support for built-in functions
28575 vector double vec_doublee (vector signed int);
28576 vector double vec_doublee (vector unsigned int);
28577 vector double vec_doublee (vector float);
28578 vector double vec_doubleh (vector signed int);
28579 vector double vec_doubleh (vector unsigned int);
28580 vector double vec_doubleh (vector float);
28581 vector double vec_doublel (vector signed int);
28582 vector double vec_doublel (vector unsigned int);
28583 vector double vec_doublel (vector float);
28584 vector double vec_doubleo (vector signed int);
28585 vector double vec_doubleo (vector unsigned int);
28586 vector double vec_doubleo (vector float);.
28587 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
28588 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
28590 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
28591 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
28592 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
28594 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
28595 vec_doublel, vec_doubleh.
28596 * doc/extend.texi: Update the built-in documentation file for the
28597 new built-in functions.
28599 2017-06-02 David Malcolm <dmalcolm@redhat.com>
28602 * ipa-inline-analysis.c (free_growth_caches): Set
28603 edge_removal_hook_holder to NULL after removing it.
28605 2017-06-02 Sudakshina Das <sudi.das@arm.com>
28607 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
28608 comparision with zero.
28610 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28611 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28612 for early expansion of vec_min and vec_max builtins.
28613 (builtin_function_type): Add min/max unsigned variants to those
28614 identified as having unsigned arguments.
28616 2017-06-02 Olivier Hainque <hainque@adacore.com>
28618 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
28620 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28622 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
28623 Use VALL_F16 iterator rather than VALL.
28625 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28627 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
28628 Emit CBNZ inside loop when doing a strong exchange and comparing
28629 against zero. Generate the CC flags after the loop.
28631 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
28633 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
28634 (dl_section_ref): New.
28635 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
28636 On AIX, append an expression to subtract the size of the
28637 section length to dl_section_ref.
28639 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28641 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28642 for early expansion of vector absolute builtins.
28644 2017-06-02 Richard Biener <rguenther@suse.de>
28646 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
28647 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
28649 2017-06-02 Richard Biener <rguenther@suse.de>
28651 PR tree-optimization/80948
28652 * tree-tailcall.c (find_tail_calls): Track stmts to move in
28653 stmt order as well.
28655 2017-06-02 Richard Biener <rguenther@suse.de>
28657 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
28659 * tree-vect-stmts.c (process_use): Do not mark backedge defs
28660 for inductions as relevant.
28662 2017-06-02 Richard Biener <rguenther@suse.de>
28664 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
28665 (vectorizable_induction): ... this. Remove dead code.
28667 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
28669 * builtins. (expand_builtin_alloca): Remove second parameter and
28670 infer its value from the first parameter instead.
28671 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
28673 2017-06-02 Jakub Jelinek <jakub@redhat.com>
28675 PR rtl-optimization/80903
28676 * loop-doloop.c (add_test): Unshare sequence.
28678 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
28680 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
28682 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
28684 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
28686 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
28687 xlogue_layout::get_instance, logue_layout::xlogue_layout,
28688 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
28689 (xlogue_layout::get_stub_rtx): Make static.
28690 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
28691 (xlogue_layout::compute_stub_managed_regs): Rename to...
28692 (xlogue_layout::count_stub_managed_regs): ...this.
28693 (xlogue_layout::is_stub_managed_reg): New function.
28694 (xlogue_layout::m_stub_names): Rename to...
28695 (xlogue_layout::s_stub_names): ...this, make static.
28696 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
28697 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
28698 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
28699 xlogue_layout::s_stub_names): Instantiate statics.
28700 (stub_managed_regs): Remove.
28701 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
28702 (disable_call_ms2sysv_xlogues): Rename to...
28703 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
28704 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
28706 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
28707 change after reload_completed.
28708 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
28710 (ix86_expand_prologue): Likewise.
28711 (ix86_expand_epilogue): Likewise.
28712 (ix86_expand_split_stack_prologue): Likewise.
28713 (ix86_compute_frame_layout): Remove frame parameter ...
28714 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
28715 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
28717 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
28718 (ix86_frame): Move from here ...
28719 * config/i386/i386.h (ix86_frame): ... to here.
28720 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
28721 complete ix86_frame data structure instead. Remove some_ld_name.
28723 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28725 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
28726 symbols that hold a DECL_VALUE_EXPR.
28728 2017-06-01 Martin Jambor <mjambor@suse.cz>
28730 PR tree-optimization/80898
28731 * tree-sra.c (process_subtree_disqualification): Removed.
28732 (disqualify_candidate): Do not acll
28733 process_subtree_disqualification.
28734 (subtree_mark_written_and_enqueue): New function.
28735 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
28736 RHS has been disqualified and re-queue LHS if necessary. Apart
28737 from that, ignore disqualified RHS.
28739 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28741 * config/s390/s390.c (s390_emit_epilogue): Disable early return
28742 address fetch for z10 or later.
28744 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28746 * config/arc/arc.md (tst_movb): Add guard when splitting.
28748 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28750 * config/arc/arc.c (arc_can_eliminate): Test against
28751 arc_frame_pointer_needed.
28753 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28755 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
28756 to prevent store reordering.
28757 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
28758 (type): Add block type.
28759 (stack_tie): Define special instruction to be used in
28762 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28764 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
28765 constraint. It is not valid for the pattern.
28766 (noncommutative_binary_comparison): Likewise.
28768 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28770 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
28773 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28775 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
28776 be used by the reg-alloc.
28778 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28780 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
28781 reg-alloc when having mul64 or mul32x16 instructions.
28782 (mulsidi3): Likewise.
28783 (umulsidi3): Likewise.
28784 (mulsi32x16): New pattern.
28785 (mulsi64): Likewise.
28786 (mulsidi64): Likewise.
28787 (umulsidi64): Likewise.
28788 (MUL32x16_REG): Define.
28789 (mul64_600): Use MUL32x16_REG.
28790 (mac64_600): Likewise.
28791 (umul64_600): Likewise.
28792 (umac64_600): Likewise.
28794 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28796 * config/arc/arc.md (mulsi3_700): Make it commutative.
28798 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
28800 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
28802 (*sign_extendsidi2_insn): Likewise for movstosw.
28804 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28806 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
28807 the type of the input discriminant value. Convert the
28808 discriminant value of signedness vary.
28810 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
28812 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
28813 Add to -Wall section.
28815 2017-06-01 Richard Biener <rguenther@suse.de>
28817 PR middle-end/66313
28818 * fold-const.c (fold_plusminus_mult_expr): If the factored
28819 factor may be zero use a wrapping type for the inner operation.
28820 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
28821 and handle moved defs.
28822 (process_assignment): Properly guard the unary op case. Return a
28823 tri-state indicating that moving the stmt before the call may allow
28824 to continue. Pass through to_move.
28825 (find_tail_calls): Handle moving unrelated defs before
28828 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
28831 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
28832 splitter result in the canonical way.
28834 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
28836 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
28837 also for 32bit target. Update insn attributes.
28838 (zero-extendsidi2 splitter): Allow all registers for operand 1.
28840 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
28842 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
28843 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
28844 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
28845 (_mm_maskz_min_ss): New intrinsics.
28847 2017-05-31 Martin Liska <mliska@suse.cz>
28849 * tree-vect-loop.c (vect_create_epilog_for_reduction):
28850 Change comment style to one we normally use.
28851 (vectorizable_reduction): Likewise.
28852 (vectorizable_induction): Likewise.
28853 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
28854 (vectorizable_call): Likewise.
28855 (vectorizable_simd_clone_call): Likewise.
28856 (vectorizable_conversion): Likewise.
28857 (vectorizable_assignment): Likewise.
28858 (vectorizable_shift): Likewise.
28859 (vectorizable_operation): Likewise.
28860 (vectorizable_store): Likewise.
28861 (vectorizable_load): Likewise.
28862 * tree-vectorizer.h: Likewise.
28864 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
28866 * passes.c (emergency_dump_function): New.
28867 * tree-pass.h (emergency_dump_function): Declare.
28868 * plugin.c (plugins_internal_error_function): Remove.
28869 * plugin.h (plugins_internal_error_function): Remove declaration.
28870 * toplev.c (internal_error_function): New static function. Use it...
28871 (general_init): ...here.
28873 2017-05-31 Graham Markall <graham.markall@embecosm.com>
28875 * config/arc/arc.c (arc_print_operand): Handle constant operands.
28876 (arc_rtx_costs): Add costs for new patterns.
28877 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
28878 * config/arc/predicates.md: Add _1_2_3_operand predicate.
28880 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28882 * tree-ssa-strlen.c (get_next_strinfo): New function.
28883 (get_stridx_plus_constant): Use it.
28884 (zero_length_string): Likewise.
28885 (adjust_related_strinfos): Likewise.
28886 (adjust_last_stmt): Likewise.
28888 2017-05-31 Richard Biener <rguenther@suse.de>
28891 * config/i386/i386.c (ix86_expand_builtin): Remove assert
28892 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
28894 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28896 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
28897 loop_vinfo argument and use of dependence distance vectors.
28898 Check instead whether the two references differ only in their
28899 initial value and assume that they have the same alignment if the
28900 difference is a multiple of the vector alignment.
28901 (vect_analyze_data_refs_alignment): Update call accordingly.
28903 2017-05-31 Martin Liska <mliska@suse.cz>
28906 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
28908 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28910 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
28911 (create_intersect_range_checks): Move from ...
28912 * tree-data-ref.c (create_intersect_range_checks_index)
28913 (create_intersect_range_checks): ... to here.
28914 (create_runtime_alias_checks): New function factored from ...
28915 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
28916 here. Call above function.
28917 * tree-data-ref.h (create_runtime_alias_checks): New function.
28919 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28921 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
28922 segment length for dr_b and compute it in wide_int.
28924 2017-05-31 Richard Biener <rguenther@suse.de>
28926 PR tree-optimization/80906
28927 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
28928 and pass through iv_map.
28929 (copy_bb_and_scalar_dependences): Adjust.
28930 (translate_pending_phi_nodes): Likewise.
28931 (copy_loop_close_phi_args): Handle code-generating IVs instead
28934 2017-05-30 David Malcolm <dmalcolm@redhat.com>
28936 * diagnostic-color.c (color_dict): Add "type-diff".
28937 (parse_gcc_colors): Update comment.
28938 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
28939 -fdiagnostics-show-template-tree and -fno-elide-type.
28940 (GCC_COLORS): Add type-diff to example.
28942 (-fdiagnostics-show-template-tree): New.
28943 (-fno-elide-type): New.
28944 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
28945 the pp_format_decoder callback. Call any m_format_postprocessor's
28947 (pretty_printer::pretty_printer): Initialize
28948 m_format_postprocessor.
28949 (pretty_printer::~pretty_printer): Delete any
28950 m_format_postprocessor.
28951 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
28952 (class format_postprocessor): New class.
28953 (struct pretty_printer::format_decoder): Document the new parameters.
28954 (struct pretty_printer::m_format_postprocessor): New field.
28955 * tree-diagnostic.c (default_tree_printer): Update for new
28956 bool and const char ** params.
28957 * tree-diagnostic.h (default_tree_printer): Likewise.
28959 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
28961 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
28962 (lwa_operand): Delete rs6000_gen_cell_microcode test.
28963 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28964 rs6000_gen_cell_microcode code.
28965 (rs6000_final_prescan_insn): Delete.
28966 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
28967 "warn-cell-microcode" entries.
28968 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
28969 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
28970 throughout. Change cc_reg_not_micro_cr0_operand to
28971 cc_reg_not_cr0_operand throughout.
28972 (*extendhi<mode>2_noload): Delete.
28973 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
28974 (mwarn-cell-microcode): Delete.
28975 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
28976 -mgen-cell-microcode and -mwarn-cell-microcode.
28978 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
28981 * config/i386/constraints.md (Yd): New constraint.
28983 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
28984 and (?Yd, r) alternatives. Update insn attributes.
28985 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
28986 and (?*Yd, r) alternatives. Update insn attributes.
28987 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
28989 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
28991 * gimplify.c (gimplify_modify_expr): Don't create a
28992 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
28995 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
28997 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
28999 2017-05-30 Richard Biener <rguenther@suse.de>
29001 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
29002 and reduc_def fields.
29003 (STMT_VINFO_REDUC_TYPE): New define.
29004 (STMT_VINFO_REDUC_DEF): Likewise.
29005 (vect_force_simple_reduction): Adjust prototype.
29006 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
29007 (vect_is_simple_reduction): Remove check_reduction argument.
29008 (vect_force_simple_reduction): Adjust and set
29009 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29010 (vectorizable_reduction): Do not re-do reduction analysis
29011 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29012 * tree-parloops.c (gather_scalar_reductions): Adjust.
29014 2017-05-30 Richard Biener <rguenther@suse.de>
29016 PR middle-end/80901
29017 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
29020 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
29022 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
29023 Introduce unknown_misalignment parameter and remove vf.
29024 (vect_peeling_hash_get_lowest_cost):
29025 Pass unknown_misalignment parameter.
29026 (vect_enhance_data_refs_alignment):
29027 Fix unsupportable data ref treatment.
29029 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29031 * tree-vect-data-refs.c (vect_get_data_access_cost):
29032 Workaround for SLP handling.
29033 (vect_enhance_data_refs_alignment):
29034 Compute costs for doing no peeling at all, compare to the best
29035 peeling costs so far and avoid peeling if cheaper.
29037 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29039 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
29040 Return peeling info and set costs to zero for unlimited cost
29042 (vect_enhance_data_refs_alignment): Also inspect all datarefs
29043 with unknown misalignment. Compute and costs for unknown
29044 misalignment, compare them to the costs for known misalignment
29045 and choose the cheapest for peeling.
29047 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29049 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
29050 (vect_get_peeling_costs_all_drs): Create function.
29051 (vect_peeling_hash_get_lowest_cost):
29052 Use vect_get_peeling_costs_all_drs.
29053 (vect_peeling_supportable): Create function.
29054 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
29056 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29058 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
29059 DR_HAS_NEGATIVE_STEP.
29060 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
29061 (vect_enhance_data_refs_alignment): Use.
29062 (vect_duplicate_ssa_name_ptr_info): Use.
29063 * tree-vectorizer.h (dr_misalignment): Use.
29064 (known_alignment_for_access_p): Use.
29066 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
29069 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
29071 (has_section_name): New function.
29073 2017-05-30 Martin Liska <mliska@suse.cz>
29076 * auto-profile.c (get_function_decl_from_block): Fix
29079 2017-05-30 Richard Biener <rguenther@suse.de>
29081 PR middle-end/80876
29082 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
29084 2017-05-30 Martin Liska <mliska@suse.cz>
29086 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
29087 * dumpfile.h (struct dump_file_info): Remove ctors.
29089 2017-05-30 Martin Liska <mliska@suse.cz>
29091 * predict.def: Fix GNU coding style.
29093 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29095 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29096 Mark 'to' argument with ATTRIBUTE_UNUSED.
29098 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29100 * config/xtensa/xtensa.c (xtensa_emit_call): Use
29101 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
29102 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
29105 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
29107 * doc/install.texi (Options specification): Restore entry of
29108 --enable-sjlj-exceptions.
29110 2017-05-27 Michael Eager <eager@eagercon.com>
29113 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
29115 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
29117 * config/microblaze/microblaze.h
29118 (FIXED_REGISTERS): Update in macro.
29119 (CALL_USED_REGISTERS): Update in macro.
29121 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
29123 * doc/install.texi: Add links to macOS binary distributions.
29125 2017-05-27 Jakub Jelinek <jakub@redhat.com>
29129 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29131 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29133 2017-05-26 Martin Liska <mliska@suse.cz>
29135 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
29137 2017-05-26 Martin Liska <mliska@suse.cz>
29139 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
29140 always leading ';; '.
29141 (dump_bb_info): Likewise.
29142 (brief_dump_cfg): Likewise.
29143 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
29144 * dumpfile.c: Remove usage of TDF_VERBOSE.
29145 * dumpfile.h (enum dump_kind): Likewise.
29146 (dump_gimple_bb_header): Do not use TDF_COMMENT.
29147 * print-tree.c (debug_verbose): Remove.
29148 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
29149 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
29150 * tree-diagnostic.c (default_tree_printer): Replace
29151 TDF_DIAGNOSTIC with TDF_SLIM.
29153 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29155 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
29156 in parameter loop, rather than loop_vinfo.
29157 (create_intersect_range_checks): Ditto.
29158 (vect_create_cond_for_alias_checks): Update call to above functions.
29160 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29162 PR tree-optimization/80815
29163 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
29164 for merging runtime alias checks. Handle negative DR_STEPs.
29166 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29168 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
29170 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
29171 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
29172 out code pruning runtime alias checks.
29173 * tree-data-ref.c (prune_runtime_alias_test_list): New function
29174 factored out from above.
29175 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29177 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29179 (prune_runtime_alias_test_list): New decalaration.
29181 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29183 * tree-vect-data-refs.c (compare_tree): Rename and move ...
29184 * tree-data-ref.c (data_ref_compare_tree): ... to here.
29185 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
29186 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
29187 (operator==, comp_dr_with_seg_len_pair): Ditto.
29188 (vect_prune_runtime_alias_test_list): Ditto.
29190 2017-05-26 Martin Liska <mliska@suse.cz>
29193 * params.def: Bound partial-inlining-entry-probability param.
29195 2017-05-26 Marek Polacek <polacek@redhat.com>
29198 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
29201 2017-05-26 Richard Biener <rguenther@suse.de>
29203 PR tree-optimization/80842
29204 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
29207 2017-05-26 Richard Biener <rguenther@suse.de>
29209 PR tree-optimization/80844
29210 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
29212 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
29214 * doc/md.texi (Machine Constraints): Update x86 family
29215 machine constraints section to match 'config/i386/constraints.md'.
29217 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
29219 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
29221 2017-05-25 Nathan Sidwell <nathan@acm.org>
29223 * doc/invoke.texi (--enable-languages): Update documentation.
29225 2017-05-25 Martin Liska <mliska@suse.cz>
29227 * dumpfile.c: Add TDF_FOLDING.
29228 * dumpfile.h (enum dump_kind): Likewise.
29229 * genmatch.c (dt_simplify::gen_1): Use it.
29231 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29233 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
29235 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29237 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29238 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
29240 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29242 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
29243 * match.pd (X == C): Rewrite it here.
29244 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
29245 with_certain_nonzero_bits2): New predicates.
29246 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
29248 2017-05-24 Nathan Sidwell <nathan@acm.org>
29250 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
29253 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
29256 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29258 * config/powerpcspe: New port. Files are copied from the rs6000
29259 port, with "rs6000" in filenames replaced by "powerpcspe".
29261 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
29263 PR rtl-optimization/80754
29264 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
29266 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
29268 * config/sparc/sparc.md (length): Return the correct value for -mflat
29269 sibcalls to match output_sibcall.
29271 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29275 * config/rs6000/rs6000.c (struct machine_function): Add new field
29277 (rs6000_get_separate_components): Init that field, use it.
29278 (rs6000_components_for_bb): Use the field.
29280 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
29282 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
29284 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
29286 PR middle-end/80823
29287 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
29289 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29292 * config/s390/s390.c (s390_check_qrst_address): Check incoming
29293 address against address_operand predicate.
29294 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
29296 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
29298 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
29299 parameters passed indirectly.
29301 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
29303 * config/i386/i386.md (*movdi_internal): Remove SSE4
29304 alternative 18 (?r, *v). Update insn attributes.
29305 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
29306 Update insn attributes.
29307 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
29308 Update insn attributes.
29309 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
29310 alternative 1 (r, v). Remove isa attribute.
29311 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
29312 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
29313 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
29315 2017-05-23 Tom de Vries <tom@codesourcery.com>
29317 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
29320 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29322 * cgraphunit.c (symbol_table::process_new_functions): Update.
29323 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
29324 (inline_generate_summary): Rename to ...
29325 (ipa_fn_summary_generate): ... this one.
29326 (inline_read_summary): Rename to ...
29327 (ipa_fn_summary_read): ... this one.
29328 (inline_write_summary): Rename to ...
29329 (ipa_fn_summary_write): ... this one.
29330 (inline_free_summary): Rename to ...
29331 (ipa_free_fn_summary): ... this one.
29332 (pass_data_local_fn_summary, pass_local_fn_summary,
29333 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
29334 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
29335 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
29336 make_pass_ipa_fn_summary): New.
29337 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
29338 inline_write_summary, inline_free_summary): Remove.
29339 (ipa_free_fn_summary) : New.
29340 * ipa-inline.c (ipa_inline): Update.
29341 (pass_ipa_inline): Do not generate summaries.
29342 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
29344 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
29345 and add pass_ipa_fn_summary.
29346 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
29348 (make_pass_inline_parameters): Remove.
29350 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
29352 * omp-low.c (struct omp_context): Remove "default_kind" member.
29355 * omp-offload.c (execute_oacc_device_lower): Remove the
29356 parallelism dimensions function attributes for unparallelized
29357 OpenACC kernels constructs.
29359 2017-05-23 Martin Liska <mliska@suse.cz>
29361 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
29363 (cgraph_edge::make_speculative): Likewise.
29364 (cgraph_edge::resolve_speculation): Likewise.
29365 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29366 (cgraph_node::dump): Likewise.
29367 * cgraph.h: Likewise.
29368 * cgraphunit.c (analyze_functions): Likewise.
29369 (symbol_table::compile): Likewise.
29370 * ipa-cp.c (print_all_lattices): Likewise.
29371 (determine_versionability): Likewise.
29372 (initialize_node_lattices): Likewise.
29373 (ipcp_verify_propagated_values): Likewise.
29374 (estimate_local_effects): Likewise.
29375 (update_profiling_info): Likewise.
29376 (create_specialized_node): Likewise.
29377 (perhaps_add_new_callers): Likewise.
29378 (decide_about_value): Likewise.
29379 (decide_whether_version_node): Likewise.
29380 (identify_dead_nodes): Likewise.
29381 (ipcp_store_bits_results): Likewise.
29382 * ipa-devirt.c (dump_targets): Likewise.
29383 (ipa_devirt): Likewise.
29384 * ipa-icf.c (sem_item::dump): Likewise.
29385 (sem_function::equals): Likewise.
29386 (sem_variable::equals): Likewise.
29387 (sem_item_optimizer::read_section): Likewise.
29388 (sem_item_optimizer::execute): Likewise.
29389 (congruence_class::dump): Likewise.
29390 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
29391 (dump_inline_summary): Likewise.
29392 (estimate_node_size_and_time): Likewise.
29393 (inline_analyze_function): Likewise.
29394 * ipa-inline-transform.c (inline_call): Likewise.
29395 * ipa-inline.c (report_inline_failed_reason): Likewise.
29396 (want_early_inline_function_p): Likewise.
29397 (edge_badness): Likewise.
29398 (update_edge_key): Likewise.
29399 (inline_small_functions): Likewise.
29400 * ipa-profile.c (ipa_profile): Likewise.
29401 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
29402 (ipa_make_edge_direct_to_target): Likewise.
29403 (remove_described_reference): Likewise.
29404 (ipa_impossible_devirt_target): Likewise.
29405 (propagate_controlled_uses): Likewise.
29406 (ipa_print_node_params): Likewise.
29407 (ipcp_transform_function): Likewise.
29408 * ipa-pure-const.c (pure_const_read_summary): Likewise.
29409 (propagate_pure_const): Likewise.
29410 * ipa-reference.c (generate_summary): Likewise.
29411 (read_write_all_from_decl): Likewise.
29412 (propagate): Likewise.
29413 (ipa_reference_read_optimization_summary): Likewise.
29414 * ipa-utils.c (ipa_merge_profiles): Likewise.
29415 * ipa.c (walk_polymorphic_call_targets): Likewise.
29416 (symbol_table::remove_unreachable_nodes): Likewise.
29417 (ipa_single_use): Likewise.
29418 * passes.c (execute_todo): Likewise.
29419 * predict.c (drop_profile): Likewise.
29420 * symtab.c (symtab_node::get_dump_name): New function.
29421 (symtab_node::dump_name): Likewise.
29422 (symtab_node::dump_asm_name): Likewise.
29423 (symtab_node::dump_references): Likewise.
29424 (symtab_node::dump_referring): Likewise.
29425 (symtab_node::dump_base): Likewise.
29426 (symtab_node::debug_symtab): Likewise.
29427 * tree-sra.c (convert_callers_for_node): Likewise.
29428 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
29429 * value-prof.c (init_node_map): Likewise.
29431 2017-05-23 Martin Liska <mliska@suse.cz>
29433 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
29434 and symtab_node::debug_symtab to symbol_table::debug.
29435 * cgraphunit.c (analyze_functions): Use the renamed function.
29436 (symbol_table::compile): Likewise.
29437 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
29438 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
29439 * passes.c (execute_todo): Likewise.
29440 * symtab.c (symbol_table::dump): New function.
29441 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
29443 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29445 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
29446 that nonconst implies exec.
29448 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29450 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29451 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29452 (inline_edge_summary_vec): Turn into ...
29453 (ipa_call_summaries): ... this one.
29454 (redirect_to_unreachable, edge_set_predicate,
29455 evaluate_properties_for_edge, inline_summary_alloc,
29456 reset_ipa_call_summary, reset_inline_summary,
29457 inline_summary_t::duplicate): Update.
29458 (inline_edge_duplication_hook): Turn to ...
29459 (ipa_call_summary_t::duplicate): ... this one.
29460 (inline_edge_removal_hook): Turn to ...
29461 (ipa_call_summary_t::remove): ... this one.
29462 (dump_inline_edge_summary): Turn to ...
29463 (dump_ipa_call_summary): ... this one.
29464 (estimate_function_body_sizes): Update.
29465 (inline_update_callee_summaries): Update.
29466 (remap_edge_change_prob): Update.
29467 (remap_edge_summaries): Update.
29468 (inline_merge_summary): Update.
29469 (do_estimate_edge_time): Update.
29470 (inline_generate_summary): Update.
29471 (inline_read_section): Update.
29472 (inline_read_summary): Update.
29473 (inline_free_summary): Update.
29474 * ipa-inline.c (can_inline_edge_p): Update.
29475 (compute_inlined_call_time): Update.
29476 (want_inline_small_function_p): Update.
29477 (edge_badness): Update.
29478 (early_inliner): Update.
29479 * ipa-inline.h (inline_edge_summary): Turn to ...
29480 (ipa_call_summary): ... this one.
29481 (ipa_call_summary_t): New class.
29482 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29483 (ipa_call_summaries): New.
29484 (inline_edge_summary): Remove.
29485 (estimate_edge_growth): Update.
29486 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29487 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29488 * ipa-split.c (execute_split_functions): Update.
29489 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29491 2017-05-23 Tom de Vries <tom@codesourcery.com>
29493 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29494 attributes): Document rdrand effective target.
29496 2017-05-23 Tom de Vries <tom@codesourcery.com>
29498 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29499 attributes): Sort alphabetically.
29501 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
29503 * config/avr/genmultilib.awk: Use gsub instead of gensub.
29505 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29508 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
29509 V2DF/V2DI splat into two separate patterns, one that handles
29510 registers, and the other that only handles memory. Drop support
29511 for splatting from a GPR on ISA 2.07 and then splitting the
29512 splat into direct move and splat.
29513 (vsx_splat_<mode>_reg): Likewise.
29514 (vsx_splat_<mode>_mem): Likewise.
29516 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
29518 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
29520 2017-05-22 Jakub Jelinek <jakub@redhat.com>
29522 PR middle-end/80809
29523 * omp-low.c (finish_taskreg_remap): New function.
29524 (finish_taskreg_scan): If unit size of ctx->record_type
29525 is non-constant, unshare the size expression and replace
29526 decls in it with possible outer var refs.
29528 PR middle-end/80809
29529 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
29530 GOVD_SHARED rather than GOVD_PRIVATE with it.
29531 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
29532 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
29534 PR middle-end/80853
29535 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
29536 as last argument to build_outer_var_ref for pointer bases of array
29537 section reductions.
29539 2017-05-19 Martin Sebor <msebor@redhat.com>
29541 * print-tree.c (print_node): Print DECL_READ_P flag.
29543 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29545 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
29546 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
29547 * cgraph.c: Likewise.
29548 * cgraphunit.c: Likewise.
29549 * gengtype.c: Likewise.
29550 * ipa-cp.c: Likewise.
29551 * ipa-devirt.c: Likewise.
29552 * ipa-icf.c: Likewise.
29553 * ipa-predicate.c: Likewise.
29554 * ipa-profile.c: Likewise.
29555 * ipa-prop.c: Likewise.
29556 * ipa-split.c: Likewise.
29558 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
29559 edge_predicate_pool, dump_inline_hints,
29560 inline_summary::account_size_time, redirect_to_unreachable,
29561 edge_set_predicate, set_hint_predicate,
29562 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
29563 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
29564 inline_summary_t::remove, remap_hint_predicate_after_duplication,
29565 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
29566 ipa_call_summary_t::remove, initialize_growth_caches,
29567 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
29568 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
29569 mark_modified, unmodified_parm_1, unmodified_parm,
29570 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
29571 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
29572 compute_bb_predicates, will_be_nonconstant_expr_predicate,
29573 will_be_nonconstant_predicate, record_modified_bb_info,
29574 get_minimal_bb, record_modified, param_change_prob,
29575 phi_result_unknown_predicate, predicate_for_phi_result,
29576 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
29577 estimate_function_body_sizes, compute_inline_parameters,
29578 compute_inline_parameters_for_curren, pass_data_inline_parameters,
29579 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
29580 inline_update_callee_summaries, remap_edge_change_prob,
29581 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
29582 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
29583 inline_analyze_function, inline_summary_t::insert,
29584 inline_generate_summary, read_ipa_call_summary, inline_read_section,
29585 inline_read_summary, write_ipa_call_summary, inline_write_summary,
29586 inline_free_summary): Move to ipa-fnsummary.h
29587 (predicate_t): Remove.
29588 * ipa-fnsummary.c: New file.
29589 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
29590 (enum inline_hints_vals, inline_hints, agg_position_info,
29591 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
29592 inline_summaries, ipa_call_summary, ipa_call_summary_t,
29593 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
29594 dump_inline_summary, dump_inline_hints, inline_generate_summary,
29595 inline_read_summary, inline_write_summary, inline_free_summary,
29596 inline_analyze_function, initialize_inline_failed,
29597 inline_merge_summary, inline_update_overall_summary,
29598 compute_inline_parameters): Move to ipa-fnsummary.h
29599 * ipa-fnsummary.h: New file.
29600 * ipa-inline-transform.h: Include ipa-inline.h.
29601 * ipa-inline.c: LIkewise.
29603 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29605 * ipa-inline.c (edge_badness): Use inlined_time instead of
29606 inline_summaries->get.
29608 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29610 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
29612 2017-05-22 Nathan Sidwell <nathan@acm.org>
29614 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
29615 (fdump-lang): Document 'raw' option.
29616 * dumpfile.h (TDI_tu): Delete.
29617 * dumpfile.c (dump_files): Remove translation-unit.
29618 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
29620 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
29622 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
29623 command option from $(AWK) call.
29624 * config/avr/genmultilib.awk: Simplify and rewrite so that it
29625 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
29626 [FORMAT]: Remove handling of variable.
29627 * config/avr/t-multilib: Regenerate.
29629 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29631 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
29633 (dump_inline_summary): Do not print self_time.
29634 (estimate_function_body_sizes): Do not set self_time.
29635 (compute_inline_parameters): Likewise.
29636 (inline_read_section, inline_write_summary): Do not stream self_time.
29637 * ipa-inline.h (inline_summary): Drop self_time.
29639 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29641 * ipa-inline-analysis.c (account_size_time): Rename to ...
29642 (inline_summary::account_size_time): ... this one.
29643 (reset_ipa_call_summary): Turn to ...
29644 (ipa_call_summary::reset): ... this one.
29645 (reset_inline_summary): Turn to ...
29646 (inline_summary::reset): ... this one.
29647 (inline_summary_t::remove): Update.
29648 (inline_summary_t::duplicate): Update.
29649 (ipa_call_summary_t::remove): Update.
29650 (dump_inline_summary): Update.
29651 (estimate_function_body_sizes): Update.
29652 (compute_inline_parameters): Update.
29653 (estimate_node_size_and_time): Update.
29654 (inline_merge_summary): Update.
29655 (inline_update_overall_summary): Update.
29656 (inline_read_section): Update.
29657 (inline_write_summary): Update.
29658 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
29659 add account_size_time and reset member functions.
29660 (ipa_call_summary): Add reset function.
29661 * ipa-predicate.h (predicate::operator &): Constify.
29663 2017-05-22 Richard Biener <rguenther@suse.de>
29665 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
29667 2017-05-19 Jason Merrill <jason@redhat.com>
29669 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
29671 2017-05-19 Marek Polacek <polacek@redhat.com>
29674 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
29675 TYPE_OVERFLOW_WRAPS checks.
29677 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
29679 * tree-core.h (enum omp_clause_default_kind): Add
29680 "OMP_CLAUSE_DEFAULT_PRESENT".
29681 * tree-pretty-print.c (dump_omp_clause): Handle it.
29682 * gimplify.c (enum gimplify_omp_var_data): Add
29683 "GOVD_MAP_FORCE_PRESENT".
29684 (gimplify_adjust_omp_clauses_1): Map it to
29685 "GOMP_MAP_FORCE_PRESENT".
29686 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
29688 * gimplify.c (oacc_default_clause): Clarify.
29690 2017-05-19 Nathan Sidwell <nathan@acm.org>
29692 LANG_HOOK_REGISTER_DUMPS
29693 * toplev.c (general_init): Call register dump lang hook.
29694 * doc/invoke.texi: Document -fdump-lang option family.
29695 * dumpfile.c (dump_files): Remove class dump here.
29696 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
29697 * dumpfile.h (tree_dump_index): Remove TDI_class.
29698 * langhooks-def.h (lhd_register_dumps): Declare.
29699 (LANG_HOOKS_REGISTER_DUMPS): Define.
29700 (LANG_HOOKS_INITIALIZER): Add it.
29701 * langhooks.c (lhd_register_dumps): Define.
29702 * langhooks.h (struct lang_hooks): Add register_dumps.
29704 2017-05-19 Nathan Sidwell <nathan@acm.org>
29706 * context.h (context::set_passes): New.
29707 * context.c (context::context): Do not create pass manager.
29708 * toplev.c (general_init): Create pass manager here.
29710 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
29712 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
29713 use this splitter if two add or or instructions would also work for
29714 the constant we want to generate.
29716 2017-05-19 Richard Biener <rguenther@suse.de>
29719 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
29720 predicate evaluation.
29722 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29724 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
29726 * ipa-inline.c (want_inline_small_function_p): Do not cast to
29729 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29731 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29732 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29733 (inline_edge_summary_vec): Turn into ...
29734 (ipa_call_summaries): ... this one.
29735 (redirect_to_unreachable, edge_set_predicate,
29736 evaluate_properties_for_edge, inline_summary_alloc,
29737 reset_ipa_call_summary, reset_inline_summary,
29738 inline_summary_t::duplicate): Update.
29739 (inline_edge_duplication_hook): Turn to ...
29740 (ipa_call_summary_t::duplicate): ... this one.
29741 (inline_edge_removal_hook): Turn to ...
29742 (ipa_call_summary_t::remove): ... this one.
29743 (dump_inline_edge_summary): Turn to ...
29744 (dump_ipa_call_summary): ... this one.
29745 (estimate_function_body_sizes): Update.
29746 (inline_update_callee_summaries): Update.
29747 (remap_edge_change_prob): Update.
29748 (remap_edge_summaries): Update.
29749 (inline_merge_summary): Update.
29750 (do_estimate_edge_time): Update.
29751 (inline_generate_summary): Update.
29752 (inline_read_section): Update.
29753 (inline_read_summary): Update.
29754 (inline_free_summary): Update.
29755 * ipa-inline.c (can_inline_edge_p): Update.
29756 (compute_inlined_call_time): Update.
29757 (want_inline_small_function_p): Update.
29758 (edge_badness): Update.
29759 (early_inliner): Update.
29760 * ipa-inline.h (inline_edge_summary): Turn to ...
29761 (ipa_call_summary): ... this one.
29762 (ipa_call_summary_t): New class.
29763 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29764 (ipa_call_summaries): New.
29765 (inline_edge_summary): Remove.
29766 (estimate_edge_growth): Update.
29767 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29768 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29769 * ipa-split.c (execute_split_functions): Update.
29770 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29772 2017-05-19 Richard Biener <rguenther@suse.de>
29774 PR middle-end/80764
29775 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
29777 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
29779 * config/rs6000/rs6000.c (struct machine_function): Add field
29780 fpr_is_wrapped_separately.
29781 (rs6000_get_separate_components): Use 64 components. Handle the
29782 new FPR components.
29783 (rs6000_components_for_bb): Handle the FPR components.
29784 (rs6000_emit_prologue_components): Handle the FPR components.
29785 (rs6000_emit_epilogue_components): Handle the FPR components.
29786 (rs6000_set_handled_components): Handle the FPR components.
29787 (rs6000_emit_prologue): Don't output prologue code for those FPRs
29788 that are already separately shrink-wrapped.
29789 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
29790 that are already separately shrink-wrapped.
29792 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29795 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
29798 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
29799 (define_peephole2 for Altivec d-form load): Add peepholes to catch
29800 cases where the register allocator uses a move and an offsettable
29801 memory operation to/from a FPR register on ISA 2.06/2.07.
29802 (define_peephole2 for Altivec d-form store): Likewise.
29804 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
29807 * config/i386/mmx.md (*mov<mode>_internal): Enable
29808 alternatives 11, 12, 13 and 14 also for 32bit targets.
29809 Remove alternatives 15, 16, 17 and 18.
29810 * config/i386/sse.md (vec_concatv2di): Change
29811 alternative (!x, *y) to (x, ?!*Yn).
29813 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
29815 * dumpfile.h (enum dump_kind): Remove stray comma.
29817 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29819 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
29820 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
29821 predicate::num_conditions
29822 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
29823 (CHANGED): turn into predicate::changed.
29824 (agg_position_info): Move to ipa-predicate.h
29825 (add_condition, predicate::add_clause, predicate::operator &=,
29826 predicate::or_with, predicate::evaluate, predicate::probability,
29827 dump_condition, dump_clause, predicate::dump,
29828 predicate::remap_after_duplication, predicate::remap_after_inlining,
29829 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
29830 (evaluate_conditions_for_known_args): Update.
29831 (set_cond_stmt_execution_predicate): Update.
29832 * ipa-inline.h: Include ipa-predicate.h
29833 (condition, inline_param_summary, conditions, agg_position_info,
29834 predicate): Move to ipa-predicate.h
29835 * ipa-predicate.c: New file.
29836 * ipa-predicate.h: New file.
29838 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
29840 * final.c (leaf_function_p): Check we are not in a sequence.
29842 2017-05-18 Martin Liska <mliska@suse.cz>
29844 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
29845 * dumpfile.c (dump_register): Use new enum dump_kind.
29846 (get_dump_file_name): Likewise.
29847 (dump_enable_all): Likewise.
29848 (dump_switch_p_1): Likewise.
29849 (enable_rtl_dump_file): Remove usage of TDF_RTL.
29850 * dumpfile.h (enum dump_kind): New enum type.
29851 (struct dump_file_info): Create constructor and
29852 format fields and comments.
29853 * passes.c (pass_manager::register_one_dump_file):
29855 * statistics.c (statistics_early_init): Likewise.
29856 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
29857 TDF_TREE with TDF_SLIM.
29858 (gather_memory_references_ref): Likewise.
29860 2017-05-18 Martin Liska <mliska@suse.cz>
29862 * vec.h (struct vnull): Use it.
29864 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29866 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
29867 (true_predicate, false_predicate, true_predicate_p,
29868 false_predicate_p): Remove.
29869 (single_cond_predicate, not_inlined_predicate): Turn to member function
29871 (add_condition): Update.
29872 (add_clause): Turn to...
29873 (predicate::add_clause): ... this one; update; allow passing NULL
29875 (and_predicates): Turn to ...
29876 (predicate::operator &=): ... this one.
29877 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
29878 (or_predicates): Turn to ...
29879 (predicate::or_with): ... this one.
29880 (evaluate_predicate): Turn to ...
29881 (predicate::evaluate): ... this one.
29882 (predicate_probability): Turn to ...
29883 (predicate::probability): ... this one.
29884 (dump_condition): Update.
29885 (dump_predicate): Turn to ...
29886 (predicate::dump): ... this one.
29887 (account_size_time): Update.
29888 (edge_set_predicate): Update.
29889 (set_hint_predicate): UPdate.
29890 (evaluate_conditions_for_known_args): Update.
29891 (evaluate_properties_for_edge): Update.
29892 (remap_predicate_after_duplication): Turn to...
29893 (predicate::remap_after_duplication): ... this one.
29894 (remap_hint_predicate_after_duplication): Update.
29895 (inline_summary_t::duplicate): UPdate.
29896 (dump_inline_edge_summary): Update.
29897 (dump_inline_summary): Update.
29898 (set_cond_stmt_execution_predicate): Update.
29899 (set_switch_stmt_execution_predicate): Update.
29900 (compute_bb_predicates): Update.
29901 (will_be_nonconstant_expr_predicate): Update.
29902 (will_be_nonconstant_predicate): Update.
29903 (phi_result_unknown_predicate): Update.
29904 (predicate_for_phi_result): Update.
29905 (array_index_predicate): Update.
29906 (estimate_function_body_sizes): Update.
29907 (estimate_node_size_and_time): Update.
29908 (estimate_ipcp_clone_size_and_time): Update.
29909 (remap_predicate): Rename to ...
29910 (predicate::remap_after_inlining): ... this one.
29911 (remap_hint_predicate): Update.
29912 (inline_merge_summary): Update.
29913 (inline_update_overall_summary): Update.
29914 (estimate_size_after_inlining): Update.
29915 (read_predicate): Rename to ...
29916 (predicate::stream_in): ... this one.
29917 (read_inline_edge_summary): Update.
29918 (write_predicate): Rename to ...
29919 (predicate::stream_out): ... this one.
29920 (write_inline_edge_summary): Update.
29921 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
29922 (clause_t): Turn to uint32_t
29923 (predicate): Turn to class; implement constructor and operators
29925 (size_time_entry): Update.
29926 (inline_summary): Update.
29927 (inline_edge_summary): Update.
29929 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
29931 * fold-const.c (fold_binary_loc): Move transformation...
29932 * match.pd (C - X CMP X): ... here.
29934 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
29936 * config/sparc/sparc.c (sparc_option_override): Set function
29937 alignment for -mcpu=niagara7 to 64 to match the I$ line.
29938 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
29940 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
29942 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
29944 2017-05-18 Marek Polacek <polacek@redhat.com>
29947 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
29948 (pass_ubsan::execute): Call gimple_assign_single_p instead of
29949 gimple_assign_load_p.
29951 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
29953 PR middle-end/80692
29954 * real.c (do_compare): Give decimal_do_compare preference over
29955 comparing just the signs.
29957 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
29959 * doc/md.texi (Canonicalization of Instructions): Describe the
29960 canonical form of instructions that inherently set a condition
29963 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
29965 PR middle-end/80775
29966 * tree-cfg.c: Move deletion of unreachable case statements to after
29967 the merging of consecutive case labels.
29969 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
29971 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
29972 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
29973 restoring of callee-saved registers.
29975 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
29977 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
29978 * config/visium/visium.c (single_set_and_flags): Likewise.
29979 * config/visium/visium.md (Substitutions): Likewise.
29981 2017-05-17 Martin Liska <mliska@suse.cz>
29983 * cfg.c: Introduce dump_flags_t type and
29984 use it instead of int type.
29986 * cfghooks.c: Likewise.
29987 * cfghooks.h (struct cfg_hooks): Likewise.
29988 * cfgrtl.c: Likewise.
29989 * cfgrtl.h: Likewise.
29990 * cgraph.c (cgraph_node::get_body): Likewise.
29991 * coretypes.h: Likewise.
29992 * domwalk.c: Likewise.
29993 * domwalk.h: Likewise.
29994 * dumpfile.c (struct dump_option_value_info): Likewise.
29995 (dump_enable_all): Likewise.
29996 (dump_switch_p_1): Likewise.
29997 (opt_info_switch_p): Likewise.
29998 * dumpfile.h (enum tree_dump_index): Likewise.
29999 (struct dump_file_info): Likewise.
30000 * genemit.c: Likewise.
30001 * generic-match-head.c: Likewise.
30002 * gengtype.c (open_base_files): Likewise.
30003 * gimple-pretty-print.c: Likewise.
30004 * gimple-pretty-print.h: Likewise.
30005 * graph.c (print_graph_cfg): Likewise.
30006 * graphite-scop-detection.c (dot_all_sese): Likewise.
30007 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
30008 * loop-unroll.c (report_unroll): Likewise.
30009 * passes.c (pass_manager::register_one_dump_file): Likewise.
30010 * print-tree.c: Likewise.
30011 * statistics.c: Likewise.
30012 * tree-cfg.c: Likewise.
30013 * tree-cfg.h: Likewise.
30014 * tree-dfa.c: Likewise.
30015 * tree-dfa.h: Likewise.
30016 * tree-dump.c (dump_function): Likewise.
30017 * tree-dump.h (struct dump_info): Likewise.
30018 * tree-pretty-print.c: Likewise.
30019 * tree-pretty-print.h: Likewise.
30020 * tree-ssa-live.c: Likewise.
30021 * tree-ssa-live.h: Likewise.
30022 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
30023 * tree-vect-loop.c: Likewise.
30024 * tree-vect-slp.c: Likewise.
30026 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
30027 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30029 PR tree-optimization/80457
30030 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
30031 of all arguments to a statement as scalar_to_vec operations.
30032 (vectorizable_call): Adjust call to vect_model_simple_cost for
30034 (vectorizable_conversion): Likewise.
30035 (vectorizable_assignment): Likewise.
30036 (vectorizable_shift): Likewise.
30037 (vectorizable_operation): Likewise.
30038 (vectorizable_comparison): Likewise.
30039 (vect_is_simple_cond): Record the def types for operands.
30040 (vectorizable_condition): Likewise, call vect_model_simple_cost.
30041 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
30042 for statement argument count.
30044 2017-05-16 Carl Love <cel@us.ibm.com>
30046 * config/rs6000/rs6000-c: Add support for built-in functions
30047 vector unsigned long long vec_bperm (vector unsigned long long,
30048 vector unsigned char)
30049 vector signed long long vec_mule (vector signed int,
30051 vector unsigned long long vec_mule (vector unsigned int,
30052 vector unsigned int)
30053 vector signed long long vec_mulo (vector signed int,
30055 vector unsigned long long vec_mulo (vector unsigned int,
30056 vector unsigned int)
30057 vector signed char vec_sldw (vector signed char,
30058 vector signed char,
30060 vector unsigned char vec_sldw (vector unsigned char,
30061 vector unsigned char,
30063 vector signed short vec_sldw (vector signed short,
30064 vector signed short,
30066 vector unsigned short vec_sldw (vector unsigned short,
30067 vector unsigned short,
30069 vector signed int vec_sldw (vector signed int,
30072 vector unsigned int vec_sldw (vector unsigned int,
30073 vector unsigned int,
30075 vector signed long long vec_sldw (vector signed long long,
30076 vector signed long long,
30078 vector unsigned long long vec_sldw (vector unsigned long long,
30079 vector unsigned long long,
30081 * config/rs6000/rs6000-c: Add support for built-in functions
30082 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
30083 * config/rs6000/altivec.h: Add defintion for vec_sldw.
30084 * doc/extend.texi: Update the built-in documentation for the
30085 new built-in functions.
30087 2017-05-16 Marek Polacek <polacek@redhat.com>
30091 * tree.c (save_expr): Don't fold the expression.
30093 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
30095 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
30096 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
30097 and (?*y,m). Update insn attributes.
30099 2017-05-16 Martin Liska <mliska@suse.cz>
30101 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
30102 flags argument of print_gimple_stmt, print_gimple_expr,
30103 print_generic_stmt and print_generic_expr.
30104 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
30105 * coretypes.h: Likewise.
30106 * except.c (dump_eh_tree): Likewise.
30107 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
30108 * gimple-pretty-print.h: Likewise.
30109 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
30110 (backprop::push_to_worklist): Likewise.
30111 (backprop::pop_from_worklist): Likewise.
30112 (backprop::process_use): Likewise.
30113 (backprop::intersect_uses): Likewise.
30114 (note_replacement): Likewise.
30115 * gimple-ssa-store-merging.c
30116 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
30117 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
30118 (pass_store_merging::execute): Likewise.
30119 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
30120 (ssa_base_cand_dump_callback): Likewise.
30121 (dump_incr_vec): Likewise.
30122 (replace_refs): Likewise.
30123 (replace_mult_candidate): Likewise.
30124 (create_add_on_incoming_edge): Likewise.
30125 (create_phi_basis): Likewise.
30126 (insert_initializers): Likewise.
30127 (all_phi_incrs_profitable): Likewise.
30128 (introduce_cast_before_cand): Likewise.
30129 (replace_one_candidate): Likewise.
30130 * gimplify.c (gimplify_expr): Likewise.
30131 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
30132 (set_rename): Likewise.
30133 (rename_uses): Likewise.
30134 (copy_loop_phi_nodes): Likewise.
30135 (add_close_phis_to_merge_points): Likewise.
30136 (copy_loop_close_phi_args): Likewise.
30137 (copy_cond_phi_args): Likewise.
30138 (graphite_copy_stmts_from_block): Likewise.
30139 (translate_pending_phi_nodes): Likewise.
30140 * graphite-poly.c (print_pdr): Likewise.
30141 (dump_gbb_cases): Likewise.
30142 (dump_gbb_conditions): Likewise.
30143 (print_scop_params): Likewise.
30144 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
30145 (build_cross_bb_scalars_use): Likewise.
30146 (gather_bbs::before_dom_children): Likewise.
30147 * hsa-dump.c (dump_hsa_immed): Likewise.
30148 * ipa-cp.c (print_ipcp_constant_value): Likewise.
30149 (get_replacement_map): Likewise.
30150 * ipa-inline-analysis.c (dump_condition): Likewise.
30151 (estimate_function_body_sizes): Likewise.
30152 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
30153 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
30154 * ipa-prop.c (ipa_dump_param): Likewise.
30155 (ipa_print_node_jump_functions_for_edge): Likewise.
30156 (ipa_modify_call_arguments): Likewise.
30157 (ipa_modify_expr): Likewise.
30158 (ipa_dump_param_adjustments): Likewise.
30159 (ipa_dump_agg_replacement_values): Likewise.
30160 (ipcp_modif_dom_walker::before_dom_children): Likewise.
30161 * ipa-pure-const.c (check_stmt): Likewise.
30162 (pass_nothrow::execute): Likewise.
30163 * ipa-split.c (execute_split_functions): Likewise.
30164 * omp-offload.c (dump_oacc_loop_part): Likewise.
30165 (dump_oacc_loop): Likewise.
30166 * trans-mem.c (tm_log_emit): Likewise.
30167 (tm_memopt_accumulate_memops): Likewise.
30168 (dump_tm_memopt_set): Likewise.
30169 (dump_tm_memopt_transform): Likewise.
30170 * tree-cfg.c (gimple_verify_flow_info): Likewise.
30171 (print_loop): Likewise.
30172 * tree-chkp-opt.c (chkp_print_addr): Likewise.
30173 (chkp_gather_checks_info): Likewise.
30174 (chkp_get_check_result): Likewise.
30175 (chkp_remove_check_if_pass): Likewise.
30176 (chkp_use_outer_bounds_if_possible): Likewise.
30177 (chkp_reduce_bounds_lifetime): Likewise.
30178 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
30179 (chkp_mark_completed_bounds): Likewise.
30180 (chkp_register_incomplete_bounds): Likewise.
30181 (chkp_mark_invalid_bounds): Likewise.
30182 (chkp_maybe_copy_and_register_bounds): Likewise.
30183 (chkp_build_returned_bound): Likewise.
30184 (chkp_get_bound_for_parm): Likewise.
30185 (chkp_build_bndldx): Likewise.
30186 (chkp_get_bounds_by_definition): Likewise.
30187 (chkp_generate_extern_var_bounds): Likewise.
30188 (chkp_get_bounds_for_decl_addr): Likewise.
30189 * tree-chrec.c (chrec_apply): Likewise.
30190 * tree-data-ref.c (dump_data_reference): Likewise.
30191 (dump_subscript): Likewise.
30192 (dump_data_dependence_relation): Likewise.
30193 (analyze_overlapping_iterations): Likewise.
30194 * tree-inline.c (expand_call_inline): Likewise.
30195 (tree_function_versioning): Likewise.
30196 * tree-into-ssa.c (dump_defs_stack): Likewise.
30197 (dump_currdefs): Likewise.
30198 (dump_names_replaced_by): Likewise.
30199 (dump_update_ssa): Likewise.
30200 (update_ssa): Likewise.
30201 * tree-object-size.c (pass_object_sizes::execute): Likewise.
30202 * tree-parloops.c (build_new_reduction): Likewise.
30203 (try_create_reduction_list): Likewise.
30204 (ref_conflicts_with_region): Likewise.
30205 (oacc_entry_exit_ok_1): Likewise.
30206 (oacc_entry_exit_single_gang): Likewise.
30207 * tree-pretty-print.h: Likewise.
30208 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
30209 (get_scalar_evolution): Likewise.
30210 (add_to_evolution): Likewise.
30211 (get_loop_exit_condition): Likewise.
30212 (analyze_evolution_in_loop): Likewise.
30213 (analyze_initial_condition): Likewise.
30214 (analyze_scalar_evolution): Likewise.
30215 (instantiate_scev): Likewise.
30216 (number_of_latch_executions): Likewise.
30217 (gather_chrec_stats): Likewise.
30218 (final_value_replacement_loop): Likewise.
30219 (scev_const_prop): Likewise.
30220 * tree-sra.c (dump_access): Likewise.
30221 (disqualify_candidate): Likewise.
30222 (create_access): Likewise.
30223 (reject): Likewise.
30224 (maybe_add_sra_candidate): Likewise.
30225 (create_access_replacement): Likewise.
30226 (analyze_access_subtree): Likewise.
30227 (analyze_all_variable_accesses): Likewise.
30228 (sra_modify_assign): Likewise.
30229 (initialize_constant_pool_replacements): Likewise.
30230 (find_param_candidates): Likewise.
30231 (decide_one_param_reduction): Likewise.
30232 (replace_removed_params_ssa_names): Likewise.
30233 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
30234 * tree-ssa-copy.c (dump_copy_of): Likewise.
30235 (copy_prop_visit_cond_stmt): Likewise.
30236 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
30237 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
30238 (record_equivalences_from_stmt): Likewise.
30239 * tree-ssa-dse.c (compute_trims): Likewise.
30240 (delete_dead_call): Likewise.
30241 (delete_dead_assignment): Likewise.
30242 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
30243 (forward_propagate_into_cond): Likewise.
30244 (pass_forwprop::execute): Likewise.
30245 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
30246 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
30248 (move_computations_worker): Likewise.
30249 (execute_sm): Likewise.
30250 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
30251 (remove_exits_and_undefined_stmts): Likewise.
30252 (remove_redundant_iv_tests): Likewise.
30253 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
30254 (adjust_iv_update_pos): Likewise.
30255 * tree-ssa-math-opts.c (bswap_replace): Likewise.
30256 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
30257 (value_replacement): Likewise.
30258 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
30259 * tree-ssa-pre.c (print_pre_expr): Likewise.
30260 (get_representative_for): Likewise.
30261 (create_expression_by_pieces): Likewise.
30262 (insert_into_preds_of_block): Likewise.
30263 (eliminate_insert): Likewise.
30264 (eliminate_dom_walker::before_dom_children): Likewise.
30265 (eliminate): Likewise.
30266 (remove_dead_inserted_code): Likewise.
30267 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
30268 * tree-ssa-reassoc.c (get_rank): Likewise.
30269 (eliminate_duplicate_pair): Likewise.
30270 (eliminate_plus_minus_pair): Likewise.
30271 (eliminate_not_pairs): Likewise.
30272 (undistribute_ops_list): Likewise.
30273 (eliminate_redundant_comparison): Likewise.
30274 (update_range_test): Likewise.
30275 (optimize_range_tests_var_bound): Likewise.
30276 (optimize_vec_cond_expr): Likewise.
30277 (rewrite_expr_tree): Likewise.
30278 (rewrite_expr_tree_parallel): Likewise.
30279 (linearize_expr): Likewise.
30280 (break_up_subtract): Likewise.
30281 (linearize_expr_tree): Likewise.
30282 (attempt_builtin_powi): Likewise.
30283 (attempt_builtin_copysign): Likewise.
30284 (transform_stmt_to_copy): Likewise.
30285 (transform_stmt_to_multiply): Likewise.
30286 (dump_ops_vector): Likewise.
30287 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
30288 (print_scc): Likewise.
30289 (set_ssa_val_to): Likewise.
30290 (visit_reference_op_store): Likewise.
30291 (visit_use): Likewise.
30292 (sccvn_dom_walker::before_dom_children): Likewise.
30293 (run_scc_vn): Likewise.
30294 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
30296 (expr_hash_elt::print): Likewise.
30297 (const_and_copies::pop_to_marker): Likewise.
30298 (const_and_copies::record_const_or_copy_raw): Likewise.
30299 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
30300 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
30301 (dump_predicates): Likewise.
30302 (find_uninit_use): Likewise.
30303 (warn_uninitialized_phi): Likewise.
30304 (pass_late_warn_uninitialized::execute): Likewise.
30305 * tree-ssa.c (verify_vssa): Likewise.
30306 (verify_ssa): Likewise.
30307 (maybe_optimize_var): Likewise.
30308 * tree-vrp.c (dump_value_range): Likewise.
30309 (dump_all_value_ranges): Likewise.
30310 (dump_asserts_for): Likewise.
30311 (register_edge_assert_for_2): Likewise.
30312 (vrp_visit_cond_stmt): Likewise.
30313 (vrp_visit_switch_stmt): Likewise.
30314 (vrp_visit_stmt): Likewise.
30315 (vrp_visit_phi_node): Likewise.
30316 (simplify_cond_using_ranges_1): Likewise.
30317 (fold_predicate_in): Likewise.
30318 (evrp_dom_walker::before_dom_children): Likewise.
30319 (evrp_dom_walker::push_value_range): Likewise.
30320 (evrp_dom_walker::pop_value_range): Likewise.
30321 (execute_early_vrp): Likewise.
30323 2017-05-16 Richard Biener <rguenther@suse.de>
30325 * dwarf2out.c (loc_list_from_tree_1): Do not create
30326 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
30328 2017-05-16 Richard Biener <rguenther@suse.de>
30330 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
30332 (note_variable_value_in_expr): If we resolved the decl ref
30333 do not push to the stack.
30335 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
30337 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
30338 operations in fast-math mode.
30339 (vaddq_f16): Likewise.
30340 (vmul_f16): Likewise.
30341 (vmulq_f16): Likewise.
30342 (vsub_f16): Likewise.
30343 (vsubq_f16): Likewise.
30344 * config/arm/neon.md (add<mode>3): New.
30346 (fma:<VH:mode>3): New. Also remove outdated comment.
30349 2017-05-16 Martin Liska <mliska@suse.cz>
30353 * ipa-devirt.c (warn_types_mismatch): Fix typo.
30354 (odr_types_equivalent_p): Likewise.
30356 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
30358 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
30360 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
30363 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
30364 non-interunit SSE move alternatives with '?'.
30365 (zero-extendsidi peephole2): New peephole to skip intermediate
30366 general register in SSE zero-extend sequence.
30368 2017-05-15 Jeff Law <law@redhat.com>
30370 * reorg.c (relax_delay_slots): Create a new variable to hold
30371 the temporary target rather than clobbering TARGET_LABEL.
30373 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
30374 missing argument to extract_bit_field call.
30375 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
30377 2017-05-15 Martin Liska <mliska@suse.cz>
30380 * gcc.c (process_command): Do not allow empty argument of -o option.
30382 2017-05-15 Renlin Li <renlin.li@arm.com>
30384 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
30385 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
30386 * config/aarch64/constraints.md (Usf): Add long call check.
30387 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
30388 (call_value): Likewise.
30389 (sibcall): Likewise.
30390 (sibcall_value): Likewise.
30392 (call_value_insn): New.
30393 (sibcall_insn): Update rtx pattern.
30394 (sibcall_value_insn): Likewise.
30395 (call_internal): Remove.
30396 (call_value_internal): Likewise.
30397 (sibcall_internal): Likewise.
30398 (sibcall_value_internal): Likewise.
30399 (call_reg): Likewise.
30400 (call_symbol): Likewise.
30401 (call_value_reg): Likewise.
30402 (call_value_symbol): Likewise.
30404 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
30407 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
30409 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
30411 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
30412 compatible with CCGOCmode and with CCZmode.
30414 2017-05-14 Martin Sebor <msebor@redhat.com>
30416 PR middle-end/77671
30417 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
30418 (gimple_fold_builtin_snprintf): Same.
30419 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
30420 (gimple_fold_builtin_snprintf): Same.
30421 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
30422 of character types.
30423 (is_call_safe): New function.
30424 (try_substitute_return_value): Call it.
30425 (try_simplify_call): New function.
30426 (pass_sprintf_length::handle_gimple_call): Call it.
30428 2017-05-14 Martin Sebor <msebor@redhat.com>
30430 PR middle-end/80669
30431 * builtins.c (expand_builtin_stpncpy): Simplify.
30433 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
30435 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
30436 * config/i386/i386.h
30437 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
30438 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
30439 (struct machine_function): Add new members call_ms2sysv,
30440 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
30441 (struct machine_frame_state): New fields sp_realigned and
30442 sp_realigned_offset.
30443 * config/i386/i386.c
30444 (enum xlogue_stub): New enum.
30445 (enum xlogue_stub_sets): New enum.
30446 (class xlogue_layout): New class.
30447 (struct ix86_frame): New fields stack_realign_allocate_offset,
30448 stack_realign_offset and outlined_save_offset. Modify comments to
30449 detail stack layout when using out-of-line stubs.
30450 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
30451 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
30452 -mcall-ms2sysv-xlogues.
30453 (stub_managed_regs): New static variable.
30454 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
30455 registers managed by out-of-line stub.
30456 (disable_call_ms2sysv_xlogues): New function.
30457 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
30458 m->call_ms2sysv when appropriate and compute frame layout for
30460 (sp_valid_at, fp_valid_at): New inline functions.
30461 (choose_basereg): New function.
30462 (choose_baseaddr): Add align parameter, use choose_basereg and modify
30464 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
30465 Use align parameter of choose_baseaddr to generated aligned SSE movs
30467 (pro_epilogue_adjust_stack): Modify to track
30468 machine_frame_state::sp_realigned.
30469 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
30470 (ix86_nsaved_sseregs): Likewise.
30471 (ix86_emit_save_regs): Likewise.
30472 (ix86_emit_save_regs_using_mov): Likewise.
30473 (ix86_emit_save_sse_regs_using_mov): Likewise.
30474 (get_scratch_register_on_entry): Likewise.
30475 (gen_frame_set): New function.
30476 (gen_frame_load): Likewise.
30477 (gen_frame_store): Likewise.
30478 (emit_outlined_ms2sysv_save): Likewise.
30479 (emit_outlined_ms2sysv_restore): Likewise.
30480 (ix86_expand_prologue): Modify stack re-alignment code and call
30481 emit_outlined_ms2sysv_save when appropriate.
30482 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
30483 parameter rtx_insn *insn, which allows the function to be used to only
30484 generate the notes.
30485 (ix86_expand_epilogue): Modify validity checks of frame and stack
30486 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
30487 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
30488 * config/i386/predicates.md
30489 (save_multiple): New predicate.
30490 (restore_multiple): Likewise.
30491 * config/i386/sse.md
30492 (save_multiple<mode>): New pattern.
30493 (save_multiple_realign<mode>): Likewise.
30494 (restore_multiple<mode>): Likewise.
30495 (restore_multiple_and_return<mode>): Likewise.
30496 (restore_multiple_leave_return<mode>): Likewise.
30497 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
30499 2017-05-14 Julia Koval <julia.koval@intel.com>
30501 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
30502 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
30503 (__builtin_ia32_xsetbv): New builtins.
30504 * config/i386/i386.c (ix86_expand_special_args_builtin):
30506 (ix86_expand_builtin): Special expand for new intrinsics.
30507 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
30508 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
30509 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
30511 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30513 * cfganal.c (inverted_post_order_compute): Change argument type
30515 * cfganal.h (inverted_post_order_compute): Adjust prototype.
30516 * df-core.c (rest_of_handle_df_initialize): Adjust.
30517 (rest_of_handle_df_finish): Likewise.
30518 (df_analyze_1): Likewise.
30519 (df_analyze): Likewise.
30520 (loop_inverted_post_order_compute): Change argument to be a vec *.
30521 (df_analyze_loop): Adjust.
30522 (df_get_n_blocks): Likewise.
30523 (df_get_postorder): Likewise.
30524 * df.h (struct df_d): Change field to be a vec.
30525 * lcm.c (compute_laterin): Adjust.
30526 (compute_available): Likewise.
30527 * lra-lives.c (lra_create_live_ranges_1): Likewise.
30528 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
30529 * tree-ssa-pre.c (compute_antic): Likewise.
30531 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30533 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
30534 (depth_first_search::depth_first_search): Change structure init
30535 function to this constructor.
30536 (depth_first_search::add_bb): Rename function to this member.
30537 (depth_first_search::execute): Likewise.
30538 (flow_dfs_compute_reverse_finish): Adjust.
30540 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30542 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
30543 (longest_simple_path): Likewise.
30544 * shrink-wrap.c (spread_components): Likewise.
30545 (disqualify_problematic_components): Likewise.
30546 (emit_common_heads_for_components): Likewise.
30547 (emit_common_tails_for_components): Likewise.
30548 (insert_prologue_epilogue_for_components): Likewise.
30550 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30552 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
30555 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30557 * df-core.c (df_set_blocks): Start using auto_bitmap.
30558 (df_compact_blocks): Likewise.
30559 * df-problems.c (df_rd_confluence_n): Likewise.
30560 * df-scan.c (df_insn_rescan_all): Likewise.
30561 (df_process_deferred_rescans): Likewise.
30562 (df_update_entry_block_defs): Likewise.
30563 (df_update_exit_block_uses): Likewise.
30564 (df_entry_block_bitmap_verify): Likewise.
30565 (df_exit_block_bitmap_verify): Likewise.
30566 (df_scan_verify): Likewise.
30567 * lra-constraints.c (lra_constraints): Likewise.
30568 (undo_optional_reloads): Likewise.
30569 (lra_undo_inheritance): Likewise.
30570 * lra-remat.c (calculate_gen_cands): Likewise.
30571 (do_remat): Likewise.
30572 * lra-spills.c (assign_spill_hard_regs): Likewise.
30573 (spill_pseudos): Likewise.
30574 * tree-ssa-pre.c (bitmap_set_and): Likewise.
30575 (bitmap_set_subtract_values): Likewise.
30577 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30579 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
30580 management with auto_bitmap.
30581 (fix_inter_tick): Likewise.
30582 (fix_recovery_deps): Likewise.
30583 * ira.c (add_store_equivs): Likewise.
30584 (find_moveable_pseudos): Likewise.
30585 (split_live_ranges_for_shrink_wrap): Likewise.
30586 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
30587 (rtx_reuse_manager::seen_def_p): Likewise.
30588 (rtx_reuse_manager::set_seen_def): Likewise.
30589 * print-rtl.h (class rtx_reuse_manager): Likewise.
30591 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30593 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
30595 (migrate_btr_def): Likewise.
30596 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
30597 * df-core.c (loop_post_order_compute): Likewise.
30598 (loop_inverted_post_order_compute): Likewise.
30599 * hsa-common.h: Likewise.
30600 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
30601 * init-regs.c (initialize_uninitialized_regs): Likewise.
30602 * ipa-inline.c (resolve_noninline_speculation): Likewise.
30603 (inline_small_functions): Likewise.
30604 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
30605 * ira.c (combine_and_move_insns): Likewise.
30606 (build_insn_chain): Likewise.
30607 * loop-invariant.c (find_invariants): Likewise.
30608 * lower-subreg.c (propagate_pseudo_copies): Likewise.
30609 * predict.c (tree_predict_by_opcode): Likewise.
30610 (predict_paths_leading_to): Likewise.
30611 (predict_paths_leading_to_edge): Likewise.
30612 (estimate_loops_at_level): Likewise.
30613 (estimate_loops): Likewise.
30614 * shrink-wrap.c (try_shrink_wrapping): Likewise.
30615 (spread_components): Likewise.
30616 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
30617 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
30618 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
30619 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
30620 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
30621 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
30622 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
30623 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
30624 (mark_threaded_blocks): Likewise.
30625 (thread_through_all_blocks): Likewise.
30626 * tree-ssa.c (verify_ssa): Likewise.
30627 (execute_update_addresses_taken): Likewise.
30628 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
30630 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30632 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
30634 (post_order_compute): Likewise.
30635 (inverted_post_order_compute): Likewise.
30636 (pre_and_rev_post_order_compute_fn): Likewise.
30638 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30640 * genrecog.c (int_set::int_set): Explicitly construct our
30641 auto_vec base class.
30642 * vec.h (auto_vec::auto_vec): New constructor.
30644 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30646 * bitmap.h (class auto_bitmap): New constructor taking
30647 bitmap_obstack * argument.
30649 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30651 * bitmap.h (class auto_bitmap): Change type of m_bits to
30652 bitmap_head, and adjust ctor / dtor and member operators.
30654 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
30656 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
30657 when returned register mode doesn't match original mode.
30659 2017-05-12 Jeff Law <law@redhat.com>
30660 Jakub Jelinek <jakub@redhat.com>
30662 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
30663 we look for cc setter after the compare-elim changes.
30664 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
30665 within the vector to match what compare-elim now expects.
30666 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
30667 (xorsi3_flags, one_cmplsi2_flags): Likewise.
30669 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
30670 after the compare-elim changes.
30671 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
30672 the vector to match what compare-elim now expects.
30673 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
30674 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
30675 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
30676 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
30677 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
30679 * config/visium/visium.c (single_set_and_flags): Fix where
30680 we look for cc setter after the compare-elim changes.
30681 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
30682 with the vector to match what compare-elim now expects.
30683 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
30684 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
30685 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
30686 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
30687 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
30688 (neg<mode>2_insn_set_overflow): Likewise.
30690 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
30692 PR middle-end/79794
30693 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
30694 maybe_expand_insn call, set ops[0].target. If still set after call,
30695 set alt_rtl. Add extra arg to recursive calls.
30696 (extract_bit_field): Add alt_rtl argument. Pass to
30698 * expmed.h (extract_bit_field): Fix prototype.
30699 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
30700 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
30701 to extract_bit_field_calls.
30702 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
30703 Pass alt_rtl to extract_bit_field calls.
30704 * calls.c (store_unaligned_arguments_into_psuedos)
30705 load_register_parameters): Pass extra NULL to extract_bit_field calls.
30706 * optabs.c (maybe_legitimize_operand): Clear op->target when call
30708 * optabs.h (struct expand_operand): Add target bitfield.
30710 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30712 * compare-elim.c (try_eliminate_compare): Canonicalize
30713 operation with embedded compare to
30714 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
30715 (set (reg) (operation)].
30717 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
30719 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30722 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
30723 cost of adding a carry flag for ADC instruction.
30724 [case MINUS]: Ignore the cost of subtracting a carry flag
30725 for SBB instruction.
30727 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
30729 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
30731 * config/rs6000/bmiintrin.h: New file.
30732 * config/rs6000/bmi2intrin.h: New file.
30733 * config/rs6000/x86intrin.h: New file.
30735 2017-05-12 Jeff Law <law@redhat.com>
30737 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
30740 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
30742 PR middle-end/80707
30743 * tree-cfg.c: Remove cfg edges of unreachable case statements.
30745 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30747 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30748 early expansion of vector divide builtins.
30749 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
30750 builtins identified as having unsigned arguments.
30752 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30754 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
30755 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
30756 expansion of vector logical operations (and, andc, or, xor,
30759 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30761 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
30762 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
30764 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30766 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30767 early GIMPLE expansion of vector multiplies.
30769 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30771 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
30772 TARGET_HAVE_MOVT conditional.
30773 (movt splitter): Likewise.
30775 2017-05-12 Richard Biener <rguenther@suse.de>
30777 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
30778 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30779 Fold all stmts not inplace.
30781 2017-05-12 Richard Biener <rguenther@suse.de>
30783 PR tree-optimization/80713
30784 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
30785 inserted_exprs bit for not removed stmts.
30787 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
30789 PR middle-end/69921
30790 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
30791 parallelized" attribute for parallelized OpenACC kernels.
30792 * omp-offload.c (execute_oacc_device_lower): Use it.
30794 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
30795 Set "oacc kernels" attribute.
30796 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
30797 parameter. Adjust all users.
30798 (oacc_fn_attrib_kernels_p): Remove function.
30799 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
30800 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
30801 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
30802 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
30803 assert "oacc kernels" attribute is set.
30805 2017-05-11 Carl Love <cel@us.ibm.com>
30807 * config/rs6000/rs6000-c: Add support for built-in functions
30808 vector unsigned char vec_popcnt (vector signed char)
30809 vector unsigned char vec_popcnt (vector unsigned char)
30810 vector unsigned short vec_popcnt (vector signed short)
30811 vector unsigned short vec_popcnt (vector unsigned short)
30812 vector unsigned int vec_popcnt (vector signed int)
30813 vector unsigned int vec_popcnt (vector unsigned int)
30814 vector unsigned long long vec_popcnt (vector signed long long)
30815 vector unsigned long long vec_popcnt (vector unsigned long long)
30816 vector signed long long vec_slo (vector signed long long,
30817 vector signed char)
30818 vector signed long long vec_slo (vector signed long long,
30819 vector unsigned char)
30820 vector unsigned long long vec_slo (vector unsigned long long,
30821 vector signed char)
30822 vector unsigned long long vec_slo (vector unsigned long long,
30823 vector unsigned char)
30824 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
30825 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
30826 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
30827 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
30828 * doc/extend.texi: Update the built-in documentation file for the
30829 new built-in functions.
30831 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
30833 * attribs.h (sorted_attr_string): Move machine independent
30834 functions for target clone support from the i386 port to common
30835 code. Rename ix86_function_versions to common_function_versions.
30836 Rename make_name to make_unique_name.
30837 (common_function_versions): Likewise.
30838 (make_unique_name): Likewise.
30839 (make_dispatcher_decl): Likewise.
30840 (is_function_default_version): Likewise.
30841 * attribs.c (attr_strcmp): Likewise.
30842 (sorted_attr_string): Likewise.
30843 (common_function_versions): Likewise.
30844 (make_unique_name): Likewise.
30845 (make_dispatcher_decl): Likewise.
30846 (is_function_default_version): Likewise.
30847 * config/i386/i386.c (attr_strcmp): Likewise.
30848 (sorted_attr_string): Likewise.
30849 (ix86_function_versions): Likewise.
30850 (make_name): Likewise.
30851 (make_dispatcher_decl): Likewise.
30852 (is_function_default_version): Likewise.
30853 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
30855 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30858 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
30859 Account for direct move costs for vec_construct of integer
30862 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
30865 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
30866 (UNSPEC_STX_ATOMIC): Ditto.
30867 (loaddi_via_sse): New insn.
30868 (storedi_via_sse): Ditto.
30869 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
30870 Update corresponding peephole2 patterns.
30871 (atomic_storedi_fpu): Ditto.
30873 2017-05-11 Julia Koval <julia.koval@intel.com>
30875 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
30876 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
30878 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
30879 (__builtin_ia32_rsqrt14ss_mask): New builtins.
30880 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
30882 2017-05-11 Nathan Sidwell <nathan@acm.org>
30884 * graphite-poly.c: Include dumpfile.h.
30886 * dumpfle.h (dump_function): Declare here ...
30887 * tree-dump.h (dump_function): ... not here.
30888 * dumpfile.c: #include tree-cfg.h.
30889 (dump_function): Move here from ...
30890 * tree-dump.c (dump_function): ... here.
30891 * gimplify.c: #include splay-tree.h, not tree-dump.h.
30892 * graphite-poly.c: Don't include tree-dump.h.
30893 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
30894 * print-tree.c: Likewise.
30895 * stor-layout.c: Likewise.
30896 * tree-nested.c: Likewise.
30898 * dumpfile.c (dump_start): Use TDF_FLAGS.
30899 (dump_enable_all): Fix TDF_KIND check thinko.
30901 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
30903 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30904 array entries to represent two legal parameterizations of the
30905 overloaded __builtin_cmpb function, as represented by the
30906 P6_OV_BUILTIN_CMPB constant.
30907 (altivec_resolve_overloaded_builtin): Add special case handling
30908 for the __builtin_cmpb function, as represented by the
30909 P6_OV_BUILTIN_CMPB constant.
30910 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
30911 (BU_P6_64BIT_2): New macro.
30912 (BU_P6_OVERLOAD_2): New macro
30913 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
30914 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
30915 (CMPB): Add overload support to represent both 32-bit and 64-bit
30916 compare-bytes function.
30917 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
30918 support for TARGET_CMPB.
30919 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
30920 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30921 documentation of the __builtin_cmpb overloaded built-in function.
30923 2017-05-11 Richard Biener <rguenther@suse.de>
30925 PR tree-optimization/80705
30926 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
30927 bases are not vectorizable.
30929 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30931 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
30932 when counting register pressure.
30934 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30936 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
30937 (iv_ca_more_deps): Renamed to ...
30938 (iv_ca_compare_deps): ... this.
30939 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
30941 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30943 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
30945 (determine_group_iv_costs): ... here.
30946 (find_inv_vars_cb): Record inv var if it's not recorded before.
30948 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30950 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
30951 (get_shiftadd_cost): Ditto.
30953 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30955 * tree-ssa-address.c: Include header file.
30956 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
30958 (add_to_parts): Refactor.
30959 (addr_to_parts): New parameter. Update use of move_hint_to_base.
30960 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
30963 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30965 PR tree-optimization/53090
30966 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
30968 (extract_cond_operands): Detect condition with IV on both sides
30969 and return COMP_IV_EXPR_2.
30970 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
30971 (rewrite_use_compare): Simplify by removing call to function
30972 extract_cond_operands.
30974 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30976 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
30977 (extract_cond_operands): Detect condition comparing against non-
30978 invariant bound and return appropriate enum value.
30979 (find_interesting_uses_cond): Update use of extract_cond_operands.
30980 Handle its return value accordingly.
30981 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
30983 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30985 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
30986 nonlinear iv_use computation in loop invariant sensitive way.
30988 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30990 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
30991 (find_iv_candidates): Call relate_compare_use_with_all_cands.
30993 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30995 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
30996 (dump_cand): Support iv_cand.inv_exprs.
30997 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
30999 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
31002 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31004 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
31006 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
31007 as local function. Include necessary header files.
31008 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
31010 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31012 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
31014 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31016 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
31017 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
31018 RSHIFT_EXPR and BIT_NOT_EXPR.
31020 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31022 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
31023 (adjust_setup_cost): New parameter supporting round up adjustment.
31024 (struct address_cost_data): Delete.
31025 (force_expr_to_var_cost): Don't bound cost with spill_cost.
31026 (split_address_cost, ptr_difference_cost): Delete.
31027 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
31028 (struct ainc_cost_data): New struct.
31029 (get_address_cost_ainc): New function.
31030 (get_address_cost, get_computation_cost): Reimplement.
31031 (determine_group_iv_cost_address): Record inv_expr for all uses of
31033 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
31034 (iv_ca_has_deps): Reimplemented to ...
31035 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
31037 (iv_ca_extend): Call iv_ca_more_deps.
31039 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31041 * tree-ssa-address.c (struct mem_address): Move to header file.
31042 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
31043 * tree-ssa-address.h (struct mem_address): Move from C file.
31044 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
31046 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31048 * tree-affine.h (aff_combination_type): New interface.
31049 (aff_combination_zero_p): Remove static.
31050 (aff_combination_const_p): New interface.
31051 (aff_combination_singleton_var_p): New interfaces.
31053 2017-05-11 Richard Biener <rguenther@suse.de>
31055 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31056 Skip unreachable blocks and destinations.
31057 (eliminate): Move stmt removal and fixup ...
31058 (fini_eliminate): ... here. Skip inserted exprs.
31059 (pass_pre::execute): Move fini_pre after fini_eliminate.
31060 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
31061 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
31062 PRE to get rid of dead code that has invalid SSA form and
31063 split critical edges again.
31065 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31067 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31069 2017-05-11 Richard Biener <rguenther@suse.de>
31071 * passes.c (execute_function_todo): Verify loops if they are
31072 said to be up-to-date.
31073 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
31074 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
31076 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
31079 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
31080 handle calling assemble_external ourself.
31083 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
31084 modes with zero size. Enhance comment.
31086 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31088 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
31089 built-ins for vec_xl and vec_xst with short and char pointer
31092 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31094 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
31095 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
31096 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
31097 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
31098 (_mm_maskz_min_round_ss): New intrinsics.
31099 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
31100 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
31101 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
31102 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
31103 (__builtin_ia32_minss_mask_round): New builtins.
31104 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31105 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31106 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
31108 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
31109 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
31111 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
31114 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31116 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
31117 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
31118 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
31119 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
31120 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
31121 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
31122 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
31123 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31124 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31125 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
31126 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
31127 (__builtin_ia32_mulss_mask_round): New builtins.
31128 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31129 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31130 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
31132 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
31133 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31135 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31138 2017-05-10 Julia Koval <julia.koval@intel.com>
31140 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
31141 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
31142 (_mm256_setr_m128i): New intrinsics.
31144 2017-05-10 Julia Koval <julia.koval@intel.com>
31146 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
31147 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
31148 (_mm_maskz_rcp14_ss): New intrinsics.
31149 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
31150 (__builtin_ia32_rcp14ss_mask): New builtins.
31151 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
31153 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
31155 PR tree-optimization/51513
31156 * tree-cfg.c (gimple_seq_unreachable_p): New function.
31157 (assert_unreachable_fallthru_edge_p): Use it.
31158 (group_case_labels_stmt): Likewise.
31159 * tree-cfg.h: Prototype it.
31160 * stmt.c: Include cfghooks.h and tree-cfg.h.
31161 (emit_case_dispatch_table) <gap_label>: New local variable.
31162 Use it to fill dispatch table gaps.
31163 Test for default_label before updating probabilities.
31164 (expand_case) <default_label>: Remove unneeded initialization.
31165 Test for unreachable default case statement and remove its edge.
31166 Set default_label accordingly.
31167 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
31169 2017-05-10 Carl Love <cel@us.ibm.com>
31171 * config/rs6000/rs6000-c: Add support for built-in functions
31172 vector signed char vec_neg (vector signed char)
31173 vector signed short int vec_neg (vector short int)
31174 vector signed int vec_neg (vector signed int)
31175 vector signed long long vec_neg (vector signed long long)
31176 vector float vec_neg (vector float)
31177 vector double vec_neg (vector double)
31178 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
31180 * config/rs6000/altivec.h: Add define for vec_neg
31181 * doc/extend.texi: Update the built-in documentation for the
31182 new built-in functions.
31184 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31186 PR tree-optimization/77644
31187 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
31189 2017-05-10 Nathan Sidwell <nathan@acm.org>
31191 * dumpfile.h (TDI_lang_all): New.
31192 (TDF_KIND): New. Renumber others
31193 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
31195 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
31197 (get_dump_file_name): Adjust suffix generation.
31198 (dump_enable_all): Use TDF_KIND.
31199 * doc/invoke.texi (-fdump-lang-all): Document.
31201 * dumpfile.h: Tabify.
31203 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
31206 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
31207 Move member access before delete.
31209 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
31211 * tree-inline.c (expand_call_inline): Split block at stmt
31214 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
31217 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
31218 are now unused after splitting mov{sf,sd}_hardfloat.
31219 (f32_lr2): Likewise.
31220 (f32_lm): Likewise.
31221 (f32_lm2): Likewise.
31222 (f32_li): Likewise.
31223 (f32_li2): Likewise.
31224 (f32_lv): Likewise.
31225 (f32_sr): Likewise.
31226 (f32_sr2): Likewise.
31227 (f32_sm): Likewise.
31228 (f32_sm2): Likewise.
31229 (f32_si): Likewise.
31230 (f32_si2): Likewise.
31231 (f32_sv): Likewise.
31232 (f32_dm): Likewise.
31233 (f32_vsx): Likewise.
31234 (f32_av): Likewise.
31235 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
31236 For movsf, order stores so the VSX stores occur before the GPR
31237 store which encourages the register allocator to use a traditional
31238 FPR instead of a GPR. For movsd, order the stores so that the GPR
31239 store comes before the VSX stores to allow the power6 to work.
31240 This is due to the power6 not having a 32-bit integer store
31241 instruction from a FPR.
31242 (movsf_hardfloat): Likewise.
31243 (movsd_hardfloat): Likewise.
31245 2017-05-09 Martin Sebor <msebor@redhat.com>
31247 PR translation/80280
31248 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
31251 PR translation/80280
31252 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
31253 data member added in r247778.
31254 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
31256 2017-05-09 Nathan Sidwell <nathan@acm.org>
31258 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
31260 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
31263 2017-05-09 Marek Polacek <polacek@redhat.com>
31265 * doc/invoke.texi: Fix typo.
31267 2017-05-09 Richard Biener <rguenther@suse.de>
31269 * tree-vrp.c (vrp_val_is_max): Adjust comment.
31270 (vrp_val_is_min): Likewise.
31271 (set_value_range_to_value): Likewise.
31272 (set_value_range_to_nonnegative): Likewise.
31273 (gimple_assign_nonzero_p): Likewise.
31274 (gimple_stmt_nonzero_p): Likewise.
31275 (vrp_int_const_binop): Likewise. Remove unreachable case.
31276 (adjust_range_with_scev): Adjust comments.
31277 (compare_range_with_value): Likewise.
31278 (extract_range_from_phi_node): Likewise.
31279 (test_for_singularity): Likewise.
31281 2017-05-09 Richard Biener <rguenther@suse.de>
31283 * tree-vrp.c (get_single_symbol): Add assert that we don't
31284 get overflowed constants as invariant part.
31285 (compare_values_warnv): Add comment before the TREE_NO_WARNING
31286 checks. Use wi::cmp instead of recursing for integer constants.
31287 (compare_values): Just ignore whether we assumed undefined
31288 overflow instead of failing the compare.
31289 (extract_range_for_var_from_comparison_expr): Add comment before the
31290 TREE_NO_WARNING sets.
31291 (test_for_singularity): Likewise.
31292 (extract_range_from_comparison): Do not disable optimization
31293 when we assumed undefined overflow.
31294 (extract_range_basic): Remove init of unused var.
31296 2017-05-09 Richard Biener <rguenther@suse.de>
31298 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
31299 (extract_range_from_multiplicative_op_1): Adjust.
31300 (extract_range_from_binary_expr_1): Use int_const_binop.
31302 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
31305 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
31306 rs6000_store_data_bypass_p in seven define_bypass directives and
31307 in several comments.
31308 * config/rs6000/rs6000-protos.h: Add prototype for
31309 rs6000_store_data_bypass_p function.
31310 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
31311 function implements slightly different (rs6000-specific) semantics
31312 than store_data_bypass_p, returning false rather than aborting
31313 with assertion error when arguments do not satisfy the
31314 requirements of store data bypass.
31315 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
31316 rs6000_store_data_bypass_p.
31318 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
31320 * config/xtensa/xtensa-protos.h
31321 (xtensa_initial_elimination_offset): New declaration.
31322 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
31323 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
31324 macro definition, add case for FRAME_POINTER_REGNUM when
31325 FRAME_GROWS_DOWNWARD.
31326 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
31327 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
31328 xtensa_initial_elimination_offset.
31330 2017-05-08 Nathan Sidwell <nathan@acm.org>
31332 * doc/invoke.texi: Alphabetize -fdump options.
31334 2017-05-08 Martin Sebor <msebor@redhat.com>
31336 PR translation/80280
31337 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
31339 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
31341 * target.def (compute_frame_layout): New optional target hook.
31342 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
31343 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
31344 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
31346 * reload1.c (verify_initial_elim_offsets): Likewise.
31347 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
31348 (use_simple_return_p): Call arm_compute_frame_layout if needed.
31349 (arm_get_frame_offsets): Split up into this ...
31350 (arm_compute_frame_layout): ... and this function.
31352 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
31354 * config/aarch64/constraints.md (Usa): New constraint.
31355 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
31357 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
31359 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
31360 with_multilib_list after it has been checked.
31362 2017-05-08 Richard Biener <rguenther@suse.de>
31364 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
31365 (bitmap_set_subtract_values): Likewise.
31367 2017-05-08 Richard Biener <rguenther@suse.de>
31369 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
31370 (gimple_assign_nonzero): ... this and remove strict_overflow_p
31372 (gimple_stmt_nonzero_warnv_p): Rename to ...
31373 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
31375 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
31376 (extract_range_basic): Adjust, do not disable propagation on
31377 strict overflow sensitive simplification.
31378 (vrp_visit_cond_stmt): Likewise.
31380 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
31382 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
31383 body size unconditionally.
31385 2017-05-07 Jeff Law <law@redhat.com>
31388 2017-05-06 Jeff Law <law@redhat.com>
31389 PR tree-optimization/78496
31390 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31393 PR tree-optimization/78496
31394 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31395 (simplify_stmt_using_ranges): Call it.
31396 (vrp_dom_walker::before_dom_children): Extract equivalences
31397 from an ASSERT_EXPR with an equality comparison against a
31400 2017-05-06 Jeff Law <law@redhat.com>
31402 PR tree-optimization/78496
31403 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31406 PR tree-optimization/78496
31407 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31408 (simplify_stmt_using_ranges): Call it.
31409 (vrp_dom_walker::before_dom_children): Extract equivalences
31410 from an ASSERT_EXPR with an equality comparison against a
31413 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31415 * lra-constraints.c (lra_copy_reg_equiv): New function.
31416 (split_reg): Use it to copy equivalence information from the
31417 original register to the spill register.
31419 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31421 PR rtl-optimization/75964
31422 * simplify-rtx.c (simplify_const_relational_operation): Remove
31423 invalid handling of comparisons of integer ABS.
31425 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
31427 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
31428 initialize to zero.
31429 (init_regs): Remove declaration.
31430 (function_arg_advance_32): Initialize error_p as boolean variable.
31432 2017-05-05 Nathan Sidwell <nathan@acm.org>
31434 * store-motion.c (remove_reachable_equiv_notes): Reformat long
31435 lines. Use for (;;).
31437 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31439 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
31440 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
31441 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
31442 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
31443 VF=2 that require versioning.
31445 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31447 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
31450 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31452 * diagnostic.h (diagnostic_override_option_index): Convert from
31453 macro to inline function.
31455 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31457 * diagnostic.c (last_module_changed_p): New function.
31458 (set_last_module): New function.
31459 (diagnostic_report_current_module): Convert macro usage to
31460 the above functions.
31461 * diagnostic.h (diagnostic_context::last_module): Strengthen
31462 from const line_map * to const line_map_ordinary *.
31463 (diagnostic_last_module_changed): Delete macro.
31464 (diagnostic_set_last_module): Delete macro.
31466 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31468 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
31469 with diagnostic_report_diagnostic.
31470 (diagnostic_n_impl_richloc): Likewise.
31471 * diagnostic.h (report_diagnostic): Delete macro.
31472 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
31473 with diagnostic_report_diagnostic.
31474 * substring-locations.c (format_warning_va): Likewise.
31476 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31478 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
31479 save/restor of format_spec. Move option-printing code to...
31480 (print_option_information): ...this new function, and
31481 reimplement by simply printing to the pretty_printer,
31482 rather than appending to the format string.
31484 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31486 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
31487 handling logic into...
31488 (update_effective_level_from_pragmas): ...this new function.
31490 2017-05-04 Andrew Waterman <andrew@sifive.com>
31492 * config/riscv/riscv.opt (mstrict-align): New option.
31493 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
31494 (SLOW_UNALIGNED_ACCESS): Define.
31495 (riscv_slow_unaligned_access): Declare.
31496 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
31498 (riscv_slow_unaligned_access): New variable.
31499 (rocket_tune_info): Set slow_unaligned_access to true.
31500 (optimize_size_tune_info): Set slow_unaligned_access to false.
31501 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
31502 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
31503 (riscv_option_override): Set riscv_slow_unaligned_access.
31504 * doc/invoke.texi: Add -mstrict-align to RISC-V.
31506 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
31508 * config/riscv/riscv.md: Unify indentation.
31510 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
31515 * config/rs6000/rs6000.md (u code attribute): Add FIX and
31517 (extendsi<mode>2): Add support for doing sign extension via
31518 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
31519 don't have ISA 3.0 instructions.
31520 (extendsi<mode>2 splitter): Likewise.
31521 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
31522 generate the normal insns since SImode can now go in vector
31523 registers. Disallow the special UNSPECs needed for previous
31524 machines to hide SImode being used. Add new insns
31525 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
31526 (fix_trunc<mode>si2_stfiwx): Likewise.
31527 (fix_trunc<mode>si2_internal): Likewise.
31528 (fixuns_trunc<mode>si2): Likewise.
31529 (fixuns_trunc<mode>si2_stfiwx): Likewise.
31530 (fctiw<u>z_<mode>_smallint): Likewise.
31531 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
31532 of floating point to 32-bit integer from doing a direct move to
31533 the GPR registers to do a store.
31534 (fctiwz_<mode>): Break long line.
31536 2017-05-05 Bin Cheng <bin.cheng@arm.com>
31538 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
31539 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
31540 (addr_list, addr_offset_valid_p): New.
31541 (split_address_groups): Check offset validity with above function.
31542 (gt-tree-ssa-loop-ivopts.h): Include header file.
31544 2017-05-05 Nathan Sidwell <nathan@acm.org>
31546 * config.gcc (arm*-*-*): Add missing 'fi'.
31548 2017-05-05 Steve Ellcey <sellcey@cavium.com>
31550 * doc/invoke.texi (-fopt-info): Explicitly say order of options
31551 included in -fopt-info does not matter.
31552 * doc/optinfo.texi (-fopt-info): Fix description of default
31553 behavour. Explicitly say order of options included in -fopt-info
31556 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
31558 * config.gcc: Allow combinations of aprofile and rmprofile values for
31559 --with-multilib-list.
31560 * config/arm/t-multilib: New file.
31561 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
31562 variables. Remove setting of ISA and floating-point ABI in
31563 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
31564 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
31565 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
31566 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
31568 * config/arm/t-rmprofile: Likewise except for the matches changes.
31569 * doc/install.texi (--with-multilib-list): Document the combination of
31570 aprofile and rmprofile values and warn about pitfalls in doing that.
31572 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31574 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
31575 (movdi_aarch64): Likewise.
31577 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31579 PR tree-optimization/80632
31580 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
31582 (build_arrays): Initialize it for virtual phis.
31583 (fix_phi_nodes): Use it for virtual phis.
31585 PR tree-optimization/80558
31586 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
31587 [x, y] op z into [x op, y op z] for op & or | if conditions
31590 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
31591 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31594 * config/arm/arm.md (use_literal_pool): Remove.
31595 (64-bit immediate split): No longer takes cost into consideration
31596 if arm_disable_literal_pool is enabled.
31597 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
31598 used when arm_disable_literal_pool is enabled.
31599 (arm_max_const_double_inline_cost): Remove use of
31600 arm_disable_literal_pool.
31601 (push_minipool_fix): Add assert.
31602 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
31603 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
31604 (no_literal_pool_sf_immediate): New.
31606 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31608 PR tree-optimization/80613
31609 * tree-ssa-dce.c (propagate_necessity): Remove cases for
31610 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
31612 2017-05-05 Richard Biener <rguenther@suse.de>
31614 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
31616 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
31618 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
31619 of this flag from insn conditions due to removal from r247495.
31621 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31623 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
31625 (arm_early_store_addr_dep_ptr): Likewise.
31626 * config/arm/aarch-common-protos.h
31627 (arm_early_load_addr_dep_ptr): Add prototype.
31628 (arm_early_store_addr_dep_ptr): Likewise.
31629 * config/arm/cortex-a53.md: Add new bypasses.
31631 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31633 * tree.c (next_type_uid): Change type to unsigned.
31634 (type_hash_canon): Decrement back next_type_uid if
31635 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
31636 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
31639 2017-05-04 Martin Sebor <msebor@redhat.com>
31641 * builtins.c: Fix a trivial typo in a comment.
31643 PR middle-end/79234
31644 * builtins.c (check_sizes): Adjust to handle reading past the end.
31645 Avoid printing excessive upper bound of ranges. Use %E to print
31646 tree nodes instead of converting them to %wu.
31647 (expand_builtin_memchr): New function.
31648 (compute_dest_size): Rename...
31649 (compute_objsize): ...to this.
31650 (expand_builtin_memcpy): Adjust.
31651 (expand_builtin_mempcpy): Adjust.
31652 (expand_builtin_strcat): Adjust.
31653 (expand_builtin_strcpy): Adjust.
31654 (check_strncat_sizes): Adjust.
31655 (expand_builtin_strncat): Adjust.
31656 (expand_builtin_strncpy): Adjust and simplify.
31657 (expand_builtin_memset): Adjust.
31658 (expand_builtin_bzero): Adjust.
31659 (expand_builtin_memcmp): Adjust.
31660 (expand_builtin): Handle memcmp.
31661 (maybe_emit_chk_warning): Check strncat just once.
31663 2017-05-04 Martin Sebor <msebor@redhat.com>
31665 PR preprocessor/79214
31666 PR middle-end/79222
31667 PR middle-end/79223
31668 * builtins.c (check_sizes): Add inlining context and issue
31669 warnings even when -Wno-system-headers is set.
31670 (check_strncat_sizes): Same.
31671 (expand_builtin_strncat): Same.
31672 (expand_builtin_memmove): New function.
31673 (expand_builtin_stpncpy): Same.
31674 (expand_builtin): Handle memmove and stpncpy.
31676 2017-05-04 Bin Cheng <bin.cheng@arm.com>
31678 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
31679 which is not used any more.
31681 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31683 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
31685 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31687 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
31688 (cortexa53_tunings): Likewise.
31689 (cortexa57_tunings): Likewise.
31690 (cortexa72_tunings): Likewise.
31691 (cortexa73_tunings): Likewise.
31693 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31695 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
31696 Set loop alignment to 8.
31698 2017-05-04 Martin Sebor <msebor@redhat.com>
31700 PR translation/80280
31701 * builtins.c (expand_builtin_object_size): Add missing quoting to
31702 %D and like directives.
31703 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
31704 (hsa_type_for_tree_type): Same.
31705 (verify_function_arguments): Same.
31706 * symtab.c (symbol_table::change_decl_assembler_name): Same.
31707 * varasm.c (get_section): Same.
31710 2017-05-04 Martin Sebor <msebor@redhat.com>
31712 PR translation/80280
31713 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
31715 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31717 * config/aarch64/aarch64.c (generic_addrcost_table):
31718 Change HI/TI mode setting.
31720 2017-05-04 Martin Jambor <mjambor@suse.cz>
31722 PR tree-optimization/80622
31723 * tree-sra.c (comes_initialized_p): New function.
31724 (build_accesses_from_assign): Only set write lazily when
31725 comes_initialized_p is false.
31726 (analyze_access_subtree): Use comes_initialized_p.
31727 (propagate_subaccesses_across_link): Assert !comes_initialized_p
31728 instead of testing for PARM_DECL.
31730 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31732 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
31733 constraint on operand 0 to allow more general addressing modes.
31734 Adjust output template.
31735 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
31737 * config/aarch64/aarch64-protos.h
31738 (aarch64_address_valid_for_prefetch_p): Declare prototype.
31739 * config/aarch64/constraints.md (Dp): New address constraint.
31740 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
31743 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
31745 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
31746 update use of estimate_ipcp_clone_size_and_time.
31747 (estimate_local_effects): Update use of
31748 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
31749 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
31750 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
31751 Return nonspecialized time.
31753 2017-05-04 Richard Biener <rguenther@suse.de>
31755 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
31756 for the last VUSE which def dominates the PHI. Directly call
31758 (get_continuation_for_phi_1): Remove.
31760 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
31762 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
31763 to explain the use of truncating division. Cap the number of
31764 iterations to the maximum given by nb_iterations_upper_bound,
31767 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31769 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
31770 * configure: Regenerate.
31771 * config.in: Regenerate.
31772 * config/i386/driver-mingw32.c: new file.
31773 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
31774 * config.host: Link driver-mingw32.o on MinGW host.
31775 * doc/install.texi: Document new --enable-mingw-wildcard configure
31778 2017-05-04 Marek Polacek <polacek@redhat.com>
31780 PR tree-optimization/80612
31781 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
31783 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31784 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
31786 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
31787 (movt splitter): Likewise.
31788 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
31789 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
31790 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
31791 block for Thumb-1 with MOVT.
31792 (thumb2_legitimate_address_p): Move code block ...
31793 (can_avoid_literal_pool_for_label_p): ... into this new function.
31794 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
31796 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
31797 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
31798 "M-profile targets with the MOVT instruction".
31800 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31802 * config/arm/arm-builtins.c (arm_init_builtins): Rename
31803 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
31804 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
31806 2017-05-04 Martin Liska <mliska@suse.cz>
31808 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
31809 variable cond_code.
31811 2017-05-04 Richard Biener <rguenther@suse.de>
31813 * tree.c (array_at_struct_end_p): Handle arrays at struct
31814 end with flexarrays more conservatively. Refactor and treat
31815 arrays of arrays or aggregates more strict. Fix
31816 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
31817 * tree.c (array_at_struct_end_p): Adjust prototype.
31818 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
31819 * gimple-fold.c (get_range_strlen): Likewise.
31820 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
31822 2017-05-04 Richard Biener <rguenther@suse.de>
31824 PR tree-optimization/31130
31825 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
31827 (supports_overflow_infinity): Likewise.
31828 (is_negative_overflow_infinity): Likewise.
31829 (is_positive_overflow_infinity): Likewise.
31830 (is_overflow_infinity): Likewise.
31831 (stmt_overflow_infinity): Likewise.
31832 (overflow_infinity_range_p): Likewise.
31833 (usable_range_p): Remove as always returning true.
31834 (make_overflow_infinity): Remove.
31835 (negative_overflow_infinity): Likewise.
31836 (positive_overflow_infinity): Likewise.
31837 (avoid_overflow_infinity): Likewise.
31838 (set_value_range): Adjust accordingly.
31839 (set_value_range_to_nonnegative): Likewise, remove now unused
31840 overflow_infinity arg.
31841 (vrp_operand_equal_p): Adjust.
31842 (update_value_range): Likewise.
31843 (range_int_cst_singleton_p): Likewise.
31844 (operand_less_p): Likewise.
31845 (compare_values_warnv): Likewise.
31846 (extract_range_for_var_from_comparison_expr): Likewise.
31847 (vrp_int_const_binop): Likewise.
31848 (zero_nonzero_bits_from_vr): Likewise.
31849 (extract_range_from_multiplicative_op_1): Likewise.
31850 (extract_range_from_binary_expr_1): Likewise.
31851 (extract_range_from_unary_expr): Likewise.
31852 (extract_range_from_comparison): Likewise.
31853 (extract_range_basic): Likewise.
31854 (adjust_range_with_scev): Likewise.
31855 (compare_ranges): Likewise.
31856 (compare_range_with_value): Likewise.
31857 (dump_value_range): Likewise.
31858 (test_for_singularity): Likewise, remove strict_overflow_p parameter
31860 (simplify_cond_using_ranges): Adjust.
31862 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31864 * brig-builtins.def: Added a builtin for class_f64.
31865 * builtin-types.def: Added a builtin type needed by class_f64.
31867 2017-05-03 Jason Merrill <jason@redhat.com>
31869 * timevar.def: Add TV_CONSTEXPR.
31871 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31873 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
31875 2017-05-03 Martin Jambor <mjambor@suse.cz>
31877 * ipa-prop.c (ipa_update_after_lto_read): Removed.
31878 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
31879 * ipa-cp.c (ipcp_propagate_stage): Do not call
31880 ipa_update_after_lto_read.
31881 * ipa-inline.c (ipa_inline): Likewise.
31883 2017-05-03 Martin Jambor <mjambor@suse.cz>
31885 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
31886 tag. Added a default constructor and a destructor.
31887 (ipa_edge_args_sum_t): New class;
31888 (ipa_edge_args_sum): Declare.
31889 (ipa_edge_args_vector): Remove declaration.
31890 (IPA_EDGE_REF): Use ipa_edge_args_sum.
31891 (ipa_free_edge_args_substructures): Remove declaration.
31892 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
31893 (ipa_edge_args_info_available_for_edge_p): Likewise.
31894 * ipa-prop.c (ipa_edge_args_vector): Removed.
31895 (edge_removal_hook_holder): Likewise.
31896 (edge_duplication_hook_holder): Likewise.
31897 (ipa_edge_args_sum): New variable.
31898 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
31899 ipa_edge_args_vector.
31900 (ipa_free_edge_args_substructures): Likewise.
31901 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
31902 ipa_edge_args_vector.
31903 (ipa_edge_removal_hook): Turned into method
31904 ipa_edge_args_sum_t::remove.
31905 (ipa_edge_duplication_hook): Turned into method
31906 ipa_edge_args_sum_t::duplicate.
31907 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
31908 registering edge hooks.
31909 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
31910 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
31911 ipa_edge_args_sum instead of ipa_edge_args_vector.
31912 * ipa-profile.c (ipa_profile): Likewise.
31914 2017-05-03 Martin Jambor <mjambor@suse.cz>
31916 * symbol-summary.h (function_summary): New method exists.
31917 (function_summary::symtab_removal): Deallocate through release.
31918 (call_summary): New class.
31919 (gt_ggc_mx): New overload.
31920 (gt_pch_nx): Likewise.
31921 (gt_pch_nx): Likewise.
31923 2017-05-03 Jeff Law <law@redhat.com>
31925 PR tree-optimization/78496
31926 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
31927 from simplify_cond_using_ranges. Split off code to walk
31928 backwards through casts into ...
31929 (simplify_cond_using_ranges_2): New function.
31930 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
31931 (execute_vrp): After identifying jump threads, call
31932 simplify_cond_using_ranges_2.
31934 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
31937 * ipa-inline.h (inline_summary): Add ctor.
31938 (create_ggc): Do not use ggc_cleared_alloc.
31940 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
31941 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31943 * gcc.c (handle_braces): Support escaping in switch matching
31945 * doc/invoke.texi (Spec Files): Document it.
31946 Remove superfluous @code markup in items.
31948 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31950 * diagnostic-show-locus.c (struct column_range): New struct.
31951 (get_affected_columns): New function.
31952 (get_printed_columns): New function.
31953 (struct correction): New struct.
31954 (correction::ensure_capacity): New function.
31955 (correction::ensure_terminated): New function.
31956 (struct line_corrections): New struct.
31957 (line_corrections::~line_corrections): New dtor.
31958 (line_corrections::add_hint): New function.
31959 (layout::print_trailing_fixits): Reimplement in terms of the new
31961 (selftest::test_overlapped_fixit_printing): New function.
31962 (selftest::diagnostic_show_locus_c_tests): Call it.
31964 2017-05-03 Nathan Sidwell <nathan@acm.org>
31966 Canonicalize canonical type hashing
31967 * tree.h (type_hash_canon_hash): Declare.
31968 * tree.c (type_hash_list, attribute_hash_list): Move into
31969 type_hash_canon_hash.
31970 (build_type_attribute_qual_variant): Break out hash code calc into
31971 type_hash_canon_hash.
31972 (type_hash_canon_hash): New. Generic type hash computation.
31973 (build_range_type_1, build_array_type_1, build_function_type,
31974 build_method_type_directly, build_offset_type, build_complex_type,
31975 make_vector_type): Call it.
31977 2017-05-03 Richard Biener <rguenther@suse.de>
31979 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
31980 When all DRs have unknown misaligned do not always peel
31981 when there is a store but apply the same costing model as if
31982 there were only loads.
31984 2017-05-03 Richard Biener <rguenther@suse.de>
31987 PR tree-optimization/80492
31988 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
31989 compare_base_decls returning dont-know properly.
31991 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
31993 * config/arm/iterators.md (CCSI): New mode iterator.
31994 (arch): New mode attribute.
31995 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
31996 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
31997 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
31998 code iterator for success result mode.
31999 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
32000 the corresponding new insn generators.
32002 2017-05-03 Bin Cheng <bin.cheng@arm.com>
32005 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32006 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32008 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
32010 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
32011 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
32012 (DDR_A): Wrap DDR argument in brackets.
32013 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
32014 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
32015 (DDR_REVERSED_P): Likewise.
32017 2017-05-03 Jakub Jelinek <jakub@redhat.com>
32019 PR tree-optimization/79472
32020 * tree-switch-conversion.c (struct switch_conv_info): Add
32021 contiguous_range and default_case_nonstandard fields.
32022 (collect_switch_conv_info): Compute contiguous_range and
32023 default_case_nonstandard fields, don't clear final_bb if
32024 contiguous_range and only the default case doesn't have the required
32026 (check_all_empty_except_final): Set default_case_nonstandard instead
32027 of failing if contiguous_range and the default case doesn't have empty
32029 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
32030 and only the default case doesn't have the required constants. Skip
32032 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
32033 if default_case_nonstandard.
32034 (build_constructors): Build constant 1 just once. Assert that default
32035 values aren't inserted in between cases if contiguous_range. Skip
32037 (build_arrays): Skip virtual phis.
32038 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
32039 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
32040 Handle virtual phis.
32041 (gen_inbound_check): Handle default_case_nonstandard case.
32042 (process_switch): Adjust check_final_bb caller. Call
32043 gather_default_values with the first non-default case instead of
32044 default case if default_case_nonstandard.
32046 2017-05-02 Nathan Sidwell <nathan@acm.org>
32048 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
32049 check. Fix formatting.
32051 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
32053 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
32054 errors when comparing specialized and unspecialized times.
32056 2017-05-02 David Malcolm <dmalcolm@redhat.com>
32058 * diagnostic-show-locus.c
32059 (layout::should_print_annotation_line_p): Make private.
32060 (layout::print_annotation_line): Make private.
32061 (layout::annotation_line_showed_range_p): Make private.
32062 (layout::show_ruler): Make private.
32063 (layout::print_source_line): Make private. Pass in line and
32064 line_width, rather than calling location_get_source_line. Drop
32066 (layout::print_leading_fixits): New method.
32067 (layout::print_any_fixits): Rename to...
32068 (layout::print_trailing_fixits): ...this, and make private.
32069 Don't print newline fixits.
32070 (diagnostic_show_locus): Move logic for printing one row into...
32071 (layout::print_line): ...this new function. Move the
32072 location_get_source_line call and error-handling from
32073 print_source_line to here. Call print_leading_fixits, and rename
32074 print_any_fixits to print_trailing_fixits.
32075 (selftest::test_fixit_insert_containing_newline): Update now that
32076 newlines are partially supported.
32077 (selftest::test_fixit_insert_containing_newline_2): New test.
32078 (selftest::test_fixit_replace_containing_newline): Update comments.
32079 (selftest::diagnostic_show_locus_c_tests): Call the new test.
32080 * edit-context.c (class added_line): New class.
32081 (class edited_line): Describe newline handling in comment.
32082 (edited_line::actually_edited_p): New method.
32083 (edited_line::print_content): Delete redundant decl.
32084 (edited_line::m_predecessors): New field.
32085 (edited_file::print_content): Call edited_line::print_content.
32086 (edited_file::print_diff): Update to support newlines.
32087 (edited_file::print_diff_hunk): Likewise.
32088 (edited_file::print_run_of_changed_lines): New function.
32089 (edited_file::print_diff_line): Convert to...
32090 (print_diff_line): ...this.
32091 (edited_file::get_effective_line_count): New function.
32092 (edited_line::edited_line): Initialize new field m_predecessors.
32093 (edited_line::~edited_line): Clean up m_predecessors.
32094 (edited_line::apply_fixit): Handle newlines.
32095 (edited_line::get_effective_line_count): New function.
32096 (edited_line::print_content): New function.
32097 (edited_line::print_diff_lines): New function.
32098 (selftest::test_applying_fixits_insert_containing_newline): New
32100 (selftest::test_applying_fixits_replace_containing_newline): New
32102 (selftest::insert_line): New function.
32103 (selftest::test_applying_fixits_multiple_lines): Add example of
32105 (selftest::edit_context_c_tests): Call the new tests.
32107 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32109 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
32110 parameter cand. Update dump information.
32111 (get_computation_cost): Update uses.
32113 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32115 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
32116 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
32117 (get_computation_at, rewrite_use_address): Update use of
32118 get_computation_aff.
32120 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32122 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
32123 (get_computation): Delete.
32124 (get_computation_cost): Implement like get_computation_cost_at.
32125 Use get_computation_at.
32126 (get_computation_cost_at): Delete.
32127 (rewrite_use_nonlinear_expr): Use get_computation_at.
32128 (rewrite_use_compare, remove_unused_ivs): Ditto.
32130 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32132 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
32134 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32136 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
32137 (ivopts_global_cost_for_size): Rename parameter and update uses.
32138 (iv_ca_recount_cost): Update uses.
32139 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
32140 candidates seperately in n_invs and n_cands.
32141 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
32143 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32145 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
32146 (find_inv_vars_cb): New.
32147 (find_depends): Renamed to ...
32148 (find_inv_vars): ... this.
32149 (add_candidate_1, force_var_cost): Call find_inv_vars.
32150 (split_address_cost, determine_group_iv_cost_cond): Ditto.
32152 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32154 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
32155 inv_vars. Add inv_exprs.
32156 (struct iv_cand): Rename depends_on to inv_vars.
32157 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
32158 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
32159 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
32160 (dump_cand): Dump inv_vars.
32161 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
32162 (record_invariant, find_depends, add_candidate_1): Ditto.
32163 (set_group_iv_cost, force_var_cost): Ditto.
32164 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
32165 (get_computation_cost_at, get_computation_cost): Ditto.
32166 (determine_group_iv_cost_generic): Ditto.
32167 (determine_group_iv_cost_address): Ditto.
32168 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
32169 (determine_group_iv_costs): Ditto.
32170 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
32171 (iv_ca_set_remove_invariants): Renamed to ...
32172 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
32173 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
32174 (iv_ca_set_add_invariants): Renamed to ...
32175 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
32176 (iv_ca_set_cp): Use iv_ca_set_add_invs.
32177 (iv_ca_has_deps): Support inv_vars and inv_exprs.
32178 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
32179 (create_new_ivs): Remove useless dump.
32181 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32183 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
32185 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
32186 (iv_ca_set_no_cp, create_new_iv): Ditto.
32188 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32190 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32192 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32194 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
32195 function tree_check2.
32197 2017-05-02 Martin Liska <mliska@suse.cz>
32199 * doc/gcov.texi: Add missing preposition.
32200 * gcov.c (function_info::function_info): Properly fill up
32201 all member variables.
32203 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32205 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
32207 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32209 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
32211 2017-05-02 Martin Liska <mliska@suse.cz>
32214 * lto-streamer-in.c (lto_read_tree_1): Remove
32215 LTO_STREAMER_DEBUG.
32216 * lto-streamer.c (struct tree_hash_entry): Likewise.
32217 (struct tree_entry_hasher): Likewise.
32218 (tree_entry_hasher::hash): Likewise.
32219 (tree_entry_hasher::equal): Likewise.
32220 (lto_streamer_init): Likewise.
32221 (lto_orig_address_map): Likewise.
32222 (lto_orig_address_get): Likewise.
32223 (lto_orig_address_remove): Likewise.
32224 * lto-streamer.h: Likewise.
32225 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
32226 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
32228 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
32230 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
32231 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
32232 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
32233 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
32234 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
32235 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
32236 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
32237 (mm_maskz_sub_ss): New intrinsics.
32238 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32239 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32240 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
32241 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
32242 (__builtin_ia32_subss_mask_round): New builtins.
32243 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32244 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32245 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
32247 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
32248 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32250 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32253 2017-05-02 Martin Jambor <mjambor@suse.cz>
32255 PR tree-optimization/78687
32256 * tree-sra.c (access): New field parent.
32257 (process_subtree_disqualification): New function.
32258 (disqualify_candidate): Call it.
32259 (build_accesses_from_assign): Reset write flag if creating an
32261 (build_access_subtree): Fill in parent field and also prpagate
32262 down grp_write flag.
32263 (create_artificial_child_access): New parameter set_grp_write, set
32264 grp_write to its value.
32265 (propagate_subaccesses_across_link): Also propagate grp_write flag
32267 (propagate_all_subaccesses): Push the closest parent back to work
32268 queue if add_access_to_work_queue returned true.
32270 2017-05-02 Richard Biener <rguenther@suse.de>
32272 * common.opt (fstrict-overflow): Alias negative to fwrapv.
32273 * doc/invoke.texi (fstrict-overflow): Remove all traces of
32274 -fstrict-overflow documentation.
32275 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
32276 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
32277 flag_strict_overflow.
32278 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
32279 * lto-opts.c (lto_write_options): Do not stream it.
32280 * lto-wrapper.c (merge_and_complain): Do not handle it.
32281 * opts.c (default_options_table): Do not set -fstrict-overflow.
32282 (finish_options): Likewise do not clear it when sanitizing.
32283 * simplify-rtx.c (simplify_const_relational_operation): Do not
32284 test flag_strict_overflow.
32286 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32288 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
32289 using enabled attribute.
32290 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
32291 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
32292 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
32293 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
32294 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
32295 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
32296 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
32297 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
32298 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
32299 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
32301 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32303 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
32305 2017-05-02 Richard Biener <rguenther@suse.de>
32307 PR tree-optimization/80591
32309 2017-04-10 Richard Biener <rguenther@suse.de>
32311 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32314 2017-05-02 Richard Biener <rguenther@suse.de>
32316 PR tree-optimization/80549
32317 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
32318 (cleanup_tree_cfg_noloop): Create forwarders to known loop
32319 headers if they do not have a preheader.
32321 2017-05-02 Martin Liska <mliska@suse.cz>
32324 * common.opt: Fix typo.
32325 * doc/invoke.texi: Likewise.
32327 2017-05-01 Jan Beulich <jbeulich@suse.com>
32329 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
32330 swapping, add (x,x,m,x,n) alternative.
32332 2017-05-01 Nathan Sidwell <nathan@acm.org>
32334 * calls.c (combine_pending_stack_adjustment_and_call): Remove
32335 unnecessary unadjusted_alignment check.
32337 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32340 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
32342 * gimplify.c (gimplify_cilk_detach): New function.
32343 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
32344 * tree-core.h: Document EXPR_CILK_SPAWN.
32345 * tree.h (EXPR_CILK_SPAWN): Define.
32347 2017-05-01 David Malcolm <dmalcolm@redhat.com>
32349 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
32350 to use new fixit_hint representation, using the "replace" logic.
32351 (get_line_span_for_fixit_hint): Likewise.
32352 (layout::print_any_fixits): Likewise.
32353 (selftest::test_one_liner_many_fixits): Rename to...
32354 (selftest::test_one_liner_many_fixits_1): ...this, and update
32355 comment and expected output to reflect that the multiple fix-it
32356 hints are now consolidated into one insertion.
32357 (selftest::test_one_liner_many_fixits_2): New test.
32358 (selftest::test_diagnostic_show_locus_one_liner): Update for
32360 (selftest::test_fixit_consolidation): Update for fix-it API
32362 * diagnostic.c (print_parseable_fixits): Likewise.
32363 * edit-context.c (edited_line::m_line_events): Convert from
32364 auto_vec <line_event *> to auto_vec <line_event>.
32365 (class line_event): Convert from abstract base class to a concrete
32366 class, taking over the role of replace_event.
32367 (class insert_event): Delete.
32368 (class replace_event): Rename to class line_event. Convert to
32370 (edit_context::add_fixits): Reimplement.
32371 (edit_context::apply_insert): Delete.
32372 (edit_context::apply_replace): Rename to...
32373 (edit_context::apply_fixit): ...this. Convert to half-open range.
32374 (edited_file::apply_insert): Delete.
32375 (edited_file::apply_replace): Rename to...
32376 (edited_file::apply_fixit): ...this.
32377 (edited_line::~edited_line): Drop deletion of events.
32378 (edited_line::apply_insert): Delete.
32379 (edited_line::apply_replace): Rename to...
32380 (edited_line::apply_fixit): ...this. Convert to half-open range.
32381 Update for change to type of m_line_events.
32382 * edit-context.h (edit_context::apply_insert): Delete.
32383 (edit_context::apply_replace): Rename to...
32384 (edit_context::apply_fixit): ...this.
32386 2017-05-01 Martin Sebor <msebor@redhat.com>
32388 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
32391 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
32394 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
32395 __get_cpuid_max returns 0.
32396 (__get_cpuid_count): Ditto.
32398 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
32400 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
32401 replacement expression is another instance of one of its arguments.
32403 2017-05-01 Jakub Jelinek <jakub@redhat.com>
32406 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
32407 check for stack push/pop autoinc.
32408 * config/i386/i386.c (ix86_agi_dependent): Return false
32409 if the only reason why modified_in_p returned true is that
32410 addr is SP based and set_insn is a push or pop.
32412 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32414 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
32417 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32420 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
32421 (account_size_time): Use two predicates - exec_pred and
32423 (evaluate_conditions_for_known_args): Compute both clause and
32425 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
32426 (inline_summary_t::duplicate): Update.
32427 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
32429 (compute_inline_parameters): Likewise.
32430 (estimate_edge_size_and_time): Update caluclation of time.
32431 (estimate_node_size_and_time): Compute both time and nonspecialized
32433 (estimate_ipcp_clone_size_and_time): Update.
32434 (inline_merge_summary): Update.
32435 (do_estimate_edge_time): Update.
32436 (do_estimate_edge_size): Update.
32437 (do_estimate_edge_hints): Update.
32438 (inline_read_section, inline_write_summary): Stream both new predicates.
32439 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
32441 (compute_inlined_call_time): Cleanup.
32442 (big_speedup_p): Update.
32443 (edge_badness): Update.
32444 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
32445 (size_time_entry): Replace predicate by exec_predicate and
32446 nonconst_predicate.
32447 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
32448 (estimate_edge_time): Return also nonspec_time.
32449 (reset_edge_growth_cache): Update.
32451 2017-04-29 Jakub Jelinek <jakub@redhat.com>
32453 PR rtl-optimization/80491
32454 * ifcvt.c (noce_process_if_block): When looking for x setter
32455 with missing else_bb, don't check only the insn right before
32456 cond_earliest, but look for the last insn that x is modified in
32457 within the same bb.
32459 PR rtl-optimization/80491
32460 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
32462 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
32464 PR tree-optimization/80487
32465 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
32467 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32469 PR tree-optimization/79697
32470 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
32471 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
32472 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
32474 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
32475 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
32477 2017-04-28 Martin Sebor <msebor@redhat.com>
32479 PR tree-optimization/80523
32480 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
32481 (init_target_to_host_charmap, target_to_host, target_strtol10): New
32483 (maybe_warn, format_directive, parse_directive): Use new functions.
32484 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
32486 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
32488 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
32490 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
32492 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
32493 target_header_dir): Set correctly.
32494 * configure: Regenerated.
32495 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32496 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32497 instead of SYSTEM_HEADER_DIR.
32499 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
32501 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
32502 (estimate_local_effects): Likewise.
32503 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
32504 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
32505 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
32506 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
32507 do_estimate_edge_time, estimate_edge_time): Likewise.
32508 * ipa-inline-analysis.c (estimate_node_size_and_time,
32509 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
32510 (estimate_time_after_inlining): Remove.
32512 2017-04-28 Martin Liska <mliska@suse.cz>
32514 * doc/gcov.texi: Enhance documentation of gcov.
32516 2017-04-28 Martin Liska <mliska@suse.cz>
32518 * doc/gcov.texi: Sort options in alphabetic order.
32519 * doc/gcov-dump.texi: Likewise.
32520 * doc/gcov-tool.texi: Likewise.
32521 * gcov.c (print_usage): Likewise.
32522 * gcov-dump.c (print_usage): Likewise.
32523 * gcov-tool.c (print_merge_usage_message): Likewise.
32524 (print_rewrite_usage_message): Likewise.
32525 (print_overlap_usage_message): Likewise.
32527 2017-04-28 Martin Liska <mliska@suse.cz>
32529 PR gcov-profile/53915
32530 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
32532 2017-04-28 Martin Liska <mliska@suse.cz>
32534 PR gcov-profile/79891
32535 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
32536 is marked by compiler as living on a line.
32537 (get_cycles_count): Remove usage of the union.
32538 (output_intermediate_file): Likewise.
32539 (find_source): Fix GNU coding style.
32540 (accumulate_line_counts): Remove old non-all block mode.
32541 (output_lines): Remove usage of the union.
32542 * profile.c (output_location): Include all BBs, even if
32543 belonging to a same line (and file) as a previous BB.
32545 2017-04-28 Martin Liska <mliska@suse.cz>
32547 * gcov.c (process_args): Handle new argument 'w'.
32548 (read_graph_file): Assign ID to BBs.
32549 (output_branch_count): Display BB # if verbose flag is set.
32550 (output_lines): Likewise for arcs.
32551 (print_usage): Add '--verbose' option help.
32552 * doc/gcov.texi: Document --verbose (-w) option.
32554 2017-04-28 Martin Liska <mliska@suse.cz>
32556 * gcov.c (struct block_location_info): New struct.
32557 (process_file): Fill up the new structure.
32558 (read_graph_file): Replace usage of encoding by the newly added
32560 (add_line_counts): Likewise.
32561 (accumulate_line_counts): Remove usage of the union.
32562 (function_info::function_info): New function.
32563 (function_info::~function_info): Likewise.
32564 (process_file): Call delete instead of release_function.
32565 (release_function): Release the function.
32566 (release_structures): Call delete instead of release_function.
32567 (solve_flow_graph): Replace usage of num_blocks.
32568 (find_exception_blocks): Likewise.
32569 (output_lines): Fix GNU coding style.
32571 2017-04-28 Martin Liska <mliska@suse.cz>
32574 * coverage.c (coverage_remove_note_file): New function.
32575 * coverage.h: Declare the function.
32576 * toplev.c (finalize): Clean if an error has been seen.
32578 2017-04-28 Martin Liska <mliska@suse.cz>
32580 PR gcov-profile/80031
32581 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
32582 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
32583 * gcov.c (read_graph_file): Read just number of blocks.
32584 * profile.c (branch_prob): Do not stream 0 flags per a basic
32587 2017-04-28 Martin Liska <mliska@suse.cz>
32589 * gcov-dump.c (tag_*): Add new argument to declarations.
32590 (dump_gcov_file): Likewise.
32591 (tag_blocks): Add and use new argument depth.
32592 (tag_arcs): Likewise.
32593 (tag_lines): Likewise.
32594 (tag_counters): Likewise.
32595 (tag_summary): Likewise.
32596 (dump_working_sets): Use depth to do a proper indentation.
32598 2017-04-28 Jakub Jelinek <jakub@redhat.com>
32601 * cgraph.h (symtab_node::debug_symtab): No longer inline.
32602 * symtab.c (symtab_node::debug_symtab): Move definition here.
32604 2017-04-28 Richard Biener <rguenther@suse.de>
32606 * lto-streamer.h (LTO_major_version): Bump to 7.
32608 2017-04-28 Richard Biener <rguenther@suse.de>
32610 * tree-vrp.c (assert_info): New struct.
32611 (add_assert_info): New helper.
32612 (register_edge_assert_for_2): Refactor to add asserts to a vector
32614 (register_edge_assert_for_1): Likewise.
32615 (register_edge_assert_for): Likewise.
32616 (finish_register_edge_assert_for): New helper actually registering
32617 asserts where live on edge.
32618 (find_conditional_asserts): Adjust.
32619 (find_switch_asserts): Likewise.
32620 (evrp_dom_walker::try_find_new_range): Generalize.
32621 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
32623 2017-04-27 Marek Polacek <polacek@redhat.com>
32626 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
32627 arg10 and arg11 to itype.
32629 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
32631 * doc/extend.texi (Object Size Checking): Improve grammar.
32633 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
32636 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
32637 that the logic for permitting reciprocal estimates matches that
32640 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32643 * tree.c (type_cache_hasher::equal): Only compare
32644 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
32645 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
32646 non-aggregate element types.
32647 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
32648 about the flag on ARRAY_TYPEs in the comment, formatting fix.
32650 2017-04-27 Richard Biener <rguenther@suse.de>
32652 PR middle-end/80533
32653 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
32654 stripping ARRAY_REFs from MEM_EXPR make sure we're not
32655 keeping a reference to a trailing array.
32657 2017-04-27 Richard Biener <rguenther@suse.de>
32659 PR middle-end/80539
32660 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
32661 being in loop-closed SSA form conservatively.
32662 (chrec_fold_multiply_poly_poly): Likewise.
32664 2017-04-27 Tamar Christina <tamar.christina@arm.com>
32666 PR middle-end/79665
32667 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
32668 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
32670 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32673 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
32674 (aarch64_function_arg_alignment): Return unsigned int again, but still
32675 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
32676 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32677 Don't emit -Wpsabi note.
32678 (aarch64_function_arg_boundary): Likewise.
32679 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32682 2017-04-26 Nathan Sidwell <nathan@acm.org>
32684 * tree.h (crc32_unsigned_n): Declare.
32685 (crc32_unsigned, crc32_unsigned): Make inline.
32686 * tree.c (crc32_unsigned_bits): Replace with ...
32687 (crc32_unsigned_n): ... this.
32688 (crc32_unsigned, crc32_byte): Remove.
32689 (crc32_string): Remove unnecessary braces.
32691 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32693 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
32694 * ipa-inline-analysis.c (MAX_TIME): Remove.
32695 (account_size_time): Use sreal for time.
32696 (dump_inline_summary): Update.
32697 (estimate_function_body_sizes): Update.
32698 (estimate_edge_size_and_time): Update.
32699 (estimate_calls_size_and_time): Update.
32700 (estimate_node_size_and_time): Update.
32701 (inline_merge_summary): Update.
32702 (inline_update_overall_summary): Update.
32703 (estimate_time_after_inlining): Update.
32704 (inline_read_section): Update.
32705 (inline_write_summary): Update.
32706 * ipa-inline.c (compute_uninlined_call_time): Update.
32707 (compute_inlined_call_time): Update.
32708 (recursive_inlining): Update.
32709 (inline_small_functions): Update.
32710 (dump_overall_stats): Update.
32711 * ipa-inline.h: Include sreal.h.
32712 (size_time_entry): Turn time to sreal.
32713 (inline_summary): Turn self_time nad time to sreal.
32715 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32717 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
32719 (sreal::stream_out, sreal::stream_in): New.
32720 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
32722 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32724 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
32727 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
32730 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
32732 (dimode_scalar_chain::compute_convert_gain): Ditto.
32733 (dimode_scalar_chain::make_vector_copies): Ditto.
32734 (dimode_scalar_chain::convert_reg): Ditto.
32735 (dimode_scalar_chain::convert_insn): Ditto.
32736 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
32737 (VI248_AVX512BW_1): New mode iterator.
32738 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
32739 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
32742 2017-04-25 Martin Sebor <msebor@redhat.com>
32744 PR tree-optimization/80497
32745 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
32746 constants are representable in HOST_WIDE_INT.
32747 (parse_directive): Ditto.
32749 2017-04-25 Martin Sebor <msebor@redhat.com>
32752 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
32753 (new_zero_array): Adjust signature.
32754 (dom_info::dom_init): Used unsigned rather that size_t.
32755 (dom_info::dom_info): Same.
32757 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32758 Jakub Jelinek <jakub@redhat.com>
32761 * config/arm/arm.c: Include gimple.h.
32762 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32763 returns negative, increment ncrn only if it returned positive.
32764 (arm_needs_doubleword_align): Return int instead of bool,
32765 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
32766 members, but if there is any such non-FIELD_DECL
32767 > PARM_BOUNDARY aligned decl, return -1 instead of false.
32768 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32769 returns negative, increment nregs only if it returned positive.
32770 (arm_setup_incoming_varargs): Likewise.
32771 (arm_function_arg_boundary): Emit -Wpsabi note if
32772 arm_needs_doubleword_align returns negative, return
32773 DOUBLEWORD_ALIGNMENT only if it returned positive.
32775 2017-04-25 Marek Polacek <polacek@redhat.com>
32778 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32779 first argument to type.
32781 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
32784 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
32785 type checks to test for compatibility instead of equality.
32787 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32788 Jakub Jelinek <jakub@redhat.com>
32791 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
32793 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
32794 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
32795 the alignment computation, but return their maximum in warn_alignment.
32796 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32797 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
32799 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
32800 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32803 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32805 * config/arc/simdext.md (dmpyh): Fix typo.
32807 2017-04-25 Richard Biener <rguenther@suse.de>
32809 PR tree-optimization/80492
32810 * alias.c (compare_base_decls): Handle registers with asm
32811 specification conservatively.
32812 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32813 compare_base_decls returning dont-know properly.
32815 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32817 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
32818 (legitimate_offset_address_p): New function.
32819 (arc_legitimate_address_p): Use above function.
32821 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32823 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
32825 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32827 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
32828 ACCH registers whenever they are available.
32830 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32832 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
32833 double regs fix when not used.
32835 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32837 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
32839 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
32840 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
32842 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32844 * config/arc/arc.c (arc_output_addsi): Check for h-register class
32845 when emitting short ADD instructions.
32847 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32849 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
32851 (cmpsi_cc_c_insn): Likewise.
32852 (cbranchsi4_scratch): Compute proper instruction length using
32853 compact_hreg_operand.
32854 * config/arc/predicates.md (compact_hreg_operand): New predicate.
32856 2017-04-25 Richard Biener <rguenther@suse.de>
32858 PR middle-end/80509
32859 * passes.c (pass_manager::pass_manager): Initialize
32860 m_name_to_pass_map.
32862 2017-04-25 Richard Biener <rguenther@suse.de>
32864 PR tree-optimization/79201
32865 * tree-ssa-sink.c (statement_sink_location): Handle calls.
32867 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32870 * config/s390/vector.md: Split MEM->GPR vector moves for
32871 non-s_operand addresses.
32873 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32876 * config/s390/predicates.md (reload_const_wide_int_operand): New
32878 * config/s390/s390.md ("movti"): Remove d/P alternative.
32879 ("movti_bigconst"): New pattern definition.
32881 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32884 * s390-protos.h (s390_expand_cs_hqi): Removed.
32885 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
32886 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
32887 modes as well as CCZ1mode and CCZmode.
32888 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
32889 signature of s390_emit_compare_and_swap.
32890 (s390_expand_cs_hqi): Likewise, make static.
32891 (s390_expand_cs_tdsi): Generate an explicit compare before trying
32892 compare-and-swap, in some cases.
32893 (s390_expand_cs): Wrapper function.
32894 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
32896 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
32897 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
32898 patterns for small and large integers. Forbid symref memory operands.
32899 Move expander to s390.c. Require cc register.
32900 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
32901 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
32902 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
32903 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
32904 symref memory operands. Remove CC mode and call s390_match_ccmode
32906 ("atomic_exchange<mode>"): Allow and implement all integer modes.
32908 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32910 * config/s390/s390.md (define_peephole2): New peephole to help
32911 combining the load-and-test pattern with volatile memory.
32913 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32915 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
32916 with CCZmode for TARGET_Z196.
32918 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32920 PR rtl-optimization/80501
32921 * combine.c (make_compound_operation_int): Set subreg_code to SET
32922 even for AND with mask of the sign bit of mode.
32924 PR rtl-optimization/80500
32925 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
32926 sum's initial value.
32928 2017-04-25 Julian Brown <julian@codesourcery.com>
32929 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32931 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
32933 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
32935 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
32937 2017-04-25 Julian Brown <julian@codesourcery.com>
32938 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32940 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
32941 (thunderx2t99_sha): New Reservation.
32943 2017-04-25 Julian Brown <julian@codesourcery.com>
32944 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32946 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
32947 type for 1-element load.
32949 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
32951 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
32953 2017-04-24 Martin Jambor <mjambor@suse.cz>
32955 PR tree-optimization/80293
32956 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
32957 char arrays not totally scalarizable if it is false.
32958 (analyze_all_variable_accesses): Pass correct value in the new
32959 parameter. Add a statistics counter.
32961 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
32963 PR middle-end/79931
32964 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
32966 2017-04-24 Richard Biener <rguenther@suse.de>
32968 PR tree-optimization/80494
32969 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
32970 out for complex types.
32972 2017-04-24 Richard Biener <rguenther@suse.de>
32974 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
32975 * tree-ssa-sccvn.c (print_scc): Print SCC size.
32976 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
32977 (DFS): Adjust and never fail.
32978 (sccvn_dom_walker::fail): Remove.
32979 (sccvn_dom_walker::before_dom_children): Adjust.
32980 (run_scc_vn): Likewise and never fail.
32981 * tree-ssa-pre.c (pass_pre::execute): Adjust.
32982 (pass_fre::execute): Likewise.
32984 2017-04-24 Richard Biener <rguenther@suse.de>
32986 PR tree-optimization/79725
32987 * tree-ssa-sink.c (statement_sink_location): Return whether
32988 failure reason was zero uses. Move that check later.
32989 (sink_code_in_bb): Deal with zero uses by removing the stmt
32992 2017-04-24 Richard Biener <rguenther@suse.de>
32995 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
32996 pointer-based references.
32998 2017-04-24 Richard Biener <rguenther@suse.de>
33001 * pass_manager.h (pass_manager::operator new): Remove.
33002 (pass_manager::operator delete): Likewise.
33003 * passes.c (pass_manager::operator new): Remove.
33004 (pass_manager::operator delete): Likewise.
33005 (pass_manager::pass_manager): Zero individual pass members.
33007 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
33010 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
33011 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
33012 Check "XEXP (src, 1)" operand here.
33013 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
33014 Check "XEXP (src, 1)" operand here.
33015 (dimode_scalar_chain::make_vector_copies): Detect count register
33016 of a shift instruction. Zero extend count register from QImode
33017 to DImode to satisfy vector shift pattern count operand predicate.
33018 Substitute vector shift count operand with a DImode copy.
33019 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
33022 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
33024 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
33025 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33026 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
33027 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33028 (UNSPEC_NOREX_MEM): Remove definition.
33030 2017-04-21 Richard Biener <rguenther@suse.de>
33032 PR tree-optimization/79547
33033 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
33034 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
33035 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
33036 without any constraints.
33038 2017-04-21 Richard Biener <rguenther@suse.de>
33040 PR tree-optimization/78847
33041 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
33043 2017-04-21 Richard Biener <rguenther@suse.de>
33045 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
33046 (build_distinct_type_copy): Likewise.
33047 (build_variant_type_copy): Likewise.
33048 * tree.c (build_qualified_type): Pass down mem-stat info.
33049 (build_distinct_type_copy): Likewise.
33050 (build_variant_type_copy): Likewise.
33052 2017-04-21 Richard Biener <rguenther@suse.de>
33054 PR tree-optimization/80237
33055 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
33057 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
33058 for a simplified result.
33060 2016-04-21 Richard Biener <rguenther@suse.de>
33062 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
33063 sth as strict as a simple_iv but a chrec without symbols and an
33064 operand defined in the loop we are peeling (and not some subloop).
33065 (propagate_constants_for_unrolling): Propagate all constants.
33067 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33070 * config/i386/i386.c (print_reg): Remove assert for disalowed
33071 regno values, call output_operand_lossage instead.
33073 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33076 * config/i386/constraints.md (Yc): New register constraint.
33077 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
33078 Use Yc constraint for alternative 2 of operand 0. Remove
33079 preferred_for_speed attribute.
33081 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33083 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
33084 lastprivate clauses in SIMT case.
33086 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
33088 * doc/invoke.texi (-Wextra-semi): Document new warning option.
33090 2017-04-20 Richard Biener <rguenther@suse.de>
33092 PR tree-optimization/57796
33093 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
33094 as N scalar stores.
33095 (vect_model_load_cost): Cost gathers as N scalar loads.
33097 2017-04-20 Richard Biener <rguenther@suse.de>
33099 * ggc-page.c (ggc_allocated_p): Rename to ...
33100 (safe_lookup_page_table_entry): ... this and return the lookup
33102 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
33104 2017-04-20 Richard Biener <rguenther@suse.de>
33106 PR tree-optimization/80453
33107 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
33108 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
33109 from the conditions.
33110 (vn_phi_eq): Pass them down.
33111 (vn_phi_lookup): Record them.
33112 (vn_phi_insert): Likewise.
33114 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
33116 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
33117 uninitialized variable warning to avoid buffer overrun.
33119 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33122 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
33123 is suppressed for '{ 0 }' in C.
33125 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33127 * BASE-VER: Set to 8.0.0.
33129 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
33131 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
33132 priority .init_array and .fini_array section with SECTION_NOTYPE
33135 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33137 PR middle-end/80423
33138 * tree.h (build_array_type): Add typeless_storage default argument.
33139 * tree.c (type_cache_hasher::equal): Also compare
33140 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
33141 (build_array_type): Add typeless_storage argument, set
33142 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
33144 (build_nonshared_array_type): Adjust build_array_type_1 caller.
33145 (build_array_type): Likewise. Add typeless_storage argument.
33147 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33148 Jakub Jelinek <jakub@redhat.com>
33150 PR tree-optimization/80426
33151 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
33152 operation on symbolic operands, also compute the overflow for the
33153 invariant part when the operation degenerates into a negation.
33155 2017-04-19 Jakub Jelinek <jakub@redhat.com>
33158 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
33159 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
33162 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
33164 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
33167 * config/avr/avr.c (tree.h): Include it.
33168 (cgraph.h): Include it.
33169 (avr_encode_section_info): Don't warn for uninitialized progmem
33170 variable if it's just an alias.
33172 2017-04-19 Richard Biener <rguenther@suse.de>
33175 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
33176 when needed by AutoPGO.
33178 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
33181 * doc/lto.texi: Remove an extra 'that'.
33183 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
33185 PR rtl-optimization/80429
33186 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
33187 are only used in debug insns.
33189 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33190 Vladimir Makarov <vmakarov@redhat.com>
33192 * config/sparc/predicates.md (input_operand): Add comment. Return
33193 true for any memory operand when LRA is in progress.
33194 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
33196 2017-04-18 Jeff Law <law@redhat.com>
33199 * mips.md ({return,simple_return}_internal): Do not overwrite
33202 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33204 PR tree-optimization/80443
33205 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
33206 instead of adding 1, subtract -1 and similarly instead of subtracting
33209 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
33211 PR rtl-optimization/80357
33212 * haifa-sched.c (tmp_bitmap): New variable.
33213 (model_recompute): Handle duplicate use records.
33214 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
33215 (free_global_sched_pressure_data): Free it.
33217 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
33220 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33221 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33222 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33223 instead of SYSTEM_HEADER_DIR.
33225 2017-04-18 Jeff Law <law@redhat.com>
33227 PR middle-end/80422
33228 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
33229 predecessors after walking up the insn chain.
33231 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33234 * dwarf2out.c (modified_type_die): Try harder not to emit internal
33235 sizetype type into debug info.
33237 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33240 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
33241 unneeded test for TARGET_UPPER_REGS_SF.
33242 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
33244 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33247 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
33248 instead of gsi_after_labels.
33250 2017-04-18 Jeff Law <law@redhat.com>
33252 * regcprop.c (maybe_mode_change): Avoid creating copies of the
33256 2017-04-13 Jeff Law <law@redhat.com>
33257 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33258 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33260 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
33263 * config/avr/avr.c (intl.h): Include it.
33264 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
33266 2017-04-18 Martin Liska <mliska@suse.cz>
33268 PR gcov-profile/78783
33269 * gcov-tool.c (gcov_output_files): Validate that destination
33270 file is either removed by the tool or by a user.
33272 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33273 Guy Benyei <guybe@mellanox.com>
33275 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
33276 block, and do not negate it, the stored id is already negative.
33278 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33280 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
33282 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33285 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
33286 masks of options that should be turned off if the VSX vector
33287 options are turned off.
33288 (OTHER_P8_VECTOR_MASKS): Likewise.
33289 (OTHER_VSX_VECTOR_MASKS): Likewise.
33290 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
33291 rs6000_disable_incompatible_switches to validate no type switches
33293 (rs6000_incompatible_switch): New function to disallow turning on
33294 other vector options if -mno-vsx, -mno-power8-vector, or
33295 -mno-power9-vector are specified.
33297 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33299 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
33301 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33303 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
33304 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
33305 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
33306 (ARG_POINTER_CFA_OFFSET): Likewise.
33308 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33310 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
33311 conditions to take advantage of various optimizations.
33313 2017-04-13 Jeff Law <law@redhat.com>
33315 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33316 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33317 (zero_extendsidi2_dext): Likewise.
33319 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33322 * fold-const.c (fold_ternary_loc): Revert
33323 use op0 instead of fold_convert_loc (loc, type, arg0) part of
33326 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
33328 PR rtl-optimization/80343
33329 * lra-remat.c (update_scratch_ops): Assign original hard reg to
33330 new scratch pseudo.
33332 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
33335 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
33336 to ubsan_encode_value.
33338 2017-04-13 Jeff Law <law@redhat.com>
33340 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
33341 appearing in DEBUG_INSNs.
33343 2017-04-13 Martin Liska <mliska@suse.cz>
33345 PR gcov-profile/80413
33346 * gcov-io.c (gcov_write_string): Copy to buffer just when
33347 allocated size is greater than zero.
33349 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33352 * dwarf2out.c (decls_for_scope): Ignore declarations of
33353 current_function_decl in BLOCK_NONLOCALIZED_VARS.
33355 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
33358 * ipa-visibility.c (non_local_p): Fix typos.
33359 (localize_node): When localizing symbol in same comdat group,
33360 dissolve the group only when we know external symbols are going
33362 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
33364 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33366 PR tree-optimization/79390
33367 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
33368 order does not result in usable sequence, retry with reversed operand
33374 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
33375 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
33376 op0 instead of fold_convert_loc (loc, type, arg0).
33378 2017-04-12 Jeff Law <law@redhat.com>
33380 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
33381 has a delay slot in the generated code.
33383 * config/cris/cris.md (cris_preferred_reload_class): Return
33384 GENNONACR_REGS rather than GENERAL_REGS.
33386 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33389 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
33390 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
33391 signedness of the result type.
33393 2017-04-12 Richard Biener <rguenther@suse.de>
33394 Jeff Law <law@redhat.com>
33396 PR tree-optimization/80359
33397 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
33398 trim stores to TARGET_MEM_REFs.
33400 2017-04-12 Richard Biener <rguenther@suse.de>
33402 PR tree-optimization/79390
33403 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
33404 threading case even more.
33406 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
33409 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
33410 for quad_address_p for TImode, instead of just not indexed_address.
33412 2017-04-12 Richard Biener <rguenther@suse.de>
33413 Bernd Edlinger <bernd.edlinger@hotmail.de>
33415 PR middle-end/79671
33416 * alias.c (component_uses_parent_alias_set_from): Handle
33417 TYPE_TYPELESS_STORAGE.
33418 (get_alias_set): Likewise.
33419 * tree-core.h (tree_type_common): Add typeless_storage flag.
33420 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
33421 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
33422 for types containing members with TYPE_TYPELESS_STORAGE.
33423 (place_field): Likewise.
33424 (layout_type): Likewise for ARRAY_TYPE.
33425 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
33426 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
33427 TYPE_TYPELESS_STORAGE.
33428 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
33430 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33433 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33434 first argument to type.
33436 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33440 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
33441 CONST0_RTX (mode) rather than const0_rtx where appropriate.
33442 (rs6000_expand_binop_builtin): Likewise.
33443 (rs6000_expand_ternop_builtin): Likewise; also add missing
33444 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
33445 vshasigma built-ins.
33446 * doc/extend.texi: Document that vec_xxpermdi's third argument
33447 must be a constant.
33449 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
33451 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
33452 Use shift_const cost parameter when calculating gain of STV shifts.
33454 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
33456 PR rtl-optimization/70478
33457 * lra-constraints.c (process_alt_operands): Check memory for
33458 disfavoring memory insn operand.
33460 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33462 PR middle-end/80100
33463 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
33464 left shift in unsigned HOST_WIDE_INT type.
33466 PR rtl-optimization/80385
33467 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
33468 (not (neg X)) into (plus X -1) for complex or non-integral modes.
33471 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
33472 if they have any depend clauses.
33474 2017-04-11 Martin Liska <mliska@suse.cz>
33477 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
33478 * ipa-split.c (split_function): Create a local comdat symbol
33479 if caller is in a comdat group.
33481 2017-04-11 Martin Liska <mliska@suse.cz>
33484 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
33487 2017-04-11 Martin Sebor <msebor@redhat.com>
33489 PR middle-end/80364
33490 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
33491 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
33493 (directive::set_width, directive::set_precision, format_character):
33495 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
33498 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
33501 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
33502 conflict, set target->arch_name instead of target->cpu_name.
33504 2017-04-11 Richard Biener <rguenther@suse.de>
33506 PR tree-optimization/80374
33507 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
33508 build_zero_cst, remove fold_convertible_p check again.
33510 2017-04-11 Martin Liska <mliska@suse.cz>
33513 * ubsan.c (instrument_object_size): Do not instrument register
33516 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33519 * config/i386/i386-builtin-types.def
33520 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
33521 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
33522 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
33523 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
33524 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
33525 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
33526 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
33527 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
33528 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
33529 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
33530 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
33531 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
33532 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
33533 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
33534 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
33535 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
33536 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
33537 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
33538 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
33539 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
33540 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
33541 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
33542 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
33543 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
33544 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
33545 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
33546 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
33547 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
33548 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
33549 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
33550 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
33551 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
33552 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
33553 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
33554 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
33555 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
33556 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
33557 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
33558 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
33559 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
33560 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
33561 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
33562 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
33563 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
33564 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
33565 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
33567 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
33568 flag to second_arg_count, handle 4 argument function type _COUNT
33569 aliases, handle second_arg_count on second argument rather than last.
33571 2017-04-10 Jeff Law <law@redhat.com>
33573 PR tree-optimization/80374
33574 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
33575 record anything if we can not convert integer_zero_node to the
33578 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33581 * config/rs6000/rs6000.c (rs6000_option_override_internal):
33582 Enhance special handling given to the TARGET_P9_MINMAX option in
33583 relation to certain other options.
33585 2017-04-10 Bin Cheng <bin.cheng@arm.com>
33587 PR tree-optimization/80153
33588 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
33589 remove POINTER_PLUS_EXPR's base part directly, rather than through
33592 2017-04-10 Richard Biener <rguenther@suse.de>
33593 Bin Cheng <bin.cheng@arm.com>
33595 PR tree-optimization/80153
33596 * tree-affine.c (aff_combination_to_tree): Get base pointer from
33597 the first element of pointer type aff_tree. Build result expr in
33599 (add_elt_to_tree): Convert to type unconditionally. Remove other
33600 fold_convert calls.
33601 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
33602 (rewrite_use_nonlinear_expr): Check invariant using iv information.
33604 2017-04-10 Richard Biener <rguenther@suse.de>
33606 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33609 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
33611 PR rtl-optimization/70478
33612 * lra-constraints.c (curr_small_class_check): New.
33613 (update_and_check_small_class_inputs): New.
33614 (process_alt_operands): Update curr_small_class_check. Disfavor
33615 alternative insn memory operands. Check available regs for small
33618 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33621 * config/mips/mips.opt (-mvirt): Update description.
33622 * doc/invoke.texi (-mvirt): Likewise.
33624 2017-04-10 Richard Biener <rguenther@suse.de>
33626 PR middle-end/80362
33627 * fold-const.c (fold_binary_loc): Look at unstripped ops when
33628 looking for NEGATE_EXPR in -A / -B to A / B folding.
33630 2017-04-10 Martin Liska <mliska@suse.cz>
33632 PR gcov-profile/80224
33633 * gcov.c (print_usage): Fix usage string.
33634 (get_gcov_intermediate_filename): Remove.
33635 (output_gcov_file): Use both for normal and intermediate format.
33636 (generate_results): Do not initialize special file for
33637 intermediate format.
33639 2017-04-10 Richard Biener <rguenther@suse.de>
33641 PR tree-optimization/80304
33642 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
33645 2017-04-10 Nathan Sidwell <nathan@acm.org>
33648 * config/rs6000/rs6000.c (rs6000_vector_type): New.
33649 (rs6000_init_builtins): Use it.
33651 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33653 * config/arm/arm.md (<mrc>): Add mode to SET source.
33654 (<mrrc>): Likewise.
33656 2017-04-10 Richard Biener <rguenther@suse.de>
33658 PR middle-end/80344
33659 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
33661 2017-04-10 Jakub Jelinek <jakub@redhat.com>
33664 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
33665 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
33666 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
33667 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
33668 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
33669 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
33670 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
33671 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
33672 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
33673 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
33674 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
33675 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
33676 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
33677 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
33678 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
33679 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
33680 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
33681 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
33682 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
33683 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
33684 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
33685 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
33686 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
33688 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
33690 PR rtl-optimization/70478
33691 * lra-constraints.c: Reverse the last patch.
33693 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
33695 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
33696 Add comment for WCHAR_T.
33698 2017-04-08 Martin Liska <mliska@suse.cz>
33701 2017-04-07 Martin Liska <mliska@suse.cz>
33704 * ipa-split.c (split_function): Add function part to a same comdat
33707 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33710 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
33712 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
33714 * rs6000/rs6000.c (vec_load_pendulum): Rename...
33715 (vec_pairing): ...to this.
33716 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
33717 (rs6000_sched_init): Adjust for name change.
33718 (struct rs6000_sched_context): Likewise.
33719 (rs6000_init_sched_context): Likewise.
33720 (rs6000_set_sched_context): Likewise.
33722 2017-04-07 Jakub Jelinek <jakub@redhat.com>
33728 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
33730 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
33731 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
33732 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
33734 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
33736 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
33738 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33740 PR rtl-optimization/70703
33741 * ira-color.c (update_conflict_hard_regno_costs): Use
33742 int64_t instead of HOST_WIDE_INT.
33744 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33746 PR rtl-optimization/70478
33747 * lra-constraints.c (process_alt_operands): Disfavor alternative
33748 insn memory operands.
33750 2017-04-07 Jeff Law <law@redhat.com>
33752 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
33753 CALL and NOTE_INSN_CALL_ARG_LOCATION.
33755 2017-04-07 Martin Liska <mliska@suse.cz>
33758 * config/aarch64/aarch64.c (aarch64_process_target_attr):
33759 Show error message instead of an ICE.
33761 2017-04-07 Martin Liska <mliska@suse.cz>
33764 * ipa-split.c (split_function): Add function part to a same comdat
33767 2017-04-07 Richard Biener <rguenther@suse.de>
33769 PR middle-end/80341
33770 * tree.c (get_unwidened): Also handle ! for_type case for
33772 * convert.c (do_narrow): Split out from ...
33773 (convert_to_integer_1): ... here. Do not pass final truncation
33774 type to get_unwidened for TRUNC_DIV_EXPR.
33776 2017-04-07 Richard Biener <rguenther@suse.de>
33778 * tree-affine.c (wide_int_ext_for_comb): Take type rather
33780 (aff_combination_const): Adjust.
33781 (aff_combination_scale): Likewise.
33782 (aff_combination_add_elt): Likewise.
33783 (aff_combination_add_cst): Likewise.
33784 (aff_combination_convert): Likewise.
33785 (add_elt_to_tree): Likewise. Remove unused argument.
33786 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
33788 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
33790 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
33792 * config/arm/arm.c (arm_default_short_enums): Use
33793 ARM_DEFAULT_SHORT_ENUMS.
33794 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
33796 2017-04-06 Jakub Jelinek <jakub@redhat.com>
33799 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
33800 members with redundant out-of-class redeclaration.
33802 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33805 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
33806 * config/i386/i386.md (*zero_extendsidi2):
33807 Add (?*x,*x) and (?*v,*v) alternatives.
33809 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33812 * config/i386/i386.c (ix86_expand_builtin)
33813 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
33814 mode from insn data. Convert operands to insn operand mode.
33815 Copy operands that don't satisfy insn predicate to a register.
33817 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
33819 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
33822 2017-04-06 Richard Biener <rguenther@suse.de>
33824 PR tree-optimization/80334
33825 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
33826 preserve alignment of accesses.
33828 2017-04-06 Richard Biener <rguenther@suse.de>
33830 PR tree-optimization/80262
33831 * tree-sra.c (build_ref_for_offset): Preserve address-space
33833 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
33834 Drop useless address-space information on MEM_REF offsets.
33836 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
33838 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
33840 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33842 PR rtl-optimization/70703
33843 * ira-color.c (update_conflict_hard_regno_costs): Use
33844 HOST_WIDE_INT instead of long.
33846 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
33849 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
33850 not defined for x86_64 target. Add -mmmx target option when __SSE2__
33852 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
33853 for x86_64 target. Handle -m3dnowa option.
33855 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33857 PR rtl-optimization/70703
33858 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
33859 (update_conflict_hard_regno_costs): Use long instead of unsigned
33860 arithmetic for cost calculation.
33862 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33863 Bernd Edlinger <bernd.edlinger@hotmail.de>
33866 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
33869 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
33872 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
33873 ptr_mode with Pmode throughout.
33874 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
33875 into probe_stack_range and use DImode.
33877 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
33880 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
33881 call_eh_return is true.
33883 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33885 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
33886 Initialize last_match_fntype_index.
33888 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33891 * tree-nvr.c: Include internal-fn.h.
33892 (pass_return_slot::execute): Ignore internal calls without
33895 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33896 Richard Biener <rguenther@suse.de>
33899 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
33900 captures used multiple times, except for the last use.
33901 * generic-match-head.c: Include gimplify.h.
33903 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33905 PR tree-optimization/79390
33906 * target.h (struct noce_if_info): Declare.
33907 * targhooks.h (default_noce_conversion_profitable_p): Declare.
33908 * target.def (noce_conversion_profitable_p): New target hook.
33909 * ifcvt.h (struct noce_if_info): New type, moved from ...
33910 * ifcvt.c (struct noce_if_info): ... here.
33911 (noce_conversion_profitable_p): Renamed to ...
33912 (default_noce_conversion_profitable_p): ... this. No longer
33914 (noce_try_store_flag_constants, noce_try_addcc,
33915 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
33916 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
33917 instead of noce_conversion_profitable_p.
33918 * config/i386/i386.c: Include ifcvt.h.
33919 (ix86_option_override_internal): Don't override
33920 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
33921 (ix86_noce_conversion_profitable_p): New function.
33922 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
33923 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
33924 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
33925 * doc/tm.texi: Regenerated.
33927 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33929 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
33932 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
33935 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
33936 instructions for small multiply cores.
33938 2017-04-04 Jeff Law <law@redhat.com>
33940 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
33942 (mips_expand_vec_perm_const): Initialize elements in orig_perm
33943 that are not set by the loop over the elements.
33945 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33948 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
33949 int mode, convert_modes it to mode as unsigned, otherwise use
33950 lowpart_subreg to mode rather than SImode.
33951 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
33952 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
33953 Use DImode instead of SImode for the shift count operand.
33954 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
33957 2017-04-04 Richard Biener <rguenther@suse.de>
33959 PR middle-end/80281
33960 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
33961 arithmetic done for the negate or the plus. Simplify.
33962 (A - (-B) -> A + B): Likewise.
33963 * fold-const.c (split_tree): Make sure to not negate pointers.
33965 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
33967 PR rtl-optimization/60818
33968 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
33969 a compare of comparisons with the thing compared if this results
33970 in a different machine mode.
33972 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
33974 * alias.c (base_alias_check): Fix typo in comment.
33975 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
33976 * cgraphunit.c (symbol_table::compile): Likewise.
33977 * collect2.c (maybe_run_lto_and_relink): Likewise.
33978 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
33979 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
33980 * config/avr/avr.c (avr_map_op_t): Likewise.
33981 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
33982 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
33983 * config/epiphany/epiphany.md (movcc): Likewise.
33984 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
33985 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
33987 * config/mips/mips.c (mips_save_restore_reg): Likewise.
33988 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
33989 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
33990 * config/sh/sh.c (sh_rtx_costs): Likewise.
33991 * fold-const.c (fold_truth_andor): Likewise.
33992 * genautomata.c (collapse_flag): Likewise.
33993 * gengtype.h (struct type::u::s): Likewise.
33994 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
33995 * input.c (FORMAT_AMOUNT): Likewise.
33996 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
33997 (known_aggs_to_agg_replacement_list): Likewise.
33998 * ipa-inline-analysis.c: Likewise.
33999 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
34000 * ipa-polymorphic-call.c
34001 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
34002 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
34003 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
34005 * modulo-sched.c (apply_reg_moves): Likewise.
34006 * omp-expand.c (build_omp_regions_1): Likewise.
34007 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
34008 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
34009 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
34010 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
34011 * value-prof.c: Likewise.
34012 * var-tracking.c (val_reset): Likewise.
34014 2017-04-03 Richard Biener <rguenther@suse.de>
34016 PR tree-optimization/80275
34017 * fold-const.c (split_address_to_core_and_offset): Handle
34020 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
34022 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
34023 descriptors is at least equal to that of functions.
34025 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34027 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
34029 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34032 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
34033 (mov<IMOD4:mode>): New expander.
34034 (*mov<IMOD4:mode>_internal): New insn and split pattern.
34036 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
34038 PR rtl-optimization/79405
34039 * fwprop.c (propagations_left): New variable.
34040 (forward_propagate_into): Decrement it.
34041 (fwprop_init): Initialize it.
34042 (fw_prop): If the variable has reached zero, stop propagating.
34043 (fwprop_addr): Ditto.
34045 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34048 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
34049 a FUNCTION_DECL, pass it as decl instead of origin to
34052 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
34054 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
34057 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
34060 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
34061 TARGET_VSX_SMALL_INTEGER.
34063 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34065 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
34066 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
34068 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34070 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
34071 extraction from odd-numbered MSA register.
34073 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34075 PR middle-end/80173
34076 * expmed.c (store_bit_field_1): Don't attempt to create
34077 a word subreg out of hard registers wider than word if they
34078 have HARD_REGNO_NREGS of 1 for their mode.
34080 PR middle-end/80163
34081 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
34082 conversions to integer types wider than word and pointer.
34085 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
34086 (rtx_equal_for_cselib_p): Pass 0 to it.
34087 * cselib.c (cselib_hasher::equal): Likewise.
34088 (rtx_equal_for_cselib_1): Add depth argument. If depth
34089 is 128, don't look up VALUE locs and punt. Increment
34090 depth in recursive calls when walking VALUE locs.
34092 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
34094 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
34095 (make_gcov_file_name): Use the canonical path name for generating
34097 (read_line): Fix handling of files with ascii null bytes.
34099 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
34101 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
34102 to initialise a vector register instead
34103 of using a const_int.
34105 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34107 PR translation/80189
34108 * gimplify.c (omp_default_clause): Use %qs instead of %s in
34109 diagnostic messages.
34111 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
34114 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
34115 (dfp_diex_<mode>): Update mode of operand 1.
34116 * doc/extend.texi (dxex, dxexq): Document change to return type.
34117 (diex, diexq): Document change to argument type.
34119 2017-03-30 Martin Jambor <mjambor@suse.cz>
34122 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
34123 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
34124 it reflects the signature changes performed at the callee side.
34125 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
34126 to cgraph_build_function_type_skip_args.
34127 (build_function_decl_skip_args): Adjust call to the above function.
34129 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34132 * config/i386/sse.md
34133 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
34134 register as dest whenever it is a MEM not rtx_equal_p to the
34135 corresponding dup operand, and when forcing into reg move the
34136 reg into the memory afterwards.
34137 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
34138 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
34139 for the force_reg mode.
34140 (avx512vl_vextractf128<mode>): Use register as dest either
34141 always when a MEM, or when it is a MEM not rtx_equal_p to the
34142 corresponding dup operand, or even not when it is a CONST_VECTOR
34143 depending on the mode and lo vs. hi.
34144 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
34146 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
34147 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
34148 Likewise. Require that operands[2] is even.
34149 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
34150 Remove extraneous parens. Require that operands[2] is a multiple
34152 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
34153 operands[0] is a MEM if <mask_applied>, the predicates/constraints
34154 disallow memory then.
34156 2017-03-30 Richard Biener <rguenther@suse.de>
34158 PR tree-optimization/77498
34159 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
34160 to non-constants over backedges.
34162 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
34164 PR rtl-optimization/80233
34165 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
34166 as last_combined_insn. Do not test for BARRIER_P separately.
34168 2017-03-29 Andreas Schwab <schwab@suse.de>
34171 * calls.c (prepare_call_address): Convert funexp to Pmode before
34172 copying to temp reg.
34174 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34176 PR tree-optimization/80158
34177 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
34178 Handle possible future case of more than one alternate
34180 (replace_rhs_if_not_dup): Likewise.
34181 (replace_one_candidate): Likewise.
34183 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
34185 PR rtl-optimization/80193
34186 * ira.c (ira): Do not check allocation for LRA.
34188 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
34190 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
34191 (nvptx_output_simt_exit): Declare.
34192 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
34193 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
34194 (init_softstack_frame): Move initialization of crtl->is_leaf to...
34195 (nvptx_declare_function_name): ...here. Emit declaration of local
34196 memory space buffer for omp_simt_enter insn.
34197 (nvptx_output_unisimt_switch): New.
34198 (nvptx_output_softstack_switch): New.
34199 (nvptx_output_simt_enter): New.
34200 (nvptx_output_simt_exit): New.
34201 * config/nvptx/nvptx.h (struct machine_function): New fields
34202 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
34203 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
34204 (UNSPECV_SIMT_EXIT): Ditto.
34205 (omp_simt_enter_insn): New insn.
34206 (omp_simt_enter): New expansion.
34207 (omp_simt_exit): New insn.
34208 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
34210 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
34211 (expand_GOMP_SIMT_ENTER_ALLOC): New.
34212 (expand_GOMP_SIMT_EXIT): New.
34213 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
34214 (GOMP_SIMT_ENTER_ALLOC): Ditto.
34215 (GOMP_SIMT_EXIT): Ditto.
34216 * target-insns.def (omp_simt_enter): New insn.
34217 (omp_simt_exit): Ditto.
34218 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
34220 (lower_rec_simd_input_clauses): Implement SIMT privatization.
34221 (lower_rec_input_clauses): Likewise.
34222 (lower_lastprivate_clauses): Handle SIMT privatization.
34224 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
34225 (ompdevlow_adjust_simt_enter): New.
34226 (find_simtpriv_var_op): New.
34227 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
34228 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
34230 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
34231 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
34232 (copy_decl_for_dup_finish): Ditto.
34234 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
34236 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
34239 * config/i386/i386.c (ix86_option_override_internal): Always
34240 allow -mpreferred-stack-boundary=3 for 64-bit targets.
34242 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34244 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
34246 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34248 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
34249 mark new edge's irreducible flag accordign to it.
34250 (vect_do_peeling): Check loop preheader edge's irreducible flag
34251 and pass it to function slpeel_add_loop_guard.
34253 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
34255 PR tree-optimization/80218
34256 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34257 Update block frequencies and counts.
34259 2017-03-28 Richard Biener <rguenther@suse.de>
34261 PR tree-optimization/78644
34262 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
34263 of a simplification result we may not use it at all.
34265 2017-03-28 Richard Biener <rguenther@suse.de>
34268 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
34269 without arguments, generate default definition of a SSA name.
34271 2017-03-28 Richard Biener <rguenther@suse.de>
34273 PR middle-end/80222
34274 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
34275 TYPE_REF_CAN_ALIAS_ALL references.
34276 * fold-const.c (fold_indirect_ref_1): Likewise.
34278 2017-03-28 Martin Liska <mliska@suse.cz>
34281 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
34282 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
34284 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34285 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
34287 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
34288 (EXTRA_SPECS): Define.
34289 (SUBTARGET_EXTRA_SPECS): Likewise.
34290 (SUBTARGET_CPP_SPEC): Likewise.
34291 * config/arc/elf.h (EXTRA_SPECS): Renamed to
34292 SUBTARGET_EXTRA_SPECS.
34293 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
34295 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34297 * config/arc/simdext.md (vst64_insn): Update pattern.
34298 (vld32wh_insn): Likewise.
34299 (vld32wl_insn): Likewise.
34300 (vld64_insn): Likewise.
34301 (vld32_insn): Likewise.
34303 2017-03-28 Marek Polacek <polacek@redhat.com>
34306 * fold-const.c (fold_comparison): Use protected_set_expr_location
34307 instead of SET_EXPR_LOCATION.
34309 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
34311 * tree.c (add_expr): Avoid name lookup warning.
34313 2017-03-27 Jeff Law <law@redhat.com>
34315 PR tree-optimization/80216
34316 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
34317 function name. Limit recursion depth.
34318 (record_temporary_equivalences): Corresponding changes.
34320 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
34322 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
34325 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34328 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
34330 * cfgcleanup.c (reg_note_cfa_p): New array.
34331 (insns_have_identical_cfa_notes): New function.
34332 (old_insns_match_p): Don't cross-jump in between /f
34333 and non-/f instructions. If both i1 and i2 are frame related,
34334 verify all CFA notes, their order and content.
34336 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
34339 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
34340 HImode and SImode with zero extend to DImode to one insn.
34341 (bswap<mode>2_extenddi): Likewise.
34342 (bswapsi2_extenddi): Likewise.
34343 (bswaphi2_extendsi): Likewise.
34344 (bswaphi2): Combine bswap HImode and SImode into one insn.
34345 Separate memory insns from swapping register.
34346 (bswapsi2): Likewise.
34347 (bswap<mode>2): Likewise.
34348 (bswaphi2_internal): Delete, no longer used.
34349 (bswapsi2_internal): Likewise.
34350 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
34351 store, and gpr<-gpr swap insns.
34352 (bswap<mode>2_store): Likewise.
34353 (bswaphi2_reg): Register only splitter, combine with the splitter.
34354 (bswaphi2 splitter): Likewise.
34355 (bswapsi2_reg): Likewise.
34356 (bswapsi2 splitter): Likewise.
34357 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
34358 the insns into load, store, and register/register insns.
34359 (bswapdi2_ldbrx): Likewise.
34360 (bswapdi2_load): Likewise.
34361 (bswapdi2_store): Likewise.
34362 (bswapdi2_reg): Likewise.
34364 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
34366 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
34367 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
34369 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
34372 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
34374 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34375 special handling for target option conflicts between dform
34376 options (-mpower9-dform, -mpower9-dform-vector,
34377 -mpower9-dform-scalar) and -mno-direct-move.
34379 2017-03-27 Richard Biener <rguenther@suse.de>
34381 PR tree-optimization/80181
34382 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
34384 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34386 * config/arc/predicates.md (move_double_src_operand): Replace the
34387 call to move_double_src_operand with a call to address_operand.
34389 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34391 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
34392 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
34393 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
34395 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34397 * config/arc/predicates.md (long_immediate_loadstore_operand):
34398 Consider scaled addresses cases.
34400 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34402 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
34403 restored when in interrupt.
34404 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
34405 doesn't have delay slot.
34407 2017-03-27 Richard Biener <rguenther@suse.de>
34410 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
34411 inlined thunk clones.
34413 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34416 * asan.c (instrument_derefs): Copy over last operand from
34417 original COMPONENT_REF to the new COMPONENT_REF with
34418 DECL_BIT_FIELD_REPRESENTATIVE.
34419 * ubsan.c (instrument_object_size): Likewise.
34421 2017-03-27 Richard Biener <rguenther@suse.de>
34423 PR tree-optimization/80170
34424 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
34425 sure DR/SCEV didnt fold in constants we do not see when looking
34426 at the reference base alignment.
34428 2017-03-27 Richard Biener <rguenther@suse.de>
34430 PR middle-end/80171
34431 * gimple-fold.c (fold_ctor_reference): Properly guard against
34432 NULL return value from canonicalize_constructor_val.
34434 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
34437 * config/i386/i386.c (ix86_expand_builtin)
34438 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
34439 flags reg setting and flags reg using instructions.
34440 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
34441 clobbering instructions to zero extend op2.
34443 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
34445 * doc/install.texi (Configuration) <--with-aix-soname>:
34446 Update link to AIX ld.
34448 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
34450 PR rtl-optimization/80160
34451 PR rtl-optimization/80159
34452 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
34453 reg_alternate_class into account.
34455 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
34458 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
34459 to consider in curr_insn_transform.
34461 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34463 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
34464 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
34465 and emit_mode_inner.
34467 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34469 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
34470 argument to the overloaded builtin variants. Use the new flag to
34471 deprecate certain builtin variants.
34472 * config/s390/s390-builtin-types.def: Add new builtin types.
34473 * config/s390/s390-builtins.h: Support new flags field for
34474 overloaded builtins.
34475 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
34476 (s390_macro_to_expand): Enable vector float data type.
34477 (s390_cpu_cpp_builtins_internal): Indicate support of the new
34478 builtins by incrementing the __VEC__ version number.
34479 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
34481 (s390_resolve_overloaded_builtin): Emit error messages depending
34482 on the builtin flags.
34483 * config/s390/s390.c (s390_expand_builtin): Support additional
34484 flags argument. Change error message to match the messages
34485 emitted in s390-c.c.
34486 * config/s390/s390.md: New UNSPEC_* constants.
34487 (op_type): Add new instruction types.
34488 * config/s390/vecintrin.h: Add new builtins and test data class
34490 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
34491 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
34492 (VEC_INEXACT, VEC_NOINEXACT): New constants.
34493 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
34494 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
34495 ("vec_mergel<mode>"): V_HW -> VEC_HW.
34497 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
34498 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
34499 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
34500 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
34502 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
34503 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
34504 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
34505 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
34507 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
34508 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
34509 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
34510 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
34511 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
34512 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
34513 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
34515 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
34516 ("vec_scatter_element<V_HW_4:mode>_DI")
34517 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
34518 ("vec_fpint<mode>", "vflls")
34519 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
34520 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
34521 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
34522 ("*vec_cmphe<mode>_cc"): ... these.
34524 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
34525 mode constant instead of magic value.
34527 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34529 * config/s390/s390.c (s390_expand_vec_compare): Support other
34530 vector floating point modes than just V2DF.
34531 (s390_expand_vcond): Likewise.
34532 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
34533 (s390_cannot_change_mode_class): Prevent mode changes between TF
34534 and V1TF in vector registers.
34535 * config/s390/s390.md (DF, SF): New mode attributes.
34536 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
34537 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
34538 SFmode support for VRs.
34539 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
34541 (VFT, VF_HW): New mode iterators.
34542 (vw, sdx): New mode attributes.
34543 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
34544 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
34545 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
34546 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
34547 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
34548 also the new vector floating point modes. Renaming to ...
34550 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
34551 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
34552 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
34553 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
34554 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
34555 ("vec_unordered<mode>"): ... these.
34557 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
34558 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
34559 ("*vec_extendv2df"): New insn definitions.
34561 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34563 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
34564 ("mulditi3_2", "*muldi3_sign"): New patterns.
34565 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
34566 rename the pattern definition.
34568 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34570 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
34572 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
34574 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34576 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
34577 instruction if possible.
34578 * config/s390/vector.md (vec_halfnumelts): New mode
34580 ("*vec_vllezlf<mode>"): New pattern.
34582 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34584 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
34585 ("popcountv4si2", "popcountv2di2"): Rename to ...
34586 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
34587 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
34589 ("popcount<mode>2_vxe"): New pattern.
34591 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34593 * common/config/s390/s390-common.c (processor_flags_table): Add
34595 * config.gcc: Add arch12.
34596 * config/s390/driver-native.c (s390_host_detect_local_cpu):
34597 Default to arch12 for unknown CPU model numbers.
34598 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
34599 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
34600 PROCESSOR_max sanity check.
34601 * config/s390/s390-opts.h (enum processor_type): Add
34603 * config/s390/s390.c (processor_table): Add arch12.
34604 (s390_expand_builtin): Add check for B_VXE flag.
34605 (s390_issue_rate): Add PROCESSOR_ARCH12.
34606 (s390_get_sched_attrmask): Likewise.
34607 (s390_get_unit_mask): Likewise.
34608 (s390_sched_score): Enable z13 scheduling for arch12.
34609 (s390_sched_reorder): Likewise.
34610 (s390_sched_variable_issue): Likewise.
34611 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
34613 (s390_tune_attr): Use z13 scheduling also for arch12.
34614 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
34615 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
34616 (TARGET_VXE_P): New macros.
34617 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
34618 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
34619 * config/s390/s390.opt: Add arch12 as processor_type.
34621 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34623 * config/s390/s390.md
34624 ("fixuns_truncdddi2", "fixuns_trunctddi2")
34625 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
34626 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
34628 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
34629 Rename expanders to ...
34631 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
34632 ("fixuns_truncdddi2_emu"): ... these.
34634 ("fixuns_trunc<mode>si2_emu"): New expander.
34636 ("*fixuns_truncdfdi2_z13"): Rename to ...
34637 ("*fixuns_truncdfdi2_vx"): ... this.
34639 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34641 * config/s390/2964.md: Remove the single element vector compare
34642 instructions which are no longer used.
34643 * config/s390/s390.c (s390_select_ccmode): Remove handling of
34645 (s390_canonicalize_comparison): Remove handling of DFmode
34647 (s390_expand_vec_compare_scalar): Remove function.
34648 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
34649 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
34651 ("*cmp<mode>_ccs"): Add wfcdb instruction.
34653 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34655 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
34657 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
34658 will anyway by matched by mov<mode>_64dfp.
34660 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34662 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
34663 vlef/vstef. Add missing operand to vleif.
34665 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34667 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
34668 pair for all vector types with 64 bit elements.
34669 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
34670 * config/s390/vector.md (V_HW_64): ... here.
34671 (V_128_NOSINGLE): New mode iterator.
34672 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
34673 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
34674 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
34675 ("*vec_load_pairv2di"): Change to ...
34676 ("*vec_load_pair<mode>"): ... this one.
34678 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34680 * config/s390/constraints.md: Add comments.
34681 (jKK): Reject element sizes > 8 bytes.
34682 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
34684 * config/s390/s390.md: Add the s_operand checks formerly in
34685 s390_split_ok_p to various splitters where they are still
34687 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
34688 for 128 bit vectors. Plus two splitters.
34690 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34692 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
34695 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34698 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
34699 error if the boundary argument is not constant.
34701 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34703 PR rtl-optimization/80112
34704 * loop-doloop.c (doloop_condition_get): Don't check condition
34705 if cmp isn't SET with IF_THEN_ELSE src.
34707 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34709 PR tree-optimization/80158
34710 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
34711 replacing a candidate statement, also replace it for the
34712 candidate's alternate interpretation.
34713 (replace_rhs_if_not_dup): Likewise.
34714 (replace_one_candidate): Likewise.
34716 2017-03-24 Richard Biener <rguenther@suse.de>
34718 PR tree-optimization/80167
34719 * graphite-isl-ast-to-gimple.c
34720 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
34722 (translate_isl_ast_to_gimple::get_rename): Likewise.
34724 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34726 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
34727 handling of certain combinations of target options, including the
34728 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
34729 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
34731 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34734 * config/arm/arm.md (*load_multiple): Add reload_completed to
34735 matching condition.
34737 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34738 Richard Biener <rguenth@suse.de>
34740 PR tree-optimization/79908
34741 PR tree-optimization/80136
34742 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34743 been cast away, gimplify_and_add suffices.
34745 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
34747 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
34749 2017-03-23 Richard Biener <rguenther@suse.de>
34751 PR tree-optimization/80032
34752 * gimplify.c (gimple_push_cleanup): Forced unconditional
34753 cleanups still have to go to the conditional_cleanups
34756 2017-03-22 Jakub Jelinek <jakub@redhat.com>
34758 PR tree-optimization/80072
34759 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
34761 (next_operand_entry_id): Change type to unsigned int.
34762 (sort_by_operand_rank): Make sure to return the right return value
34763 even if unsigned fields are bigger than INT_MAX.
34764 (struct oecount): Change cnt and id type to unsigned int.
34765 (oecount_hasher::equal): Formatting fix.
34766 (oecount_cmp): Make sure to return the right return value
34767 even if unsigned fields are bigger than INT_MAX.
34768 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
34771 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
34772 TREE_READONLY on result if writing it more than once.
34775 * doc/invoke.texi (-fsanitize=thread): Document that with
34776 -fnon-call-exceptions atomics are not able to throw
34780 * tsan.c: Include tree-eh.h.
34781 (instrument_builtin_call): Call maybe_clean_eh_stmt or
34782 maybe_clean_or_replace_eh_stmt where needed.
34783 (instrument_memory_accesses): Add cfg_changed argument.
34784 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
34785 if it returned true.
34786 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
34788 PR rtl-optimization/63191
34789 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
34790 wrapper function, moved the whole old content into ...
34791 (ix86_delegitimize_address_1): ... this. New inline function.
34792 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
34793 true as last argument instead of ix86_delegitimize_address.
34795 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34797 * config/aarch64/aarch64.c (generic_branch_cost): Copy
34798 cortexa57_branch_cost.
34800 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34802 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
34804 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34807 * doc/md.texi (Constraints): Document wA constraint.
34808 * config/rs6000/constraints.md (wA): New.
34809 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
34810 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
34811 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
34812 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
34814 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
34817 * gimplify.c (is_oacc_declared): New function.
34818 (oacc_default_clause): Use it to set default flags for acc declared
34819 variables inside parallel regions.
34820 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
34821 declared variables.
34822 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
34823 declare attribute to any decl as necessary.
34825 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
34828 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
34829 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
34830 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
34831 (arm_arch_lpae): This.
34832 * config/arm/arm.c (arm_arch7ve): Rename into ...
34833 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
34834 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
34837 2017-03-22 Martin Liska <mliska@suse.cz>
34840 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
34841 error message instead of an ICE.
34843 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34845 * doc/extend.texi (6.11 Additional Floating Types): Revise.
34847 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
34849 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34851 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34854 2017-03-21 Martin Sebor <msebor@redhat.com>
34856 * doc/extend.texi: Use "cannot" instead of "can't."
34857 * doc/hostconfig.texi: Same.
34858 * doc/install.texi: Same.
34859 * doc/invoke.texi: Same.
34860 * doc/loop.texi: Same.
34861 * doc/md.texi: Same.
34862 * doc/objc.texi: Same.
34863 * doc/rtl.texi: Same.
34864 * doc/tm.texi: Same.
34865 * doc/tm.texi.in: Same.
34866 * doc/trouble.texi: Same.
34868 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
34871 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
34872 (collect_checksum_attributes): Set it.
34873 (die_checksum_ordered): Use it.
34875 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34877 PR tree-optimization/79908
34878 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
34879 change: For a VA_ARG whose LHS has been cast away, use
34880 force_gimple_operand to construct the side effects.
34882 2017-03-21 David Malcolm <dmalcolm@redhat.com>
34884 PR translation/80001
34885 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
34886 more amenable to translation.
34887 (oacc_loop_auto_partitions): Likewise.
34889 2017-03-21 Marek Polacek <polacek@redhat.com>
34890 Martin Sebor <msebor@redhat.com>
34892 PR tree-optimization/80109
34893 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
34894 on INTEGRAL_TYPE_P.
34896 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34897 Segher Boessenkool <segher@kernel.crashing.org>
34900 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
34901 check reg_used_between_p between insn and one of succ or succ2
34902 depending on if succ is artificial insn not inserted into insn
34905 2017-03-21 Martin Liska <mliska@suse.cz>
34907 PR gcov-profile/80081
34908 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
34909 * doc/gcc.texi: Include gcov-dump stuff.
34910 * doc/gcov-dump.texi: New file.
34912 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
34914 PR rtl-optimization/79150
34915 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
34916 conditional jump, if the jump is the last insn of the loop.
34918 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34919 Richard Biener <rguenth@suse.de>
34921 PR tree-optimization/79908
34922 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34923 been cast away, use force_gimple_operand to construct the side
34926 2017-03-21 Martin Liska <mliska@suse.cz>
34928 PR libfortran/79956
34929 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
34932 2017-03-21 Brad Spengler <spender@grsecurity.net>
34935 * plugin.c (htab_hash_plugin): New function.
34936 (add_new_plugin): Use it and adjust.
34937 (parse_plugin_arg_opt): Adjust.
34938 (init_one_plugin): Likewise.
34940 2017-03-21 Richard Biener <rguenther@suse.de>
34942 PR tree-optimization/80032
34943 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
34944 if set force the cleanup to happen unconditionally.
34945 (gimplify_target_expr): Push inserted clobbers with force_uncond
34946 to avoid them being removed by control-dependent DCE.
34948 2017-03-21 Richard Biener <rguenther@suse.de>
34950 PR tree-optimization/80122
34951 * tree-inline.c (copy_bb): Do not expans va-arg packs or
34952 va_arg_pack_len when the inlined call stmt requires pack
34954 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
34956 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34959 * tsan.c (instrument_builtin_call): If the memory model argument
34960 is not a constant, assume it is valid.
34963 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
34966 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34968 PR rtl-optimization/79910
34969 * combine.c (can_combine_p): Do not allow combining an I0 or I1
34970 if its dest is used by an insn before I2 (other than the combined
34971 insns themselves, which are properly handled already).
34973 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34976 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34978 * combine.c (record_used_regs): New static function.
34979 (try_combine): Handle situations where there is an additional
34980 instruction between I2 and I3 which needs to have a LOG_LINK
34984 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34986 * combine.c (try_combine): Delete redundant i1 test. Call
34987 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34989 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34992 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
34993 alternatives 13/14.
34995 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34997 PR tree-optimization/80054
34998 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
34999 the optimization if a PHI or any of its arguments is not dominated
35000 by the candidate's basis. Use gphi* rather than gimple* as
35002 (replace_profitable_candidates): Clean up a gimple* variable that
35003 should be a gphi* variable.
35005 2017-03-20 Martin Sebor <msebor@redhat.com>
35008 * doc/extend.texi (attribute constructor): Document present limitation.
35010 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
35013 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
35014 __POWER9_VECTOR__ #ifdef control, change template definition to
35015 use Power9-specific built-in function.
35016 (vec_any_eq): Likewise.
35017 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
35018 to control outcomes from this test.
35019 (vector_ae_<mode>p): For VEC_F modes, likewise.
35021 2017-03-20 Ian Lance Taylor <iant@google.com>
35023 * config/i386/i386.c (ix86_function_regparm): Save an extra
35024 register for -fsplit-stack with DECL_STATIC_CHAIN.
35026 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35029 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
35030 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
35032 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35034 * config/riscv/riscv.c (riscv_print_operand): Use "fence
35036 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
35039 2017-03-20 Marek Polacek <polacek@redhat.com>
35042 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
35044 2017-03-20 Richard Biener <rguenther@suse.de>
35046 PR tree-optimization/80113
35047 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
35048 allocate extra SSA name for PHI def.
35049 (add_close_phis_to_outer_loops): Likewise.
35050 (add_close_phis_to_merge_points): Likewise.
35051 (copy_loop_close_phi_args): Likewise.
35052 (copy_cond_phi_nodes): Likewise.
35054 2017-03-20 Martin Liska <mliska@suse.cz>
35056 PR middle-end/79753
35057 * tree-chkp.c (chkp_build_returned_bound): Do not build
35058 returned bounds for a LHS that's not a BOUNDED_P type.
35060 2017-03-20 Martin Liska <mliska@suse.cz>
35064 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
35065 COMPLEX_CST and VECTOR_CST.
35067 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35070 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
35071 target operand. A new splitter adds the clobber statement in case
35072 the target operand is dead anyway.
35074 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
35076 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
35077 to age-old versions of binutils and glibc.
35079 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
35081 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
35083 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35085 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
35087 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35089 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
35090 requirement for binutils 2.13.
35092 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35094 * combine.c (try_combine): Delete redundant i1 test. Call
35095 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35097 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
35099 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
35100 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
35102 <riscv64-*-elf>: Re-arrange section
35103 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
35104 <riscv32-*-linux>: Likewise.
35105 <riscv64-*-elf>: Likewise
35106 <riscv64-*-linux>: Likewise.
35108 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
35111 * aarch64.opt(verbose-cost-dump): Fix typo.
35113 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
35116 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
35117 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
35119 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35121 * reload.c (find_reloads): When reloading a nonoffsettable address,
35122 use RELOAD_OTHER for it and its address reloads.
35124 PR rtl-optimization/79910
35125 * combine.c (record_used_regs): New static function.
35126 (try_combine): Handle situations where there is an additional
35127 instruction between I2 and I3 which needs to have a LOG_LINK
35130 2017-03-17 Jeff Law <law@redhat.com>
35132 PR tree-optimization/71437
35133 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
35134 conditional in the hash table first.
35135 (vrp_dom_walker::before_dom_children): Extract condition from
35136 ASSERT_EXPR. Record condition, its inverion and any implied
35137 conditions as well.
35139 2017-03-17 Marek Polacek <polacek@redhat.com>
35140 Markus Trippelsdorf <markus@trippelsdorf.de>
35142 PR tree-optimization/80079
35143 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
35146 2017-03-17 Richard Biener <rguenther@suse.de>
35148 PR middle-end/80075
35149 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
35150 Properly verify the LHS before the RHS possibly claims to be
35152 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
35155 2017-03-17 Martin Jambor <mjambor@suse.cz>
35157 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
35158 (List of -O2 options): Likewise.
35159 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
35162 2017-03-17 Tom de Vries <tom@codesourcery.com>
35164 * gcov-dump.c (print_usage): Print bug_report_url.
35166 2017-03-17 Richard Biener <rguenther@suse.de>
35168 PR middle-end/80050
35169 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
35170 (parser::peek): Likewise.
35172 2017-03-17 Richard Biener <rguenther@suse.de>
35174 PR tree-optimization/80048
35175 * sese.c (free_sese_info): Properly release rename_map and
35176 copied_bb_map elements.
35178 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
35180 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
35181 Add linked-list forward and backlinks. Insert on
35182 construction, remove on destruction.
35183 (class pass_store_merging): Add m_stores_head field.
35184 (pass_store_merging::terminate_and_process_all_chains):
35185 Iterate over m_stores_head list.
35186 (pass_store_merging::terminate_all_aliasing_chains):
35188 (pass_store_merging::execute): Check for debug stmts first.
35189 Push new chains onto the m_stores_head stack.
35191 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
35194 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
35195 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
35196 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
35198 2017-03-16 Jeff Law <law@redhat.com>
35200 PR tree-optimization/71437
35201 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
35202 member function. Implementation moved into after_dom_children
35203 member function and into the threader's thread_outgoing_edges
35205 (dom_opt_dom_walker::after_dom_children): Simplify by moving
35206 some code into new thread_outgoing_edges.
35207 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
35208 definition. Simplify marker handling (do it here). Assume we always
35209 have the available expression and the const/copies tables.
35210 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
35212 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
35213 * tree-vrp.c (equiv_stack): No longer file scoped.
35214 (vrp_dom_walker): New class.
35215 (vrp_dom_walker::before_dom_children): New member function.
35216 (vrp_dom_walker::after_dom_children): Likewise.
35217 (identify_jump_threads): Setup domwalker. Use it rather than
35218 walking edges in a random order by hand. Simplify setup/finalization.
35219 (finalize_jump_threads): Remove.
35220 (vrp_finalize): Do not call identify_jump_threads here.
35221 (execute_vrp): Do it here instead and call thread_through_all_blocks
35224 PR tree-optimization/71437
35225 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
35227 (simplify_stmt_for_jump_threading): Add basic_block argument. All
35229 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
35230 (dom_opt_dom_walker::thread_across_edge): Remove
35231 handle_dominating_asserts argument. All callers changed.
35232 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
35233 changes. Remove calls to lhs_of_dominating_assert. Other
35234 uses of handle_dominating_asserts turn into unconditional code
35235 (simplify_control_stmt_condition_1): Likewise.
35236 (simplify_control_stmt_condition): Likewise.
35237 (thread_through_normal_block, thread_across_edge): Likewise.
35238 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
35239 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
35240 object if it is not an SSA_NAME.
35241 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
35242 before calling into the VRP specific simplifiers.
35243 (identify_jump_threads): Remove handle_dominating_asserts
35246 2017-03-16 Jakub Jelinek <jakub@redhat.com>
35249 * tree-diagnostic.c (default_tree_printer): No longer static.
35250 * tree-diagnostic.h (default_tree_printer): New prototype.
35252 2017-03-16 Tamar Christina <tamar.christina@arm.com>
35254 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
35255 Change ins into fmov.
35257 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35259 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
35260 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
35261 Use h_con constraint for operand 1.
35262 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
35263 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
35265 2017-03-15 Jeff Law <law@redhat.com>
35267 PR tree-optimization/71437
35268 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
35269 (record_temporary_equivalences): Use it.
35271 PR tree-optimization/71437
35272 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
35273 tree-ssa-scopedtables.
35274 (lookup_avail_expr, build_and_record_new_cond): Likewise.
35275 (record_conditions, record_cond, vuse_eq): Likewise.
35276 (record_edge_info): Adjust to API tweak of record_conditions.
35277 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
35278 (record_temporary_equivalences, optimize_stmt): Likewise.
35279 (eliminate_redundant_computations): Likewise.
35280 (record_equivalences_from_stmt): Likewise.
35281 * tree-ssa-scopedtables.c: Include options.h and params.h.
35282 (vuse_eq): New function, moved from tree-ssa-dom.c
35283 (build_and_record_new_cond): Likewise.
35284 (record_conditions): Likewise. Accept vector of conditions rather
35285 than edge_equivalence structure for first argument.
35286 for the first argument.
35287 (avail_exprs_stack::lookup_avail_expr): New member function, moved
35288 from tree-ssa-dom.c.
35289 (avail_exprs_stack::record_cond): Likewise.
35290 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
35291 from tree-ssa-dom.c.
35292 (avail_exprs_stack): Add new member functions lookup_avail_expr
35294 (record_conditions): Declare.
35296 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
35299 * lra-constraints.c (process_alt_operands): Increase reject for
35300 reloading an input/output operand.
35302 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
35305 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
35306 insns to convert from signed/unsigned char/short to IEEE 128-bit
35308 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
35310 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
35313 * config/i386/i386.c (ix86_vector_duplicate_value): Create
35314 subreg of inner mode for values already in registers.
35316 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
35318 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
35319 iteration reg is used after the loop.
35321 2017-03-14 Martin Sebor <msebor@redhat.com>
35323 PR tree-optimization/79800
35324 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
35325 precision in negative-positive range.
35326 (format_floating): Call non-const overload with adjusted precision.
35328 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
35331 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
35334 2017-03-14 Martin Sebor <msebor@redhat.com>
35336 PR middle-end/80020
35337 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
35338 * builtins.def (aligned_alloc): Use it.
35341 * Makefile.in (GTFILES): Add calls.c.
35342 * calls.c: Include "gt-calls.h".
35344 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
35346 PR rtl-optimization/79728
35347 * regs.h (struct target_regs): New field
35348 x_contains_allocatable_regs_of_mode.
35349 (contains_allocatable_regs_of_mode): New macro.
35350 * reginfo.c (init_reg_sets_1): Initialize it, and change
35351 contains_reg_of_mode so it includes global regs as well.
35352 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
35353 rather than contains_regs_of_mode.
35355 2017-03-14 Martin Liska <mliska@suse.cz>
35357 * doc/invoke.texi: Document options that can't be combined with
35358 -fcheck-pointer-bounds.
35360 2017-03-14 Martin Liska <mliska@suse.cz>
35362 PR middle-end/79831
35363 * doc/invoke.texi (-Wchkp): Document the option.
35365 2017-03-14 Martin Liska <mliska@suse.cz>
35367 * Makefile.in: Install gcov-dump.
35369 2017-03-14 Martin Liska <mliska@suse.cz>
35371 * multiple_target.c (expand_target_clones): Bail out for
35372 an invalid attribute.
35374 2017-03-14 Richard Biener <rguenther@suse.de>
35376 * alias.c (struct alias_set_entry): Pack properly.
35377 * cfgloop.h (struct loop): Likewise.
35378 * cse.c (struct set): Likewise.
35379 * ipa-utils.c (struct searchc_env): Likewise.
35380 * loop-invariant.c (struct invariant): Likewise.
35381 * lra-remat.c (struct cand): Likewise.
35382 * recog.c (struct change_t): Likewise.
35383 * rtl.h (struct address_info): Likewise.
35384 * symbol-summary.h (function_summary): Likewise.
35385 * tree-loop-distribution.c (struct partition): Likewise.
35386 * tree-object-size.c (struct object_size_info): Likewise.
35387 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
35388 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
35389 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
35390 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
35391 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
35392 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
35393 (struct _stmt_vec_info): Likewise.
35395 2017-03-14 Martin Liska <mliska@suse.cz>
35398 * multiple_target.c (create_dispatcher_calls): Check that
35399 a target can create a function dispatcher.
35401 2017-03-14 Martin Liska <mliska@suse.cz>
35404 * multiple_target.c (expand_target_clones): Drop local.local
35405 flag for default implementation.
35407 2017-03-14 Richard Biener <rguenther@suse.de>
35409 PR tree-optimization/80030
35410 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
35412 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
35414 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
35415 gcc_fallthrough() instead of __attribute__((fallthrough));
35417 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35419 * doc/gcc.texi: Remove "up" link to (DIR).
35420 * doc/gccint.texi: Ditto.
35422 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35424 * doc/install.texi (Specific) <avr>: Remove reference to
35427 2017-03-13 Jeff Law <law@redhat.com>
35429 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
35430 attribute rather than comments.
35432 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
35433 match_scratch operand is highest.
35435 2017-03-13 Martin Liska <mliska@suse.cz>
35437 PR middle-end/78339
35438 * ipa-pure-const.c (warn_function_noreturn): If the declarations
35439 is a CHKP clone, use original declaration.
35441 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35443 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
35444 (arc_conditional_register_usage): Use a different allocation order
35445 when optimizing for size.
35446 * common/config/arc/arc-common.c (arc_option_optimization_table):
35447 Section anchors default on when optimizing for size.
35449 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35451 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
35453 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35455 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
35456 * config/arc/arc.md (cpu_facility): Add cd variant.
35457 (*movqi_insn): Add code density variant.
35458 (*movhi_insn): Likewise.
35459 (*movqi_insn): Likewise.
35460 (*addsi3_mixed): Likewise.
35461 (subsi3_insn): Likewise.
35463 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35465 * config/arc/arc.md (movsi_cond_exec): Update constraint.
35467 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35469 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
35470 expressions with MINUS and UNARY ops.
35472 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35475 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
35477 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
35478 between vec_select and vector argument.
35479 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
35480 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
35481 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
35482 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
35483 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
35484 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
35486 2017-03-13 Richard Biener <rguenther@suse.de>
35489 * params.def (vect-max-peeling-for-alignment): Fix typo.
35491 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35493 * doc/install.texi (Specific) <mips-*-*>: Remove description of
35494 issue that only occurred with binutils below 2.18.
35496 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35498 * doc/install.texi (Specific) <cris-axis-elf>: No longer
35499 refer to binutils 2.11/2.12 minimum.
35501 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35503 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
35504 ftp.kernel.org and simplify binutils requirement.
35506 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
35508 * doc/invoke.texi (Warning Options): Fix spelling of link-time
35510 (Optimize Options): Ditto. Also remove redundancy.
35512 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35514 PR translation/79848
35515 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
35517 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
35518 to G_ to avoid double translation.
35520 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35522 PR translation/79923
35523 * auto-profile.c (get_combined_location): Convert leading
35524 character of diagnostics to lower case and remove trailing period.
35525 (read_profile): Likewise for various diagnostics.
35526 * config/arm/arm.c (arm_option_override): Remove trailing period
35527 from various diagnostics.
35528 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
35529 (msp430_expand_delay_cycles): Likewise.
35531 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35534 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
35535 full command-line argument, rather than just "str".
35536 (aarch64_validate_march): Likewise.
35537 (aarch64_validate_mtune): Likewise.
35539 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
35541 PR rtl-optimization/78911
35542 * lra-assigns.c (must_not_spill_p): New function.
35543 (spill_for): Use it.
35545 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35547 PR tree-optimization/79981
35548 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
35549 ATOMIC_COMPARE_EXCHANGE ifn result.
35550 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
35551 IFN_ATOMIC_COMPARE_EXCHANGE.
35553 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35556 * opts.c (parse_sanitizer_options): Add missing question mark to
35557 "did you mean" message.
35559 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35561 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
35563 (VMULEUH_UNS): Likewise.
35564 (VMULOUB_UNS): Likewise.
35565 (VMULOUH_UNS): Likewise.
35566 * config/rs6000/rs6000.c (builtin_function_type): Remove
35567 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
35569 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35572 * read-rtl-function.c (function_reader::read_rtx_operand): Update
35573 x with result of extra_parsing_for_operand_code_0.
35574 (function_reader::extra_parsing_for_operand_code_0): Convert
35575 return type from void to rtx, returning x. When reading
35576 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
35577 larger size containing struct block_symbol.
35579 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
35581 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
35582 -mfloat128-hardware without -m64.
35584 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
35587 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
35588 entries to the case statement that marks unsigned arguments to
35589 overloaded functions.
35591 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
35593 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
35594 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
35596 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
35599 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
35600 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
35602 2017-03-10 Martin Liska <mliska@suse.cz>
35606 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
35607 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
35610 2017-03-10 Olivier Hainque <hainque@adacore.com>
35612 * tree-switch-conversion (array_value_type): Start by resetting
35613 candidate type to it's main variant.
35615 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35617 PR rtl-optimization/79909
35618 * combine.c (try_combine): Use simplify_replace_rtx on individual
35619 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
35620 of the whole CALL_INSN_FUNCTION_USAGE.
35622 PR tree-optimization/79972
35623 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
35624 get_range_info on SSA_NAMEs. Formatting fixes.
35626 2017-03-10 Richard Biener <rguenther@suse.de>
35627 Jakub Jelinek <jakub@redhat.com>
35629 PR tree-optimization/77975
35630 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
35631 edge to be constant.
35632 (get_val_for): For constant x return it. Formatting fix.
35633 (loop_niter_by_eval): Avoid pointless looping if the next iteration
35634 would use the same bases as the current one.
35636 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35638 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
35639 instead of vec_select for V1TImode.
35640 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
35642 (VSX_LE_128): Add V1TI to this mode iterator.
35643 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
35644 (*vsx_le_perm_store_<mode>): Likewise.
35645 (pre-reload splitter for VSX stores): Likewise.
35646 (post-reload splitter for VSX stores): Likewise.
35647 (*vsx_xxpermdi2_le_<mode>): Likewise.
35648 (*vsx_lxvd2x2_le_<mode>): Likewise.
35649 (*vsx_stxvd2x2_le_<mode>): Likewise.
35651 2017-03-09 Michael Eager <eager@eagercon.com>
35653 Correct failures with --enable-checking=yes,rtl.
35655 * config/microblaze/microblaze.c (microblaze_expand_shift):
35656 Replace GET_CODE test with CONST_INT_P and INTVAL test with
35657 test for const0_rtx.
35658 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
35659 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
35661 2017-03-09 Richard Biener <rguenther@suse.de>
35663 PR tree-optimization/79977
35664 * graphite-scop-detection.c (scop_detection::merge_sese):
35665 Handle the case of extra exits to blocks dominating the entry.
35667 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
35669 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
35672 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
35674 PR rtl-optimization/79949
35675 * lra-constraints.c (process_alt_operands): Check memory when
35676 trying to predict a cycle. Print about the overall increase.
35678 2017-03-09 Richard Biener <rguenther@suse.de>
35680 PR middle-end/79971
35681 * gimple-expr.c (useless_type_conversion_p): Preserve
35682 TYPE_SATURATING for fixed-point types.
35684 2017-03-09 Richard Biener <rguenther@suse.de>
35687 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
35688 alignment of BLKmode params.
35690 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35693 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
35694 (VALL_NO_V2Q): Likewise.
35696 * config/aarch64/aarch64-simd.md
35697 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
35699 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
35700 VALL_NO_V2Q mode iterator.
35701 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
35703 2017-03-09 Martin Liska <mliska@suse.cz>
35705 PR tree-optimization/79631
35706 * tree-chkp-opt.c (chkp_is_constant_addr): Call
35707 tree_int_cst_sign_bit just for INTEGER constants.
35709 2017-03-09 Martin Liska <mliska@suse.cz>
35713 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
35716 2017-03-09 Marek Polacek <polacek@redhat.com>
35719 * tree.c (inchash::add_expr): Handle TREE_VEC.
35721 2017-03-09 Martin Liska <mliska@suse.cz>
35724 (chkp_narrow_size_and_offset): New function.
35725 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
35726 (void chkp_parse_bit_field_ref): New function.
35727 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
35728 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
35730 2017-03-09 Martin Liska <mliska@suse.cz>
35733 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
35734 (chkp_find_bounds_1): Remove gcc_unreachable.
35736 2017-03-09 Jakub Jelinek <jakub@redhat.com>
35739 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
35740 BUILT_IN_SYNC*, determine the access type from the size suffix and
35741 always build a MEM_REF with that type. Handle forgotten
35742 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
35745 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
35746 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
35747 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
35748 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
35749 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
35750 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
35751 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
35752 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
35753 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
35754 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
35755 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
35756 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
35757 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
35758 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
35759 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
35760 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
35761 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
35762 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
35763 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
35764 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
35765 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
35766 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
35767 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
35768 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
35769 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
35770 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
35771 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
35772 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
35773 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
35774 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
35775 definitions outside of __OPTIMIZE__ guarded section.
35778 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
35779 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
35780 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
35781 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
35784 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35786 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
35787 ("vfenez<mode>"): Add missing constraints.
35789 2017-03-08 Martin Sebor <msebor@redhat.com>
35792 * config/nds32/nds32.c (nds32_option_override):
35793 Fix misspelled diagnostic.
35795 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35798 * gimplify.c (gimplify_omp_for): Replace index var in outer
35799 taskloop statement with an artificial variable and add
35800 OMP_CLAUSE_PRIVATE clause for it.
35802 2017-03-08 Richard Biener <rguenther@suse.de>
35804 PR tree-optimization/79955
35805 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
35806 for accesses that are completely outside of the variable.
35808 2017-03-08 Andrew Haley <aph@redhat.com>
35810 PR tree-optimization/79943
35811 * tree-ssa-loop-split.c (compute_new_first_bound): When
35812 calculating the new upper bound, (END-BEG) should be added, not
35815 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35817 * config/avr/avr.md (setmemhi): Make sure match_dup
35818 operand number comes before match_scratch.
35820 2017-03-08 Richard Biener <rguenther@suse.de>
35822 PR tree-optimization/79920
35823 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
35824 with ncopies == 1 to ...
35825 (vect_transform_slp_perm_load): ... here. Properly compute
35826 all element loads by iterating VF times over the group. Do
35827 not handle ncopies (computed in a broken way) in
35828 vect_create_mask_and_perm.
35830 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35833 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
35834 is a uniform vector, use uniform_vector_p return value instead of
35835 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
35837 2017-03-07 Marek Polacek <polacek@redhat.com>
35839 PR middle-end/79809
35840 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
35841 (alloca_call_type): Likewise.
35843 2017-03-07 Martin Liska <mliska@suse.cz>
35845 * gcov.c (process_args): Put comment to correct location.
35847 2017-03-07 Martin Liska <mliska@suse.cz>
35849 PR middle-end/68270
35850 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
35851 Use array_at_struct_end_p instead of DECL_CHAIN (field).
35852 (chkp_narrow_bounds_for_field): Likewise.
35853 (chkp_parse_array_and_component_ref): Pass one more argument to
35856 2017-03-07 Richard Biener <rguenther@suse.de>
35858 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
35861 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
35863 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
35864 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
35866 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35869 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
35870 to end of description.
35871 (PARAM_MAX_STORES_TO_MERGE): Likewise.
35873 2017-03-07 Jakub Jelinek <jakub@redhat.com>
35875 PR rtl-optimization/79901
35876 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
35878 (*avx512f_<code><mode>3<mask_name>): ... this.
35879 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
35880 iterator instead of VI8_AVX2_AVX512BW.
35882 PR rtl-optimization/79901
35883 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
35884 min/max expander, expand it using expand_vec_cond_expr.
35887 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
35890 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35893 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
35894 to void * for PCH reasons.
35895 * dwarf2out.c (output_loc_operands, output_die): Cast
35896 v.val_vec.array to unsigned char *.
35898 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
35901 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
35904 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
35906 PR rtl-optimization/79571
35907 * lra-constraints.c (process_alt_operands): Calculate static
35908 reject and subtract it from overall when only addresses will be
35911 2017-03-06 Julia Koval <julia.koval@intel.com>
35914 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
35915 incoming stack boundary to 128 for 64-bit targets.
35917 2017-03-06 Richard Biener <rguenther@suse.de>
35919 PR tree-optimization/79894
35920 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
35921 to NULL after folding it.
35923 2017-03-06 Richard Biener <rguenther@suse.de>
35925 PR tree-optimization/79824
35926 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
35927 check disabling peeling for gaps.
35929 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
35931 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
35932 attributes): Document gettimeofday.
35934 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35936 * config/s390/s390.c (s390_option_override_internal): Set
35937 PARAM_MIN_VECT_LOOP_BOUND
35939 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35941 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
35942 * config/s390/s390.md: Likewise.
35944 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35947 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
35948 (<avx2_avx512>_perm<mode>): Rename to ...
35949 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
35951 (<avx512>_perm<mode>_mask): Rename to ...
35952 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
35954 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
35955 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
35956 instead of VI8F_256_512.
35957 (avx512f_perm<mode>): New define_expand.
35958 (avx512f_perm<mode>_mask): Likewise.
35959 (avx512f_perm<mode>_1<mask_name>): New define_insn.
35960 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
35962 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35964 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
35965 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
35967 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
35969 2017-03-06 Martin Liska <mliska@suse.cz>
35972 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
35973 when having a SSA NAME w/o VAR_DECL assigned to it.
35975 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35977 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
35978 msa_dpsub_<su>_d): Fix MODE for vec_select.
35980 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35982 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
35984 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
35986 2017-03-06 Richard Biener <rguenther@suse.de>
35988 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
35989 * plugin.c (register_plugin_info): Likewise.
35990 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
35992 2017-03-05 Jakub Jelinek <jakub@redhat.com>
35994 * config/i386/sse.md (sse_storehps, sse_storelps,
35995 avx_<castmode><avxsizesuffix>_<castmode>,
35996 avx512f_<castmode><avxsizesuffix>_<castmode>,
35997 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
35998 in condition that at least one operand is not a MEM.
36000 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36002 PR middle-end/79805
36003 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
36004 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
36006 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
36007 gimple_call_nothrow_p flag based on whether original builtin can throw.
36008 If it can, emit following stmts on the fallthrough edge.
36009 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
36010 don't create new bb if inserting just debug stmts on the edge, try to
36011 insert them on the fallthru bb or just reset debug stmts.
36013 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
36016 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
36017 restore recog_data (including the operand rtxes inside it) around
36018 the call to get_insn_template.
36020 2017-03-03 Martin Sebor <msebor@redhat.com>
36022 PR tree-optimization/79699
36023 * context.c (context::~context): Free MPFR caches to avoid
36024 a memory leak on program exit.
36026 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36028 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
36029 Use wide_int::ulow () instead of .elt (0).
36031 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36033 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
36034 (*pushxf): Limit oF constraint to 32bit targets and add oC
36035 constraint for 64bit targets.
36036 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
36037 (*pushdf): Change rmF constraint to rmC.
36039 2017-03-03 Martin Liska <mliska@suse.cz>
36041 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
36042 Remove unused variable.
36044 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36047 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
36048 is a memory operand, increase num_memory.
36049 (ix86_expand_args_builtin): Likewise.
36051 2017-03-03 Jan Hubicka <jh@suse.cz>
36054 * ipa-devirt.c (maybe_record_node): Properly handle
36055 __cxa_pure_virtual visibility.
36057 2017-03-03 Martin Liska <mliska@suse.cz>
36059 PR tree-optimization/79803
36060 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
36062 (pass_loop_prefetch::execute): Disabled optimization if an
36063 assumption about L1 cache size is not met.
36065 2017-03-03 Martin Liska <mliska@suse.cz>
36067 PR rtl-optimization/79574
36068 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
36069 (hash_scan_set): Likewise.
36070 (dump_hash_table): Likewise.
36071 (hoist_code): Likewise.
36073 2017-03-03 Richard Biener <rguenther@suse.de>
36075 * fixed-value.c (fixed_from_string): Restore use of elt (1)
36076 in place of uhigh ().
36077 (fixed_convert_from_real): Likewise.
36079 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36082 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
36084 2017-03-03 Richard Biener <rguenther@suse.de>
36086 PR middle-end/79818
36087 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
36088 TYPE_OVERFLOW_UNDEFINED check.
36090 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36092 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
36094 (vector_ae_<mode>_p): Likewise.
36095 (vector_nez_<mode>_p): Likewise.
36096 (vector_ne_v2di_p): Likewise.
36097 (vector_ae_v2di_p): Likewise.
36098 (vector_ne_<mode>_p): Likewise.
36099 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
36101 (vsx_tsqrt<mode>2_fe): Likewise.
36103 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
36106 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
36108 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36110 PR rtl-optimization/79780
36111 * cprop.c (one_cprop_pass): When second and further conditional trap
36112 in a single basic block is turned into an unconditional trap, turn it
36113 into a deleted note to avoid RTL verification failures.
36115 2017-03-02 Richard Biener <rguenther@suse.de>
36117 * fold-const.c (const_binop): Use ulow () instead of elt (0).
36119 2017-03-02 Richard Biener <rguenther@suse.de>
36121 PR tree-optimization/79345
36123 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
36124 param and abort the walk, returning -1 if it is hit.
36125 (walk_aliased_vdefs): Take a limit param and pass it on.
36126 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
36127 defaulting to 0 and return a signed int.
36128 * tree-ssa-uninit.c (struct check_defs_data): New struct.
36129 (check_defs): New helper.
36130 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
36131 about uninitialized memory.
36132 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
36133 bogus uninitialized warning.
36134 (fixed_convert_from_real): Likewise.
36136 2017-03-02 Bin Cheng <bin.cheng@arm.com>
36138 PR tree-optimization/66768
36139 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
36140 iv_use if base object can't be determined.
36142 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36144 PR tree-optimization/79345
36145 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
36146 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
36147 (get_pattern_stats): Initialize it.
36148 * genemit.c (gen_expand): Verify match_scratch numbers come after
36149 match_operand/match_dup numbers.
36150 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
36151 match_scratch numbers.
36152 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
36154 * config/s390/s390.md (trunctdsd2): Likewise.
36156 2017-03-02 Richard Biener <rguenther@suse.de>
36158 * wide-int.h (wide_int_storage::operator=): Implement in terms
36161 2017-03-02 Richard Biener <rguenther@suse.de>
36163 PR tree-optimization/79777
36164 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
36165 the to insert expression to sth existing.
36167 2017-03-01 Martin Sebor <msebor@redhat.com>
36169 PR middle-end/79692
36170 * gimple-ssa-sprintf.c
36171 (directive::known_width_and_precision): New function.
36172 (format_integer): Use it.
36173 (get_mpfr_format_length): Consider the full range of precision
36174 when computing %g output with the # flag. Set the likely byte
36175 count to 3 rather than 1 when precision is indeterminate.
36176 (format_floating): Correct the lower bound of precision.
36178 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36180 * doc/invoke.texi: Document default code model for 64-bit Linux.
36182 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36185 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
36186 udiv rather than div since input pattern is unsigned.
36188 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
36190 * config/i386/i386.c (print_reg): Warn for values of
36191 unsupported size in integer register.
36193 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
36196 * config/rs6000/predicates.md (current_file_function_operand): Do
36197 not allow self calls to be local if the function is replaceable.
36199 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
36202 * config/rs6000/altivec.h (vec_ctz and others): Change the
36203 preprocessor macro that controls conditional compilation from
36204 _ARCH_PWR9 to __POWER9_VECTOR__.
36205 (vec_all_ne): Change parameterization of __altivec_scalar_pred
36206 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
36207 control (instead of _ARCH_PWR9 control) so that template
36208 definition uses power9-specific function.
36209 (vec_any_eq): Likewise.
36210 (vec_all_ne): Change macro definition to use a power9-specific
36211 expansion under #ifdef __POWER9_VECTOR__ control (instead of
36212 _ARCH_PWR9 control).
36213 (vec_any_eq) Likewise.
36214 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
36215 expansion for CMPNEF to remove support for xvcmpnesp instruction.
36216 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
36217 support for xvcmpnedp instruction.
36218 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
36219 macro expansion so that Power9 implementation of vec_all_ne does
36220 not use the AltiVec predicate framework.
36221 (VCMPNEH_P): Likewise.
36222 (VCMPNEW_P): Likewise.
36223 (VCMPNED_P): Likewise.
36224 (VCMPNEFP_P): Likewise.
36225 (VCMPNEDP_P): Likewise.
36226 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
36227 implementation of vec_any_eq to not use AltiVec predicate
36229 (VCMPAEH_P): Likewise.
36230 (VCMPAEW_P): Likewise.
36231 (VCMPAED_P): Likewise.
36232 (VCMPAEFP_P): Likewise.
36233 (VCMPAEDP_P): Likewise.
36234 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
36235 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
36236 not use the AltiVec predicate framework.
36237 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
36238 of vec_any_eq to not use AltiVec predicate framework.
36239 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36240 support for predefined __POWER9_VECTOR__ macro to indicate that
36241 Power9 instruction selection is enabled.
36242 (altivec_overloaded_builtins): Remove extraneous
36243 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
36244 function argument types RS6000_BTI_bool_V16QI and
36245 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
36246 entry for overloaded function argument types RS6000_BTI_bool_V4SI
36247 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
36248 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
36249 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
36250 Power9 for implementations of vec_cmpne. Change the signature for
36251 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
36252 (representing vec_all_ne) to remove the previously described first
36253 argument of type RS6000_BTI_INTSI, as this was an artifact of
36254 reliance on the AltiVec predicate framework, which is no longer
36255 used in the implementation of these functions. Add
36256 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
36257 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
36258 since, unlike the AltiVec predicate framework implementation, we
36259 do not share function descriptors between vec_alle and vec_anyeq.
36260 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
36261 set of modes that receive special treatment even when
36262 TARGET_P9_VECTOR is true. The special treatment emits code that
36263 does not depend on Power9 instructions.
36264 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
36265 define_expand to not rely on AltiVec predicate framework.
36266 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36268 (vector_ne_v2di_p): Change this define_expand to not rely on
36269 AltiVec predicate framework.
36270 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
36272 (vector_ne_<mode>_p): Change this define_expand to not rely on
36273 AltiVec predicate framework.
36274 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36276 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
36277 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
36278 define_insn pattern.
36279 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
36280 define_insn pattern because the xvcmpne<VSs>. instruction is not
36282 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
36283 instruction is not supported.
36285 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36287 * config/nvptx/nvptx.c: Include intl.h.
36289 2017-03-01 Martin Jambor <mjambor@suse.cz>
36292 * ipa-prop.h (ipa_bits): Removed field known.
36293 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
36294 to pointers. Adjusted their comments to warn about their sharing.
36295 (ipcp_transformation_summary): Change bits to a vector of pointers.
36296 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
36297 (ipa_get_ipa_bits_for_value): Declare.
36298 * tree-vrp.h (value_range): Mark as GTY((for_user)).
36299 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
36300 (ipa_bits_hash_table): Likewise.
36301 (ipa_vr_ggc_hash_traits): Likewise.
36302 (ipa_vr_hash_table): Likewise.
36303 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
36304 being pointers and vr_known being removed.
36305 (ipa_set_jf_unknown): Likewise.
36306 (ipa_get_ipa_bits_for_value): New function.
36307 (ipa_set_jfunc_bits): Likewise.
36308 (ipa_get_value_range): New overloaded functions.
36309 (ipa_set_jfunc_vr): Likewise.
36310 (ipa_compute_jump_functions_for_edge): Use the above functions to
36311 construct bits and vr parts of jump functions.
36312 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
36313 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36315 (ipcp_grow_transformations_if_necessary): Also allocate
36316 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36318 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
36319 them. Fix too long lines.
36320 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
36321 vr_known being removed.
36322 (ipa_read_jump_function): Use new setter functions to construct bits
36323 and vr parts of jump functions or set them to NULL.
36324 (write_ipcp_transformation_info): Adjust for bits being pointers.
36325 (read_ipcp_transformation_info): Likewise.
36326 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
36328 Include gt-ipa-prop.h.
36329 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
36331 (ipcp_store_bits_results): Likewise.
36332 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
36333 Do not write to existing jump functions but use a temporary instead.
36335 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36338 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
36339 attempt to use its first operand as BIT_FIELD_REF base.
36341 2017-03-01 Richard Biener <rguenther@suse.de>
36343 PR middle-end/79721
36344 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
36345 interpolating formula in wrapping arithmetic.
36346 (chrec_apply): Convert chrec_evaluate return value to wanted type.
36348 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36350 PR tree-optimization/79734
36351 * tree-vect-generic.c (expand_vector_condition): Optimize
36352 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
36353 Handle VEC_COND_EXPR where comparison has different inner width from
36354 type's inner width.
36356 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
36358 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
36359 markup, and similar issues. Remove @opindex entries for things
36360 that aren't options. Add missing -mmpy-option entries.
36362 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36364 PR tree-optimization/79737
36365 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
36366 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
36367 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
36368 instead of byte_size. Formatting fix.
36369 (shift_bytes_in_array_right): Formatting fix.
36371 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
36374 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
36375 condition on optimize for the leaf function test.
36377 2017-02-28 Martin Liska <mliska@suse.cz>
36380 * read-rtl-function.c (function_reader::handle_unknown_directive):
36381 Bail out when one uses -flto.
36383 2017-02-28 Martin Liska <mliska@suse.cz>
36385 * common.opt: Replace space with tabular for options of <number>
36387 * config/i386/i386.opt: Show <number> value for
36388 -mlarge-data-threshold.
36389 * opts.c (print_filtered_help): Do not display number in hexadecimal
36392 2017-02-28 Martin Liska <mliska@suse.cz>
36394 * common.opt: Fix --help=option -Q for options which are of
36397 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
36399 * config/i386/i386.c (print_reg): Error out for values
36400 of 8-bit size in invalid integer register.
36402 2017-02-28 Martin Sebor <msebor@redhat.com>
36404 PR tree-optimization/79691
36405 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
36407 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36410 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
36411 gcc_unreachable with output_operand_lossage.
36413 2017-02-28 Richard Biener <rguenther@suse.de>
36415 PR tree-optimization/79740
36416 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
36418 (visit_nary_op): Insert the nary into the hashtable if we
36419 pattern-matched sth.
36420 * tree-ssa-pre.c (eliminate_insert): Robustify.
36422 2017-02-28 Richard Biener <rguenther@suse.de>
36424 PR middle-end/79731
36425 * fold-const.c (decode_field_reference): Reject out-of-bound
36428 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36430 * config/i386/i386.c: Include intl.h.
36431 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
36432 instead of just cond ? "..." : "...".
36433 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
36434 * coverage.c (read_counts_file): Likewise.
36435 * omp-offload.c: Include intl.h.
36436 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
36437 of just cond ? "..." : "...".
36438 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
36439 of just cond ? "..." : "...".
36441 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
36444 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
36446 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
36447 'tune for' CPU name.
36448 * config/arm/arm-cpu-data.h: Regenerated.
36450 2017-02-28 Richard Biener <rguenther@suse.de>
36452 PR tree-optimization/79732
36453 * tree-inline.c (expand_call_inline): Do not shadow var.
36455 2017-02-28 Richard Biener <rguenther@suse.de>
36457 PR tree-optimization/79723
36458 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
36459 address-space properly.
36461 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
36463 * doc/optinfo.texi (Optimization groups): Fix option used for
36465 * doc/invoke.texi (-fopt-info): Document "omp".
36466 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
36467 (OPTGROUP_ALL): Add OPTGROUP_OMP.
36468 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
36469 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
36470 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
36472 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
36474 * dumpfile.c (optgroup_options): Instead of "openmp", associate
36475 OPTGROUP_OMP with "omp".
36477 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
36480 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
36481 for arithmetic shift of unsigned V2DI.
36483 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
36485 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
36486 arc/linux.h headers.
36487 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
36488 (LINK_SPEC): Likewise.
36489 (ARC_TLS_EXTRA_START_SPEC): Likewise.
36490 (EXTRA_SPECS): Likewise.
36491 (STARTFILE_SPEC): Likewise.
36492 (ENDFILE_SPEC): Likewise.
36493 (LIB_SPEC): Likewise.
36494 (TARGET_SDATA_DEFAULT): Likewise.
36495 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
36496 (MULTILIB_DEFAULTS): Likewise.
36497 (DWARF2_UNWIND_INFO): Likewise.
36498 * config/arc/big.h: New file.
36499 * config/arc/elf.h: Likewise.
36500 * config/arc/linux.h: Likewise.
36501 * config/arc/t-uClibc: Remove.
36503 2017-02-27 Bin Cheng <bin.cheng@arm.com>
36505 PR tree-optimization/77536
36506 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
36507 (tree_transform_and_unroll_loop): Use above function to compute the
36508 estimated niter of unrolled loop and use it when scaling profile.
36509 Also use count info rather than frequency if it's non-zero.
36510 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
36511 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
36512 (vect_transform_loop): Call above function.
36514 2017-02-27 Richard Biener <rguenther@suse.de>
36516 PR tree-optimization/45397
36517 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
36518 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
36519 (visit_nary_op): Add pattern matching for CSEing sign-changed
36520 or truncated operations with wider ones.
36522 2017-02-27 Richard Biener <rguenther@suse.de>
36524 PR tree-optimization/79690
36525 * tree-vect-stmts.c (vectorizable_store): Use vector type
36526 built from the DR with address-space.
36528 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
36530 * doc/invoke.texi (Optimize Options): Refine the description
36531 of asan-use-after-return.
36533 2017-02-25 Alan Modra <amodra@gmail.com>
36535 PR rtl-optimization/79584
36536 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
36537 base, not ad->base_term, the reg within base. Remove assertion
36538 that ad->base == ad->base_term. Replace gen_int_mode using
36539 bogus mode with const0_rtx.
36541 2017-02-25 Jakub Jelinek <jakub@redhat.com>
36543 PR middle-end/79396
36544 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
36545 FMA_EXPR like tcc_binary or tcc_unary.
36547 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
36550 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
36552 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
36553 (output_loc_operands): Handle DW_OP_call_ref and
36554 DW_OP_GNU_variable_value.
36555 (struct variable_value_struct): New type.
36556 (struct variable_value_hasher): Likewise.
36557 (variable_value_hash): New variable.
36558 (string_types): Remove.
36559 (copy_loc_descr): New function.
36560 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
36561 (prepend_loc_descr_to_each): New function.
36562 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
36563 instead of add_loc_descr_to_each if the first argument is single
36564 location list and the second has multiple.
36565 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
36566 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
36567 when looking for variable value which doesn't have other location info.
36568 (loc_list_from_tree): Formatting fix.
36569 (gen_array_type_die): Simplify DW_AT_string_length handling.
36570 (adjust_string_types): Remove.
36571 (gen_subprogram_die): Don't call adjust_string_types nor test/set
36572 string_types. Call resolve_variable_values.
36573 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
36574 (resolve_addr_in_expr): Likewise. Add A argument.
36575 (copy_deref_exprloc): Remove deref argument. Adjust for the
36576 original expression being DW_OP_GNU_variable_value with optionally
36577 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
36578 optionally after it.
36579 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
36580 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
36581 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
36582 (variable_value_hasher::hash, variable_value_hasher::equal): New
36584 (resolve_variable_value_in_expr, resolve_variable_value,
36585 resolve_variable_values, note_variable_value_in_expr,
36586 note_variable_value): New functions.
36587 (dwarf2out_early_finish): Call note_variable_value on all toplevel
36590 2017-02-24 Jakub Jelinek <jakub@redhat.com>
36593 * opts.h (handle_generated_option): Add GENERATED_P argument.
36594 * opts-common.c (handle_option): Adjust function comment.
36595 (handle_generated_option): Add GENERATED_P argument, pass it to
36597 (control_warning_option): Pass false to handle_generated_option
36599 * opts.c (maybe_default_option): Pass true to handle_generated_option
36601 * optc-gen.awk: Likewise.
36603 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36605 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
36606 a REG, look at the REG it is a SUBREG of.
36607 (splitter for cmpeqsi_t): Ditto.
36609 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36611 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
36612 the special USEs with the pattern of the insn, not the insn itself.
36614 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
36617 * doc/invoke.texi: Document -mload-store-pairs.
36619 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36620 Sandra Loosemore <sandra@codesourcery.com>
36622 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
36623 argument isn't a CONST_INT.
36624 (nios2_alternate_compare_const): Assert op is a CONST_INT.
36625 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
36626 (nios2_validate_compare): Bypass alternate compare logic if *op2
36627 is not a CONST_INT.
36628 (ldstwm_operation_p): Return false if first_base is not a REG or
36629 if first_offset is not a CONST_INT.
36631 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36633 * config/cris/cris.md: Use correct operand in a define_peephole2.
36635 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36637 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
36639 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36641 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
36642 this_insn if it is an INSN or JUMP_INSN.
36643 (force_offsettable): Look at base, not at addr.
36644 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
36645 on things that aren't necessarily CONST_INTs.
36647 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
36649 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
36650 -mfpmath=sse is the default also for x86-32 targets with SSE2
36651 instruction set when @option{-ffast-math} is enabled
36653 2017-02-24 Jeff Law <law@redhat.com>
36655 PR rtl-optimizatoin/79286
36656 * ira.c (update_equiv_regs): Drop may_trap_p exception to
36659 2017-02-24 Richard Biener <rguenther@suse.de>
36661 PR tree-optimization/79389
36662 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
36665 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
36667 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
36668 function comment to reflect reality.
36669 (loop_exits_before_overflow): Fix typo in function description.
36671 2017-02-24 Richard Biener <rguenther@suse.de>
36673 PR tree-optimization/79389
36674 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
36675 properly that a threading opportunity exists. Detect conditional
36676 copy/constant propagation opportunities.
36678 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
36680 * config/visium/visium.md (type): Add trap.
36681 (b): New mode attribute.
36682 (*btst): Rename into...
36683 (*btst<mode>): ...this and adjust.
36684 (*cbranchsi4_btst_insn): Rename into...
36685 (*cbranch<mode>4_btst_insn): ...this and adjust.
36686 (trap): New define_insn.
36688 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36690 PR tree-optimization/79389
36691 * ifcvt.c (struct noce_if_info): Add rev_cond field.
36692 (noce_reversed_cond_code): New function.
36693 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
36694 reversed_comparison_code. Formatting fix.
36695 (noce_try_store_flag): Test rev_cond != NULL in addition to
36696 reversed_comparison_code.
36697 (noce_try_store_flag_constants): Likewise.
36698 (noce_try_store_flag_mask): Likewise.
36699 (noce_try_addcc): Use rev_cond if non-NULL instead of
36700 reversed_comparison_code.
36701 (noce_try_cmove_arith): Likewise. Formatting fixes.
36702 (noce_try_minmax, noce_try_abs): Clear rev_cond.
36703 (noce_find_if_block): Initialize rev_cond.
36704 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
36705 instead of false as last argument never attempt to reverse it
36708 2017-02-23 Bin Cheng <bin.cheng@arm.com>
36710 PR tree-optimization/79663
36711 * tree-predcom.c (combine_chains): Process refs in reverse order
36712 only for ZERO length chains, and add explaining comment.
36714 2017-02-23 Jeff Law <law@redhat.com>
36716 PR tree-optimization/79578
36717 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
36718 in call to operand_equal_p.
36720 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
36723 * config/i386/cpuid.h: Fix another undefined behavior.
36725 2017-02-23 Richard Biener <rguenther@suse.de>
36727 PR tree-optimization/79683
36728 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
36729 vector types for data-refs.
36731 2017-02-23 Martin Liska <mliska@suse.cz>
36733 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
36735 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36737 PR middle-end/79665
36738 * internal-fn.c (get_range_pos_neg): Moved to ...
36739 * tree.c (get_range_pos_neg): ... here. No longer static.
36740 * tree.h (get_range_pos_neg): New prototype.
36741 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
36742 are known to be in between 0 and signed maximum inclusive, try to
36743 expand both unsigned and signed divmod and use the cheaper one from
36746 2017-02-22 Jeff Law <law@redhat.com>
36748 PR tree-optimization/79578
36749 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
36750 to compare base operands.
36752 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
36755 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
36756 gpc_reg_operand instead of fpr_reg_operand.
36758 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
36760 * config/mips/mips.c (mips_return_in_memory): Force FP
36761 vector types to be returned in memory for o32 ABI.
36763 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36765 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
36766 instead of DW_TAG_member for static data member declarations and don't
36767 set no_linkage_name for static inline data members.
36768 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
36771 2017-02-22 Martin Liska <mliska@suse.cz>
36773 * doc/invoke.texi: Replace inequality signs with square brackets
36776 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
36779 * lra-constraints.c (simplify_operand_subreg): Handle
36780 WORD_REGISTER_OPERATIONS targets.
36782 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36785 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
36786 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
36787 elimination by swapping fld*.
36789 2017-02-22 Richard Biener <rguenther@suse.de>
36791 PR tree-optimization/79673
36792 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
36793 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
36794 irrelevant address-space qualifiers and avoiding a
36795 ADDR_SPACE_CONVERT_EXPR from fold_convert.
36797 2017-02-22 Richard Biener <rguenther@suse.de>
36799 PR tree-optimization/79666
36800 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
36801 to not symbolically negate if that may introduce undefined
36804 2017-02-22 Martin Liska <mliska@suse.cz>
36807 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
36808 * data-streamer-out.c (streamer_write_gcov_count_stream):
36810 * value-prof.c (stream_out_histogram_value): Make assert more
36811 precise based on type of counter.
36813 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
36816 * config/i386/i386.md (standard_x87sse_constant_load splitter):
36817 Use nonimmediate_operand instead of memory_operand for operand 1.
36818 (float-extend standard_x87sse_constant_load splitter): Ditto.
36820 2017-02-21 Jeff Law <law@redhat.com>
36822 PR tree-optimization/79621
36823 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
36824 blocks with edges to themselves.
36826 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36829 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
36830 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
36831 Use gimple_call_builtin_p.
36834 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
36835 on temporarily removed DEBUG_INSNs.
36837 PR tree-optimization/79649
36838 * tree-loop-distribution.c (classify_partition): Give up on
36839 non-generic address space loads/stores.
36841 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
36843 * doc/loop.texi (Loop manipulation): Remove nonexistent
36844 tree_ssa_loop_version from the documentation.
36845 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
36847 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36850 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
36851 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
36852 * config/rs6000/rs6000.c: Include except.h.
36853 (rs6000_expand_split_stack_prologue): Call
36854 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
36856 2017-02-21 Martin Jambor <mjambor@suse.cz>
36859 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
36860 have been analyzed.
36862 2017-02-21 Martin Jambor <mjambor@suse.cz>
36864 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
36865 for backward compatibility only.
36866 * doc/invoke.texi (Option Summary): Remove all references to
36867 -fipa-cp-alignment.
36869 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
36873 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36875 * lra-constraints.c (curr_insn_transform): Handle
36876 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36878 2017-02-21 Martin Liska <mliska@suse.cz>
36880 * config/i386/i386.opt: Replace -masm-dialect with -masm.
36882 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
36884 PR translation/79638
36885 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
36887 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
36890 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
36891 (arm_function_ok_for_sibcall): Return false for an indirect call by
36892 descriptor if all the argument registers are used.
36893 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
36894 alignment of the function.
36896 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36898 PR tree-optimization/61441
36899 * simplify-rtx.c (simplify_const_unary_operation): For
36900 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
36901 the sNaN unmodified.
36903 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
36905 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
36906 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
36907 instead of SYSTEM_HEADER_DIR.
36909 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
36910 Martin Liška <mliska@suse.cz>
36912 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
36913 Fix typos and grammar, use active voice, and clarify.
36915 2017-02-20 Marek Polacek <polacek@redhat.com>
36917 PR middle-end/79537
36918 * gimplify.c (gimplify_expr): Handle unused *&&L;.
36921 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
36923 2017-02-20 Jakub Jelinek <jakub@redhat.com>
36926 * config/i386/i386.c (ix86_expand_builtin): Handle
36927 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
36928 ix86_builtins_isa[fcode].isa as a requirement of those
36929 flags and any other flag in the bitmask.
36930 (ix86_init_mmx_sse_builtins): Use 0 instead of
36931 ~OPTION_MASK_ISA_64BIT as mask.
36932 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
36933 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
36934 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
36935 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
36937 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36940 * lra-constraints.c (split_reg): Check requested split mode
36941 is supported by the register.
36943 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36945 * lra-constraints.c (simplify_operand_subreg): Remove early
36948 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36951 * lra-constraints.c (curr_insn_transform): Tighten condition
36952 for converting SUBREG reloads from OP_OUT to OP_INOUT.
36954 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36957 * lra-constraints.c (curr_insn_transform): Handle
36958 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36960 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
36963 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
36965 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
36967 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
36970 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
36973 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36975 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
36978 2017-02-18 Jakub Jelinek <jakub@redhat.com>
36980 * final.c (last_columnnum, override_columnnum): New variables.
36981 (final_start_function): Set last_columnnum, pass it to begin_prologue
36982 hook and pass 0 to dwarf2out_begin_prologue.
36983 (final_scan_insn): Update override_columnnum. Pass last_columnnum
36984 to source_line debug hook.
36985 (notice_source_line): Compute last_columnnum and for debug_column_info
36986 return true on column changes.
36987 * debug.h (struct gcc_debug_hooks): Add column argument to
36988 source_line and begin_prologue hooks.
36989 (debug_nothing_int_charstar_int_bool): Remove prototype.
36990 (debug_nothing_int_int_charstar,
36991 debug_nothing_int_int_charstar_int_bool): New prototypes.
36992 (dwarf2out_begin_prologue): Add column argument.
36993 * debug.c (do_nothing_debug_hooks): Adjust source_line and
36994 begin_prologue hooks.
36995 (debug_nothing_int_charstar_int_bool): Remove.
36996 (debug_nothing_int_int_charstar,
36997 debug_nothing_int_int_charstar_int_bool): New functions.
36998 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
36999 through to dwarf2out_source_line.
37000 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
37001 (dwarf2out_source_line): Add column argument, emit it if requested.
37002 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
37004 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37005 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37006 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
37007 through to dwarf2out_begin_prologue.
37008 (vmsdbgout_source_line): Add column argument, pass it through to
37009 dwarf2out_source_line.
37010 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
37011 dbxout_source_line caller.
37012 (dbxout_source_line): Add column argument.
37014 * common.opt (gno-column-info, gcolumn-info): New options.
37015 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
37016 (check_die): Also test for multiple DW_AT_decl_column attributes.
37017 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
37018 DW_AT_decl_column if requested.
37019 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
37021 (gen_variable_die): Likewise.
37022 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
37023 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
37026 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
37027 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
37028 (ix86_handle_option): Handle OPT_m3dnowa.
37029 * doc/invoke.texi (-m3dnowa): Document.
37030 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
37031 -m3dnowa instead of -m3dnow -march=athlon.
37034 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
37035 instead of gcc_assert for K, r and R code checks. Formatting fixes.
37037 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37040 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
37041 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
37042 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
37043 generator for vsx_xxpermdi_<mode>_be.
37044 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
37045 force big-endian semantics.
37046 (vsx_xxpermdi_<mode>_be): New define_expand with same
37047 implementation as previous version of vsx_xxpermdi_<mode>.
37049 2017-02-17 Jakub Jelinek <jakub@redhat.com>
37051 PR tree-optimization/79327
37052 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
37053 variable, its initialization and use.
37055 2017-02-17 Julia Koval <julia.koval@intel.com>
37057 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
37058 (OPTION_MASK_ISA_PKU_UNSET): New.
37059 (ix86_handle_option): Handle -mrdpid.
37060 * config/i386/cpuid.h (bit_RDPID): New.
37061 * config/i386/driver-i386.c (host_detect_local_cpu):
37062 Detect RDPID feature.
37063 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
37064 * config/i386/i386-c.c (ix86_target_macros_internal):
37066 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
37067 (ix86_valid_target_attribute_inner_p): Add "rdpid".
37068 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
37069 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
37070 * config/i386/i386.md (define_insn "rdpid"): New.
37071 * config/i386/i386.opt Add -mrdpid.
37072 * config/i386/immintrin.h (_rdpid_u32): New.
37074 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
37076 PR rtl-optimization/79541
37077 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
37078 instead of transforming it into USE.
37080 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
37082 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
37083 If HONOR_SNANS (SFmode) force the input to a register.
37084 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
37085 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
37086 an frsp or similar insn.
37088 2017-02-17 Martin Liska <mliska@suse.cz>
37090 PR rtl-optimization/79577
37091 * params.def (selsched-max-sched-times): Increase minimum to 1.
37093 2017-02-17 Martin Liska <mliska@suse.cz>
37095 PR rtl-optimization/79574
37096 * gcse.c (want_to_gcse_p): Prevent integer overflow.
37098 2017-02-17 Martin Liska <mliska@suse.cz>
37100 PR tree-optimization/79529
37101 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
37102 ssa_defined_default_def_p to handle cases which are implicitly
37104 * tree-ssa.c (ssa_defined_default_def_p): New function.
37105 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
37106 which are implicitly defined.
37107 * tree-ssa.h (ssa_defined_default_def_p): Declare.
37109 2017-02-17 Richard Biener <rguenther@suse.de>
37111 PR middle-end/79576
37112 * params.def (max-ssa-name-query-depth): Limit to 10.
37114 2017-02-17 Richard Biener <rguenther@suse.de>
37116 PR tree-optimization/79552
37117 * tree-ssa-structalias.c (visit_loadstore): Properly verify
37120 2017-02-17 Richard Biener <rguenther@suse.de>
37123 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
37125 2017-02-17 Marek Polacek <polacek@redhat.com>
37127 PR middle-end/79536
37128 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
37129 (fold_negate_expr): New wrapper.
37131 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
37133 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
37134 Correct terminology and de-emphasize pre-standard behavior.
37136 2017-02-16 Alan Modra <amodra@gmail.com>
37138 PR rtl-optimization/79286
37139 * ira.c (def_dominates_uses): New function.
37140 (update_equiv_regs): Don't create an equivalence for insns that
37141 may trap where the register def does not dominate the use.
37143 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
37145 PR rtl-optimization/78127
37146 * lra.c (lra): Call lra_eliminate before finish the loop after
37149 2017-02-16 Richard Biener <rguenther@suse.de>
37151 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
37153 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
37154 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
37155 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
37156 (isl_val_int_from_wi): New function.
37157 (extract_affine_gmp): Rename to ...
37158 (extract_affine_wi): ... this, take a widest_int.
37159 (extract_affine_int): Just wrap extract_affine_wi.
37160 (add_param_constraints): Use isl_val_int_from_wi.
37161 (add_loop_constraints): Likewise, and extract_affine_wi.
37163 2017-02-15 Jeff Law <law@redhat.com>
37165 PR middle-end/79521
37166 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
37167 ira_init_register_move_cost_if_necessary.
37169 2017-02-15 Martin Sebor <msebor@redhat.com>
37171 PR middle-end/32003
37172 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
37173 removed in a prior commit.
37175 2017-02-15 Bin Cheng <bin.cheng@arm.com>
37177 PR tree-optimization/79347
37178 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
37179 counters during peeling.
37181 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
37183 * Makefile.in (site.exp): Remove "set ISLVER".
37185 2017-02-15 Jakub Jelinek <jakub@redhat.com>
37188 * real.c (real_from_integer): Call real_convert even for decimal.
37190 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
37193 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
37195 2017-02-14 Andrew Pinski <apinski@cavium.com>
37197 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
37198 cores and change the partno/implementer to be correct.
37199 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
37200 the 'B" as the implementer.
37201 * config/aarch64/aarch64-tune.md: Regenerate.
37203 2017-02-14 Carl Love <cel@us.ibm.com>
37205 * config/rs6000/rs6000.c: Add case statement entry to make the
37206 xvcvuxdsp built-in argument unsigned.
37207 * config/rs6000/vsx.md: Fix the source and return operand types so they
37208 match the instruction definitions from the ISA document. Fix typo
37209 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
37212 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
37215 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
37216 member early_clobber_alts.
37217 * lra-lives.c (reg_early_clobber_p): New.
37218 (process_bb_lives): Use it.
37219 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
37220 (debug_operand_data): Initialize early_clobber_alts.
37221 (setup_operand_alternative): Set up early_clobber_alts.
37222 (collect_non_operand_hard_regs): Ditto. Pass early clobber
37223 alternatives to new_insn_reg.
37224 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
37226 (lra_update_insn_regno_info): Pass the new arg.
37228 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37230 PR middle-end/79505
37231 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
37232 (new_oacc_loop_raw): Don't clear already cleared fields.
37235 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
37236 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
37237 _mm512_prefetch_i64gather_ps): New inline functions and macros.
37239 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
37242 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
37244 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
37247 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
37248 the extra instruction to the right place to store 128-bit constant
37251 2017-02-14 Martin Sebor <msebor@redhat.com>
37253 PR middle-end/79448
37254 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
37255 warning for strings of unknown length.
37257 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
37259 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
37261 2017-02-14 Jeff Law <law@redhat.com>
37264 * ira-costs.c (scan_one_insn): Initialize register move costs
37265 for pseudos seen in USE/CLOBBER insns.
37267 PR tree-optimization/79095
37268 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
37269 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
37270 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
37271 if the operands are known to be not equal, then the resulting range
37273 (intersect_ranges): If the new range is ~[0,0] and the old range is
37274 wide, then prefer ~[0,0].
37275 * tree-vrp.c (overflow_comparison_p_1): New function.
37276 (overflow_comparison_p): New function.
37277 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
37278 if NAME is used in an overflow test.
37279 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
37280 overflow check that can be expressed as an equality test, then adjust
37281 ops to be that equality test.
37283 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37285 * config/s390/s390-builtin-types.def: Remove flags argument.
37286 * config/s390/s390.c (s390_init_builtins): Likewise.
37288 2017-02-14 Martin Liska <mliska@suse.cz>
37290 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
37291 vector. Fix trailing white spaces.
37293 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
37295 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
37298 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37300 PR rtl-optimization/68664
37301 * config/arm/arm.c (arm_sched_can_speculate_insn):
37302 New function. Declare prototype.
37303 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37305 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37307 PR rtl-optimization/68664
37308 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
37310 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37312 2017-02-14 Amit Pawar <amit.pawar@amd.com>
37314 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
37315 max skip bytes for function, loop and jump.
37317 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37319 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
37320 ABS_EXPR for gimple dump.
37322 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37325 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
37327 PR tree-optimization/79408
37328 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
37329 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
37330 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
37331 also if rhs1 is INTEGER_CST.
37333 2017-02-14 Richard Biener <rguenther@suse.de>
37335 PR middle-end/79432
37336 * tree-into-ssa.c (insert_phi_nodes): When the function can
37337 have abnormal edges rewrite SSA names with broken use-def
37338 dominance out of SSA and register them for PHI insertion.
37340 2017-02-13 Martin Sebor <msebor@redhat.com>
37342 PR middle-end/79496
37343 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
37344 clearing info.nowrite flag when snprintf size argument is a range.
37346 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37348 * cprop.c (cprop_jump): Add missing space in string literal.
37349 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
37350 (get_constraint_for_component_ref): Likewise.
37351 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
37352 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
37353 * lra-constraints.c (process_alt_operands): Likewise.
37354 * ipa-inline.c (inline_small_functions): Likewise.
37355 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
37356 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
37357 * trans-mem.c (diagnose_tm_1_op): Likewise.
37358 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
37359 (grid_parallel_clauses_gridifiable): Likewise.
37361 * config/nvptx/mkoffload.c (process): Add space in between
37364 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
37365 "MOD4_SSE_REGS" and "ALL_REGS".
37367 * spellcheck.c (test_data): Add , in between "foo" and "food".
37369 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37372 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
37373 boundary crossing check and subsequent code generation agree.
37375 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37377 * config/aarch64/aarch64.c (has_memory_op): Delete.
37378 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
37381 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37383 PR rtl-optimization/79388
37384 PR rtl-optimization/79450
37385 * combine.c (distribute_notes): When removing TEM_INSN for which
37386 corresponding dest has last value recorded, invalidate that last
37389 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37391 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
37392 of explicit '@'. Add missing assembly comment marker on branch costs
37395 2017-02-13 Nathan Sidwell <nathan@acm.org>
37397 * gengtype-lex.l (<in_struct>): Add '/'.
37399 2017-02-13 Martin Liska <mliska@suse.cz>
37402 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
37404 2017-02-13 Richard Biener <rguenther@suse.de>
37406 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
37408 * configure: Re-generate.
37409 * config.in: Likewise.
37410 * graphite-dependences.c: Simplify as if
37411 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
37412 * graphite-isl-ast-to-gimple.c: Likewise.
37413 * graphite-optimize-isl.c: Likewise.
37414 * graphite-poly.c: Likewise.
37415 * graphite-sese-to-poly.c: Likewise.
37416 * graphite.h: Likewise.
37417 * toplev.c: Include isl/version.h and use isl_version () for
37418 printing the ISL version.
37419 * doc/install.texi: Update ISL requirement.
37421 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37423 * doc/standards.texi (Standards): Update reference to
37426 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37428 * doc/extend.texi (Named Address Spaces): sourceware.org now
37430 * doc/install.texi (Binaries): Ditto.
37433 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37435 * doc/cpp.texi: Replace "stringify"/"stringification" with C
37436 standard terminology "stringize"/"stringizing" throughout.
37437 * doc/cppinternals.texi: Likewise.
37439 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37441 * doc/extend.texi: Fix some spelling mistakes and typos.
37442 * doc/invoke.texi: Likewise.
37444 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37447 * params.def (inline-min-speedup) Change from 10 to 8.
37449 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37451 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
37454 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37457 * ipa-inline-analysis.c (get_minimal_bb): New function.
37458 (record_modified): Use it.
37459 (remap_edge_change_prob): Handle also ancestor functions.
37461 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
37463 * doc/contrib.texi (Contributors): Remove broken link into
37464 the Mauve CVS repository.
37466 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37468 PR middle-end/79454
37469 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
37470 result computation whenever lhs doesn't have vector mode, not
37471 just when it has BLKmode.
37473 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37475 * doc/makefile.texi (profiledbootstrap): Refer to the
37476 installation instructions only in textual form.
37478 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37481 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
37483 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37485 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
37486 (Specific): Update mingw-w64 reference.
37488 (Specific): Remove broken link to Renesas RX processor.
37490 2017-02-10 Richard Biener <rguenther@suse.de>
37492 * toplev.c (process_options): Do not mention obsolete graphite
37493 options when printing sorry message about missing graphite support.
37494 Mention -floop-nest-optimize.
37496 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
37498 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
37499 (vtst_p16): Likewise.
37500 (vtstq_p8): Likewise.
37501 (vtstq_p16): Likewise.
37503 (vtstq_p64): Likewise.
37504 * config/arm/arm_neon.h (vgetq_lane_p64): New.
37505 (vset_lane_p64): New.
37506 (vsetq_lane_p64): New.
37508 2017-02-10 Jakub Jelinek <jakub@redhat.com>
37510 PR tree-optimization/79411
37511 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
37512 stmt operands are SSA_NAMEs used in abnormal phis.
37513 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
37516 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37519 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
37522 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37524 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
37526 2017-02-09 Jakub Jelinek <jakub@redhat.com>
37528 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
37532 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
37533 not arbitrary TREE_CONSTANT.
37536 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
37537 "omp declare target link" attribute unless is_global_var.
37538 * omp-offload.c (find_link_var_op): Likewise.
37540 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
37541 Chung-Lin Tang <cltang@codesourcery.com>
37543 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
37545 (gimplify_adjust_omp_clauses): Don't delete TILE.
37546 (gimplify_omp_for): Deal with TILE.
37547 * internal-fn.c (expand_GOACC_TILE): New function.
37548 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
37550 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
37551 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
37553 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
37554 avoid DIV for outermost collapse var.
37555 (expand_oacc_for): Insert tile element loop as needed. Adjust.
37556 Remove out of date comments, fix whitespace.
37557 * omp-general.c (omp_extract_for_data): Deal with tiling.
37558 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
37559 adjust OLF_DIM_BASE value.
37560 (struct omp_for_data): Add tiling field.
37561 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
37562 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
37563 for auto loops. Remove default auto determining, moved to
37564 oacc_loop_fixed_partitions.
37565 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
37566 stmts, add e_mask field.
37567 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
37568 (oacc_thread_numbers): Use oacc_dim_call.
37569 (oacc_xform_tile): New.
37570 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
37571 (finish_oacc_loop): Adjust for ifns vector.
37572 (oacc_loop_discover_walk): Append loop abstraction sites to list,
37573 add case for GOACC_TILE fns.
37574 (oacc_loop_xform_loop): Delete.
37575 (oacc_loop_process): Iterate over call list directly, and add
37576 handling for GOACC_TILE fns.
37577 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
37579 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
37580 vector partitioning to outer loops. Assign 2 partitions to loops
37581 when available. Add TILE handling.
37582 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
37583 (execite_oacc_device_lower): Process GOACC_TILE fns,
37584 ignore unknown specs.
37585 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
37586 * tree.c (omp_clause_num_ops): Adjust TILE ops.
37587 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
37589 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
37591 * configure.ac (ACX_BUGURL): Update.
37592 * configure: Regenerate.
37594 2017-02-09 Richard Biener <rguenther@suse.de>
37596 PR tree-optimization/69823
37597 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
37598 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
37600 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37602 * config/arc/arc-c.def: Add __NPS400__ definition.
37603 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
37604 (TARGET_NPS400): Define.
37606 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37608 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
37610 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
37611 pointer, arch_info.
37612 (arc_cpu_types): Fill the arch_info field with a pointer into the
37613 arc_arch_types table.
37614 (arc_selected_cpu): Declare.
37615 * config/arc/arc.c (arc_selected_cpu): Make global.
37616 (arc_selected_arch): Delete.
37617 (arc_base_cpu): Delete.
37618 (arc_override_options): Remove references to deleted variables,
37619 update access to arch information.
37620 (ARC_OPT): Update access to arch information.
37621 (ARC_OPTX): Likewise.
37622 * config/arc/arc.h (arc_base_cpu): Remove declaration.
37623 (TARGET_ARC600): Update access to arch information.
37624 (TARGET_ARC601): Likewise.
37625 (TARGET_ARC700): Likewise.
37626 (TARGET_EM): Likewise.
37627 (TARGET_HS): Likewise.
37628 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
37631 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
37634 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
37635 condition/operands for integer GE/LE/GEU/LEU operations.
37637 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
37639 PR translation/79397
37640 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
37643 2017-02-08 Martin Jambor <mjambor@suse.cz>
37646 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
37647 whether allocation happened.
37648 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
37649 nothing was allocated.
37651 2017-02-08 Jakub Jelinek <jakub@redhat.com>
37653 PR tree-optimization/79408
37654 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
37655 constant, but SSA_NAME with a known integer range, use the minimum
37656 of that range instead of op1 to determine if modulo can be replaced
37657 with its first operand.
37659 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37661 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
37663 2017-02-08 Richard Biener <rguenther@suse.de>
37665 PR tree-optimization/71824
37666 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37667 Check all loops contained in the merged region.
37669 2017-02-07 Andrew Pinski <apinski@cavium.com>
37671 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
37673 2017-02-07 Andrew Pinski <apinski@cavium.com>
37675 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
37676 (thunderxt88): Likewise.
37677 (thunderxt81): Disable LSE and change v8.1 to v8.
37678 (thunderxt83): Likewise.
37680 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37681 Richard Biener <rguenther@suse.de>
37683 PR middle-end/79399
37684 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
37685 type from int to size_t.
37686 * ira-costs.c (struct_costs_size): Change type from int to size_t.
37688 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37690 PR rtl-optimization/79386
37691 * cprop.c (bypass_conditional_jumps): Initialize
37692 bypass_last_basic_block already before splitting bbs after
37693 unconditional traps...
37694 (bypass_conditional_jumps): ... rather than here.
37697 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
37698 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
37699 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
37700 fix -masm=intel patterns.
37702 2017-02-07 Richard Biener <rguenther@suse.de>
37704 PR tree-optimization/79256
37705 PR middle-end/79278
37706 * builtins.c (get_object_alignment_2): Use min_align_of_type
37707 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
37708 and ADJUST_FIELD_ALIGN.
37710 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
37712 * doc/tm.texi: Regenerate.
37713 * stor-layout.c (layout_decl): Adjust.
37714 (update_alignment_for_field): Likewise.
37715 (place_field): Likewise.
37716 (min_align_of_type): Likewise.
37717 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
37718 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
37719 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
37720 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
37721 * config/frv/frv.c (frv_adjust_field_align): Likewise.
37722 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
37723 * config/i386/i386.c (x86_field_alignment): Likewise.
37724 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
37725 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
37726 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
37727 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
37728 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
37729 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
37733 2017-01-30 Richard Biener <rguenther@suse.de>
37735 PR tree-optimization/79256
37736 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37737 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37740 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
37742 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
37743 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
37744 builtins to SImode and emit a zero-extend, if necessary.
37746 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37748 * docs/invoke.texi (RISC-V Options): Alphabetize.
37750 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37752 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
37755 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37757 * config/riscv/riscv.c: New file.
37758 * common/config/riscv/riscv-common.c: Likewise.
37759 * config.gcc: Likewise.
37760 * config/riscv/constraints.md: Likewise.
37761 * config/riscv/elf.h: Likewise.
37762 * config/riscv/generic.md: Likewise.
37763 * config/riscv/linux.h: Likewise.
37764 * config/riscv/multilib-generator: Likewise.
37765 * config/riscv/peephole.md: Likewise.
37766 * config/riscv/pic.md: Likewise.
37767 * config/riscv/predicates.md: Likewise.
37768 * config/riscv/riscv-builtins.c: Likewise.
37769 * config/riscv/riscv-c.c: Likewise.
37770 * config/riscv/riscv-ftypes.def: Likewise.
37771 * config/riscv/riscv-modes.def: Likewise.
37772 * config/riscv/riscv-opts.h: Likewise.
37773 * config/riscv/riscv-protos.h: Likewise.
37774 * config/riscv/riscv.h: Likewise.
37775 * config/riscv/riscv.md: Likewise.
37776 * config/riscv/riscv.opt: Likewise.
37777 * config/riscv/sync.md: Likewise.
37778 * config/riscv/t-elf-multilib: Likewise.
37779 * config/riscv/t-linux: Likewise.
37780 * config/riscv/t-linux-multilib: Likewise.
37781 * config/riscv/t-riscv: Likewise.
37782 * configure.ac: Likewise.
37783 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
37784 Waterman as RISC-V maintainers.
37785 * doc/install.texi: Add RISC-V entries.
37786 * doc/invoke.texi: Add RISC-V options section.
37787 * doc/md.texi: Add RISC-V constraints section.
37788 * configure: Regenerated.
37790 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
37793 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
37794 false values to be constant vectors with all 0 or all 1 bits set.
37795 (vcondu<mode><mode>): Likewise.
37796 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
37798 (fpmask_comparison_operator): Update comment.
37799 (vecint_comparison_operator): New predicate.
37800 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
37801 vector conditionals when the true and false values are constant
37802 vectors with all 0 bits or all 1 bits set.
37804 2017-02-06 Martin Sebor <msebor@redhat.com>
37806 PR tree-optimization/79376
37807 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
37809 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
37811 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
37812 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
37813 to simplify split condition.
37815 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37817 * omp-expand.c (oxpand_omp_atomic_fetch_op,
37818 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
37821 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
37823 PR rtl-optimization/68664
37824 * target.def (can_speculate_insn): New hook.
37825 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
37826 * doc/tm.texi: Regenerate.
37827 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
37828 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
37829 (rs6000_sched_can_speculate_insn): New function.
37831 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37833 PR tree-optimization/79284
37834 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
37835 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
37836 vectorizable_mask_load_store, vectorizable_operation,
37837 vect_is_simple_cond, get_same_sized_vectype): Use it instead
37838 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
37839 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
37840 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
37841 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
37842 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
37843 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
37844 is_gimple_assign (stmt). Replace another such test with
37845 is_gimple_assign (stmt).
37847 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37850 * config/avr/avr.c (rtl-iter.h): Include it.
37851 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
37852 (avr_legitimate_combined_insn): ...and implementation.
37854 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
37856 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
37857 * config/s390/s390.c (s390_const_operand_ok)
37858 (s390_canonicalize_comparison, s390_extract_part)
37859 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
37860 (s390_contiguous_bitmask_p, s390_rtx_costs)
37861 (legitimize_pic_address): Likewise.
37862 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
37863 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
37864 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
37865 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
37866 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
37868 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37870 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
37871 REGNO($0) == REGNO($1).
37873 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37875 * config/s390/linux.h(SIZE_TYPE): Add comment.
37877 2017-02-06 Julian Brown <julian@codesourcery.com>
37878 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37879 Virendra Pathak <virendra.pathak@broadcom.com>
37881 * config/aarch64/aarch64-cores.def: Change the scheduler
37883 * config/aarch64/aarch64.md: Include thunderx2t99.md.
37884 * config/aarch64/thunderx2t99.md: New file.
37886 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37888 * doc/standards.texi (Go Language): Update link to language
37891 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37893 * tree-eh.c (lower_resx): Sanitize profile.
37894 (cleanup_empty_eh_move_lp): Likewise.
37896 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37899 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
37901 * cfgloopmanip.h (loop_version): Update prototype.
37902 * modulo-sched.c (sms_schedule): Update call of loop_version.
37903 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
37904 * tree-parloops.c (gen_parallel_loop): Likewise.
37905 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
37906 * tree-ssa-loop-split.c (split_loop): Likewise.
37907 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
37908 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
37910 2017-02-05 Martin Liska <mliska@suse.cz>
37913 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
37915 (print_operand_address): Initialize a struct to zero.
37917 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37919 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
37920 garbage collector only in textual form.
37922 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37924 * doc/extend.texi (x86 specific memory model extensions for
37925 transactional memory): Simplify a phrase.
37927 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
37930 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
37931 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
37932 (atomic_storedi_1): Likewise.
37934 2017-02-04 Jakub Jelinek <jakub@redhat.com>
37936 PR tree-optimization/79338
37937 * tree-parloops.c (gather_scalar_reductions): Don't call
37938 vect_analyze_loop_form for loop->inner before destroying loop's
37941 2017-02-03 Martin Sebor <msebor@redhat.com>
37943 PR tree-optimization/79327
37944 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
37945 when precision has resulted in leading zeros.
37946 (format_integer): Adjust the likely counter to assume an unknown
37947 argument that may be zero is non-zero.
37949 2017-02-03 Jason Merrill <jason@redhat.com>
37952 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
37953 avoid copying non-taken branch.
37955 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37957 PR tree-optimization/79340
37958 * tree-vect-loop.c (vectorizable_reduction): Release
37959 vec_defs elements after safe_splicing them into other vectors.
37962 PR tree-optimization/79327
37963 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
37964 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
37966 (format_integer): Use wide_int_to_tree instead of build_int_cst
37967 + to_?hwi. If argmin is NULL, just set argmin and argmax to
37968 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
37969 of shortest and longest sequence.
37971 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
37973 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
37974 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
37976 2017-02-03 Walter Lee <walt@tilera.com>
37979 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
37980 after initial stackframe link reg save.
37981 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
37983 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37986 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
37987 wu for stxssp alternative.
37989 2017-02-03 Martin Sebor <msebor@redhat.com>
37991 PR tree-optimization/79352
37992 * gimple-fold.c (get_range_strlen): Add argument.
37993 (get_range_strlen): Change return type to bool.
37994 (get_maxval_strlen): Pass in a dummy argument.
37995 * gimple-fold.h (get_range_strlen): Change return type to bool.
37996 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
37997 * tree.h (array_at_struct_end_p): Add argument.
37998 * tree.c (array_at_struct_end_p): Handle it.
38000 2017-02-03 Martin Liska <mliska@suse.cz>
38003 * multiple_target.c (create_dispatcher_calls): Redirect edge
38004 from a caller of a dispatcher.
38005 (expand_target_clones): Make the clones local.
38006 (ipa_target_clone): Do both target clones and resolvers.
38007 (ipa_dispatcher_calls): Remove the pass.
38008 (pass_dispatcher_calls::gate): Likewise.
38009 (make_pass_dispatcher_calls): Likewise.
38010 * passes.def (pass_target_clone): Put as very first IPA early
38013 2017-02-03 Martin Liska <mliska@suse.cz>
38015 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
38016 in case of a function with ifunc attribute.
38018 2017-02-03 Martin Liska <mliska@suse.cz>
38020 * cgraph.c (cgraph_node::dump): Dump function version info.
38021 * symtab.c (symtab_node::dump_base): Add missing new line.
38023 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38025 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
38026 (ifcombine_ifandif): Use it.
38028 2017-02-03 Martin Liska <mliska@suse.cz>
38030 * doc/invoke.texi: Document default value for
38031 use-after-scope-direct-emission-threshold.
38033 2017-02-03 Martin Liska <mliska@suse.cz>
38035 PR tree-optimization/79339
38036 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
38037 (format_floating): Likewise.
38039 2017-02-03 Martin Liska <mliska@suse.cz>
38042 * ipa-prop.c (ipa_node_params_t::insert): Remove current
38044 (ipa_node_params_t::remove): Likewise.
38045 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
38046 initialization from removed ipa_node_params_t::insert.
38047 (ipa_node_params::~ipa_node_params): Move from removed
38048 ipa_node_params_t::release.
38049 * symbol-summary.h (symbol_summary::m_released): New member.
38050 Do not release a summary twice. Do not allow to call finalizer
38051 for types of a summary that live in GGC memory.
38053 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38055 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
38058 2017-02-02 Martin Sebor <msebor@redhat.com>
38060 PR middle-end/79275
38061 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
38062 (format_string): Tighten up the range of output for non-constant
38063 strings and correct the expected range for wide non-constant strings.
38065 2017-02-02 Martin Sebor <msebor@redhat.com>
38067 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
38069 PR middle-end/32003
38070 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
38072 (-fdump-tree-@var): Add to index and document how to come up
38073 with pass-specific option and dump file names.
38074 (-fdump-passes): Clarify where to look for output.
38076 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38078 PR middle-end/77445
38079 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
38080 statistics of the analyzed path; allow threading for speed when
38081 any of BBs along the path are optimized for speed.
38083 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
38085 PR middle-end/78468
38086 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
38087 settings of the virtual registers.
38090 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
38092 * explow.c (get_dynamic_stack_size): Take known alignment of stack
38093 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
38096 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38098 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
38099 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
38101 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38103 * config/s390/s390.md: Add missing comments with the expanded
38105 * config/s390/vector.md: Likewise.
38106 * config/s390/vx-builtins.md: Likewise.
38108 2017-02-02 Jakub Jelinek <jakub@redhat.com>
38111 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
38112 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
38113 conditions on a single line.
38115 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38117 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38118 __S390_VX__ to __VX__.
38120 2017-02-01 Andrew Pinski <apinski@cavium.com>
38122 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
38123 stmt_info to record_stmt_cost.
38124 (vect_get_known_peeling_cost): Pass stmt_info if known to
38126 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
38127 cpu_vector_cost field into
38128 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
38129 field into vec_int_stmt_cost and vec_fp_stmt_cost.
38130 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
38131 splitting of scalar_stmt_cost and vec_stmt_cost.
38132 (thunderx_vector_cost): Likewise.
38133 (cortexa57_vector_cost): LIkewise.
38134 (exynosm1_vector_cost): Likewise.
38135 (xgene1_vector_cost): Likewise.
38136 (thunderx2t99_vector_cost): Improve after the splitting of the two
38138 (aarch64_builtin_vectorization_cost): Update for the splitting of
38139 scalar_stmt_cost and vec_stmt_cost.
38141 2017-02-01 Torvald Riegel <triegel@redhat.com>
38142 Richard Henderson <rth@redhat.com>
38144 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
38145 conditional on existance of a fast atomic load.
38146 * optabs-query.c (can_atomic_load_p): New function.
38147 * optabs-query.h (can_atomic_load_p): Declare it.
38148 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
38149 no fast atomic load is available for the particular size of access.
38150 (expand_atomic_compare_and_swap): Likewise.
38151 (expand_atomic_load): Likewise.
38152 (expand_atomic_store): Likewise.
38153 (expand_atomic_fetch_op): Likewise.
38154 * testsuite/lib/target-supports.exp
38155 (check_effective_target_sync_int_128): Remove x86 because it provides
38156 no fast atomic load.
38157 (check_effective_target_sync_int_128_runtime): Likewise.
38159 2017-02-01 Richard Biener <rguenther@suse.de>
38161 * graphite.c: Include tree-vectorizer.h for find_loop_location.
38162 (graphite_transform_loops): Provide opt-info for optimized nests.
38163 * tree-parloop.c (parallelize_loops): Provide opt-info for
38164 parallelized loops.
38166 2017-02-01 Richard Biener <rguenther@suse.de>
38168 PR middle-end/79315
38169 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
38170 was not set before.
38172 2017-02-01 Richard Biener <rguenther@suse.de>
38174 PR tree-optimization/71824
38175 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38176 Verify the loops are valid in the merged SESE region.
38177 (scop_detection::can_represent_loop_1): Check analyzing the
38178 evolution of the number of iterations in the region succeeds.
38180 2017-01-31 Ian Lance Taylor <iant@golang.org>
38182 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
38183 REG_ARGS_SIZE note to 32-bit push insns and call insn.
38185 2017-01-31 David Malcolm <dmalcolm@redhat.com>
38187 PR preprocessor/79210
38188 * input.c (get_substring_ranges_for_loc): Replace line_width
38189 assertion with error-handling.
38191 2017-01-31 Richard Biener <rguenther@suse.de>
38193 PR tree-optimization/77318
38194 * graphite-sese-to-poly.c (extract_affine): Fix assert.
38195 (create_pw_aff_from_tree): Take loop parameter.
38196 (add_condition_to_pbb): Pass loop of the condition to
38197 create_pw_aff_from_tree.
38199 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38201 * config/s390/s390.c (s390_asan_shadow_offset): New function.
38202 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
38204 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
38208 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
38210 (convert_int_to_float128): Likewise.
38211 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
38212 (convert_int_to_float128): Likewise.
38213 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
38214 (UNSPEC_IEEE128_CONVERT): Likewise.
38215 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
38216 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
38217 Use local variables for IBM extended format.
38218 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
38219 (fix_trunc<mode>si2_fprs): Likewise.
38220 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
38221 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
38222 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
38223 to know that we can now have integers of all sizes in vector
38225 (fix<uns>_<mode>di2_hw): Likewise.
38226 (float<uns>_<mode>si2_hw): Likewise.
38227 (fix_<mode>si2_hw): Likewise.
38228 (fixuns_<mode>si2_hw): Likewise.
38229 (float<uns>_<mode>di2_hw): Likewise.
38230 (float_<mode>di2_hw): Likewise.
38231 (float_<mode>si2_hw): Likewise.
38232 (floatuns_<mode>di2_hw): Likewise.
38233 (floatuns_<mode>si2_hw): Likewise.
38234 (xscvqp<su>wz_<mode>): Delete, no longer used.
38235 (xscvqp<su>dz_<mode>): Likewise.
38236 (xscv<su>dqp_<mode>): Likewise.
38237 (ieee128_mfvsrd_64bit): Likewise.
38238 (ieee128_mfvsrd_32bit): Likewise.
38239 (ieee128_mfvsrwz): Likewise.
38240 (ieee128_mtvsrw): Likewise.
38241 (ieee128_mtvsrd_64bit): Likewise.
38242 (ieee128_mtvsrd_32bit): Likewise.
38244 2017-01-31 Martin Liska <mliska@suse.cz>
38247 * ipa-prop.c (ipa_free_all_node_params): Call release method
38248 instead of ~sumbol_summary to not to trigger double times
38251 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
38253 PR tree-optimization/71691
38254 * bitmap.h (class auto_bitmap): New.
38255 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
38256 is_maybe_undefined instead of ssa_undefined_value_p.
38258 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38260 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38261 __S390_ARCH_LEVEL__ to __ARCH__.
38263 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38265 PR tree-optimization/79267
38266 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
38267 if should_remove_lhs_p is true.
38269 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
38272 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
38273 (add_alignment_attribute): New.
38274 (base_type_die): Add alignment attribute.
38275 (subrange_type_die): Likewise.
38276 (modified_type_die): Likewise.
38277 (gen_array_type_die): Likewise.
38278 (gen_descr_array_type_die: Likewise.
38279 (gen_enumeration_type_die): Likewise.
38280 (gen_subprogram_die): Likewise.
38281 (gen_variable_die): Likewise.
38282 (gen_field_die): Likewise.
38283 (gen_ptr_to_mbr_type_die): Likewise.
38284 (gen_struct_or_union_type_die): Likewise.
38285 (gen_subroutine_type_die): Likewise.
38286 (gen_typedef_die): Likewise.
38287 (base_type_cmp): Compare alignment attribute.
38289 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38292 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
38293 (setb_unsigned) New pattern for setb with CCUNS.
38294 * config/rs6000/rs6000.c (expand_block_compare): Use a different
38295 subfc./subfe sequence to avoid overflow problems. Generate a
38296 shorter sequence with cmpld/setb for power9.
38297 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
38298 for generating subfc. instruction.
38299 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
38300 now uses this instruction.
38302 2017-01-30 Ian Lance Taylor <iant@google.com>
38305 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
38306 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
38308 2017-01-30 Martin Sebor <msebor@redhat.com>
38310 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
38311 Move constant to the right of a relational operator.
38312 (get_mpfr_format_length, format_character, format_string): Ditto.
38313 (should_warn_p, maybe_warn): Same.
38315 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
38317 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
38320 * asan.c (get_translation_unit_decl): Remove function.
38321 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
38323 2017-01-30 Martin Liska <mliska@suse.cz>
38325 PR gcov-profile/79259
38326 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
38327 -fprofile-generate.
38329 2017-01-30 Martin Liska <mliska@suse.cz>
38332 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
38333 Initialize variables with NULL value.
38335 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
38338 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
38340 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
38342 2017-01-30 Richard Biener <rguenther@suse.de>
38344 PR tree-optimization/79276
38345 * tree-vrp.c (process_assert_insertions): Properly adjust common
38346 when removing a duplicate.
38348 2017-01-30 Richard Biener <rguenther@suse.de>
38350 PR tree-optimization/79256
38351 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38352 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38354 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
38356 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
38359 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
38360 ("*r<noxa>sbg_<mode>_sll_bitmask")
38361 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
38362 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
38363 Use contiguous_bitmask_nowrap_operand.
38365 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38368 * config/rs6000/altivec.h (vec_xl): Revise #define.
38369 (vec_xst): Likewise.
38371 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
38373 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
38375 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
38377 PR rtl-optimization/79194
38378 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
38379 traps before call to bypass_conditional_jumps.
38381 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38383 PR tree-optimization/71374
38384 * lra-constraints.c (check_conflict_input_operands): New.
38385 (match_reload): Use it.
38387 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38390 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
38391 account to calculate conflict_set.
38393 2017-01-27 Bin Cheng <bin.cheng@arm.com>
38395 PR rtl-optimization/78559
38396 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
38397 other_insn in combine.
38399 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
38401 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
38402 uint16_type_node for BT_UINT16.
38404 2017-01-27 David Malcolm <dmalcolm@redhat.com>
38406 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
38407 "RTL Tests" to menu.
38408 (GIMPLE Tests): New node.
38409 (RTL Tests): New node.
38411 2017-01-27 Richard Biener <rguenther@suse.de>
38413 PR tree-optimization/79245
38414 * tree-loop-distribution.c (distribute_loop): Apply cost
38415 modeling also to detected patterns.
38417 2017-01-27 Richard Biener <rguenther@suse.de>
38419 PR tree-optimization/71433
38420 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
38421 (compare_assert_loc): New function.
38422 (process_assert_insertions): Sort and optimize assert locations
38423 to remove duplicates and push down identical assertions on
38424 edges to their destination block.
38426 2017-01-27 Richard Biener <rguenther@suse.de>
38428 PR tree-optimization/79244
38429 * tree-vrp.c (remove_range_assertions): Forcefully propagate
38430 out SSA names even if abnormal.
38432 2017-01-27 Jakub Jelinek <jakub@redhat.com>
38434 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
38435 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
38436 instead of MPFR_RNDN.
38438 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
38441 * arm.c (arm_option_override): Don't call build_target_option_node
38442 until after doing all option overrides.
38443 (arm_valid_target_attribute_tree): Likewise.
38445 2017-01-27 Martin Liska <mliska@suse.cz>
38447 * doc/invoke.texi (-fprofile-arcs): Document profiling support
38448 for {cd}tors and C++ {cd}tors.
38450 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
38452 * config/s390/s390.md ("*setmem_long_and")
38453 ("*setmem_long_and_31z"): Use zero_extend instead of and.
38455 2017-01-26 Martin Sebor <msebor@redhat.com>
38457 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
38460 2017-01-26 Martin Sebor <msebor@redhat.com>
38462 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
38463 HAVE_DFmode before using XFmode or DFmode.
38464 (parse_directive): Avoid using the z length modifier to avoid
38465 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
38467 PR middle-end/78703
38468 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
38469 to accept adjustment as an array.
38470 (get_int_range): New function.
38471 (struct directive): Make width and prec arrays.
38472 (directive::set_width, directive::set_precision): Call get_int_range.
38473 (format_integer, format_floating): Handle width and precision ranges.
38474 (format_string, parse_directive): Same.
38476 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38479 * dwarf2out.c (generate_skeleton_bottom_up): For children with
38480 comdat_type_p set, just clone them, but keep the children in the
38484 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
38485 which have direct callers with -fvar-tracking-assignments enabled
38487 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
38488 inside of type units.
38490 2017-01-26 Martin Sebor <msebor@redhat.com>
38492 PR middle-end/78703
38493 * gimple-ssa-sprintf.c (struct result_range): Add likely and
38495 (struct format_result): Replace number_chars, number_chars_min,
38496 and number_chars_max with a single member of struct result_range.
38498 (format_result::operator+=): Adjust.
38499 (struct fmtresult): Remove bounded. Handle likely and unlikely
38501 (fmtresult::adjust_for_width_or_precision): New function.
38502 (fmtresult:type_max_digits): New function.
38503 (bytes_remaining): Handle likely and unlikely counters.
38504 (min_bytes_remaining): Remove.
38505 (format_percent): Simplify.
38506 (format_integer, format_floating): Set likely and unlikely counters.
38507 (get_string_length, format_character, format_string): Same.
38508 (format_plain, should_warn_p): New function.
38509 (maybe_warn): Call should_warn_p. Update diagnostic messages
38510 and handle those for all directives, including plain strings.
38511 (format_directive): Handle likely and unlikely counters.
38512 Remove unnecessary quoting from diagnostics. Add an informational
38514 (add_bytes): Remove.
38515 (pass_sprintf_length::compute_format_length): Simplify.
38516 (try_substitute_return_value): Handle likely and unlikely counters.
38518 2017-01-26 Carl Love <cel@us.ibm.com>
38520 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
38521 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
38523 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
38526 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
38527 endianess for subregs into account.
38528 * lra-constraints.c (lra_constraints): Do risky transformations
38529 always on the first iteration.
38530 * lra-lives.c (check_pseudos_live_through_calls): Add arg
38531 last_call_used_reg_set.
38532 (process_bb_lives): Define and use last_call_used_reg_set.
38533 * lra.c (lra): Always continue after lra_constraints on the first
38536 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
38538 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
38540 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
38542 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38544 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
38545 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
38546 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
38547 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
38548 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
38549 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
38550 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
38551 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
38552 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
38554 2017-01-26 Marek Polacek <polacek@redhat.com>
38557 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
38558 for the third operand.
38560 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38562 PR middle-end/79236
38563 * omp-low.c (struct omp_context): Add simt_stmt field.
38564 (scan_omp_for): Return omp_context *.
38565 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
38566 context to the _simt_ SIMD stmt.
38567 (lower_omp_for): For combined SIMD with sibling _simt_
38568 SIMD, make sure to use the same decls in _looptemp_
38569 clauses as in the sibling.
38571 2017-01-26 David Sherwood <david.sherwood@arm.com>
38573 PR middle-end/79212
38574 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
38577 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38580 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
38581 emit fld b; fld a; if possible.
38583 * brig-builtins.def: Update copyright years.
38584 * config/arm/arm_acle_builtins.def: Update copyright years.
38586 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
38589 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
38590 constraint instead of o for the stxsd instruction.
38592 2017-01-25 Carl Love <cel@us.ibm.com>
38594 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
38595 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
38597 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
38599 * doc/invoke.texi (C++ Dialect Options): Fix typo.
38601 2017-01-25 Richard Biener <rguenther@suse.de>
38603 PR tree-optimization/69264
38604 * target.def (vector_alignment_reachable): Improve documentation.
38605 * doc/tm.texi: Regenerate.
38606 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
38608 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
38609 earlier changes with respect to TYPE_USER_ALIGN.
38610 (vector_alignment_reachable_p): Likewise. Improve dumping.
38612 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38615 * config/arm/arm.md (xordi3): Force constant operand into a register
38618 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38620 * doc/invoke.texi (-fstore-merging): Correct default optimization
38621 levels at which it is enabled.
38622 (-O): Move -fstore-merging from list to...
38625 2017-01-25 Richard Biener <rguenther@suse.de>
38628 * omp-expand.c: Include debug.h.
38629 (expand_omp_taskreg): Make sure to generate early debug before
38630 outlining anything from a function.
38631 (expand_omp_target): Likewise.
38632 (grid_expand_target_grid_body): Likewise.
38634 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
38637 * asan.c (get_translation_unit_decl): New function.
38638 (asan_add_global): Extract modules file name from globals
38639 TRANSLATION_UNIT_DECL name.
38641 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
38644 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
38645 for long calls with APCS frame and VFP.
38647 2017-01-24 David Malcolm <dmalcolm@redhat.com>
38649 * cfg.c (original_copy_tables_initialized_p): New function.
38650 * cfg.h (original_copy_tables_initialized_p): New decl.
38651 * cfgrtl.c (relink_block_chain): Guard the call to
38652 free_original_copy_tables with a call to
38653 original_copy_tables_initialized_p.
38654 * cgraph.h (symtab_node::native_rtl_p): New decl.
38655 * cgraphunit.c (symtab_node::native_rtl_p): New function.
38656 (symtab_node::needed_p): Don't assert for early assembly output
38657 for __RTL functions.
38658 (cgraph_node::finalize_function): Set "force_output" for __RTL
38660 (cgraph_node::analyze): Bail out early for __RTL functions.
38661 (analyze_functions): Update assertion to support __RTL functions.
38662 (cgraph_node::expand): Bail out early for __RTL functions.
38663 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
38665 * function.h (struct function): Update comment for field
38667 * gimple-expr.c: Include "tree-pass.h".
38668 (gimple_has_body_p): Return false for __RTL functions.
38669 * Makefile.in (OBJS): Add run-rtl-passes.o.
38670 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
38672 (gcc::pass_manager::get_clean_slate): New accessor.
38673 * passes.c: Include "insn-addr.h".
38674 (should_skip_pass_p): Add logging. Update logic for running
38675 "expand" to be compatible with both __GIMPLE and __RTL. Guard
38676 property-provider override so it is only done for gimple passes.
38678 (skip_pass): New function.
38679 (execute_one_pass): Call skip_pass when skipping passes.
38680 * read-md.c (md_reader::read_char): Support filtering
38681 the input to a subset of line numbers.
38682 (md_reader::md_reader): Initialize fields
38683 m_first_line and m_last_line.
38684 (md_reader::read_file_fragment): New function.
38685 * read-md.h (md_reader::read_file_fragment): New decl.
38686 (md_reader::m_first_line): New field.
38687 (md_reader::m_last_line): New field.
38688 * read-rtl-function.c (function_reader::create_function): Only
38689 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
38690 curr_properties. Set DECL_INITIAL to a dummy block.
38691 (read_rtl_function_body_from_file_range): New function.
38692 * read-rtl-function.h (read_rtl_function_body_from_file_range):
38694 * run-rtl-passes.c: New file.
38695 * run-rtl-passes.h: New file.
38697 2017-01-24 Jeff Law <law@redhat.com>
38699 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
38702 2017-01-24 Bin Cheng <bin.cheng@arm.com>
38704 PR tree-optimization/79159
38705 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
38706 (record_nonwrapping_iv): Improve boundary using above function if no
38707 value range information.
38709 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
38710 Martin Jambor <mjambor@suse.cz>
38712 * brig-builtins.def: New file.
38713 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
38714 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
38715 (DEF_HSAIL_SAT_BUILTIN): Likewise.
38716 (DEF_HSAIL_INTR_BUILTIN): Likewise.
38717 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
38718 * builtin-types.def (BT_INT8): New.
38719 (BT_INT16): Likewise.
38720 (BT_UINT8): Likewise.
38721 (BT_UINT16): Likewise.
38722 (BT_FN_ULONG): Likewise.
38723 (BT_FN_UINT_INT): Likewise.
38724 (BT_FN_UINT_ULONG): Likewise.
38725 (BT_FN_UINT_LONG): Likewise.
38726 (BT_FN_UINT_PTR): Likewise.
38727 (BT_FN_ULONG_PTR): Likewise.
38728 (BT_FN_INT8_FLOAT): Likewise.
38729 (BT_FN_INT16_FLOAT): Likewise.
38730 (BT_FN_UINT32_FLOAT): Likewise.
38731 (BT_FN_UINT16_FLOAT): Likewise.
38732 (BT_FN_UINT8_FLOAT): Likewise.
38733 (BT_FN_UINT64_FLOAT): Likewise.
38734 (BT_FN_UINT16_UINT32): Likewise.
38735 (BT_FN_UINT32_UINT16): Likewise.
38736 (BT_FN_UINT16_UINT16_UINT16): Likewise.
38737 (BT_FN_INT_PTR_INT): Likewise.
38738 (BT_FN_UINT_PTR_UINT): Likewise.
38739 (BT_FN_LONG_PTR_LONG): Likewise.
38740 (BT_FN_ULONG_PTR_ULONG): Likewise.
38741 (BT_FN_VOID_UINT64_UINT64): Likewise.
38742 (BT_FN_UINT8_UINT8_UINT8): Likewise.
38743 (BT_FN_INT8_INT8_INT8): Likewise.
38744 (BT_FN_INT16_INT16_INT16): Likewise.
38745 (BT_FN_INT_INT_INT): Likewise.
38746 (BT_FN_UINT_FLOAT_UINT): Likewise.
38747 (BT_FN_FLOAT_UINT_UINT): Likewise.
38748 (BT_FN_ULONG_UINT_UINT): Likewise.
38749 (BT_FN_ULONG_UINT_PTR): Likewise.
38750 (BT_FN_ULONG_ULONG_ULONG): Likewise.
38751 (BT_FN_UINT_UINT_UINT): Likewise.
38752 (BT_FN_VOID_UINT_PTR): Likewise.
38753 (BT_FN_UINT_UINT_PTR: Likewise.
38754 (BT_FN_UINT32_UINT64_PTR): Likewise.
38755 (BT_FN_INT_INT_UINT_UINT): Likewise.
38756 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
38757 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
38758 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
38759 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
38760 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
38761 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
38762 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
38763 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
38764 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
38765 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
38766 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
38767 * doc/frontends.texi: List BRIG FE.
38768 * doc/install.texi (Testing): Add BRIG tesring requirements.
38769 * doc/invoke.texi (Overall Options): Mention BRIG.
38770 * doc/standards.texi (Standards): Doucment BRIG HSA version.
38772 2017-01-24 Richard Biener <rguenther@suse.de>
38774 PR translation/79208
38775 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
38777 2017-01-24 Martin Jambor <mjambor@suse.cz>
38780 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
38781 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
38782 and known_contexts.
38784 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
38786 PR middle-end/79123
38787 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
38788 casts from signed to unsigned really don't have a range.
38790 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
38792 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
38793 GMP_RNDx for compatiblity.
38795 2017-01-24 Martin Liska <mliska@suse.cz>
38798 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
38799 that would prevent us to call alloca with -1 as argument.
38801 2017-01-24 Jakub Jelinek <jakub@redhat.com>
38803 * dwarf2out.c (output_compilation_unit_header, output_file_names):
38804 Avoid -Wformat-security warning.
38806 2017-01-23 Andrew Pinski <apinski@cavium.com>
38808 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
38811 2017-01-23 Martin Sebor <msebor@redhat.com>
38813 PR middle-end/78703
38814 * gimple-ssa-sprintf.c (warn_level): New global.
38815 (format_integer): Use it here and throughout the rest of the file.
38816 Use the same switch to compute sign as base.
38817 (maybe_warn): New function.
38818 (format_directive): Factor out warnings into maybe_warn.
38819 Add debugging output. Use warn_level.
38820 (add_bytes): Use warn_level.
38821 (pass_sprintf_length::compute_format_length): Add debugging output.
38822 (try_substitute_return_value): Same.
38823 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
38825 PR middle-end/78703
38826 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
38827 (struct fmtresult, format_integer, format_floating): Adjust.
38828 (fmtresult::fmtresult): Set max correctly in two argument ctor.
38829 (get_string_length, format_string,format_directive): Same.
38830 (pass_sprintf_length::compute_format_length): Same.
38831 (try_substitute_return_value): Simplify slightly.
38833 PR middle-end/78703
38834 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
38835 (fmtresult::operator+=): Outlined.
38836 (struct fmtresult): Add ctors.
38837 (struct conversion_spec): Rename...
38838 (struct directive): ...to this. Add and remove data members.
38839 (directive::set_width, directive::set_precision): New functions.
38840 (format_percent): Use fmtresult ctor.
38841 (get_width_and_precision): Remove.
38842 (format_integer): Make naming changes. Avoid computing width and
38844 (format_floating): Same. Adjust indentation.
38845 (format_character, format_none): New functions.
38846 (format_string): Moved character handling to format_character.
38847 (format_directive): Remove arguments, change return type.
38848 (parse_directive): New function.
38849 (pass_sprintf_length::compute_format_length): Move directive
38850 parsing to parse_directive.
38852 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38854 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
38855 (assign_assembler_name_if_needed): ... this.
38856 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
38857 (assign_assembler_name_if_needed): ... this.
38858 (free_lang_data_in_cgraph): Adjust callers.
38859 * cgraphunit.c (cgraph_node::analyze): Likewise.
38860 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
38863 2017-01-23 Richard Biener <rguenther@suse.de>
38865 PR tree-optimization/79088
38866 PR tree-optimization/79188
38867 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
38868 resetting loop bounds after last path deletion. Reset loop
38869 bounds of the target loop, make code match the comments.
38870 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
38871 Make sure loops need no fixups.
38873 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
38875 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
38876 exponent support with double type for first argument.
38877 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
38878 type returned by __builtin_vec_extract_sig,
38879 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
38880 functions from "vector int" to "vector unsigned int" or from
38881 "vector long long int" to "vector unsigned long long int".
38882 Changed type returned by __builtin_vec_extract_exp,
38883 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
38884 functions from "vector int" to "vector unsigned int" or from
38885 "vector long long int" to "vector unsigned long long int".
38886 Changed return type of __builtin_vec_test_data_class,
38887 __builtin_vec_test_data_class_sp, and
38888 __builtin_vec_test_data_class_dp from "vector int" to
38889 "vector bool int" or from "vector long long int" to "vector bool
38890 long long int" and changed second argument type from "unsigned
38891 int" to "int". Added new overloaded function forms "vector float
38892 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
38893 "vector float __builtin_vec_insert_exp_sp (vector float, vector
38894 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
38895 double, vector unsigned long long int)" and "vector double
38896 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
38897 long int)". Changed return type of
38898 __builtin_scalar_test_data_class and
38899 __builtin_scalar_test_data_class_sp and
38900 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
38901 int" and changed second argument from "unsigned int" to "int".
38902 Changed type returned by __builtin_scalar_test_neg,
38903 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
38904 from "int" to "bool int". Added new overloaded function form
38905 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
38906 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
38907 exponent double-precision with floating point first argument.
38908 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
38909 documentation of scalar_test_data_class, scalar_test_neg,
38910 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
38911 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
38912 vec_test_data_class built-in functions to reflect refinements in
38913 their type signatures.
38915 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
38917 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
38919 (aarch64_elf_asm_destructor): Likewise.
38921 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
38923 PR rtl-optimization/78634
38924 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
38925 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
38926 * ifcvt.c (noce_try_cmove): Add missing cost check.
38928 PR rtl-optimization/71724
38929 * combine.c (if_then_else_cond): Look for situations where it is
38930 beneficial to undo the work of one of the recursive calls.
38932 2017-01-23 Bin Cheng <bin.cheng@arm.com>
38934 PR tree-optimization/70754
38935 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
38936 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
38937 combined stmt before it if not NULL.
38938 (combine_chains): Process refs reversely and compute dominance point
38941 2017-01-23 Martin Liska <mliska@suse.cz>
38943 PR tree-optimization/79196
38944 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
38945 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
38947 (strlen_optimize_stmt): Call the renamed function.
38949 2017-01-23 Michael Matz <matz@suse.de>
38951 PR tree-optimization/78384
38952 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
38954 2017-01-23 Richard Biener <rguenther@suse.de>
38956 PR tree-optimization/79186
38957 * tree-vrp.c (register_new_assert_for): Make sure we've seen
38958 both incoming edges before moving an assert.
38960 2017-01-23 Martin Jambor <mjambor@suse.cz>
38962 * ipa-prop.c (load_from_param_1): Removed.
38963 (load_from_unmodified_param): Bits from load_from_param_1 put back
38965 (load_from_param): Removed.
38966 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
38967 with stmt. Reverted back to use of load_from_unmodified_param.
38969 2017-01-23 Martin Jambor <mjambor@suse.cz>
38972 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
38973 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
38974 field a pointer to garbage collected vector, mark lattices and
38975 ipcp_orig_node with GTY((skip)).
38976 (ipa_get_param_count): Adjust to descriptors being a pointer.
38977 (ipa_get_param): Likewise.
38978 (ipa_get_type): Likewise.
38979 (ipa_get_param_move_cost): Likewise.
38980 (ipa_set_param_used): Likewise.
38981 (ipa_get_controlled_uses): Likewise.
38982 (ipa_set_controlled_uses): Likewise.
38983 (ipa_is_param_used): Likewise.
38984 (ipa_node_params_t): Move into garbage collector. New methods insert
38986 (ipa_node_params_sum): Annotate wth GTY(()).
38987 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
38989 (ipa_load_from_parm_agg): Adjust declaration.
38990 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
38991 * ipa-profile.c (ipa_profile): Likewise.
38992 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
38993 (ipa_populate_param_decls): Make descriptors parameter garbage
38995 (ipa_dump_param): Adjust to descriptors being a pointer.
38996 (ipa_alloc_node_params): Likewise.
38997 (ipa_initialize_node_params): Likewise.
38998 (load_from_param_1): Make descriptors parameter garbage collected.
38999 (load_from_unmodified_param): Likewise.
39000 (load_from_param): Likewise.
39001 (ipa_load_from_parm_agg): Likewise.
39002 (ipa_node_params::~ipa_node_params): Removed.
39003 (ipa_free_all_node_params): Remove call to delete operator.
39004 (ipa_node_params_t::insert): New.
39005 (ipa_node_params_t::remove): Likewise.
39006 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
39007 copy known_csts and known_contexts vectors.
39008 (ipa_read_node_info): Adjust to descriptors being a pointer.
39009 (ipcp_modif_dom_walker): Make m_descriptors field garbage
39011 (ipcp_transform_function): Make descriptors variable garbage
39014 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
39016 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
39017 * config/i386/avx512dqintrin.h: Ditto.
39018 * config/i386/avx512fintrin.h: Ditto.
39019 * config/i386/i386.c: Handle new builtins.
39020 * config/i386/i386-builtin.def: Add new builtins.
39021 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
39022 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
39024 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39025 Martin Liska <mliska@suse.cz>
39027 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
39028 * asan.c (asan_expand_poison_ifn): Support stores and use
39029 appropriate ASAN report function.
39030 * internal-fn.c (expand_ASAN_POISON_USE): New function.
39031 * internal-fn.def (ASAN_POISON_USE): Declare.
39032 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
39033 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
39034 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
39035 ASAN_POISON calls w/o LHS.
39036 * tree-ssa.c (execute_update_addresses_taken): Create clobber
39037 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
39038 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
39039 * gimplify.c (asan_poison_variables): Add attribute
39040 use_after_scope_memory to variables that really needs to live
39042 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
39043 having the attribute.
39045 2017-01-23 Martin Liska <mliska@suse.cz>
39047 * asan.c (create_asan_shadow_var): New function.
39048 (asan_expand_poison_ifn): Likewise.
39049 * asan.h (asan_expand_poison_ifn): New declaration.
39050 * internal-fn.c (expand_ASAN_POISON): Likewise.
39051 * internal-fn.def (ASAN_POISON): New builtin.
39052 * sanopt.c (pass_sanopt::execute): Expand
39053 asan_expand_poison_ifn.
39054 * tree-inline.c (copy_decl_for_dup_finish): Make function
39056 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
39057 * tree-ssa.c (is_asan_mark_p): New function.
39058 (execute_update_addresses_taken): Rewrite local variables
39059 (identified just by use-after-scope as addressable) into SSA.
39061 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39063 * doc/install.texi (Specific): opensource.apple.com uses https
39064 now. Remove trailing slash.
39066 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39068 * README.Portability: Remove note on an Irix compatibility issue.
39070 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
39072 * gcov.c (INCLUDE_ALGORITHM): Define.
39073 (INCLUDE_VECTOR): Define.
39074 No longer include <vector> and <algorithm> directly.
39076 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
39078 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
39080 * doc/invoke.texi (Code Gen Options): Ditto.
39082 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
39085 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
39087 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
39089 rtl-optimization/79125
39090 * cprop.c (local_cprop_pass): Handle cases where we make an
39091 unconditional trap.
39093 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
39097 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
39098 read from, for big endian.
39100 2017-01-20 Jiong Wang <jiong.wang@arm.com>
39102 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
39103 register pauth builtins for LP64 only.
39105 2017-01-20 Marek Polacek <polacek@redhat.com>
39108 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
39111 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39113 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
39115 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
39116 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
39117 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
39119 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39122 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
39123 in big-endian mode when they are not a single duplicated value.
39125 2017-01-20 Richard Biener <rguenther@suse.de>
39127 * BASE-VER: Bump to 7.0.1.
39129 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39131 * omp-low.c (omplow_simd_context): New struct. Use it...
39132 (lower_rec_simd_input_clauses): ...here and...
39133 (lower_rec_input_clauses): ...here to hold common data. Adjust all
39134 references to idx, lane, max_vf, is_simt.
39136 2017-01-20 Graham Markall <graham.markall@embecosm.com>
39138 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
39141 2017-01-20 Martin Jambor <mjambor@suse.cz>
39143 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
39144 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
39146 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
39147 (GTFILES): Rename hsa.c to hsa-common.c.
39148 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
39149 * hsa-dump.c: Likewise.
39150 * hsa-gen.c: Likewise.
39151 * hsa-regalloc.c: Likewise.
39152 * ipa-hsa.c: Likewise.
39153 * omp-expand.c: Likewise.
39154 * omp-low.c: Likewise.
39155 * toplev.c: Likewise.
39157 2017-01-20 Marek Polacek <polacek@redhat.com>
39160 * doc/invoke.texi: Document -Wduplicated-branches.
39161 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
39162 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
39163 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
39164 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
39165 return 0 only when not OEP_LEXICOGRAPHIC.
39166 (fold_build_cleanup_point_expr): Use the expression
39167 location when building CLEANUP_POINT_EXPR.
39168 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
39169 * tree.c (add_expr): Handle error_mark_node.
39171 2017-01-20 Martin Liska <mliska@suse.cz>
39174 * tree-profile.c (init_ic_make_global_vars): Do not call
39176 (gimple_init_gcov_profiler): Likewise.
39178 2017-01-20 Martin Liska <mliska@suse.cz>
39181 * cgraph.h (maybe_create_reference): Remove argument and
39183 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
39185 * ipa-cp.c (create_specialized_node): Likewise.
39186 * symtab.c (symtab_node::maybe_create_reference): Handle
39187 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
39189 2017-01-20 Martin Liska <mliska@suse.cz>
39191 * read-rtl-function.c (function_reader::create_function): Use
39192 build_decl instread of build_decl_stat.
39194 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
39196 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
39197 * config/i386/avx512dqintrin.h: Ditto.
39198 * config/i386/avx512fintrin.h: Ditto.
39199 * config/i386/i386-builtin-types.def: Add new types.
39200 * config/i386/i386.c: Handle new types.
39201 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
39202 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
39203 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
39204 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
39205 (__builtin_ia32_kshiftridi): New.
39206 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
39208 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
39212 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
39213 define to rs6000_init_stack_protect_guard.
39214 (rs6000_init_stack_protect_guard): New function.
39216 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39217 Yunqiang Su <yunqiang.su@imgtec.com>
39219 * config.gcc (supported_defaults): Add madd4.
39220 (with_madd4): Add validation.
39221 (all_defaults): Add madd4.
39222 * config/mips/mips.opt (mmadd4): New option.
39223 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39225 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39227 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
39228 (ISA_HAS_FUSED_MADD4): Likewise.
39229 * doc/invoke.texi (-mmadd4): Document the new option.
39230 * doc/install.texi (--with-madd4): Document the new option.
39232 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39234 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
39235 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
39236 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
39237 (aarch64_init_pauth_hint_builtins): New.
39238 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
39239 (aarch64_expand_builtin): Expand new builtins.
39241 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39243 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
39244 * combine-stack-adj.c (no_unhandled_cfa): Handle
39245 REG_CFA_TOGGLE_RA_MANGLE.
39246 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
39247 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
39248 info for return address signing.
39249 (aarch64_expand_epilogue): Likewise.
39251 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39253 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
39254 * config/aarch64/aarch64-protos.h
39255 (aarch64_return_address_signing_enabled): New declaration.
39256 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
39258 (aarch64_expand_prologue): Sign return address before it's pushed onto
39260 (aarch64_expand_epilogue): Authenticate return address fetched from
39262 (aarch64_override_options): Sanity check for ILP32 and ISA level.
39263 (aarch64_attributes): New function attributes for "sign-return-address".
39264 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
39265 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
39266 ("*do_return"): Generate combined instructions according to key index.
39267 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
39268 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
39270 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
39271 * config/aarch64/aarch64.opt (msign-return-address=): New.
39272 * doc/extend.texi (AArch64 Function Attributes): Documents
39273 "sign-return-address=".
39274 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
39276 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39278 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
39279 overall option summary.
39281 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39283 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
39284 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
39285 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
39286 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
39288 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
39290 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
39291 -mpower9-minmax by default for -mcpu=power9.
39292 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
39293 128-bit floating point.
39295 2017-01-20 Alan Modra <amodra@gmail.com>
39297 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
39298 optimizing for size.
39300 2017-01-20 Alan Modra <amodra@gmail.com>
39303 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
39304 for strcmp and strncmp from corresponding builtin decl.
39306 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39308 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
39309 instead of i386/rtems-64.h.
39310 * config/i386/rtems-64.h: Remove.
39312 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39316 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
39318 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
39320 2017-01-19 Tamar Christina <tamar.christina@arm.com>
39322 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
39323 Change int to HOST_WIDE_INT.
39324 * config/aarch64/aarch64-protos.h
39325 (aarch64_simd_gen_const_vector_dup): Likewise.
39326 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
39328 2017-01-19 David Malcolm <dmalcolm@redhat.com>
39330 * langhooks-def.h (lhd_type_for_size): New decl.
39331 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
39332 * langhooks.c (lhd_type_for_size): New function, taken from
39335 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
39337 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
39338 define_bypass for CR latency.
39339 (power9-cracked-alu): Update bypass latency and remove power9-branch.
39340 (power9-alu2): Add define_bypass for CR latency.
39342 (power9-mul): Update insn latency.
39343 (power9-mul-compare): Update insn latency, bypass latency and remove
39346 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39348 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
39350 * config/aarch64/aarch64.md
39351 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
39352 aarch64_nopcrelative_literal_loads.
39353 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
39355 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
39357 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
39358 TARGET_LOONGSON_3A.
39359 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
39361 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
39364 * config.gcc (supported_defaults): Add lxc1-sxc1.
39365 (with_lxc1_sxc1): Add validation.
39366 (all_defaults): Add lxc1-sxc1.
39367 * config/mips/mips.opt (mlxc1-sxc1): New option.
39368 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39370 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39371 __mips_no_lxc1_sxc1.
39372 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
39373 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
39374 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
39376 2017-01-19 Richard Biener <rguenther@suse.de>
39378 PR tree-optimization/72488
39379 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
39380 sure to restore SSA info.
39381 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
39383 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
39385 PR rtl-optimization/79121
39386 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
39387 of the inner type when shifting an extended value.
39389 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39392 * symtab.c (symtab_node::equal_address_to): Fix comparing of
39393 interposable aliases.
39395 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
39398 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
39399 Use the evmergelohi instruction.
39400 (mov_si<mode>_e500_subreg4_2_le): Likewise.
39401 (mov_sitf_e500_subreg8_2_be): Likewise.
39402 (mov_sitf_e500_subreg12_2_le): Likewise.
39403 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
39404 (mov_si<mode>_e500_subreg4_2_be): Likewise.
39405 (mov_sitf_e500_subreg8_2_le): Likewise.
39406 (mov_sitf_e500_subreg12_2_be): Likewise.
39408 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39410 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
39411 attribute from vecsimple to vecperm.
39412 (altivec_vbpermq2): Likewise.
39414 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39417 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
39419 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39420 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
39421 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
39422 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
39423 case where N arg is SIZE_MAX.
39424 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
39425 (cmpstrsi): Add pattern.
39427 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
39429 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
39430 __builtin_vec_revb builtins.
39431 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
39432 built-in functions to support generation of the ISA 3.0 XXBR<x>
39433 vector byte reverse instructions.
39434 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
39435 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
39436 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
39437 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
39438 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
39439 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
39440 (P9V_BUILTIN_VEC_REVB): Likewise.
39441 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
39442 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
39443 (p9_xxbrq_v16qi): Likewise.
39444 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
39445 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
39446 (p9_xxbrh_v8hi): Likewise.
39447 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
39448 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
39449 vec_revb built-in functions.
39451 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
39453 PR rtl-optimization/78952
39454 * config/i386/i386.md (any_extract): New code iterator.
39455 (*insvqi_2): Use any_extract for source operand.
39456 (*insvqi_3): Use any_shiftrt for source operand.
39458 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
39460 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
39462 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
39464 2017-01-18 Matthias Klose <doko@ubuntu.com>
39466 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
39468 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39470 * config/rs6000/altivec.h (vec_bperm): Change #define.
39471 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
39472 (altivec_vbpermq2): New define_insn.
39473 (altivec_vbpermd): Likewise.
39474 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
39475 function interface.
39476 (VBPERMD): Likewise.
39477 (VBPERM): New polymorphic function interface.
39478 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
39479 Add entries for P9V_BUILTIN_VEC_VBPERM.
39480 * doc/extend.texi: Add interfaces for vec_bperm.
39482 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39484 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
39485 first letter of error messages.
39486 (s390_resolve_overloaded_builtin): Likewise.
39487 * config/s390/s390.c (s390_expand_builtin): Likewise.
39488 (s390_invalid_arg_for_unprototyped_fn): Likewise.
39489 (s390_valid_target_attribute_inner_p): Likewise.
39490 * config/s390/s390.md ("tabort"): Likewise.
39492 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
39494 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
39495 (ISA_AVOID_DIV_HILO): New macro.
39496 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
39497 (ISA_HAS_DDIV): Likewise.
39499 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39501 * doc/invoke.texi (fabi-version): Correct number of occurrences.
39503 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39505 * doc/invoke.texi (fabi-version): Spelling fix.
39507 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39510 * doc/invoke.texi (fabi-version): Mention mangling fix for
39513 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39516 * doc/invoke.texi (fabi-version): Document discriminator mangling.
39518 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
39521 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
39522 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
39524 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
39526 (stack_protect_test): Ditto.
39527 * config/rs6000/rs6000.opt (mstack-protector-guard=,
39528 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
39530 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
39531 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
39532 -mstack-protector-guard-offset=.
39533 (RS/6000 and PowerPC Options): Ditto.
39535 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39537 * config/i386/i386.h (MASK_CLASS_P): New define.
39538 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
39539 there are no registers from different register sets also when
39540 mask registers are used. Update function comment.
39541 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
39542 to (*k/*r) and (*k/*km) alternatives.
39544 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
39546 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
39547 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
39548 (EH_RETURN_HANDLER_RTX): New define.
39549 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
39550 Force frame pointer in EH return functions.
39551 (aarch64_expand_epilogue): Add barrier for eh_return.
39552 (aarch64_final_eh_return_addr): Remove.
39553 (aarch64_eh_return_handler_rtx): New function.
39554 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
39556 (aarch64_eh_return_handler_rtx): New prototype.
39558 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39560 * config/rs6000/altivec.h (vec_rlmi): New #define.
39561 (vec_vrlnm): Likewise.
39562 (vec_rlnm): Likewise.
39563 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
39564 (UNSPEC_VRLNM): Likewise.
39565 (VIlong): New mode iterator.
39566 (altivec_vrl<VI_char>mi): New define_insn.
39567 (altivec_vrl<VI_char>nm): Likewise.
39568 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
39570 (VRLDNM): Likewise.
39571 (RLNM): New polymorphic function entry.
39572 (VRLWMI): New monomorphic function entry.
39573 (VRLDMI): Likewise.
39574 (RLMI): New polymorphic function entry.
39575 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
39576 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
39577 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
39580 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39583 * dwarf2out.c (field_byte_offset): Restore the
39584 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
39585 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
39586 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
39589 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39592 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
39594 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39597 * dwarf2out.c (add_data_member_location_attribute): For constant
39598 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
39599 instead of DW_AT_data_member_location, DW_AT_bit_offset and
39600 DW_AT_byte_size attributes.
39602 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39604 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
39605 after forcing to constant memory when the code model is medium.
39607 2017-01-17 Julia Koval <julia.koval@intel.com>
39610 * config/i386/avx512fintrin.h
39611 (_mm512_i32gather_ps): Change __addr type to void const*.
39612 (_mm512_mask_i32gather_ps): Ditto.
39613 (_mm512_i32gather_pd): Ditto.
39614 (_mm512_mask_i32gather_pd): Ditto.
39615 (_mm512_i64gather_ps): Ditto.
39616 (_mm512_mask_i64gather_ps): Ditto.
39617 (_mm512_i64gather_pd): Ditto.
39618 (_mm512_mask_i64gather_pd): Ditto.
39619 (_mm512_i32gather_epi32): Ditto.
39620 (_mm512_mask_i32gather_epi32): Ditto.
39621 (_mm512_i32gather_epi64): Ditto.
39622 (_mm512_mask_i32gather_epi64): Ditto.
39623 (_mm512_i64gather_epi32): Ditto.
39624 (_mm512_mask_i64gather_epi32): Ditto.
39625 (_mm512_i64gather_epi64): Ditto.
39626 (_mm512_mask_i64gather_epi64): Ditto.
39627 (_mm512_i32scatter_ps): Change __addr type to void*.
39628 (_mm512_mask_i32scatter_ps): Ditto.
39629 (_mm512_i32scatter_pd): Ditto.
39630 (_mm512_mask_i32scatter_pd): Ditto.
39631 (_mm512_i64scatter_ps): Ditto.
39632 (_mm512_mask_i64scatter_ps): Ditto.
39633 (_mm512_i64scatter_pd): Ditto.
39634 (_mm512_mask_i64scatter_pd): Ditto.
39635 (_mm512_i32scatter_epi32): Ditto.
39636 (_mm512_mask_i32scatter_epi32): Ditto.
39637 (_mm512_i32scatter_epi64): Ditto.
39638 (_mm512_mask_i32scatter_epi64): Ditto.
39639 (_mm512_i64scatter_epi32): Ditto.
39640 (_mm512_mask_i64scatter_epi32): Ditto.
39641 (_mm512_i64scatter_epi64): Ditto.
39642 (_mm512_mask_i64scatter_epi64): Ditto.
39643 * config/i386/avx512pfintrin.h
39644 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
39645 (_mm512_mask_prefetch_i32gather_ps): Ditto.
39646 (_mm512_mask_prefetch_i64gather_pd): Ditto.
39647 (_mm512_mask_prefetch_i64gather_ps): Ditto.
39648 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
39649 (_mm512_prefetch_i32scatter_ps): Ditto.
39650 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
39651 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
39652 (_mm512_prefetch_i64scatter_pd): Ditto.
39653 (_mm512_prefetch_i64scatter_ps): Ditto.
39654 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
39655 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
39656 * config/i386/avx512vlintrin.h
39657 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
39658 (_mm_mmask_i32gather_ps): Ditto.
39659 (_mm256_mmask_i32gather_pd): Ditto.
39660 (_mm_mmask_i32gather_pd): Ditto.
39661 (_mm256_mmask_i64gather_ps): Ditto.
39662 (_mm_mmask_i64gather_ps): Ditto.
39663 (_mm256_mmask_i64gather_pd): Ditto.
39664 (_mm_mmask_i64gather_pd): Ditto.
39665 (_mm256_mmask_i32gather_epi32): Ditto.
39666 (_mm_mmask_i32gather_epi32): Ditto.
39667 (_mm256_mmask_i32gather_epi64): Ditto.
39668 (_mm_mmask_i32gather_epi64): Ditto.
39669 (_mm256_mmask_i64gather_epi32): Ditto.
39670 (_mm_mmask_i64gather_epi32): Ditto.
39671 (_mm256_mmask_i64gather_epi64): Ditto.
39672 (_mm_mmask_i64gather_epi64): Ditto.
39673 (_mm256_i32scatter_ps): Change __addr type to void*.
39674 (_mm256_mask_i32scatter_ps): Ditto.
39675 (_mm_i32scatter_ps): Ditto.
39676 (_mm_mask_i32scatter_ps): Ditto.
39677 (_mm256_i32scatter_pd): Ditto.
39678 (_mm256_mask_i32scatter_pd): Ditto.
39679 (_mm_i32scatter_pd): Ditto.
39680 (_mm_mask_i32scatter_pd): Ditto.
39681 (_mm256_i64scatter_ps): Ditto.
39682 (_mm256_mask_i64scatter_ps): Ditto.
39683 (_mm_i64scatter_ps): Ditto.
39684 (_mm_mask_i64scatter_ps): Ditto.
39685 (_mm256_i64scatter_pd): Ditto.
39686 (_mm256_mask_i64scatter_pd): Ditto.
39687 (_mm_i64scatter_pd): Ditto.
39688 (_mm_mask_i64scatter_pd): Ditto.
39689 (_mm256_i32scatter_epi32): Ditto.
39690 (_mm256_mask_i32scatter_epi32): Ditto.
39691 (_mm_i32scatter_epi32): Ditto.
39692 (_mm_mask_i32scatter_epi32): Ditto.
39693 (_mm256_i32scatter_epi64): Ditto.
39694 (_mm256_mask_i32scatter_epi64): Ditto.
39695 (_mm_i32scatter_epi64): Ditto.
39696 (_mm_mask_i32scatter_epi64): Ditto.
39697 (_mm256_i64scatter_epi32): Ditto.
39698 (_mm256_mask_i64scatter_epi32): Ditto.
39699 (_mm_i64scatter_epi32): Ditto.
39700 (_mm_mask_i64scatter_epi32): Ditto.
39701 (_mm256_i64scatter_epi64): Ditto.
39702 (_mm256_mask_i64scatter_epi64): Ditto.
39703 (_mm_i64scatter_epi64): Ditto.
39704 (_mm_mask_i64scatter_epi64): Ditto.
39705 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
39706 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
39707 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
39708 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
39709 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
39710 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
39711 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
39712 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
39713 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
39714 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
39715 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
39716 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
39717 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
39718 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
39719 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
39720 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
39721 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
39722 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
39723 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
39724 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
39725 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
39726 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
39727 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
39728 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
39729 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
39730 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
39731 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
39732 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
39733 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
39734 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
39735 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
39736 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
39737 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
39738 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
39739 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
39740 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
39741 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
39742 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
39743 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
39744 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
39745 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
39746 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
39747 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
39748 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
39749 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
39750 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
39751 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
39752 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
39753 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
39754 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
39755 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
39756 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
39757 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
39758 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
39759 definitions accordingly.
39761 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
39762 Kuan-Lin Chen <kuanlinchentw@gmail.com>
39765 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
39768 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
39771 * ira-conflicts.c (ira_build_conflicts): Update total conflict
39772 hard regs for inner regno.
39774 2017-01-17 Martin Liska <mliska@suse.cz>
39777 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
39778 assumption and add comment.
39780 2017-01-17 Nathan Sidwell <nathan@acm.org>
39782 * ipa-visibility.c (localize_node): New function, broken out of ...
39783 (function_and_variable_visibility): ... here. Call it.
39785 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39787 PR middle-end/77445
39788 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
39789 correctly set frequency of oudgoing edge.
39790 (duplicate_thread_path): Fix profile updating.
39792 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39795 * configure.ac: Add GCC_BASE_VER.
39796 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
39797 version from BASE-VER file.
39798 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
39799 (gcc.o): Depend on $(BASEVER).
39800 * common.opt (dumpfullversion): New option.
39801 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
39802 * doc/invoke.texi: Document -dumpfullversion.
39803 * doc/install.texi: Document --with-gcc-major-version-only.
39804 * configure: Regenerated.
39806 2017-01-17 Richard Biener <rguenther@suse.de>
39808 PR tree-optimization/71433
39809 * tree-vrp.c (register_new_assert_for): Merge same asserts
39810 on all incoming edges.
39811 (process_assert_insertions_for): Handle insertions at the
39814 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
39816 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
39817 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
39819 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
39822 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
39825 2017-01-17 Alan Modra <amodra@gmail.com>
39828 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
39829 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
39830 symbolic stack limit when pic.
39832 2017-01-16 Martin Sebor <msebor@redhat.com>
39834 PR tree-optimization/78608
39835 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
39837 2017-01-16 Jeff Law <law@redhat.com>
39840 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
39841 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
39842 for several include directories that may be relative to sysroot.
39843 * config/i386/x-mingw32 (gplus_includedir): Define.
39844 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
39845 (native_system_includedir): Likewise.
39846 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
39847 override if TARGET_SYSTEM_ROOT is defined.
39848 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
39850 PR tree-optimization/79090
39851 PR tree-optimization/33562
39852 PR tree-optimization/61912
39853 PR tree-optimization/77485
39854 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
39855 and computed trims into the dump file.
39857 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39859 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
39861 2017-01-16 Jakub Jelinek <jakub@redhat.com>
39864 * gimplify.c (gimplify_init_constructor): If want_value and
39865 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
39869 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
39870 sequence. Formatting fixes.
39871 (doloop_optimize): Formatting fixes.
39874 * gcc.c (debug_level_greater_than_spec_func): New function.
39875 (static_spec_functions): Add debug-level-gt spec function.
39876 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
39878 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39879 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39880 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
39881 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
39882 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
39883 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
39885 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
39887 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
39888 QImode fixups to general and mask registers only.
39890 2017-01-16 Carl Love <cel@us.ibm.com>
39892 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
39893 for built-in functions
39894 vector signed char vec_nabs (vector signed char)
39895 vector signed short vec_nabs (vector signed short)
39896 vector signed int vec_nabs (vector signed int)
39897 vector signed long long vec_nabs (vector signed long long)
39898 vector float vec_nabs (vector float)
39899 vector double vec_nabs (vector double)
39900 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
39902 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
39903 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
39904 * doc/extend.texi: Update the documentation file for the new built-in
39907 2017-01-16 Martin Sebor <msebor@redhat.com>
39909 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
39912 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39914 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
39915 UNSPEC_VSX__XXSPLTD to require special splat handling.
39917 2017-01-16 David Malcolm <dmalcolm@redhat.com>
39920 * system.h: Poison strndup.
39922 2017-01-16 Alan Modra <amodra@gmail.com>
39925 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
39928 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
39930 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
39932 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
39934 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
39935 call recog here. Assert that INSN_CODE (insn) is non-negative.
39937 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39940 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
39942 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
39943 in the currently scheduled RTL fragment.
39945 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39947 PR rtl-optimization/78751
39948 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
39951 2017-01-14 Jeff Law <law@redhat.com>
39953 PR tree-optimization/79090
39954 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
39955 variable length stores.
39956 (compute_trims): Delete dead assignment to *trim_tail.
39957 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
39960 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
39962 PR rtl-optimization/78626
39963 PR rtl-optimization/78727
39964 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
39965 of a block, and split such blocks after everything else is finished.
39967 2017-01-14 Alan Modra <amodra@gmail.com>
39970 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
39971 target legitimate_combined_insn.
39972 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
39973 (rs6000_legitimate_combined_insn): New function.
39974 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
39976 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
39977 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
39978 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
39980 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
39982 * doc/frontends.texi (G++ and GCC): Remove references to Java.
39984 2017-01-13 Jeff Law <law@redhat.com>
39986 PR tree-optimization/33562
39987 PR tree-optimization/61912
39988 PR tree-optimization/77485
39989 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
39991 (delete_dead_assignment): Likewise.
39992 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
39993 statement to delete_dead_call and delete_dead_assignment.
39995 2017-01-13 David Malcolm <dmalcolm@redhat.com>
39998 * substring-locations.c (format_warning_va): Strengthen case 1 so
39999 that both endpoints of the substring must be within the format
40000 range for just the substring to be printed.
40002 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
40004 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
40005 * config/i386/i386.c (ix86_target_string): Add missing options
40006 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
40007 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
40008 flags_other and ix86_target_other to flags2_other. Display unknown
40010 (ix86_valid_target_attribute_inner_p): Add missing options and
40011 reorder options by implied ISAs, as in ix86_target_string.
40013 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40015 * hash-table.h (hash_table::too_empty_p): New function.
40016 (hash_table::expand): Use it.
40017 (hash_table::traverse): Likewise.
40018 (hash_table::empty_slot): Use sizeof (value_type) instead of
40019 sizeof (PTR) to convert bytes to elements. Shrink the table
40020 if the current size is excessive for the current number of
40023 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40025 * ira-costs.c (record_reg_classes): Break from the inner loop
40026 early once alt_fail is known to be true. Update outer loop
40027 handling accordingly.
40029 2017-01-13 Jeff Law <law@redhat.com>
40031 * tree-ssa-dse.c (decrement_count): New function.
40032 (increment_start_addr, maybe_trim_memstar_call): Likewise.
40033 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
40034 when we know the partially dead statement is a mem* function.
40036 PR tree-optimization/61912
40037 PR tree-optimization/77485
40038 * tree-ssa-dse.c: Include expr.h.
40039 (maybe_trim_constructor_store): New function.
40040 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
40042 PR tree-optimization/33562
40043 PR tree-optimization/61912
40044 PR tree-optimization/77485
40045 * doc/invoke.texi: Document new dse-max-object-size param.
40046 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
40047 * tree-ssa-dse.c: Include params.h.
40048 (dse_store_status): New enum.
40049 (initialize_ao_ref_for_dse): New, partially extracted from
40051 (valid_ao_ref_for_dse, normalize_ref): New.
40052 (setup_live_bytes_from_ref, compute_trims): Likewise.
40053 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
40054 (maybe_trim_partially_dead_store): Likewise.
40055 (maybe_trim_complex_store): Likewise.
40056 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
40057 Track what bytes live from the original store. Return tri-state
40058 for dead, partially dead or live.
40059 (dse_dom_walker): Add constructor, destructor and new private members.
40060 (delete_dead_call, delete_dead_assignment): New extracted from
40062 (dse_optimize_stmt): Make a member of dse_dom_walker.
40063 Use initialize_ao_ref_for_dse.
40065 PR tree-optimization/33562
40066 PR tree-optimization/61912
40067 PR tree-optimization/77485
40068 * sbitmap.h (bitmap_count_bits): Prototype.
40069 (bitmap_clear_range, bitmap_set_range): Likewise.
40070 * sbitmap.c (bitmap_clear_range): New function.
40071 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
40073 2017-01-13 Martin Liska <mliska@suse.cz>
40076 * function.c (set_cfun): Add new argument force.
40077 * function.h (set_cfun): Likewise.
40078 * ipa-inline-transform.c (inline_call): Use the function when
40079 strict alising from is dropped for function we inline to.
40081 2017-01-13 Richard Biener <rguenther@suse.de>
40083 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
40084 for dumping GIMPLE INTEGER_CSTs.
40086 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
40088 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
40089 to 201112L since C++17.
40091 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
40094 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
40095 if -fsanitize=kernel-address is present.
40097 2017-01-13 Richard Biener <rguenther@suse.de>
40099 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
40100 as _Literal ( type ) number in case usual suffixes do not
40101 preserve all information.
40103 2017-01-13 Richard Biener <rguenther@suse.de>
40105 PR tree-optimization/77283
40106 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
40107 and ssa-iterators.h.
40108 (is_feasible_trace): Implement a cost model based on joiner
40111 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
40114 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
40115 char or short to __float128/_Float128 directly.
40117 2017-01-12 Martin Sebor <msebor@redhat.com>
40119 to -Wformat-overflow.
40120 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
40121 (min_bytes_remaining): Same.
40122 (get_string_length): Same.
40123 (format_string): Same.
40124 (format_directive): Same.
40126 (pass_sprintf_length::handle_gimple_call): Same.
40128 2017-01-12 Jakub Jelinek <jakub@redhat.com>
40130 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
40131 info.nowrite calls with no lhs that can't throw. Return bool
40132 whether gsi_remove has been called or not.
40133 (pass_sprintf_length::handle_gimple_call): Return bool whether
40134 try_substitute_return_value called gsi_remove. Formatting fix.
40135 (pass_sprintf_length::execute): Don't use gsi_remove if
40136 handle_gimple_call returned true.
40139 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
40140 be removed due to side-effects, don't remove following barrier nor
40141 turn the successor edge into fallthru edge.
40143 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40146 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
40147 element-reversing loads and stores as not swappable.
40149 2017-01-12 Nathan Sidwell <nathan@acm.org>
40150 Nicolai Stange <nicstange@gmail.com>
40152 * combine.c (try_combine): Don't ignore result of overlap checking
40153 loop. Combine overlap & asm check into single loop.
40155 2017-01-12 Richard Biener <rguenther@suse.de>
40157 * tree-pretty-print.c (dump_generic_node): Provide -gimple
40158 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
40160 2017-01-12 Richard Biener <rguenther@suse.de>
40162 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
40163 and TS_TARGET_OPTION directly derive from TS_BASE.
40164 * tree-core.h (tree_optimization_option): Derive from tree_base.
40165 (tree_target_option): Likewise.
40167 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
40169 * config/i386/i386.c (memory_address_length): Increase len
40170 only when rip_relative_addr_p returns false.
40172 2017-01-11 Julia Koval <julia.koval@intel.com>
40174 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
40175 (OPTION_MASK_ISA_SGX_SET): New.
40176 (ix86_handle_option): Handle OPT_msgx.
40177 * config.gcc: Added sgxintrin.h.
40178 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
40179 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
40180 * config/i386/i386.c (ix86_target_string): Add -msgx.
40182 (ix86_option_override_internal): Handle new options.
40183 (ix86_valid_target_attribute_inner_p): Add sgx.
40184 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
40185 * config/i386/i386.opt: Add msgx.
40186 * config/i386/sgxintrin.h: New file.
40187 * config/i386/x86intrin.h: Add sgxintrin.h.
40189 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40192 * fold-const.c (maybe_nonzero_address): Return 1 for function
40194 (tree_single_nonzero_warnv_p): Don't handle function local objects
40198 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
40201 2017-01-11 David Malcolm <dmalcolm@redhat.com>
40204 * opts.c: Include "spellcheck.h"
40205 (struct string_fragment): New struct.
40206 (struct edit_distance_traits<const string_fragment &>): New
40208 (get_closest_sanitizer_option): New function.
40209 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
40211 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40213 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
40215 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
40216 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
40217 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
40218 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
40219 for initial die_offset if dwarf_split_debug_info.
40220 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
40221 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
40223 (output_skeleton_debug_sections): Formatting fix. Use
40224 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
40225 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
40227 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
40229 * config/arm/cortex-a53.md: Add bypasses for
40230 cortex_a53_r2f_cvt.
40231 (cortex_a53_r2f): Only use for transfers.
40232 (cortex_a53_f2r): Likewise.
40233 (cortex_a53_r2f_cvt): Add reservation for conversions.
40234 (cortex_a53_f2r_cvt): Likewise.
40236 2017-01-11 Tamar Christina <tamar.christina@arm.com>
40238 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
40239 to all inlined functions, change static to extern.
40241 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
40244 * config/arm/arm.c (legitimize_pic_address): Handle reference to
40246 (arm_assemble_integer): Likewise.
40248 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
40250 * config.gcc: Use new awk script to check CPU, FPU and architecture
40251 parameters for --with-... options.
40252 * config/arm/parsecpu.awk: New file
40253 * config/arm/arm-cpus.in: New file.
40254 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
40256 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
40258 * config/arm/t-arm: Update dependency rules.
40259 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
40260 of processing .def files.
40261 * config/arm/genopt.sh: Deleted.
40262 * config/arm/gentune.sh: Deleted.
40263 * config/arm/arm-cores.def: Deleted.
40264 * config/arm/arm-arches.def: Deleted.
40265 * config/arm/arm-fpus.def: Deleted.
40266 * config/arm/arm-tune.md: Regenerated.
40267 * config/arm/arm-tables.opt: Regenerated.
40268 * config/arm/arm-cpu.h: New generated file.
40269 * config/arm/arm-cpu-data.h: New generated file.
40270 * config/arm/arm-cpu-cdata.h: New generated file.
40272 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
40275 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
40277 (input_varpool_node): Unpack dynamically_initialized bit.
40279 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
40281 PR rtl-optimization/79032
40282 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
40283 the alignment of the adjusted memory reference against that of MODE,
40284 instead of the alignment of the original memory reference.
40286 2017-01-11 Martin Jambor <mjambor@suse.cz>
40288 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
40290 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
40291 decorated functions.
40293 2017-01-11 Richard Biener <rguenther@suse.de>
40295 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
40296 set range/nonnull info for PHI results. Do not set it on
40297 stmts marked for removal.
40299 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
40301 * expr.c (store_field): In the bitfield case, fetch the return value
40302 from the registers before applying a single big-endian adjustment.
40303 Always do a final load for a BLKmode value not larger than a word.
40305 2017-01-10 David Malcolm <dmalcolm@redhat.com>
40308 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40309 that we correctly handle column numbers greater than
40310 LINE_MAP_MAX_COLUMN_NUMBER.
40312 2017-01-10 Martin Sebor <msebor@redhat.com>
40314 PR middle-end/78245
40315 * gimple-ssa-sprintf.c (get_destination_size): Call
40316 {init,fini}object_sizes.
40317 * tree-object-size.c (addr_object_size): Adjust.
40318 (pass_through_call): Adjust.
40319 (pass_object_sizes::execute): Adjust.
40320 * tree-object-size.h (fini_object_sizes): Declare.
40322 2017-01-10 Martin Sebor <msebor@redhat.com>
40324 PR tree-optimization/78775
40325 * builtins.c (get_size_range): Move...
40326 * calls.c: ...to here.
40327 (alloc_max_size): Accept zero argument.
40328 (operand_signed_p): Remove.
40329 (maybe_warn_alloc_args_overflow): Call get_size_range.
40330 * calls.h (get_size_range): Declare.
40332 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
40334 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
40335 from TI's devices.csv file as of September 2016.
40336 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
40338 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
40340 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
40341 * doc/invoke.texi: Likewise.
40342 * doc/md.texi: Likewise.
40343 * doc/objc.texi: Likewise.
40345 2017-01-10 Joshua Conner <joshconner@google.com>
40347 * config/arm/fuchsia-elf.h: New file.
40348 * config/fuchsia.h: New file.
40349 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
40350 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
40352 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
40354 2016-01-10 Richard Biener <rguenther@suse.de>
40356 PR tree-optimization/79034
40357 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
40358 Propagate out degenerate PHIs in the joiner.
40360 2017-01-10 Martin Liska <mliska@suse.cz>
40362 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
40363 (sort_congruence_classes_by_decl_uid): Likewise.
40364 (sort_congruence_class_groups_by_decl_uid): Likewise.
40365 (sem_item_optimizer::merge_classes): Sort class, groups in these
40366 classes and members in the groups by DECL_UID of declarations.
40367 This would make merge operations stable.
40369 2017-01-10 Martin Liska <mliska@suse.cz>
40371 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
40372 usage of m_classes_vec.
40373 (sem_item_optimizer::~sem_item_optimizer): Likewise.
40374 (sem_item_optimizer::get_group_by_hash): Likewise.
40375 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
40376 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
40377 (sem_item_optimizer::verify_classes): Likewise.
40378 (sem_item_optimizer::process_cong_reduction): Likewise.
40379 (sem_item_optimizer::dump_cong_classes): Likewise.
40380 (sem_item_optimizer::merge_classes): Likewise.
40381 * ipa-icf.h (congruence_class_hash): Rename from
40382 congruence_class_group_hash. Remove declaration of m_classes_vec.
40384 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
40386 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
40387 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
40388 * config.gcc: Add avx512vpopcntdqintrin.h.
40389 * config/i386/avx512vpopcntdqintrin.h: New.
40390 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
40391 * config/i386/i386-builtin-types.def: Add new types.
40392 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
40393 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
40394 __builtin_ia32_vpopcountq_v8di_mask): New.
40395 * config/i386/i386-c.c (ix86_target_macros_internal): Define
40396 __AVX512VPOPCNTDQ__.
40397 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
40398 (PTA_AVX512VPOPCNTDQ): Define.
40399 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
40400 TARGET_AVX512VPOPCNTDQ_P): Define.
40401 * config/i386/i386.opt: Add mavx512vpopcntdq.
40402 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
40403 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
40405 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40407 PR middle-end/77484
40408 * predict.def (PRED_CALL): Set to 67.
40410 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
40412 * expr.c (store_field): In the bitfield case, if the value comes from
40413 a function call and is of an aggregate type returned in registers, do
40414 not modify the field mode; extract the value in all cases if the mode
40415 is BLKmode and the size is not larger than a word.
40417 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
40420 * config/i386/cpuid.h: Fix undefined behavior.
40422 2017-01-04 Jeff Law <law@redhat.com>
40424 PR tree-optimization/79007
40425 PR tree-optimization/67955
40426 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
40427 conservative for pt.null when flag_non_call_exceptions is on.
40429 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40431 PR translation/79019
40432 PR translation/79020
40433 * params.def (PARAM_INLINE_MIN_SPEEDUP,
40434 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
40435 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
40437 * config/avr/avr.opt (maccumulate-args): Likewise.
40438 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
40439 * common.opt (freport-bug): Likewise.
40440 * cif-code.def (CIF_FINAL_ERROR): Likewise.
40441 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
40442 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
40443 translatable string.
40444 * config/i386/i386.c (function_value_32): Likewise.
40445 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
40446 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
40448 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
40449 * common/config/msp430/msp430-common.c (msp430_handle_option):
40451 * symtab.c (symtab_node::verify_base): Likewise.
40452 * opts.c (set_debug_level): Likewise.
40453 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
40454 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
40455 missing whitespace to translatable strings.
40456 * config/avr/avr.md (bswapsi2): Fix typo in comment.
40457 * config/sh/superh.h: Likewise.
40458 * config/i386/xopintrin.h: Likewise.
40459 * config/i386/znver1.md: Likewise.
40460 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
40461 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
40462 * double-int.h (struct double_int): Likewise.
40463 * double-int.c (div_and_round_double): Likewise.
40464 * wide-int.cc: Likewise.
40465 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
40466 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
40467 * cfgcleanup.c (crossjumps_occured): Renamed to ...
40468 (crossjumps_occurred): ... this.
40469 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
40472 PR tree-optimization/78899
40473 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
40474 returning bool return struct loop *, NULL for failure and the new
40476 (versionable_outer_loop_p): Don't version outer loop if it has
40477 dont_vectorized bit set.
40478 (tree_if_conversion): When versioning outer loop, ensure
40479 tree_if_conversion is performed also on the inner loop of the
40480 non-vectorizable outer loop copy.
40481 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
40482 LOOP_VECTORIZED in inner loop of the scalar outer loop and
40483 prevent vectorization of it.
40484 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
40485 the outer loop vectorization of the non-scalar version is attempted
40486 before vectorization of the inner loop in scalar version. If
40487 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
40488 vectorization of its inner loop.
40489 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
40490 has 2 inner loops, rename also on edges from bb whose single pred
40491 is outer_loop->header. Fix typo in function comment.
40493 2017-01-09 Martin Sebor <msebor@redhat.com>
40496 * asan.c (asan_emit_stack_protection): Increase local buffer size
40497 to avoid snprintf truncation warning.
40499 2017-01-09 Andrew Pinski <apinski@cavium.com>
40501 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
40502 to reference thunderx2t99 for the tuning structure
40503 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
40505 (thunderx2t99_extra_costs): This.
40506 * config/aarch64/aarch64-tune.md: Regenerate.
40507 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
40508 (vulcan_addrcost_table): This.
40509 (vulcan_regmove_cost): Rename to ...
40510 (thunderx2t99_regmove_cost): This.
40511 (vulcan_vector_cost): Rename to ...
40512 (thunderx2t99_vector_cost): this.
40513 (vulcan_branch_cost): Rename to ...
40514 (thunderx2t99_branch_cost): This.
40515 (vulcan_tunings): Rename to ...
40516 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
40517 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
40519 2017-01-09 Martin Jambor <mjambor@suse.cz>
40523 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
40524 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
40525 (propagate_vr_accross_jump_function): Use the above function for all
40526 value range computations for pass-through jump functions and type
40527 converasion from explicit value range values.
40528 (ipcp_propagate_stage): Do not attempt to deduce types of formal
40529 parameters from TYPE_ARG_TYPES.
40530 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
40531 (ipa_write_node_info): Stream type of the actual argument.
40532 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
40534 2017-01-09 Martin Liska <mliska@suse.cz>
40537 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
40538 (lookup_compiler): Do not show error message with have_E.
40540 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40542 PR tree-optimization/78938
40543 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
40544 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
40545 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
40546 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
40549 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40551 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
40554 2017-01-09 Richard Biener <rguenther@suse.de>
40556 PR tree-optimization/78997
40557 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
40558 name condition properly.
40560 2017-01-09 Richard Biener <rguenther@suse.de>
40563 * dwarf2out.c (is_cxx): New overload with context.
40564 (is_naming_typedef_decl): Use it.
40566 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40568 * invoke.texi (Option Summary): Correct spacing in option lists
40569 and add line breaks to fix over-long lines.
40571 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40573 PR middle-end/17660
40575 * extend.texi (Common Variable Attributes): Add xref to GCC
40576 Internals manual to explain mode attribute keywords.
40578 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40581 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
40582 and Preprocessor Options.
40583 (Options for Linking): Document -pthread here....
40584 (RS/6000 and PowerPC Options): ...not here.
40585 (Solaris 2 Options): ...or here.
40586 * doc/cppopts.texi: Document -pthread.
40588 2017-01-08 Martin Sebor <msebor@redhat.com>
40590 PR middle-end/77708
40591 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
40592 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
40593 New member functions.
40594 (format_directive): Used them.
40596 (pass_sprintf_length::handle_gimple_call): Same.
40597 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
40598 to avoid truncation for any argument.
40599 (extract_affine_mul): Same.
40600 * tree.c (get_file_function_name): Same.
40602 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40604 PR middle-end/77484
40605 * predict.def (PRED_INDIR_CALL): Set to 86.
40607 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40609 PR preprocessor/54124
40610 * doc/cppopts.texi: Reformat -d subtable to list the full name
40611 of the options. Add cross-reference to the docs for the general
40612 compiler -d options.
40613 * doc/invoke.texi (Developer Options): Add cross-reference to the
40614 preprocessor-specific -d option documentation.
40616 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40618 PR preprocessor/13498
40619 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
40620 redudant material, and reflect new command-line options.
40621 (System Headers): Likewise.
40623 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40625 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
40626 -isystem, and -idirafter. Copy-edit.
40627 * doc/cppopts.texi: Copy-edit. Remove contradiction about
40628 default for -ftrack-macro-expansion. Delete obsolete and
40629 badly-formatted implementation details about -fdebug-cpp output.
40630 * doc/cppwarnopts.texi: Copy-edit.
40632 2017-01-07 David Malcolm <dmalcolm@redhat.com>
40635 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40636 that the transition from a max line width >= 1<<10 to narrower
40637 lines works correctly.
40639 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
40641 * doc/options.texi (PerFunction): New.
40642 * opt-functions.awk (switch_flags): Map both Optimization and
40643 PerFunction to CL_OPTIMIZATION.
40644 * opth-gen.awk: Test for PerFunction flag along with
40646 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
40647 it only when the latter is present. Skip those that don't in
40648 the hash function generator.
40649 * common.opt (fvar-tracking): Mark as PerFunction instead of
40651 (fvar-tracking-assignments): Likewise.
40652 (fvar-tracking-assignments-toggle): Likewise.
40653 (fvar-tracking-uninit): Likewise.
40655 2017-01-07 Jakub Jelinek <jakub@redhat.com>
40657 PR translation/79018
40658 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
40661 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
40664 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
40665 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
40666 TARGET_LONG_JUMP_TABLE_OFFSETS.
40667 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
40668 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
40669 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
40670 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
40671 * config/m68k/m68k.md (tablejump expander): Likewise.
40672 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
40673 TARGET_LONG_JUMP_TABLE_OFFSETS.
40674 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
40675 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
40677 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
40678 David Holsgrove <david.holsgrove@xilinx.com>
40680 * common/config/microblaze/microblaze-common.c
40681 (TARGET_EXCEPT_UNWIND_INFO): Remove.
40682 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
40684 * config/microblaze/microblaze.c (microblaze_must_save_register)
40685 (microblaze_expand_epilogue, microblaze_return_addr): Handle
40687 (microblaze_eh_return): New function.
40688 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
40689 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
40690 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
40691 * config/microblaze/microblaze.md (eh_return): New pattern.
40693 2017-01-06 Jakub Jelinek <jakub@redhat.com>
40695 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
40696 GCC_DIAGNOSTIC_STRINGIFY): Define.
40698 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
40700 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40702 * config/arm/arm.md (<mcrr>): New.
40704 * config/arm/arm.c (arm_arch5te): New.
40705 (arm_option_override): Set arm_arch5te.
40706 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
40708 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
40709 (arm_mcrr_qualifiers): ... this. New.
40710 (MRRC_QUALIFIERS): Define to...
40711 (arm_mrrc_qualifiers): ... this. New.
40712 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
40714 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
40715 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
40716 (MRRCI, mrrc, MRRC): New.
40717 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
40718 VUNSPEC_MRRC2): New.
40720 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40722 * config/arm/arm.md (<mcr>): New.
40724 * config/arm/arm.c (arm_coproc_builtin_available): Add
40725 support for mcr, mrc, mcr2 and mrc2.
40726 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
40727 (arm_mcr_qualifiers): ... this. New.
40728 (MRC_QUALIFIERS): Define to ...
40729 (arm_mrc_qualifiers): ... this. New.
40730 (MCR_QUALIFIERS): Define to ...
40731 (arm_mcr_qualifiers): ... this. New.
40732 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
40734 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
40735 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
40736 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
40737 VUNSPEC_MRC2): New.
40739 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40741 * config/arm/arm.md (*ldc): New.
40745 * config/arm/arm.c (arm_coproc_builtin_available): Add
40746 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
40747 (arm_coproc_ldc_stc_legitimate_address): New.
40748 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
40749 'qualifier_const_pointer'.
40750 (LDC_QUALIFIERS): Define to...
40751 (arm_ldc_qualifiers): ... this. New.
40752 (STC_QUALIFIERS): Define to...
40753 (arm_stc_qualifiers): ... this. New.
40754 * config/arm/arm-protos.h
40755 (arm_coproc_ldc_stc_legitimate_address): New.
40756 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
40757 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
40758 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
40759 stc2, stcl, stc2l): New.
40760 * config/arm/constraints.md (Uz): New.
40761 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
40762 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
40763 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
40764 VUNSPEC_STC2L): New.
40766 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40768 * config/arm/arm.md (<cdp>): New.
40769 * config/arm/arm.c (neon_const_bounds): Rename this ...
40770 (arm_const_bounds): ... this.
40771 (arm_coproc_builtin_available): New.
40772 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
40773 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
40774 (CDP_QUALIFIERS): Define to...
40775 (arm_cdp_qualifiers): ... this. New.
40777 (arm_expand_builtin_args): Add case for 6 arguments.
40778 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
40779 (arm_const_bounds): ... this.
40780 (arm_coproc_builtin_available): New.
40781 * config/arm/arm_acle.h (__arm_cdp): New.
40783 * config/arm/arm_acle_builtins.def (cdp): New.
40785 * config/arm/iterators.md (CDPI,CDP,cdp): New.
40786 * config/arm/neon.md: Rename all 'neon_const_bounds' to
40787 'arm_const_bounds'.
40788 * config/arm/types.md (coproc): New.
40789 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
40790 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
40791 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
40792 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
40794 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40796 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
40797 (UBINOP_QUALIFIERS): New.
40799 (acle_builtin_data): New. Change comment.
40800 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
40801 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
40802 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
40803 arm_acle_builtins.def.
40804 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
40805 (arm_init_acle_builtins): New.
40806 (CRC32_BUILTIN): Remove.
40807 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
40808 crc32cb, crc32ch and crc32cw.
40809 (arm_init_crc32_builtins): Remove.
40810 (arm_init_builtins): Use arm_init_acle_builtins rather
40811 than arm_init_crc32_builtins.
40812 (arm_expand_acle_builtin): New.
40813 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
40814 * config/arm/arm_acle_builtins.def: New.
40816 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40818 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
40819 (arm_builtin_datum): ... this.
40820 (arm_init_neon_builtin): Rename to ...
40821 (arm_init_builtin): ... this. Add a new parameters PREFIX
40822 and USE_SIG_IN_NAME.
40823 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
40824 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
40825 'arm_builtin_datum'.
40826 (arm_init_vfp_builtins): Likewise.
40827 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
40828 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
40829 (arm_expand_neon_args): Rename to ...
40830 (arm_expand_builtin_args): ... this. Rename builtin_arg
40831 enum values and differentiate between ARG_BUILTIN_MEMORY
40832 and ARG_BUILTIN_NEON_MEMORY.
40833 (arm_expand_neon_builtin_1): Rename to ...
40834 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
40835 values, arm_expand_builtin_args and add bool parameter NEON.
40836 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
40837 (arm_expand_vfp_builtin): Likewise.
40838 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
40840 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40842 PR middle-end/77484
40843 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
40844 * predict.c (tree_estimate_probability_bb): Reverse direction of
40845 polymorphic call predictor.
40847 2017-01-06 David Malcolm <dmalcolm@redhat.com>
40849 * passes.c (execute_one_pass): Split out pass-skipping logic into...
40850 (determine_pass_name_match): ...this new function and...
40851 (should_skip_pass_p): ...this new function.
40853 2017-01-06 Nathan Sidwell <nathan@acm.org>
40855 * ipa-visibility.c (function_and_variable_visibility): Reformat
40856 comments and long lines. Remove extrneous if.
40857 * symtab.c (symtab_node::make_decl_local): Fix code format.
40858 (symtab_node::set_section_for_node): Fix comment typo.
40860 2017-01-06 Martin Liska <mliska@suse.cz>
40863 * lra-constraints.c: Rename invariant to lra_invariant.
40864 * predict.c (set_even_probabilities): Initialize e to NULL.
40866 2017-01-05 Martin Sebor <msebor@redhat.com>
40868 PR tree-optimization/78910
40869 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
40870 (format_integer): Correct off-by-one error in the handling
40871 of precision with negative numbers in signed conversions..
40873 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
40875 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
40877 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40879 PR tree-optimization/71016
40880 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
40881 factor_out_conditional_conversion. Formatting fix.
40882 (factor_out_conditional_conversion): Add cond_stmt argument.
40883 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
40884 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
40887 2017-01-05 David Malcolm <dmalcolm@redhat.com>
40889 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
40890 read-rtl-function.o, and selftest-rtl.o.
40891 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
40892 (selftest::aarch64_test_loading_full_dump): New function.
40893 (selftest::aarch64_run_selftests): New function.
40894 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
40895 selftest::aarch64_run_selftests.
40896 * config/i386/i386.c
40897 (selftest::ix86_test_loading_dump_fragment_1): New function.
40898 (selftest::ix86_test_loading_call_insn): New function.
40899 (selftest::ix86_test_loading_full_dump): New function.
40900 (selftest::ix86_test_loading_unspec): New function.
40901 (selftest::ix86_run_selftests): Call the new functions.
40902 * emit-rtl.c (maybe_set_max_label_num): New function.
40903 * emit-rtl.h (maybe_set_max_label_num): New decl.
40904 * function.c (instantiate_decls): Guard call to
40905 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
40906 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
40908 * gensupport.c (gen_reader::gen_reader): Pass "false"
40909 for new "compact" param of rtx_reader.
40910 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
40911 rather than an empty string for NULL strings.
40912 * read-md.c: Potentially include config.h rather than bconfig.h.
40913 Wrap include of errors.h with #ifdef GENERATOR_FILE.
40914 (have_error): New global, copied from errors.c.
40915 (md_reader::read_name): Rename to...
40916 (md_reader::read_name_1): ...this, adding "out_loc" param,
40917 and converting "missing name or number" to returning false, rather
40919 (md_reader::read_name): Reimplement in terms of read_name_1.
40920 (md_reader::read_name_or_nil): New function.
40921 (md_reader::read_string): Handle "(nil)" by returning NULL.
40922 (md_reader::md_reader): Add new param "compact".
40923 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
40924 (md_reader::read_file): New method.
40925 * read-md.h (md_reader::md_reader): Add new param "compact".
40926 (md_reader::read_file): New method.
40927 (md_reader::is_compact): New accessor.
40928 (md_reader::read_name): Convert return type from void to file_location.
40929 (md_reader::read_name_or_nil): New decl.
40930 (md_reader::read_name_1): New decl.
40931 (md_reader::m_compact): New field.
40932 (noop_reader::noop_reader): Pass "false" for new "compact" param
40934 (rtx_reader::rtx_reader): Add new "compact" param.
40935 (rtx_reader::read_rtx_operand): Make virtual and convert return
40936 type from void to rtx.
40937 (rtx_reader::read_until): New decl.
40938 (rtx_reader::handle_any_trailing_information): New virtual function.
40939 (rtx_reader::postprocess): New virtual function.
40940 (rtx_reader::finalize_string): New virtual function.
40941 (rtx_reader::m_in_call_function_usage): New field.
40942 (rtx_reader::m_reuse_rtx_by_id): New field.
40943 * read-rtl-function.c: New file.
40944 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
40945 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
40946 (selftest::verify_three_block_rtl_cfg): New decl.
40947 * read-rtl-function.h: New file.
40948 * read-rtl.c: Potentially include config.h rather than bconfig.h.
40949 For host, include function.h, memmodel.h, and emit-rtl.h.
40950 (one_time_initialization): New function.
40951 (struct compact_insn_name): New struct.
40952 (compact_insn_names): New array.
40953 (find_code): Handle insn codes in compact dumps.
40954 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
40955 (bind_subst_iter_and_attr): Likewise.
40956 (add_condition_to_string): Likewise.
40957 (add_condition_to_rtx): Likewise.
40958 (apply_attribute_uses): Likewise.
40959 (add_current_iterators): Likewise.
40960 (apply_iterators): Likewise.
40961 (initialize_iterators): Guard usage of apply_subst_iterator with
40962 #ifdef GENERATOR_FILE.
40963 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
40964 (md_reader::read_mapping): Likewise.
40965 (add_define_attr_for_define_subst): Likewise.
40966 (add_define_subst_attr): Likewise.
40967 (read_subst_mapping): Likewise.
40968 (check_code_iterator): Likewise.
40969 (rtx_reader::read_rtx): Likewise. Move one-time initialization
40971 (one_time_initialization): New function.
40972 (rtx_reader::read_until): New method.
40973 (read_flags): New function.
40974 (parse_reg_note_name): New function.
40975 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
40976 Handle reuse_rtx ids.
40977 Wrap iterator lookup within #ifdef GENERATOR_FILE.
40978 Add parsing support for RTL dumps, mirroring the special-cases in
40979 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
40980 values, and calling handle_any_trailing_information.
40981 (rtx_reader::read_rtx_operand): Convert return type from void
40982 to rtx, returning return_rtx. Handle case 'e'. Call
40983 finalize_string on XSTR and XTMPL fields.
40984 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
40985 "(nil)" values were omitted. Call the postprocess vfunc on the
40987 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
40988 class ctor. Initialize m_in_call_function_usage. Call
40989 one_time_initialization.
40990 * rtl-tests.c (selftest::test_uncond_jump): Call
40991 set_new_first_and_last_insn.
40992 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
40993 * selftest-rtl.c: New file.
40994 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
40995 (selftest::get_insn_by_uid): New decl.
40996 * selftest-run-tests.c (selftest::run_tests): Call
40997 read_rtl_function_c_tests.
40998 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
40999 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
41002 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
41004 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
41005 operands in a special way. Assert that pos+len <= mode precision.
41007 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41009 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
41010 3 argument Alias with unlimited for the negative form.
41011 (fno-vect-cost-model): Removed.
41013 2017-01-05 Martin Liska <mliska@suse.cz>
41015 * hsa-gen.c (gen_hsa_divmod): New function.
41016 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
41018 2017-01-05 Martin Liska <mliska@suse.cz>
41021 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
41024 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41026 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
41027 small constant length operands.
41029 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41031 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
41032 between loop iterations.
41034 2017-01-05 Martin Liska <mliska@suse.cz>
41037 * gimplify.c (gimplify_decl_expr): Compare to
41038 asan_poisoned_variables instread of checking flags.
41039 (gimplify_target_expr): Likewise.
41040 (gimplify_expr): Likewise.
41041 (gimplify_function_tree): Conditionally initialize
41042 asan_poisoned_variables.
41044 2017-01-04 Jeff Law <law@redhat.com>
41046 PR tree-optimizatin/78812
41047 * rtl.h (contains_mem_rtx_p): Prototype.
41048 * ifcvt.c (containts_mem_rtx_p): Move from here to...
41049 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
41050 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
41051 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
41052 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
41054 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41056 * input.c (assert_char_at_range): Default-initialize actual_range.
41058 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41060 * df-scan.c (df_ref_create_structure): Make regno unsigned,
41061 to match the caller.
41063 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41065 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
41066 insns after final jump in test to emit dummy move.
41068 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41070 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
41071 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
41073 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41075 * multiple_target.c (create_dispatcher_calls): Init e_next.
41076 * tree-ssa-loop-split.c (split_loop): Init border.
41077 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
41080 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
41085 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
41086 (altivec_register_operand): Do not return true if the operand
41087 contains a SUBREG mixing SImode and SFmode.
41088 (vsx_register_operand): Likewise.
41089 (vsx_reg_sfsubreg_ok): New predicate.
41090 (vfloat_operand): Do not return true if the operand contains a
41091 SUBREG mixing SImode and SFmode.
41092 (vint_operand): Likewise.
41093 (vlogical_operand): Likewise.
41094 (gpc_reg_operand): Likewise.
41095 (int_reg_operand): Likewise.
41096 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
41097 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
41098 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
41100 (rs6000_emit_move_si_sf_subreg): New helper function.
41101 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
41102 fixup SUBREGs involving SImode and SFmode.
41103 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
41104 numbers for the new peephole2 optimization.
41105 (peephole2 for SFmode unions): New peephole2 to optimize cases in
41106 the GLIBC math library that do AND/IOR/XOR operations on single
41107 precision floating point.
41108 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
41109 target macros to say whether we need to avoid SUBREGs mixing
41111 (TARGET_ALLOW_SF_SUBREG): Likewise.
41112 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
41113 (UNSPEC_SI_FROM_SF): Likewise.
41114 (iorxor): Change spacing.
41115 (and_ior_xor): New iterator for AND, IOR, and XOR.
41116 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
41117 (movdi_from_sf_zero_ext): Likewise.
41118 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
41119 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
41120 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
41121 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
41122 (fms<mode>4): Likewise.
41123 (fnma<mode>4): Likewise.
41124 (fnms<mode>4): Likewise.
41125 (nfma<mode>4): Likewise.
41126 (nfms<mode>4): Likewise.
41128 2017-01-04 Marek Polacek <polacek@redhat.com>
41131 * doc/invoke.texi: Document -Wpointer-compare.
41133 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41135 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
41138 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
41139 descriptions for -gdwarf-5 and emit them as uleb128 instead of
41142 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
41145 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
41146 documentation of the powerpc_popcntb_ok attribute.
41147 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
41148 code to issue warning messages if a requested CPU configuration is
41149 not supported by the binary (assembler and loader) toolchain.
41150 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
41151 made to define a built-in function that has been disabled.
41152 (paired_init_builtins): Add assertion to prevent ICE if attempt is
41153 made to define a built-in function that has been disabled.
41154 (altivec_init_builtins): Add comment explaining why definition
41155 of the DST built-in functions is not preceded by an assertion
41156 check. Add assertions to prevent ICE if attempts are made to
41157 define an altivec predicate or an abs* built-in function that has
41159 (htm_init_builtins): Add comment explaining why definition of the
41160 htm built-in functions is not preceded by an assertion check.
41162 2017-01-04 Jeff Law <law@redhat.com>
41164 PR tree-optimizatin/67955
41165 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
41166 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
41167 the points-to solution does not include pt_null. Use DECL_PT_UID
41170 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
41172 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
41173 Use gen_int_mode instead of gen_lopwart for const_int operands.
41175 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41177 PR tree-optimization/71563
41178 * match.pd: Simplify X << Y into X if Y is known to be 0 or
41179 out of range value - has low bits known to be zero.
41181 2017-01-04 Alan Modra <amodra@gmail.com>
41183 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
41184 * configure: Regenerate.
41185 * config.in: Regenerate.
41187 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41190 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
41191 a substring of the message, but strcmp with the whole message. Ifdef
41192 ENABLE_NLS, translate the message first using dgettext.
41194 2017-01-03 Jeff Law <law@redhat.com>
41196 PR tree-optimizatin/78856
41197 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
41198 (mark_threaded_blocks): Remove code to truncate thread paths that
41199 cross multiple loop headers. Instead invalidate the cached loop
41200 iteration information and handle case of a thread path walking
41201 into an irreducible region.
41203 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
41206 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
41207 assertions. Add support for doing the signbit if the IEEE 128-bit
41208 floating point value is in a GPR.
41209 * config/rs6000/rs6000.md (Fsignbit): Delete.
41210 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
41211 Update the length attribute if the value is in a GPR.
41212 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
41213 the sign or zero extension instruction, since the value is always 0/1.
41214 (signbit<mode>2_dm2): Delete using <Fsignbit>.
41217 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
41218 extracting SImode to a GPR register so that we can generate a
41219 store, limit the vector to be in a traditional Altivec register
41220 for the vextuwrx instruction.
41222 2017-01-03 Ian Lance Taylor <iant@google.com>
41224 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
41226 2017-01-03 Martin Sebor <msebor@redhat.com>
41228 PR tree-optimization/78696
41229 * gimple-ssa-sprintf.c (format_floating): Correct handling of
41230 precision. Use MPFR for %f for greater fidelity. Correct handling
41232 (pass_sprintf_length::compute_format_length): Set width and precision
41233 specified by asrerisk to void_node for vararg functions.
41234 (try_substitute_return_value): Adjust dump output.
41236 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
41238 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
41240 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
41242 * doc/invoke.texi (SPARC options): Document -mlra as the default.
41243 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
41244 -mlra/-mno-lra was passed to the compiler.
41246 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
41248 PR rtl-optimization/65618
41249 * emit-rtl.c (try_split): Move initialization of "before" and
41250 "after" to just before the call to emit_insn_after_setloc.
41252 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
41254 * doc/md.texi (Standard Names): Remove reference to Java frontend.
41256 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
41258 * dwarf2out.c (gen_enumeration_type_die): When
41259 -gno-strict-dwarf, add a DW_AT_encoding attribute.
41261 2017-01-03 Jakub Jelinek <jakub@redhat.com>
41263 PR tree-optimization/78965
41264 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
41265 Change first argument from const call_info & to call_info &. For %n
41266 set info.nowrite to false.
41268 PR middle-end/78901
41269 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
41270 possibly throwing calls.
41272 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
41273 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
41274 and fns handling, rather than in a separate case SSA_NAME.
41276 2017-01-02 Jeff Law <law@redhat.com>
41278 * config/darwin-driver.c (darwin_driver_init): Const-correctness
41279 fixes for first_period and second_period variables.
41281 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
41284 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
41285 (*insvqi_1): New insn pattern.
41286 (*insvqi_1_mem_rex64): Ditto.
41287 (*insvqi_2): Ditto.
41288 (*insvqi_3): Rename from *insvqi.
41290 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
41292 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
41294 * doc/cfg.texi (Edges): Remove reference to Java.
41295 (Maintaining the CFG): Ditto.
41297 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41299 PR middle-end/77674
41300 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
41301 transparent aliases.
41303 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41305 PR middle-end/77484
41306 * predict.def (PRED_CALL): Update hitrate.
41307 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
41308 * predict.c (tree_estimate_probability_bb): Split CALL predictor
41309 into direct/indirect/polymorphic variants.
41311 2017-01-01 Jakub Jelinek <jakub@redhat.com>
41313 Update copyright years.
41315 * gcc.c (process_command): Update copyright notice dates.
41316 * gcov-dump.c (print_version): Ditto.
41317 * gcov.c (print_version): Ditto.
41318 * gcov-tool.c (print_version): Ditto.
41319 * gengtype.c (create_file): Ditto.
41320 * doc/cpp.texi: Bump @copying's copyright year.
41321 * doc/cppinternals.texi: Ditto.
41322 * doc/gcc.texi: Ditto.
41323 * doc/gccint.texi: Ditto.
41324 * doc/gcov.texi: Ditto.
41325 * doc/install.texi: Ditto.
41326 * doc/invoke.texi: Ditto.
41328 Copyright (C) 2017 Free Software Foundation, Inc.
41330 Copying and distribution of this file, with or without modification,
41331 are permitted in any medium without royalty provided the copyright
41332 notice and this notice are preserved.